July 1994: Defensive Medicine and Medical Malpractice

Defensive Medicine and Medical Malpractice July 1994
OTA-H-602 NTIS order #PB94-193257 GPO stock #052-003-01377-1
Recommended Citation: U.S. Congress, Office of Technology Assessment, Defensive Medicine and Medical Malpractice, OTA-H--6O2 (Washington, DC: U.S. Government Printing Office, July 1994).
Foreword T
he medical malpractice system has frequently been cited as a contributor to increasing health care costs and has been targeted in many health care reform proposals as a potential source of savings. The medical malpractice system can add to the costs of health care directly through increases in malpractice insurance premiums, which may be passed on to consumers and thirdparty payers in the form of higher fees. However, total direct costs of the medical malpractice system represent less than 1 percent of overall health care costs in the United States. The medical malpractice system may also increase costs indirectly by encouraging physicians to practice defensive medicine. In this assessment, the Office of Technology Assessment first examines the nature of defensive medicine, adopting a working definition of defensive medicine that embraces the complexity of the problem from both the physician and broader public policy perspectives. It then presents and critically examines existing as well as new evidence on the extent of defensive medicine. Finally, it comments on the potential impact of a variety of medical malpractice reforms on the practice of defensive medicine. This assessment was prepared in response to a request by the House Committee on Ways and Means and the Senate Committee on Labor and Human Resources. The report was prepared by OTA staff, but OTA gratefully acknowledges the contributions of the assessment advisory panel, numerous researchers who did work under contract to OTA, and many other individuals who provided valuable information and reviewed preliminary drafts. As with all OTA documents, the final responsibility for the content of the assessment rests with OTA.
Director
Advisory Panel
R. Randall Bovbjerg
Panel Chair Senior Research Associate The Urban Institute Washington, DC

John Ball Executive Vice President American College of Physicians Philadelphia, PA James Blumstein Professor of Law Vanderbilt University Law School Nashville, TN Troyen Brennan Associate Professor Department of Medicine Harvard Medical School Boston, MA Brad Cohn President Physician Insurers Association of America San Francisco, CA Edward David Chairman Maine Board of Registration in Medicine Bangor, ME
Richard Frank Professor Department of Health Policy and Management School of Hygiene and Public Health The Johns Hopkins University Baltimore, MD Pamela Gilbert D irector Public Citizen Congress Watch Washington, DC Rodney Hayward Assistant Professor Department of Internal Medicine University of Michigan School of Medicine Ann Arbor, MI Richard Kravitz Assistant Professor of Medicine University of California, Davis Sacramento, CA George Malkasian Department of Obstetrics and Gynecology Mayo Clinic Rochester, MN
Barry Manuel Associate Dean Boston University College of Medicine Boston, MA J. Douglas Peters Charfoos and Christensen Attorneys at Law Detroit, MI Richmond Prescott Former Associate Executive Director The Permanante Medical Group, Inc. San Francisco, CA David Sundwall Vice president and Medical Director American Healthcare Systems Institute Washington, DC Laurence Tancredi Private Consultant Ncw York, NY James Todd Executive Vice President American Medical Association
Chicago, IL
Note: OTA appreciatess and is grateful for the valuable assistance and thoughtful critiques provided by the advisory panel members. The panel does not. however-, necessarily disapprove, or endorse this report. OTA assumes fu!l responsibility for the rcport and the accuracy of its contents.
iv
Preject Staff
Clyde J. Behney Assistant Director, OTA Sean R. Tunis Health Program Director PROJECT STAFF Judith L. Wagner Project Director Jacqueline A. Corrigan Scnior Analyst David Klingman Senior Analyst Leah Wolfe Analyst Philip T. Polishuk Research Analyst ADMINISTRATIVE STAFF Beckie Erickson Office Administrator Daniel B. Carson P.C. Specialist Carolyn Martin Word Processing Specialist
PRINCIPAL CONSULTANTS Russell Localio Pcnnsylvania State University Jeremy Sugarman Duke University
CONTRACTORS Laura-Mae Baldwin University of Washington Kevin Grumbach University of California/San Francisco Eleanor Kinney Indiana Univcrsity Laura Morlock The Johns Hopkins Univcrsity Pony Ehrenhaft Consultant Mark Hall Wake Forest School of Law Harold S. Luft University of California/San Francisco John Rolph RAND Gloria Ruby Consultant Peter Glassman RAND Peter Jacobson RAND Thomas Metzloff Duke University John Rosenquist University of California/Davis v
c
1 Findings and Policy Options 1
Defining Defining Defensive Medicine 3 Extent of Dcfcnsive Medicine 3 Recent Factors Affecting the Amount of Defensive
The Medicine 9 The Impact of Malpractice Reform
ontents
on Defensive Medicine 1 0 Defensive Medicine in an Era of Health Care Reform 1 5 Policy Options 16
2 Defensive Medicine: Definition and Causes 21

Defining Defensive Medicine 21 The Sources of Defensive Medicine 26 Conclusions 36
3 Summary of the Evidence on Defensive Medicine 39

Evidcncc of the Extent of Defensive Medicine 43 Conclusions 74
4 Impact of Malpractice Reform on Defensive Medicine 75

The Impact of Conventional Malpractice Reforms on Direct Malpractice Costs 76
Impact of Newer Malpractice Reformas on Defensive Medicine 81 Defensive Medicine and Health Care Reform 91 Conclusions 92
APPENDIXES A B Method of Study 95 Acknowledgments 101 The Impact of Nonclinical Factors on Physicians Use of Resources 104 Methods Used in the OTA Clinical Scenario Surveys 106 Detailed Results of the OTA Clinical Scenario Surveys 118 Estimates of the Costs of Selected Defensive Medical Procedures 128 Summary of State Studies on Tort Reforms 133 Clinical Practice Guidelines and Malpractice Liability 140 Description of 32 Direct Physician Surveys of Defensive Medicine Reviewed by OTA 149 Detailed Critique of Reynolds et al. and Lewin-VHl Estimates 154 Glossary 160
c
D E F G H I
J K
REFERENCES 165 INDEX 179
Findings and Policy Options

SUMMARY OF FINDINGS
. Defensive medicine occurs when doctors order tests, procedures, or visits, or avoid certain high-risk patients or procedures, primarily (but not necessarily solely) because of concern about malpractice liability.
q
Most defensive medicine is not of zero benefit. Instead, fear of liability pushes physicians tolerance for medical uncertainty to low levels, where the expected benefits are very small and the costs are high.
Many physicians say they would order aggressive diagnostic
procedures in cases where conservative management is considered medically acceptable by professional expert panels. Most physicians who practice in this manner would do so primarily because they believe such procedures are medically indicated, not primarily because of concerns about liability.
s
It is impossible to accurately measure the overall level and national cost of defensive medicine. The best that can be done is to develop a rough estimate of the upper limits of the extent of certain components of defensive medicine.
Overall, a small percentage of diagnostic procedures--certainly less than 8 percentis likely to be caused primarily by conscious concern about malpractice liability. This estimate is based on physicians responses to hypothetical clinical scenarios that were designed to be malpractice-sensitive; hence, it overestimates the rate at which defensive medicine is consciously practiced in diagnostic situations.
2 | Defensive Medicine and Medical Malpractice
INTRODUCTION
For more than two decades many physicians. researchers, and government officials have claimed that the most damaging and costly result of the medical malpractice system as it has evolved in the United States is the practice of defensive medicine: the ordering of tests, procedures, and visits, or avoidance of certain procedures or patients, due to concern about malpractice liability risk. Calls for reform of the medical malpractice system have rested partly on arguments that such reforms would save health care costs by reducing doctors incentives to practice defensively. Such an argument even found its way into the 1992 presidential debates, when President Bush contended that the malpractice ...trial lawyers lawsuits ...are running the costs of medical care up $25 to $50 billion. (35) Such claims notwithstanding, the extent of defensive medicine and its impact on health care costs remain a matter of controversy. Some critics claim that defensive medicine is nothing more than a convenient explanation for practices that physicians would engage in even if there were no malpractice law or malpractice lawyers. This Office of Technology Assessment (OTA) study of defensive medicine grew out of congressional interest in understanding the extent to which defensive medicine does. indeed, influence medical practice and how various approaches to reforming the malpractice system might alter these behaviors. The assessment was first requested by Congressman Bill Archer, Ranking Republican Member of the Committee on Ways and Means, and Senator Orrin Hatch, a member of OTAs Technology Assessment Board. Other members of OTA's Technology Assessment Board also requested that OTA examine these issues, including Senator Edward M. Kennedy, Chairman of the Committee on Labor and Human Resources: Congressman John D. Dingell, Chairman of the Committee on Energy and Commerce: and Senators Charles E. Grassley and Dave Durenberger. OTA addressed the following questions:
Physicians are very conscious of the risk of being sued and tend to overestimate that risk. A large number of physicians believe that being sued will adversely affect their professional, financial. and emotional status. The role of the malpractice system as a deterrent against too little or poor-quality care--one of its intended purposeshas not been carefully studied.
One malpractice reform that directly targets wasteful and low-benefit defensive medicine is to enhance the evidentiary status in malpractice court cases of selected clinical practice guidelines that address situations in which defensive medicine is a major problem. The overall effects of this reform on health care costs would probably be small, however, because only a few clinical situat ions represent clear cases of wasteful or lowbenefit defensive medicine. The fee-for-service system both empowers and encourages physicians to practice very lowrisk medicine. Health care reform may change financial incentives toward doing fewer rather than more tests and procedures. If that happens, concerns about malpractice liability may act to check potential tendencies to provide too few services.
Chapter 1 Findings and Policy Options | 13
What is defensive medicine and how can it be measured? What are the causes of defensive medicine? How widespread is defensive medicine today? What effect will current proposals for malpractice reform have on the practice of defensive medicine? What are the implications of other aspects of health care reform for the practice of defensive medicine? OTA also published a background paper in September 1993, Impact of Legal Reforms on Medical Malpractice Costs, which summarizes the current status of malpractice law reforms in the 50 states and evaluates the best available evidence on the effect of malpractice system reforms on physicians malpractice insurance premiums.
Most importantly, defensive medicine is not always bad for patients. Although political or media references to defensive medicine almost always imply unnecessary and costly procedures, OTAs definition does not exclude practices that may benefit patients. Rather, OTA concluded that a high percentage of defensive medical procedures are ordered to minimize the risk of being wrong when the medical consequences of being wrong are severe:
OTA asked panels of experts in three medical
specialties-cardiology, obstetrics/gynecology (OB/GYN), and surgery-to identify clinical scenarios in which they would expect the threat of a malpractice suit to play a major role in their own or their colleagues clinical decisions.
The groups
identified over 75 scenarios, all of which involved a patient presenting with a probable minor condition but with a small chance for a potentially very serious or fatal condition.
DEFINING DEFENSIVE MEDICINE

OTA defines defensive medicine as follows: Defensive medicine occurs when doctors order tests, procedures, or visits, or avoid high-risk patients or procedures, primarily (but not necessarily soley) to reduce their exposure to mal practice liability. When physicians do extra tests or procedures primarily to reduce malpractice liability, they are practicing positive defensive medicine. When they avoid certain patients or procedures, they are practicing negative defensive medicine. Under this definition, a medical practice is defensive even if it is done for other reasons (such as belief in a procedure effectiveness, desire to reduce medical uncertainty, or financial incentives), provided that the primary motive is to avoid malpractice risk. Also, the motive need not be conscious. Over time some medical practices may become so ingrained in customary practice that physicians are unaware that liability concerns originally motivated their use.
Thus, concern about malpractice liability pushes physicians tolerance for uncertainty about medical outcomes to very low levels. Stated another way, concerns about liability drive doctors to order tests, procedures, and specialist consultations whose expected benefits are very low. Using such medical technologies and services to reduce risk to the lowest possible level is likely to be very costly even when the price of the procedure is low, because for every case where its performance makes the life-or-death difference, there will be many additional cases where its performance is clinically inconsequential.
THE EXTENT OF DEFENSIVE MEDICINE

OTA searched for evidence of defensive medicine in the existing literature and also conducted and contracted for new analyses where feasibility and
Physicians may stop performing certain tests or procedures if by doing so they can ellminatc the need for costly or hard-to-find malpractice insurance to cover these activities, The most frequently citcd examples of negative defensive medicine are decisions by family practitioners and even some obstetrlcim-gynecologists to stop providing obstetric services. These decisions may be a result of higher malpractice insurance premiums for physicians who deliver babies.
costs permitted. One conclusion from these efforts is that accurate measurement of the extent of this phenomenon is virtually impossible. There are only two possible approaches to estimating how often doctors do (or do not do) procedures for defensive reasons: ask them directly in surveys, or link differences in their actual procedure utilization rates to differences in their risk of liability. Both of these approaches have serious limitations. If physicians are asked how often they practice defensive medicine in survey questionnaires, they may be inclined to respond with the answer most likely to elicit a favorable political response and thus exaggerate their true level of concern about malpractice. Even when physicians are asked in a more neutral instrument what they would do in certain clinical situations and why, they might be prompted if one of the potential listed reasons relates to concern about malpractice suits. On the other hand, without listed reasons from which to choose, physicians may respond as if the survey is a medical board examination and justify their choices on purely clinical grounds when other factors do in fact operate. In addition, surveys cannot uncover defensive practices performed unconsciously by physicians. In short, surveys can elicit responses that are biased in either direction. These obvious problems suggest that it might be better to start with actual behavior as recorded in data on utilization of procedures and try to ascertain the percentage of use that arises from fear of malpractice suits. The only way to measure such a percentage is to relate variations in utilization across physicians to variations in the strength of the malpractice signal across physicians. For example, physicians practicing in hospitals or communities with high rates of malpractice claims or high malpractice premiums might be more sensitive to malpractice risks and alter their practices accordingly. Statistical analyses of such variations could pick up these differential effects. To take this tack, data must be available to control for other factors that can account for differences among physicians in their utilization of ser-
vices, including the health status of the patient population. Often such data are unavailable. Even more troublesome is the fact that this approach can pick up only the incremental effects of stronger versus weaker malpractice signals. It cannot accurately assess the generalized baseline level of defensive medicine that may exist in all physicians practices. Professional society newsletters and other national media often report on especially large or unusual jury verdicts. Physicians may react to these news items as vigorously as they would to their own or their colleagues experience with malpractice claims. Physicians may be almost as defensive if they face a small risk of being sued as they are if they face a higher risk. This is especially likely if they have the power, with no negative and sometimes positive financial consequences, to order tests and procedures that reduce medical risks to their lowest feasible level. Despite these problems, OTA undertook new analyses that offered the best chance, within time and budgetary constraints, of adding to the current state of knowledge about the scope of defensive medical practice while acknowledging the methodological problems described above. OTA-initiated studies included the following: Four separate physician surveys (conducted jointly with three medical specialty societies) containing hypothetical clinical scenarios that asked respondents to indicate what clinical actions they would take and the reasons for them. The survey materials contained no references to suggest that OTAs purpose was to study malpractice or defensive medicine, though malpractice concern was one of five reasons listed for each possible course of action. An analysis of the relationship between the use of prenatal care services in low-risk pregnancy and the level of malpractice risk facing doctors in Washington State. An analysis of the relationship between New Jersey physicians responses on a clinical scenario survey and their personal malpractice claim history.
An analysis relating changes in New York State physicians obstetric malpractice insurance premiums to decisions to abandon the practice of obstetrics. These analyses join a small preexisting literature and discussions with experts in the area to form the basis for OTAs findings. The following studies were particularly important evidence because of their relatively strong research designs:
s s
clusion in the four surveys involved clinical encounters requiring some diagnostic judgment or action.2 Virtually all of the clinical scenarios involved patients whose presenting signs and symp-
A study by Localio and colleagues of the relationship between Caesarean delivery rates and malpractice risk in New York State hospitals ( 128). s A survey of physicians responses to c1inical scenarios conducted by a Duke Law Journal project on medical malpractice (58).
Other studies, including the ninny direct physician surveys conducted over the years by national. state, and specialty medical societies. are reviewed by OTA in this report. Their results are highly suspect, however, because they invariably prompt responding physicians to consider malpractice liability as a factor in their practice choices.
toms would suggest only minor injury or a selflimiting problem, with a very small outside chance of a debilitating or life-threatening illness. Although the panelists were not asked to assess the appropriateness of different clinical actions or procedures, implicit in their creation of each scenario was the idea that conservative treatment was an acceptable course of action. Across the scenarios, between 5 and 29 percent of all responding physicians cited malpractice concern as the primary reason for choosing at least one clinical action (figure 1-1 ), Yet, in six of the nine scenarios, defensive medicine was cited by less than 10 percent of all physicians as the primary reason for choosing at least one clinical action. The scenario with the greatest evidence of defensive medicine was a case of a 15-year-old boy with a minor head injury resulting from a skateboard accident. In that case, almost one-half of all respondents reported that they would order a computed tomography (CT) scan, and 45 percent of those who said they would order it would do so primarily out of concern for malpractice. Figure 1-2 shows the specific clinical actions with the highest reported rates of defensive medicine. These procedures constitute only 23 out of the 54 "interventionist actions in the nine scenarios (i.e.. other than waiting or doing nothing). Physicians who reported they would order the procedure said they would do so primarily out of concern about malpractice between 11 and 53 percent of the time. Yet. the percentage of responses in which the procedure would be ordered out of concern for malpractice seldom exceeded 5 percent, because relatively few physicians reported that they would choose the procedure at all. Across all possible actions in the nine scenarios, excluding waiting or doing nothing, a me-
NOTE Results are weighted to reflect the total population of professional society members on which the survey sample was based Numbers reflect responses to case verslons of the scenarious only (see ch 3) See table 3-2 for confidence intervals of these proportions SOURCE Off Ice of Technology Assessment, 1994
dian 3 of 8 percent of those who chose the procedure or hospital admission said they would do so primarily because of malpractice concerns (see table 3-3 in chapter 3).
The surveys covered only three medical specialties, at least two of which have relatively high exposure to malpractice liability. Also, the level of defensive medicine recorded in these scenarios is
3 That is, one-half of the procedures had a percentage score higher than the median percentage; one-half had a percentage score that was lower than the median.
Percent of respondents Percent of respondents

Scenario Clinical action choosing clinical action 7.6 66.3 26.5 23.1 7.8 50.2 Admit & obtain ECG 22.4 21.5 13.3 l middle-aged man Colonoscopy Head injury in a 15-year-old boy Skull x-ray ( Cervical spine x-ray Back pain in a 52-year-old man . CT of head 24.4 3.4 1 2 . 6 45.6 21.1 48.8 3.4 1.0 2.0 11.2 26.2 33.7 8.4 19.2 II I
Of clinical actions chosen,
choosing clinical action primarily percent done primarily for malpractice concerns for malpractice concerns
I 1.5
-203
7.2 108 137 m 1 8.6 1 1 i 8 7 9 d . . . 139 2 2 4 118 1 10 2 9 9 . 6 0 . 3 5 4 2 5
I I 3.6 3.4 1.4
u
II II 3
2.7 2.1 2.3
4.4
5.0
Lumbosacral x-ray I L CT scan MRI
139 2 1 9 6 123 2 2 11 1 109 1 4 5 . 8
Breast lump -
-- Mammography Refer to surgeon
u
l
5.6 6.3
Complicated delivery Perimenopausal bleeding
Caesarean delivery Pregnancy test D&C 4 . 2
23.8
6 5.5
0.5
KEY MRI magnetIc resonance image EEG - electroencephalogram ECG = electrocarcjlogram CT computed tomography D&C dilation ar~d curettage NOTES A frequent occurrence was defined as when at least 10 percent of physicians who would take the cilnlc.al actlorl would do so prlmarlly beca~se of malpractice concerns Twenty-three out of a total of 54 c1 nlcal options (excludng walhng or doing notblng) In the OTA scenaros met this crlterlon (case scenarios only) See table 3-3 for complete results SOURCE Off Ice of Technology Assessment 1994 Data analyzed m collaboration with Dr Russell Locallo of Pennsylvania State Umversty
likely to be above average for diagnostic encounters, since the scenarios were explicitly designed to evoke concern about liability. Thus, a relativel y small proportion of diagnostic procedures overall--certainly less than 8 percentis likely to be caused by conscious concern about malpractice liability. In virtually all of the scenarios, many physicians chose aggressive patient management styles even though conservative management was considered medical] y acceptable by the expert panels. In most cases, however, it was medical indications, not malpractice concern, that motivated the interventions:
For example, almost two-thirds of all cardiologists reported that they would hospitalize
The annual national cost of defensive Caesarean deliveries in cases of prolonged or dysfunctional labor in women between 30 and 39 years of age is approximately $8.7 million. The annual national cost of defensive radiologic procedures (CT scans, skull x-rays, and cervical spine x-rays) in children between 5 and 24 years of age arriving in emergency rooms with apparently minor head injuries is roughly $45 million. Although these estimates in and of themselves represent a miniscule percentage of total health care costs, they cover on] y a few procedures performed in very specific clinical situations, and they reflect only that portion of defensive medicine that physicians practice consciously. The numbers suggest, however, that if conscious defensive medicine is costly in the aggregate, it would have to operate in a very large number of clinical situations, each contributing a relatively small amount to total costs.
50-year-old
woman who had fainted in a hot church with no other serious problems, but only 10.8 percent of those would do so primarily out of concern for malpractice risk. instead, the vast majority of those who would hospitalize a patient of this kind reported that they would do so primarily because it was medically indicated.
Procedure Utilization Studies

OTAs review of the evidence relating actual use of services to measures of malpractice risk, including the OTA-sponsored studies using this approach, found only limited evidence that defensive medicine exists. The strongest evidence was produced in a study by Localio and colleagues of Caesarean deliveries in New York State ( 128): New
York State obstetricians who practice in hospitals with high malpractice claim frequency and premiums do more Caesarean deliveries than do obstetricians practicing in areas with low malpractice claim frequency and premiums. The odds of a Caesarean delivery in a hospital with the highest frequency of obstetric malpractice claims were 32 percent higher than the odds of a Caesarean delivery in a hospital with the lowest frequency of obstetric malpractice claims (128). Two OTA-sponsored research contracts that attempted to relate physicians utilization rates to
Thus, if malpractice risk is a major factor influencing physicians actions in general, it is not conscious, but works indirectly over time through changes in physicians assessments of appropriate care. It is impossible to use these very specific clinical scenarios to estimate overall health care costs that are due to defensive medicine. First, the scenarios were selected to heighten the probability of finding defensive practices. Second, they involve very specific presenting signs and symptoms. Slight changes in the scenarios might yield large changes in the kinds of procedures chosen and their consequent costs. OTA did estimate the national cost of defensive medicine for selected procedures in two scenarios: Caesarean delivery in a difficult labor, and diagnostic radiology in a young emergency room patient with minor head injury.
their actual or perceived malpractice risks failed to find significant relationships between the risk of malpractice and physician behavior:
A study of 1,963 low-risk pregnancies managed by 209 physicians in Washington State failed to find a significant relationship between physicians personal malpractice suit history or the malpractice claims rate in the county and the use of selected services, such as diagnostic ultrasound early in pregnancy, referrals to specialists, and Caesarean delivery (10). A study of 835 New Jersey surgeons, cardiologists, obstetrician/gynecologists, and internal medicine specialists failed to find a significant relationship between physicians personal malpractice suit history and their use of services as reported in their responses to hypothetical clinical scenarios (73)
RECENT FACTORS AFFECTING THE AMOUNT OF DEFENSIVE MEDICINE

OTA staff talked with over 100 physicians and health care professionals about their beliefs regarding the existence and frequency of defensive medicine. These conversations reinforced the findings of opinion surveys that many physicians believe defensive medicine is an important and growing phenomenon that distorts their medical judgment in ways they find very troubling.
Both of these studies were based on a small number of cases; consequently. failure to find a significant relationship could mean either that no relationship exists or that the studies lacked the statistical power to identify a significant relationship. Also, the New Jersey study did not examine the malpractice signal that physicians may receive because they practice in a high-risk locality. Nevertheless, if doctors do react to the strength of the malpractice signals measured in these studies, the changes are not large enough to be detectable in studies of the size reported here. OTA commissioned one study of negative defensive medicinethe decision not to provide a service because of concern about the risk of malpractice liability or the availability or cost of malpractice insurance. That study also failed to find significant effects:
Doctors active in obstetrics in New York State in 1980 who experienced rapid increases in malpractice insurance premiums between 1980 and 1989
Perceptions of increasing risk may arise from the continual development of new diagnostic techniques and improved therapies for serious conditions. Both of these technological trends could make the consequences of not testing more serious. The availability of more accurate or early tests or new therapies changes a natural riskfor example, the risk of death from diseaseinto a preventable risk, and places a new burden on the physician to correctly interpret the results of the test. When a medical technology is new, physicians may have greater uncertainty about the appropriate indications for its use and therefore more conscious concern about the potential for liability:
A urologlst interviewed by OTA described hls practice of ordering a prostate specific antigen (PSA) test, a screening test for prostate cancer first available in 1990, on all men over age 5 0 who come to his office, regardless of their complaint, and despite his belief that the test may, in the end, do more harm than good A cardiology fellow who makes daily decisions about the choice of clot-dissolving drugs in heart attack patients described the difficulty she and her colleagues are having evaluating the evidence on the relative effectiveness of newer versus older drugs under specific conditions of u s e and in different kinds of patients She and her colleagues openly discuss the potential for a malpractice suit if a patient dies when the less costly thrombolytic agent is used
were
NOT found to be more likely than phy-
sicians with lower premium increases to withdraw from obstetrics practice during the same period (81).
The fear of malpractice does not operate alone to stimulate the diffusion of new technologies, however. As with all medical practices, a complex array of factors influences physicians decisions to adopt new technologies:
In an OTA-sponsored study of low osmolality contrast agents (LOCAs), a new kind of contrast media injected in patients undergoing certain diagnostic x-ray examinations, Jacobson and Rosenquist found that legal concerns ranked seventh out of 11 possible factors in decisions on whether or not to use this expensive new technology. Clinical factors, such as patient safety and comfort, were ranked as the most important determinants by the responding physicians (105).
THE IMPACT OF MALPRACTICE REFORM ON DEFENSIVE MEDICINE

OTA assessed the impact of malpractice reforms on the practice of defensive medicine. Other impacts of malpractice reform may be as or even more important than defensive medicine, including impacts on: the quality of care, the physician-patient relationship, q access to the legal system, s the adequacy of compensation for medical injuries.
s s
Another reason for growing concern about the malpractice system is that the negative consequences to physicians of being sued appear to be on the rise. For the majority of physicians, a single malpractice suit does not have a significant impact on personal finances or professional status. Recent federal and state laws requiring reporting of malprtictice claims to a central repository. however, may increase the professional and financial significance of even a single lawsuit in the minds of physicians. Since 1990, federal law has required malpractice insurers to report all payments on behalf of a physician to a National Practitioner Data Bank (NPDB). The NPDB maintains a short narrative on the incident. and this information must be accessed by hospitals when hiring new staff and every two years for review of current staff (45 C.F. R. Sec. 60. 10). It can also be accessed by other potential employers. Some states also have malpractice reporting requirements tied to licensing or disciplinary processes. None of the federal or state databanks currently in place is open to the general public. Yet the ongoing debate as to whether to allow public access to the federal NPDB ( 165) may have already increased physicians anxiety about being sued.
These other impacts of malpractice reform have been reviewed extensively elsewhere (12,21,37, 102,122,191 ,208a,243) and are not discussed at length in this report. Predicting the impact of any malpractice reform on defensive medicine is very difficult, because there is little understanding of which specific aspects of the malpractice system actually drive physicians to practice defensively. Is it simply distaste for having ones clinical actions called into question? Is it distaste for having ones actions judged by lay juries? Is it a desire to avoid court trials? Is it a fear, however unfounded, of being financially ruined? Or is it the belief that the legal standard of care is so capricious that the system offers no clear guidelines for how to avoid liability? The relative importance of each of these factors in explaining motivations for defensive medicine will determine the effect of specific malpractice reforms on defensive medicine. For example, if physicians are afraid only of the extremely low chance of financial ruin, then reforms that eliminate the possibility of such an event might reduce defensive medicine even with no major changes in the system. But if physicians abhor the prospect of having to defend their judgment in any forum, then malpractice reformers would have to find ways to substantially reduce the frequency with which claims are brought, regardless of the process for resolving those claims. OTA assessed how different kinds of tort reforms would address the various aspects of the malpractice system that might motivate physi-
shortening the statute of limitations (the time period in which a suit can be brought), limiting plaintiffs attorney fees, requiring or allowing pretrial screening of claims, placing caps on damages, amending the collateral source rule (requiring or letting the jury reduce the award by the amount received from health or disability insurance), and periodic payment of damages (instead of upfront lump-sum payment). Although some of these reforms effectively limit the direct costs of malpractice (i.e., malpractice insurance premiums) (236), evidence of their effect on defensive medicine is weak.
The best evidence that physicians behavior can be altered by reducing the frequency with which plaintiffs sue, or the amounts that can be recovered when they do, comes from a study of the impact of malpractice risk on Caesarean delivery rates in New York State ( 128, 129). That study, which found a systematic relationship between the strength of various malpractice risk measures (i.e., claim frequency and insurance premiums) and Caesarean delivery rates, is consistent with the hypothesis that tort reforms that reduce claim frequency or malpractice premiums will reduce defensive behavior. Yet. it is unknown how far Localios findings for obstetricians and Caesarean rates can be generalized to other states, specialties. clinical situations, or procedures-especially in light of the failure of other studies funded by OTA to find a correlation between malpractice risk and clinical behavior. To the extent that physicians respond not to the absolute risk of suit but to their inability to predict what kinds of behavior will lead to a suit, they may behave defensively even in the face of very low malpractice risks. Malpractice reforms that limit damages or reduce claim frequency without making the system more predictable may not have much effect on defensive behavior. In the early 1970s, when malpractice claim frequency and premiums were quite low compared with todays levels, there was still considerable concern about defensive medicine ( 13, 14,20,58,243). Some experts have suggested that states (or the federal government) develop compensation guidelines to help juries determine a fair award for noneconomic damages (i.e.. pain and suffering) (23a). The guidelines would be keyed to characteristics of the plaintiff and his or her injuries. including age and type or level of disability. This approach would be less punishing to seriously injured plaintiffs than a single cap on damages applicable to all cases, and it would also promote consistency in amounts awarded across juries and jurisdictions.
greater use of clinical practice guidelines as the standard of cam, enterprise liability, alternative dispute resolution (ADR), and selective no-fault malpractice systems.
Clinical Practice Guidelines

A larger role for clinical practice guidelines in medical malpractice litigation is being tested in a small number of states. The State of Maines ongoing experimental program has become a model
for such efforts. In Maine, selected guidelines can be used as an affirmative defense (i.e., a complete defense if it can be shown that the defendant adhered to the guidelines). The state has recently adopted guidelines in areas of practice thought to involve substantial defensive medicine (e. g., Caesarean deliveries, cervical spine x-rays for head injury, preoperative testing). The Maine guidelines were written in part to reduce defensive medical practice. For example, Maines guideline for cervical spine x-rays prov ides physicians with explicit criteria for when it is not necessary to obtain such an examination. If these guidelines are upheld in court, physicians may be able to rely on them for legal protection when they decline to perform such a test. There is some evidence that the Maine initiative has reduced defensive medicine in some Select procedures (e.g., cervical spine x-rays in emergency rooms). Because the number of clinical situations in which such guidelines can be applied is limited, however, these approaches may not have much of an impact overall on medical practice or health care costs. Even under the current legal system, where guidelines carry no greater legal weight than other expert testimony, the continued development of clinical practice guidelines by professional groups and governments might reduce defensive medicine in certain areas if they help clarify the legal standard of care. The greatest potential benefit for increasing the use of guidelines in the tort system is that they offer a method for selectively addressing problems of defensive medicine by differentiating procedures that are appropriate from those that are not worth their medical risks and costs. They can also address instances in which defensive medicine is practiced unconsciously by alerting physicians to the new standard of care as reflected in the guidelines.
It is worth noting, however, that guidelines are generally developed by panels of experts (usually dominated by physicians) who, for a variety of reasons, may recommend aggressive use of diagnostic and therapeutic interventions without consideration of the implications for health care costs. For example, prior to the 1992 reauthorization of the federal government new guideline development program, the expert groups developing the guidelines were advised to consider only medical effectiveness and risks, and not the cost, of interventions (241 ). Moreover, when there is a great deal of uncertainty about the relative effectiveness of alternative courses of action, the developers of guidelines often demur from taking a stand and instead provide an array of diagnostic and treatment options, leaving it to the physician to make the choice. Thus, the net impact of the general trend toward more development of practice guidelines on defensive medicine is unclear.
Enterprise Liability
The main feature of enterprise liability is that the physician would no longer be personal] y liable for his or her malpractice. Instead, the institution in which the physician practices, or the health plan responsible for paying for the services, would assume the physicians liability. Enterprise liability promises certain efficiencies; for example, eliminating the costs of suits involving multiple defendants and thereby facilitating settlement. It could also promote better quality control within institutions and health plans while relieving physicians of some of the psychological burdens of a malpractice suit. Although the physician would not be named in the suit and may not have as great a role in the pretrial discovery process, if the case does go to trial, the physician would probably be the primary witness. (Presently, only 10 to 20 percent of malpractice cases go to trial.) Thus, although there maybe some psychological benefit to physicians of not being held personally liable, they may still feel
burdened by the prospect of having to defend their actions in court. The number of claims against health plans or institutions could go up under enterprise liability if patients feel more comfortable suing institutions than suing their own doctors. If doctors find themselves being witnesses in a larger number of suits, and subject to greater oversight and possibly disciplinary action by the institution in which they practice, they could become even more fearful of malpractice and, hence, practice more defensive medicine. The enterprise that assumes the liability would have incentives to limit potential suits and improve the quality of care. Enterprise liability may not, however, lead to a reduction in the kinds of defensive medicine whose costs are high in relation to their potential benefits unless the organization also has incentives to limit health care costs. If the organization that assumes liability has no financial incentive to control health care costs, it may target its quality control efforts to eliminate all adverse events and charge patients or their insurers for defensive procedures with low benefits and high costs.
Alternative Dispute Resolution

ADR can take many forms, but a common attribute of most such programs is that the dispute is heard or decided by one or more arbitrators or mediators rather than by a jury. The ADR proceeding is often less formal, less costly, and less public than a judicial trial. ADR can be nonbinding or binding. For nonbinding ADR, the case can still proceed to trial. Therefore, if physicians practice defensively out of anxiety about court trials, binding ADR may be the better approach to reduce defensive medicine. The most feasible approach to binding ADR is voluntary pretreatment contracts between patients and providers (or between patients and health plans) in which the parties agree prior to treatment to arbitrate any malpractice suit that might arise from that treatment. This approach has not been
tried very often because of present uncertainty about the enforceability of such contracts.6 To the extent that physicians believe an ADR system is more fair than the judicial system, they might practice less defensively. Also, cases would not go to public trial under binding ADR, so if physicians abhor the publicity of a trial, they would be relieved of that concern. On the other hand, arbitrators may be more likely to reach compromise decisions rather than completely exonerate the physician. Physicians might find they are held liable more often in arbitration than in trial. An increase in liability findings could make physicians more defensive. Finally, ADR may increase the frequency of suits, because the cost of bringing a claim should be lower and plaintiffs may find arbitration less intimidating than civil litigation. To the extent that physicians react to increasing claim frequency by becoming more defensive, this feature of ADR could increase the practice of defensive medicine. Like the traditional malpractice reforms, any effect of ADR on defensive medicine would be general; ADR could not provide specific guidance about which defensive medical practices are, and which are not, worth their costs.
source limitations. For plaintiffs, the plan offers easier filing of claims and free legal services once a claim is judged to have merit. Most cases would probably be decided by a claims investigator, a single physician, or a hearing examiner, depending on the stage at which they are resolved. Although the proposal would eliminate physicians anxiety about court trials, linking malpractice claim resolution with medical licensing could make physicians apprehensive in another way. In addition, if the AMA is correct in its prediction that many more injured patients would file claims under such a system, physicians could find themselves named in more claims. Both of these factorshigher claims frequency and the increased link between malpractice claims and formal disciplinary bodies--could increase incentives to practice defensive medicine. On the other hand, if the determinations of the medical boards improve the consistency of findings of negligence, physicians may get clearer signals about which kinds of defensive medicine will protect them from disciplinary actions. Thus, the system may differentiate better than the present system between good and bad defensive medicine.
The American Medical Association/ Specialty Society Medical Liability Project

Another ADR model has been proposed by the American Medical Association and 31 national medical specialty societies (AMA/S SMLP). Each states medical licensing board would have exclusive authority to hear and decide malpractice claims. The newly expanded medical licensing boards would consist of seven members, with no more than three coming from the health professions, The AMA/SSMLP proposal outlines in detail the process for claim resolution and proposes certain revisions in the legal rules to be used, including a cap on damages and a change in the legal standard of care to more explicitly recognize re-
Selective No-Fault
Under a selective no-fault system, medical experts would identify categories of medical injuries that would be compensable without a determination of fault on the part of the physician. When these injuries occur, patients would be compensated through some kind of administrative system. Claims not involving these injuries would still be compensated through either a judicial system or an ADR system, retaining negligence as the liability standard. Virginia and Florida have implemented nofault systems for a selected set of severe birth-related injuries. These injuries were chosen because the issue of causality is very muddled in these cases (i.e., it is difficult to prove that an injury did not result from the birth process). Although the
6 The courts often scrutinize the fairness of such contracts, because the health care provider usually has superior bargaining p)wer.
two programs have been operational for close to five years, no studies have documented whether these programs have increased the availability of obstetric care or changed the use of any obstetric procedures. A selective no-fault system with broader application across a wide array of clinical situations has been proposed by researchers since the early 1970s (2, 19,22 1). The developers of this proposal have identified about 150 accelerated compensation events (ACES), defined by adverse outcomes resulting from certain clinical actions or omissions. These adverse outcomes should be avoidable with good medical care. Under their proposal, injuries falling into an ACE category would be compensated quickly and with no inquiry into negligence. Selective no-fault goes further than enterprise liability in relieving the physician of personal liability; it should therefore reduce some pressures to practice defensively, Yet compensation under an ACE may still carry a personal stigma for the physician. ACES can and probably would be used to monitor the quality of care as well as to determine compensation, and physicians might be disciplined if they are implicated in a large number of ACES. Some ACES involve failure to diagnose a fatal condition, such as breast cancer. If, as OTA contends, a substantial proportion of defensive medicine involves extra tests and procedures to avoid very unlikely but serious consequences, physicians may feel as compelled to practice defensively to avoid an ACE as they do to avoid a malpractice suit.
DEFENSIVE MEDICINE IN AN ERA OF HEALTH CARE REFORM

Positive defensive medicine as it is practiced today evolved in the context of a fee-for-service
health care system in which physicians for the most part faced little or no financial penalty and sometimes were financially rewarded when they ordered or performed extra tests and procedures. Even the growth of health maintenance organizations (HMOs), which put plans at risk of exceeding their capitated budgets, has not changed this reality for most of the health care system. 7 As noted above, OTA concluded that most defensive medicine practices are not completely wasteful but instead reflect the tendency of liability concerns to push physicians tolerance for medical risks of a bad outcome to extremely low levels. The fee-for-service system of third-party payment both empowers and encourages physicians to practice very low-risk medicine. A new health care delivery system may evolve in the coming years as a consequence of health care reform. Whether the new system actually changes the financial incentives to order or perform tests and procedures remains to be seen, but some proposals clear] y do envision a new set of incentives. In particular, proposals that embody managed competition as a governing framework for the organization of the health care system would create incentives for health plans to reduce the number of procedures used by their members. Just as the malpractice system may push doctors tolerance for medical risks to low levels, managed competition may provide a countervailing force to raise it back up. Indeed, a critical question regarding managed competition is how quali(y of care will be monitored and enforced in plans where incentives to cut costs are strong. For all its problems, the medical malpractice system is designed to hold the medical profession to an acceptable level of quality by deterring negligence. Whether the current malpractice system is effective in achieving this objective is a matter
x ~anaged ~onll)cl;tl{)n in thl~ rew)rt rcfer~ to ii s}stcn~ ITI w h]ch each ct)nsumer cht~(~scs ammg ctmlpctmg health plans that offer a Slandard set of herwfits at different prices ( I.e., prcmlunw ). Ct)n~pet]t Itm ammg plans f~w patlcnts {In the b;is I~ t)l prlcc as w cII a~ qual Ily would presumably force plans to l(N)k f[)ropp)rtunl[lcs toel]nllnatc wasteful t)ronl} nlarginall> useful sm Ices. In acid ItI~)n, the Acinl]nlstratlon s propml Imposes caps (m increases in health Insurance prcn)lunl~. I t I\ CR Pcctccl th;i[ plans will c~crt greater Intlucncc on thclr participant Ing d(xtt)rs and hospitals to k more cost c{mscl(ws in making clln]u<il dckli](ms.
of debate. OTA found only one study that tested the deterrent effect of the malpractices system, and that study failed to show an effect:
In an attempt to estimate the deterrent effect of the malpractice system, researchers at Harvard University recently analyzed the relationship between the number of malpractice claims per negligent injury and the rate of negligent injury in New York State hospitals in 1984. They failed to
demonstrate a statistically significant relationship between malpractice claim activity and the rate of negligent injury in a hospital (254).9
Nevertheless, given new incentives to do less rather than more in a reformed health care system, major reforms of the medical malpractice system that reduce or remove incentives to practice defensively could reduce or remove a deterrent to providing too little care at the very time that such mechanisms are most needed. Ultimately two questions must be answered as the United States moves to a new health care system: what level of medical risk are the American people willing to bear for the sake of cost containment? what quality assurance mechanisms should be used to decide on and enforce adherence to that level? Under the malpractice system as it is currently configured, juries help decide the acceptable level of medical risk in at least some cases. Better methods may exist, but until such alternatives are tried and tested, the advisability of major changes in the malpractice system is a policy issue that deserves careful consideration.
practices are performed, how costly they are, or how much they affect the quality of care. Although physicians do not appear to consciously practice defensive medicine as often as they say they do, the malpractice system may have a subtle and cumulative effect over time on what physicians believe is the appropriate level of care. This unconscious component of defensive medicine may comprise a large part of the defensive medicine problem. Yet, an unknown proportion of both conscious and unconscious defensive medicine improves the outcomes of patient care. A reasonable goal of federal policy would ~be to reduce physicians ability or incentives to engage (either consciously or unconsciously) in defensive practices whose benefits to patients are not worth their costs. Finding specific policies that move the health care system toward that goal is not so easy, however. Below are four specific options for addressing the problem of defensive medicine. Each is imperfect, some more so than others. OTA has provided a rationale for suggesting that certain of these options provide a sharper scalpel than others for excising the bad practices while retaining the good. Finally, each policy option has different implications for fairness and equity to patients. These implications are laid out in the discussion following each option. Reduce the strength of the malprac tice signal by mandating traditional tort reforms that limit plaintiffs access to the courts or potential compensation. Some traditional tort reforms, particularly caps on noneconomic damages and elimination of the collateral source rule, have been shown to reduce malpractice premiums consistently in a number of studies. Any tort reform that makes it more difficult to prove liability or less potentially remunerative for a plaintiff to file and pursue a malpractice case should reduce claim frequency or payouts.
POLICY OPTIONS
OTAs assessment of the extent of defensive medicine will not close the debate on how often such
9 Lack {~f statistically significant findings in this case may result from the small sample of ht)spi[als in the study. The estimated effect of [he malpractice system (m negligent injuries was rwg:itive. though not statistically significant.
That malpractice premiums are lower in the presence of these reforms is therefore not surprising. The evidence linking frequency of claims and malpractice premiums to the frequency with which physicians practice defensive medicine is sparse, consisting of one study showing that lower claims frequency and lower premiums are associated with lower rates of Caesarean deliveries (128). (Smaller studies of other procedures commissioned by OTA failed to find an effect. ) That study did not address the effect of differences in Caesarean delivery rates on patient outcomes. Thus, while the very limited existing evidence supports the notion that defensive medicine might be sensitive to the general strength of the malpractice signal, the existence of the effect across different procedures and the impact on the quality of care are unknown. The main problem with using the traditional reforms to reduce defensive medicine is that they do not target the practices that are likely to be least medically beneficial. In reducing physicians general anxiety about being sued or having unlimited financial exposure, the y may also weaken whatever deterrence value the current malpractice system provides, with no quality assurance system offered in its place to otherwise hold physicians accountable for the care they render. Some traditional tort reforms, particularly those that limit potential compensation (e.g., caps on damages or mandatory periodic payment of damages), affect the vety small minority of plaintiffs who receive high damage awards. These are disproportionately those with the most severe injuries. Not only does this raise the issue of fairness to victims of negligence, but it ~UISO sends a signal to physicians that the most serious results of malpractice will have more limited financial consequences.
assess the incentives and quality assurance mechanisms inherent in health care reform before changing the basic structure of the malpractice system. While this approach would avoid the potential for removing whatever "deterrence value the current malpractice system offers before alternative quality assurance mechanisms are in place, it could also put the malpractice system in direct conflict with the incentives inherent in health care reform. In particular, under health care reform. physicians may feel pressure to make cost-benefit tradeoffs in their clinical choices. Yet the current legal standard of care does not explicitly recognize cost concerns as a legitimate input into c1inical decisionmaking. Over time, cost-benefit tradeoffs may become integrated into the customary standard of care and the courts will defer to this new standard of care. However, there is likely to be a transition period in which the physician will be pushed to conserve resources but will not be provided legal protection for those decisions. This could lead to new tensions among physicians. patients, and patients health plans.
form has been settled. A "go-slow" approach to malpractice reform would permit state and federal policy makers to
One kind of malpractice reform that will be useful regardless of the shape of health care reform is the development and enhanced use as evidence in the courts of clinical practice guidelines covering situations in which defensive medicine plays a substantial role. OTA found that Caesarean deliveries and head injuries in emergency rooms are two c1inical situations in which defensive medicine is a major problem. Other possible subjects for guideline development include procedures for followup of routine mammography (see chapter 2) and routine preoperative testing ( 125). The federal government already has the administrative mechanisms in place to sponsor guideline development efforts in areas identified as high potential sources of inappropriate defensive prac-
tices. The Agency for Health Care Policy and Researchs Office of the Forum for Quality and Effectiveness in Health Care could sponsor the development of such guidelines and dissemination to the states. It could also act as a clearinghouse for similar defensive-medicine targeted guidelines developed at the state level. The development and dissemination of guidelines linked to specific problems of defensive medicine may be enough to encourage states to adopt legislation that would give them greater weight in court and thus help clarify the standard of care. Alternatively, the federal government could mandate changes in state civil procedure to make it easy to introduce such guidelines as evidence or to enhance their evidentiary weight. Constitutional issues would have to be considered in designing any such federal legislation. The impact of this approach on defensive medicine is more predictable than other reforms, because guidelines would be targeted to specific areas where defensive medical practice is prevalent and widely agreed to promote medical practices with low expected benefits and high costs. The overall impact on health care practices and costs is likely to be small, however. There are probably a very limited number of clinical situations in which such guidelines could be developed with sufficient specificity to provide clear-cut clinical guidance and legal protection. In addition, even if clinical practice guidelines do indicate when a procedure need not be ordered, there is no guarantee that physicians will substantially change their behavior to conform to such guidelines. It must also be recognized that such guidelines, when legislatively mandated for use in malpractice cases, are implicitly setting upper limits on the cost that society is willing to bear for small improvements in health outcomes. Who makes these decisions (e.g., physician groups, broadly representative public commissions) may affect the acceptability of guidelines to practicing physicians,
their legal status, and the degree to which they reflect societys true preferences.
q q Establish demonstration projects of malpractice reforms that either remove or limit the physicians involvement in the litigation proc en.
Physicians express dissatisfaction with many aspects of the legal system, for example, large noneconomic damages, the jurys ability to determine the standard of care, and the quality of expert witnesses. Although traditional tort reforms may reduce physicians anxieties about being sued or financially ruined, they do not eliminate the threat of being sued and do nothing to clarify the standard of care. Reforms that relieve the physician of personal liability may be more likely to reduce defensive medicine. The two most promising reforms from this perspective are: s selective no-fault compensation systems using ACES, and s enterprise liability. If personal liability is retained, then reforms that significantly alter the nature of the physicians interaction with the legal system to provide greater consistency in outcomes and payouts may have some impact on defensive medic inc. Such reforms include: programs to encourage the use of binding arbitration, and s the AMA/SSMLP administrative proposal.
s
The impact of these reforms on defensive medicine is unknown. However, any reform that relieves the physician of personal liability could also have an adverse impact on the quality of care. To counter this effect, quality control systems would need to be in place. If these systems used sanctions to ensure quality, they could also prompt defensive medical practice. Much would depend on whether physicians perceive new quali-
t y control systems as rational and fairtwo adjectives rarely used by physicians to describe the tort system. Because of the many uncertainties about the impact of these reforms on defensive medicine and the quality of care, state-level demonstrations may be warranted to evaluate these more innovative alternatives before full-scale commitment to any particular model.
Finally, the savings generated through reductions in defensive medicine, which are likely to be modest overall, are unlikely to offset the additional costs of some of these reforms. In particular, a selective no-fault system and the AMA/SSMLP administrative proposal will probably substantially increase net expenditures for medical injury compensation.
Defensive Medicine: Definition and Causes

espitc widespread use of the term in the current health policy debate, there is limited understanding of-let alone consensus on-- the true nature of defensive medicine. This chapter explores the concept of defensive medic inc. First, it sets forth the Office of Technology Assessment (OTAs) definition and compares it with alternative approaches to defining defensive medicine. Second. it explores the sources of defensive medicine: why physicians want to avoid lawsuits: what types of signals the malpractice system sends to physicians; the role of institutional risk management and quality assurance activities in defensive medicine; and finally, the role of graduate medical education in promoting defensive medicine.
DEFINING DEFENSIVE MEDICINE

OTAS definition of defensive medicine, adapted from several sources ( 71 ,252,260), is as follows:
21
Note that this definition includes only those practice changes affecting the rate of use of medical services, Changes in practice style, such as spending more time with patients, giving more attention to careful documentation of the medical record, or making greater efforts to communicate or obtain informed consent, are not defensive medical practices under OTAs definition. Documenting the extent of these changes in practice style would be very difficult, and their positive implicat ions for the quality of care are less equivocal than are the implications of doing more or fewer procedures. OTAs definition raises three important issues of interpretation. Each is discussed below.
Conscious vs. Unconscious Defensive Medicine

The first question is whether the desire to limit malpractice liability must be conscious in order for a practice to be labeled defensive medicine. OTAs definition permits a practice to be defined as defensive even if the physician is not consciously motivated by a concern about liability. How can physicians practice defensively without knowing that the y do? Over time, many procedures originally performed out of conscious concern about liability may become so ingrained in customary practice that physicians are no longer aware of the original motivation for doing them and come to believe that such practices are medically indicated. Medical training may incorporate such customs without explicitly communicating to interns and residents the medicolegal considerations behind them. Thus, although physicians may practice conscious defensive medicine in a limited set of clinical situations, additional defensive practices may result from the cumulative response of the medical profession to signals from the malpractice system.
Defensive Medicine: Good, Bad, or Both?

OTAs definition does not specify whether the defensive action is good or bad for the patient; it requires only that the physicians primary motivation to act is the desire to reduce the risk of liability. Thus, some defensive medical practices may be medically justified and appropriate while others are medically inappropriate.
2 For example, Dr. James Todd, cxecuti~e vice president of the American Medical Associatitm, recently defined defensive medicine as (~bjcct]vc rmxurcs taken to d(xument clinical judgment in case there is a lawsuit... (226). Lewin-VH1, Inc., adopted a similar definition in a nxxmt study funded by MMI, Inc. ( 12S ).
Chapter 2 Defensive Medicine: Definition and Causes | 23
This definition conflicts with other definitions of defensive medicine. The Secretarys Commission on Medical Malpractice, for example, defined defensive medicine to include only those medical practices performed primarily to prevent or defend against the threat of liability that are not medically justified (243). This definition is consistent with the widely accepted pejorative view of doctors ordering unnecessary and cost] y procedures because of the malpractice system. OTA rejected this definition for two reasons. First, measuring the extent of defensive medicine under such a definition would require judgments about the appropriateness of all medical practicesa task far beyond the scope of this study and infeasible given the current state of medical knowledge. Second, malpractice reforms that reduce physicians propensity to engage in inappropriate defensive medicine may also reduce their use of appropriate practices. Analysis of the impact of malpractice reforms on defensive medicine should include explicit consideration of their impact on both kinds of behavior. One explicit goal of the medical malpractice system is to deter doctors and other health care providers from putting patients at excessive risk of bad outcomes. To the extent that it exists, defensive medicine that improves outcomes contributes to the deterrence goal. In the process of improving outcomes, good defensive medicine may raise or lower health care costs. But the malpractice system may also encourage physicians to order risky tests or procedures that both raise health care costs and on balance do more harm than good for patients. These practices are clearly both inappropriate and wasteful of health care dollars. Figure 2-1 gives a simple schematic of four kinds of defensive medicine. classified according to their impact on health care outcomes and costs. Box A includes practice changes that are unquestionable y good for the health care system and its pa-
Effect on health care costs
Cost reducing) quality raising
Cost raising quality raising
Cost reducing/ quality reducing
Cost raising j quality reducing
tients, because patients do better and health care costs are reduced. Box C includes practices that are unquestionabl y bad. Boxes B and D, however , represent situations involving tradeoffs between health care quality and health care costs. All defensive practices in boxes A and D would contribute to the deterrent effect of the malpractice system, because patients do better when they have access to them. Which practices in box D are medically appropriate, however, is a matter of judgment. Is an expensive test justified for a patient who has one chance in 15,()()0 of having the disease in question? What if the chance of a positive test is one in 100,000? What if the disease in question is not very serious? Judgments about questions such as these determine the dividing line be tween appropriate and inappropriate medical procedures.
OTA has no evidence on the frequency of these four different kinds of defensive medicine.3 Not only is it difficult to measure the frequency of defensive medicine overall, but when instances of defensive medicine are found it is also difficult to categorize them according to their ultimate impact on costs and health outcomes. The following two examples illustrate this point.
Number of Percent biopsies malignant a

Mammograms interpreted at Brigham and Womens Hospital Mammograms Interpreted at
a
b
280
26.1%
other hospitals and officesb Lobular carcinomas considered benign
981
16.7C
Example #1: Referrals for Breast Biopsy After Screening Mammography

The Physicians Insurance Association of America recently reported that delayed diagnosis of breast malignancy was the second most common cause of malpractice claims and accounted for the greatest percentage of money awarded to plaintiffs ( 184). It would not be surprising, then, if it were discovered that radiologists responsible for interpreting screening mammograms practice defensively by referring for- biopsy any patient whose mammogram contained a suspicious finding, no matter how equivocal. A study by Meyer and colleagues at Brigham and Womens Hospital, a large teaching hospital in Boston, suggests that community-based radiologists are more aggressive in their recommendations for followup of suspicious mammograms than are hospital radiologists ( 160). Table 2-1 contrasts the positive biopsy rate for mammograms interpreted by staff radiologists at the teaching hospital with that of mammograms referred for biopsy by radiologists practicing at other institutions or in the community. Whereas 26.1 percent of the biopsies performed on cases originating at the hospital were positive, only 16.7 percent of biopsies for cases originating in other settings were positive. 4
There were 73 separate hospitals and offices Statistical significance of difference in percent malignant = p< 05
SOURCE J E Meyer, T Eberleln P Stomper, and M Sonnenfeld, Biopsy of Occult Breast Lesions Analysis of 1261 Abnormalities, Jour(17) 2341-2343, 1990 nal of the American Medical Association 263,
Meyer and colleagues did not study whether the difference was due to defensive medicine on the part of the community radiologists versus other factors such as skill or patient differences, Even if it were possible (o conclude that the entire difference is due to defensive medicine, however, it Would still be impossible to classify it according to the schematic of figure 2-1. On the one hand, the community radiologists followed a diagnostic process that presumably would find more cancers, most likely at an earlier and more easily treatable stage. On the other hand, breast biopsy is painful and scarring, which not only distresses patients but also makes future diagnosis of malignancy in a patient with a negative biopsy more difficult (27). Some experts advocate mammographic followup in 6 to 12 months in cases where the first mammogram is interpreted as most likely benign (28). However, in a retrospective study of 400 breast biopsies from screening mammograms, researchers found that eliminating 126 of the least suspicious findings from the group referred for biopsy would have missed five cancers, four of
3 At present, there arc aln]{)st m) studIcs of the e~tcnt tt) which the malpractices) stem. as it IS presen[ly configured, deters physicians frtm] pr(widmg care [~f I(m qual]ty. OTA IS aware of (rely (me study addressing th[s Issue in a h[~spital inpatient p)pulat]tm. Researchers at Harvard Llnly ~rsl[} recent] } anal) /cd the rc]atlonship bclwccn the number of malpractice cl:i Ims pc.r ncgl[gent injury( and the r:ite of negllgt!nt injuries in Ncw Y(wh State ht)spltals in 1984. They fallcxi to denltmstra[c a significant rclatit)nship hctwccn ii hi)spitals malpractice claim activit} and its rate ()! negl]gcnl lnjur} (254).
procedures classified as medicolegalan undisplaced navicular (hand) fracture-did treatment change as a result of the x-ray. The study did not explore the extent to which the emergency room physicians who ordered these x-rays were practicing defensive medicine. Other motivations may have entered into ordering procedures. The study authors suggested that the emergency room physicians, most of whom were interns and residents, may not have had the experience or appropriate training to discriminate adequately among cases. The high percentage of medicolegal spine and skull x-rays (see table 2-2) suggests that physicians tend to be aggressive in their test ordering when the medical consequences of being wrong are very serious.
Region Cervical spine Pelvis Skull Sacrum Lumbar spine Other

a
Percent of all procedures 1 %40 10 19
Percent classified medico legal

7 8 %
71
70
05
4 80
69
62 39
Total number of procedures was 2,359 Some patients underwent more than one procedure
SOURCE M Eilastam, E Rose, and H Jones, Utlllzatlon of Dlagnostlc Radiologic Examinations Journal of Trauma 20(1) 61-66, 1980
5 Medic(}Iega] was a name given after the study was completed to all cases not mce[]ng the cl inlcal cntcria for fracture In the other three categories.
rule out a remote but potential] y very serious or fatal condition. When the same experts were asked to alter the clinical scenarios to remove defensive medicine as a motive, they virtually always added signs and symptoms that increased the probability that the patient had a serious disease. Figure 2-2 illustrates the general relationship between the probability that the patient has the disease(s) or condition(s) being tested for and the probability that a physician will order a test. As the severity of the suspected disease or condition increases, the desire to test increases at any given probability of disease. In certain cases, concern about liability might decrease physicians tolerance for uncertainty and cause them to order tests more frequently when the probability of disease is very low or very high (see figure 2-2). When the probability of disease is very low, the physician may want to rule out its possibility. When the probability of disease is very high, the physician may be concerned about documentation of the condition for protection against potential claims of misdiagnosis. At more intermediate probabilities. the effect of malpractice liability on physicians test ordering might not be so great, since uncertainty is already high. Again, one might expect defensive medicine to be most pronounced when the probability of a positive test is very low but the consequences of not finding the disease are catastrophic.
THE SOURCES OF DEFENSIVE
Not all t)f thtx miswl diagntwx result fr(ml omissi(ms in testing. Missed dtagm)ses may (K-CUr as a result ~~f fwlurc to c(mlpletc a physical
exarmnat Itm, I ncxwrcct mlcrprctatitm of a diagm)stlc test, or delay in fol hm m: up [m a P )S itivc tindlng. on] Issiims in test]ng prt)babl~ represent
a n]int}r]ty of a]l casts of mIsscd diagnosis (26, I I 9).
100% -
- -
- -
- - - - -
Disease with
severe health consequences
Relative frequency
of test ordering
Disease with moderate health consequences
0%
Probability of disease
1000!0
Frequency of testing in absence of fear of malpractice . . . . . . Frequency of testing in presence of fear of malpractice
SOURCE Office of Technology Assessment 1994
veys reveal that an overwhelming majority believe that most malpractice claims are unwarranted and that the present system for resolving claims is unfair (38, 180). Although some of these beliefs may not be well-founded, they are real and pervasive in the physician community. Evidence has also shown that, across all specialties, physicians tend to substantially overestimate their risk of being sued ( 123) (see table 2-3).
Financial Consequences
For the vast majority of physicians, a malpractice suit does not have a major impact on personal finances or professional status, mainly because most physicians have adequate malpractice insur-
ance. Some physicians report that lawsuits damage their reputation or reduce the demand for their services, but most classify such losses as minor, and physicians who have already been sued are less likely than those who have not to report these effects ( 180). Physicians do incur some personal financial costs when they are named in a malpractice suit. These costs are primarily in the form of lost days of practice, although sometimes physicians retain personal counsel. (Physicians are usually represented by their insurers counsel.) Survey-based estimates of physician time and income lost in defending against malpractice claims range from 2.7 to 5 days of practice and
7 The best available emp]ncal cl Klcncc ]ndlcattx [hat -U) to 60 percent ()( malpractice claims are n{mmerit(wi(ms, bu[ rm~st of these suits are eliminated early ]n the prt)ccss (68,222.235). In addltl(m. rctr(npcctl$ c studws of CI(MCI clmn M suggest that pa} ment of malpractice claims, whether through settlement or a trial, IS mlt haphazard-the vast may)ril) of lndcfenslhle claims are paid, and the substantial majtmi(y of defensible claims are dropped (40,68,222). (Defensibll]ty of a claim was Judged either by an insurer, physlcm panel, or ht)spital. ) On the other hand, the studies also docunkml [hat mls[akes arc SO JIKIIJWS made txjth m find]n: ph~ slc]ans rregl]gcn[ ~ ho JnCI the standard of care and in failing U) c(mqxnsate victims t)f medical negligence.
Physician characteristic
Perceived risk: percent of physicians sued per yeara 12 .I% 234

3 4 3
Actual risk: percent of physicians sued in 1986 3870 109

2 0 8
Ratio of perceived risk to actual risk 3.2 21

1 6
Specialty group Low-risk internal medicineb Medium-risk general surgeryc High-risk obstetrics, orthopedics, neurosurgery Suit status Never sued
149
Sued at least once

Overall
a
238
195 6.6 3.0
The question asked of physicians m this 1989 survey was In your opmlon, for every 100 physicians m your speclalfy m New York State, how many do you think WIII be sued at least once this year? b includes associated Speclaltles such as family prachce, gastroenterology, and neurology c Includes associated specialties such as ophthalmology, plastlc surgery, and urology SOURCE AdaDted from A G Lawthers A R Locallo, N M Lalrdet al , Phvslcwns PerceDtlons of the Risk of Bema Sued, Journa/ o//-/th/th Po/1(/cs. Pohcy and Law 17(3) 462-482, fall 1992
from $2,400 to $5,600 in lost income per claim (123,194). In a 1989 survey of New York physicians, six percent of those sued reported that they had retained their own counsel and incurred between $1,000 and $5,000 in out-of-pocket expenses; three percent of sued physicians reported paying out-of-pocket settlement costs, with one percent reporting expenses greater than $25,000 (123). Physicians anxiety about being sued may result from misperceptions about the potential financial consequences of a lawsuit. Numerous examples exist of multimillion dollar malpractice verdictsverdicts that far exceed most physicians insurance limit .8 But physicians almost never pay any damages above their policy limits because such awards are usually either covered by several defendants or reduced in post-trial negotiation among the parties (45). Individuals perceptions of risk, however, do not always agree with objective measures of risk. Recent federal and state laws requiring repotting of malpractice claims to central repositories may change the perceived importance of even a
single lawsuit in the minds of physicians. Since 1990, federal law has required all payments for malpractice made by or on behalf of a physician to be reported to a new National Practitioner Data Bank (NPDB). The NPDB maintains a short narrative on the incident, including any response filed by the physician (246). This information must be reviewed by hospitals when hiring new staff and every 2 years for current staff (45 C.F.R. Sec. 60.10). It can also be accessed by a limited number of other potential employers. Some states have their own malpractice reporting requirements. In California, for example, a report to the medical licensing board is required whenever a payment of $30,000 or more is made on behalf of a physician (Cal. Bus. & Prof. Code Sees. 801,802,803 (1989)). The purpose of federal and state reporting systems is to improve monitoring of physician quality and conduct. In California, for example, reports of malpractice awards are reviewed by the licensing board to determine if disciplinary action is warranted (153,224). The overwhelming majority of claims are reviewed by contract physi-
Most physicians carry policies of between $1 million to $2 million per occurrence and $3 million to $6 million per year(211 ).
cians and closed. Only those with evidence of gross negligence or incompetence are referred to regional offices for further action (224). Disciplinary actions in these few cases are almost always relatively minor; for example, being called in for a conference with a regional medical consultant. In rare cases, the Board may issue a restraining order or suspend a physician medical license (152). None of the federal or state databanks current] y in place are open to the general public. However, an ongoing debate over whether to allow public access to the Federal NPDB has probably increased physicians anxiety about being sued (165). The financial burden of malpractice premiums may be substantial for certain physicians in highrisk specialties or living in certain geographic areas. Malpractice insurance prcmiums vary by specialty and geographic area and can be very high in some localities. In 1987. obstetricians/ gynecologists (O B/GYNs ) in Dade and Broward Counties, Florida. paid $165,300 per year for standard coverage, compared with $69.300 for OB/GYNS outside of those counties, and $19,400 for family practitioners in Dade and Broward Counties (176). Physicians reactions to premium costs may sometimes be exacerbated by the fact that premiums are generally not volume-sensitive; OB/ GYNs with coverage for high-risk deliveries pay the same premium regardless of how many deli\eries they perform ( 2 100). 9 While malpractice insurance rates arc generally insensitive to personal malpractice history (21 0), the physician malpractice claim history can lead to denial or termination of coverage 206.207). In addition, a very smal1 percentage of physic i ans may incur some kind of financial or profcssiona sanction from their malpractice insurers if they have been named in negligence suits (207).
Psychological Consequences
Although the financial and professional costs of malpractice liability are real, the primary impact on physicians may be psychological. Physicians report that a malpractice claim causes short-term losses of self-esteem, and in two physician surveys. between 20 and 40 percent reported symptoms of clinical depression, anger, fatigue, or irritability (37,38).10 In another survey, 50 percent of physicians felt there would be a short-term decrease in self-esteem, and about one-third felt a suit could lead to long-term behavioral or personality changes, or physical illness. However, physicians who had already been sued reported these adverse effects at a rate about half of that for non-sued physicians, suggesting a worried well effect among physicians who have not been sued ( 180). The anxiety caused by a lawsuit may continue for a long time. The average time between filing of a claim and its resolution is approximately 33 months, although it may take longer than 48 months ( 186). Moreover, a claim is often not filed until 20 months after the incident ( 186), leaving the physician much time to speculate as to whether a particular patient will bring a suit after an adverse outcome.
A central goal of the tort system is to deter negligent behavior and hence improve the quality of medical care (253 ). At least two conditions must be met for the tort system to effective y deter poor quality care: first. the malpractice system must provide physicians with information as to what care is acceptable; second, physicians must be able to improve the quality of care they offer. The malpractice system, however, may not always
send a clear signal to physicians about the standard of care the legal system demands (221).
Physicians Interpretation of
the Legal Standard of Care
Physicians often express frustration with the malpractice system and, in particular, with the legal 11 In onversations with OTA. standard of care. many physicians claimed that the legal standard of care does not reflect medical practice but is instead a legal construct divorced from the practice of medicine. Some of this frustration may stem from the fact that it is difficult for physicians to predict from previous cases the standard of care expected in the future. The legal standard of care is developed anew in each case. which is not surprising, since each patient has unique medical and other characteristics. In addition, the practice of medicine changes rapidly. This de novo approach to each case. however. may appear to physicians as unpredictable, despite the fact that the legal standard of care is always based on expert testimony about the prevailing standard in the profession. Physicians also express concern about the qual ity of expert witnesses who establish the standard of care. An expert witness is required to have knowledge and skill above that of a lay person, but there is generally no requirement that an expert have education, training, and experience similar to that of the defendant ( 185). According to the American Medical Association (AMA), experts have been permitted to testify when they do not have specific cxperience in the relevant area of practice (9). In some cases, the expert had not yet entered the profession at the time of the incident (9). Although a witnesss qualifications may be challenged to prevent admission of testimony before the jury, once the testimony is admitted, the jury decides whether the testimony is credible.
The courts recognize that there is variation in medical practice, and a physician will not be held 1 i able for following a practice if a q respectable minority of physicians also follows the practice (134). But the jury must resolve any disagreements among experts on whether a physician Should have made a particular diagnosis or performed a certain procedure. Physicians believe that lay juries are poorly equipped to resolve complicated clinical judgment issues (9). If physicians believe that the legal system is unpredictable and incapable of accurately judging the quality of medical care (a conclusion not fully supported by recent empirical researchsee footnote 7), then physicians are not receiving a clear signal about the standard of care demanded by the legal system. Consequently, physicians may conclude that the only way to avoid a suit is to do everything possible to avoid an adverse outcome, no matter how unlikely the bad outcome is or how costly the intervention. A key area of concern is the potential liability for missed or delayed diagnosis, Suits alleging missed or delayed diagnosis appear to be increasing in severity. Data obtained from St. Paul's Fire and Marine Insurance Company showed that although "failure-to-diagnose claims did not increase as a percent of total claims between 1980 and 1993, there was a statistically significant in-
crease in the amount paid for these claims. In 1984, payments for failure-to-diagnose claims ac counted for 25 percent of all payouts, compared \vith 34 percent in 1993 (228). The increasing relative importance of failuret o-diagnose claims may result from a combination of- better diagnostic techniques and improved outcomes when serious medical conditions are detected earlier. Both of these technological trends could make the consequences of not testing more serious. As technology changes, the legal standard
of care evolves, and physicians may feel especially vulnerable if they are not aggressive n diagnosis.
Changing Legal Doctrines

Changes in legal doctrines that alter the boundary between negligence and non-negligence may also confuse physicians. Recent changes in the legal doctrine called loss of chance in some states have put physicians at greater risk of being held negligent for not providing a diagnosis or treatment even when the chance of recovery from the condition are low. In cases involving the loss-of-chance doctrine. the plaintiff usually has a serious or fatal condition but, if properly treated, has a chance of longer survival or cure. A patient (or the patients estate ) can sue for malpractice, claiming that a physicians negligent act. rather than the underlying disease. was the proximate cause of the plaintiff death or increased suffering. The questions of whether the physician caused the injury and whether the underlying disease was responsible are decided by the jury. However, the judge does not allow the jury to consider questions of causality and negligence unless there is sufficient evidcncc that the physician~ action could be the proximate cause of the patient injury or death. In general, to have sufficient evidence, the plaintiff must prove that it is more likely than not that, in the absence of the physician ncgligence, he or she would have survived or had a better outcome (96, 110, 178). To meet this standard, the courts have traditionally required that the plain tiff chance of survival with proper diagnosis or treatment would have been better than 50 percent (96,1 10).
A minority of courts have abandoned the strict 51 percent rule and instead allows the jury to determine whether a physician was negligent when the physicians conduct is determined to be a substantial factor in causing the plaintiff's harm ( 178).12 The physician may be held liable when his or her negligence eliminated a 35 or 40 percent chance of survival or recovery (96). In one often -cited case, the jury was allowed to consider whether a health maintenance organization (HMO) could be held liable for the patients death from lung cancer when his physicians negligence in diagnosing the cancer reduced the patient chance of survival from 39 to 25 percent. 13 The court went on to say, however, that the defendant was not liable for full damages resulting from the plaintiffs death. but only for those damages directly related to the delay in diagnosis caused by the physician negligence. 14 A number of courts that allow recovery when the chance of survival is less than 50 percent limit the damages according] y (96, 110,151 ). Physicians may find these cases troubling because the courts are willing to hold the physician liable when his or her conduct diminishes the patients chances for survival by only a small percentage. Physicians may feel they are being unfairly held accountable for an inevitable injury or death, given the patient underlying medical condition. As one court noted, when dealing with causation, it can never be known with certainty whether a different course of treatment would have avoided the adverse consequenccs. 15 Finally, predicting surviva l rates is not an exact science, which leaves room for conflicting expert testimony. If sufficient numbers of physicians respond to missed diagnosis cases by beginning to screen for
serious conditions in low-risk populations, then the standard of care in the profession may change. If ordering diagnostic tests on low-risk patients becomes more common, plaintiffs will have an easier time establishing that the failure to order the test was negligent, because more medical experts will be willing to testify that such testing is the standard of care. Gradually, the standard of care will be ratcheted up as physicians respond to the increasing threat of malpractice for failure to diagnose. Eventually, physicians may cease to characterize or even think about their actions as defensive.
Hospitals, HMO's, and malpractice insurers often have risk management and quality assurance programs that seek to minimize the number of adverse events and malpractice suits and improve the quality of care by changing physician behavior. Many risk management activities are directed toward nonphysician hospital employees (e.g., nursing staff) (41 ), but risk management programs are increasingly focusing on reducing the risk of injury in clinical care (41, 120.163, 167). Because risk management is an administrative function, risk managers are unlikely to be clinical ly trained. Recently, however, nurses have played a more active role in risk management (41 ,237). Risk managers do not typically develop clinical protocols for physicians but instead spend much of their time working with the hospital and legal personnel to address existing and potential claims. Larger risk management programs provide educational information on the kinds of suits that are brought and analysis of how these suits might be prevented+. g., through better communication with patients, better informed consent, and implementation of systems designed to minimize human error (46, 181,182.183,184, 196,237), The most common recommendations of risk managers are to document the record completely and to obtain informed consent (5,36,46). Sys-
tems can also be set up to prevent mistakes that can lead to injuries. For example, protocols are often set up to account for all sponges and instruments after surgery, or to ensure that the correct heart valve is selected during surgery (163,237). OTA learned in interviews with risk managers that they may also recommend removing technology if the staff does not know how to use it properly; for example, removing fetal monitors from an emergency room, closing underequipped or understaffed faci1ities, or referring difficult cases to specialists. How physicians respond to information promulgated through risk management programs has not been studied. Although risk managers stress documenting the chart, communicating with the patient, and obtaining informed consent, physicians preferred method of documenting diagnosis may sometimes be to perform additional tests and procedures (46,86). For example, in a risk management study of Erbs Palsy and shoulder dystocia conducted by the Risk Management Foundation of the Harvard Medical Institutions, physicians were told:
although shoulder dystocia occurs infrequently and largely unexpectedly, assessing risk factors such as maternal diabetes or large fetus (4000 grams or more) may help obstetricians anticipate shoulder dystocia . . . Obstetricians should document any evaluation performed for these conditions as well as their conclusions and followup. (217)
This guidance appeared with a review of malpractice claims that included an allegation of failure to do an ultrasound to evaluate cephalopelvic disproportion (2 17). Physicians could interpret such information as a suggestion that they perform routine intrapartum ultrasound to evaluate fetal size. .A trend in recent years is the linkage of risk management with quality assurance activities. The Joint Commission on Accreditation of Health Care Organizations requires that hospitals seeking accreditation have programs linking risk management with quality assurance ( 167). American Health Care Systems Inc., has published a model
program for integrating quality and risk management activities in multihospital systems (4). Quality assurance in hospitals or other institutions is usually overseen by physicians (42,46, 163). The quality assurance process is often triggered by reports from the risk management department (41,1 63). In some quality assurance programs, protocols are designed specifically to reduce the number of malpractice claims. For example. several clinical departments of the Harvard University-affiliated medical institutions use protocols for anesthesia, obstetrics, and radiology that were designed to address problems identified in reviews of malpractice claims (99). These guidelines primarily address proper documentation, prompt and accurate communication of clinical data among staff. informed consent, and monitoring of patients.16 The guidelines are voluntary, but they have been widely adopted within the Harvard Medical Institutions (99). Certain malpractice insurersmainly physician--owned companies-develop guidelines to prevent malpractice claims ( 19,223). Some insurer guidelines are mandatory clinical protocols that physicians must follow to maintain coverage. although physicians may deviate from the guidelines with proper documentation (19.43,154,). These protocols are often developed through a consensus development process among physicians using medical literature and expert consultants. If these guidelines and protocols improve outcomes of care and minimize errors, then they may be an appropriate response to the signals from the malpractice system, even if they involve increasing the number of procedures or services provided. That is, they may promote quality-enhancing rather than wasteful defensive medicine.
Risk managers contacted by OTA and others who were involved in quality control consistently stated that their quality assurance programs did not promote unnecessary tests and procedures (80.163.237). However, risk management and quality assurance programs may at times encourage broader use of certain tests and procedures in order to avoid the potential for serious. but remote, adverse outcomes. Whether these measures are unnecessary is a value judgment. If the risk management process is insulated from pressures to control healh care spending. recommendations are unlikely to reflect a balancing of cost and outcome considerations. In contrast to risk management and quality assurance programs, the individual physician does not undertake a specific review of claims but instead reacts to a less orgamozed signal and tries to anticipate future suits. This reactive and emoional process may be even more likely to lead to defensive medicine than the systematic claims review and guideline development done by hospitals, HMOs. and malpractice insurers.
type of programs studied, it is difficult to draw any broad generalizations from the interviews about the teaching of defensive medicine during graduate medical training. However, responses to the interviews suggested the following findings regarding the role of graduate medical education in promoting defensive medicine: Malpractice concerns were noted by residents and faculty in all four (mining programs, but the extent of concern varied greatly across department specialty, geographic location, and individual attending physician. Concern appeared to be more pervasive in obstetrics/gynecology than in internal medicine and more heightened in the metropolitan training center than at the training center in a small city (see box 2-1 ). Limited formal instruction on malpractice Issues in organized classes and conferences does exist, but defensive medicine is not taught explicitly at these seminars. In general, residents are exposed to many different practice styles during their training. The extent to which they are exposed to defensive medicine practices depends in large part on the practice styles of the faculty with whom they work most closely. Some faculty and senior residents in each of the four centers acknowledge that they teach some defensive practices to junior residents; others claim they do not. [formation about defensive medicine is conveyed not only consciously but also unknowingly by faculty and senior residents. Recordkeeping, patient communication, informed consent, hospital admissions, referrals and consultations, and use of additional tests and procedures were all cited by faculty and residents as examples of defensive practices
Obstetrics and Gynecology Training Program, Medical Center A Faculty "[It is] very difficult for residents to escape sensing concern [about malpractice] Nonetheless everyone here has as a first goal to do the right thing by the patient I do not think that anyone IS cold enough to reduce Iiability at the expense of mistreating or not adequately treating the patient a second concern, and a close second is creating a scenario that makes it less Iikely that the patient will sue A lot of defensive procedures that are incorporated in our practice are not consciously acknowledged to be defensive procedures. If I have a patient with a gastrointestinal complaint and I think I know what it is I may still be inclined to refer her to a specialist even though I can treat it myself I know that there iS back-up here I have not explicitly taught this to residents but they get a sense of it The minor purpose of the chart [I e
the medical record] iS to inform other practitioners about the care
Internal Medicine Training Program, Medical Center B Faculty.

s
1 do not discuss, implicitly explicitly, a defensive posture with patients I view the concept of defensive
medicine as poor medical practice. You are doing something unnecessary to cover yourself and we do not stress for our residents that we should do that But I have had residents say I think we are going to be sued, and my usual response iS to shrug my shoulders and say do the right thing." s 1 cannot say that after or during a case I do not consider the legal ramifications, but I still try to make my decisions based on the patient and not on the legal system Residents.
q
If someone iS explicit [about teaching defensive rnedicine], it makes me question it more and say that
iS a stupid reason and you should not do it If it is implicit, it iS insidious
Center A IS in a large metropolitan area center B
IS in
a small
city
taught to varying degrees during residency. Among these examples, the most commonly mentioned was documentation of patient care. Most residents leave training thinking they have to protect themselves against medical malpractice litigation when they go into practice. The effects of graduate medical education on the subsequent practice of defensive medicine by trained physicians vary depending on the degree to which they were exposed to it during training and the length of time elapsed since completion of training.
training program (69). Because it is unclear what type of practice settingacademic, hospitalbased, community-based-is most conducive to the practice of defensive medicine, it is difficult to predict whether a shift from one setting to another would on balance increase or decrease the teaching of defensive medicine.
CONCLUSIONS
Under OTAs definition, defensive medicine occurs when doctors order tests, procedures, or visits, or avoid high-risk patients or procedures, primarily (but not necessarily so/e/}) to reduce their exposure to malpractice liability. This definition recognizes that practices regarded as defensive may be motivated by other factors in addition to liability concerns (e.g., medical benefit, financial incentives) and may be either quality-enhancing or quality-reducing. Due to lack of information on the relative effectiveness of many medical interventions. as well as lack of consensus on what level of risk individuals or society are willing to accept. it is difficult if not impossible to classify most instances of defensive medicine as purely "good" or "bad". I n add i t ion, a substantial proportion of defensive medicine may occur unconsciously-i.e., physicians may follow practices that initially evolved out of liability concerns but later became customary practice.
For some time now, there has been a movement afoot to restructure residency programs (247). It is unclear exactly what direction these reforms might take; however, to the extent that any future reforms affect the relationships between and among hospitals, teaching faculty, and residents, they may also affect the channels through which defensive practices are currently taught to young physicians in training. For example, if more of residency training is shifted to ambulatory care settings, the role of the large medical institution as a source of the standards and values of a resident future professional career may be diminished. OTAs interviews, as well as literature on the sociology of medical education, suggest that the molding of a students practice style depends heavily on the practice style of his or her mentor as well as the general culture of the particular
Physicians receive signals from the malpractice system in a variety of ways, including personal litigation experience, the experience of their colleagues, the media, risk management and quality assurance activities, and their malpractice insurance premiums. Although it is unclear whether and to what extent these malpractice signals affect physician practice, it has been documented that physicians consistently overestimate their own and their colleagues risk of being sued. Physicians are concerned about the professional, fi-
nancial, and psychological consequences of litigation but, on balance, they tend to overestimate the risk of these effects as well. Young physicians in residency training maybe particularly susceptible to learning defensive practices-either explicitly or implicitlyfrom their supervisors and faculty. Graduate medical education may thus help perpetuate defensive medicine at both the conscious and unconscious levels.
Summary of the Evidence on Defensive Medicine

or more than two decades, news stories, interest groups, and witnesses at congressional hearings have quoted estimates of the extent of defensive medicine and its impact on health care costs. Often these statements have been based on anecdotes, which mayor may not represent the general experience of physicians in the United States. This chapter reviews the evidence regarding the extent of defensive medicine in the United States, including new evidence developed as part of this Office of Technology Assessment (OTA) study. It begins by outlining the major strengths and weaknesses of methods used to measure defensive medicine. It then summarizes the findings of many studies conducted over the past two decades. Some studies surveyed physicians directly about the extent of their defensive behavior; others used objective data and more sophisticated statistical analyses. To expand the base of knowledge in this area, OTA undertook four physician surveys and commissioned three additional empirical studies.
APPROACHES TO MEASURING THE EXTENT OF DEFENSIVE MEDICINE

A challenge facing all approaches to measuring the extent of defensive medicine is to isolate the precise contribution that concern about malpractice liability makes to medical practice decisions. Defensive medicine typically operates in tandem with other forces to motivate clinical practice decisions. Figure 3-1 presents a model of the many influences on physician test ordering or treatment decisions. Some of these influences are clinical: I 39
of disease
for uncertainty
SOURCE Off Ice of Technology Assessment, 1994 Adapted from unpublished work of Richard Kravltz, MD, Ass~stant Professor of Medlcme, Unlverslty of Cahforn[a, Daws, School of Med[clne, Sacramento, CA
Chapter 3 Summary of the Evidence on Defensive Medicine | 41
patient symptoms, seriousness of the suspected disease, degree of certainty about diagnosis, accuracy of the available diagnostic tests, and risks and benefits of treatment. Other influences, in addition to the fear of malpractice liability, are nonclinical: 1 availability of technology, physician specialty and training, practice organization (solo, group, hospitalbased), familiarity with the patient, awareness of and sensitivity to test costs, financial incentives, patient expectations, and insurance status of the patient. Sometimes these other factors dominate malpractice liability concerns; some, such as patients insurance coverage and financial incentives under fee-for-service medicine, may enable physicians to act on their fear of liability. There are four major methodologic approaches to measuring defensive medicine:
risk patients or procedures or quit medical practice altogether (negative defensive medicine). The major problem with this approach is that people do not always accurately report what they do. Most physician surveys of this sort inadvertently prompt respondents to think about malpractice liability and its potential effects on their medical practices. This prompting may lead physicians to respond in ways they would not if they were simply asked how and why their practices have changedwithout asking directly about liability concerns. For example, the attention paid to defensive medicine by physic i an organizations, the news media, and policy makers might cause physicians to exaggerate the impact of liability concerns on their practices in the hope of eliciting a favorable political response, An additional problem of most surveys of this kind is that they do not ask about the extent to which respondents practice defensive medicine only whether or not they practice it.
The strengths and weaknesses of each of these approaches are discussed below.
The simplest way to gauge the extent of defensive medicine is to ask physicians how their medical practices have been affected by the threat of malpractice liability. Questions typically asked in such surveys include whether malpractice concerns have caused the physician generally to use additional diagnostic or therapeutic procedures (positive defensive medicine) or to avoid high-
A clinical scenario survey typically presents physicians with a description of a simulated patient and asks them to choose specified clinical actions. Respondents then indicate which of a list of reasons influenced their choices, with one of the choices being malpractice liability concerns. One advantage of this approach over the more general surveys described above is that prompting may be less direct if malpractice liability is only one among many reasons. Another advantage is that scenarios can focus in on areas where defensive medicine is thought to be a major concern. Finally, because they ask more concrete and precise questions about particular clinical situations, scenarios may permit more reliable estimates of the extent of defensive medicine in those particular areas. Only one previously published study, conducted by the Duke Law Journal Project in 1970
[ See appendix C ff~r a rck lew of the ev]dencc IlnKlng these and other noncl]nlcal fact(ws tt~ the utll Izati{m of scrk Icei.
(58), has used this approach. OTA conducted four clinical scenario surveys of the memberships of three medical professional societies and contracted for a study of defensive medicine in New Jersey that used this approach. To succeed in measuring defensive medicine, a clinical scenario survey must succinctly yet thoroughly describe the key features of the simulated case, provide lists of all likely clinical choices and meaningful reasons for making those choices, and blind the respondents to the purpose of the survey. An open question is whether clinical scenarios that include malpractice liability concerns among potential reasons for choice, without any other references to defensive medicine, sufficiently blind respondents to the purpose of the survey. But not including a list of reasons (i.e., asking respondents to list their own reasons for each clinical choice) also runs the risk of biased responses. Physicians may regard such an open-ended instrument as a test of their medical knowledge and cite only clinical factors. A critical limitation of clinical scenario surveys is that their results cannot be generalized beyond the specific scenarios, and results of different scenarios cannot be directly compared with one another. Indeed, the more clinical and demographic detail given in a scenario, the less generalizable its results are to other clinical situations. Finally, clinical scenario surveys capture only those defensive practices of which the physician is consciously aware.
yses, can control for other factors that might also influence physicians behavior (e.g., patient age and health status, hospital characteristics, socioeconomic factors). These studies usually use existing utilization data gathered for other purposes, such as hospital discharge records or physician health insurance claims. The unit of analysis can be the individual physician, the hospital, or the geographic area. The major strengths of this approach include the use of more objective data, the potential for large sample sizes, and the ability to control for many different influences on physician behavior. Typical problems confronting such studies include: limited generalizability due to the availability of data only for certain health care providers or localities, incomplete control for relevant factors other than malpractice liability (e.g., clinical indications), limited or problematic data on both independent and dependent variables, and small numbers of physicians or hospitals in certain categories or geographic areas. To the extent that these limitations can be minimized, multivariate studies can provide strong evidence regarding the incrcmental impact of differences in malpractice liability risk on physicians use of procedures. They cannot, however, provide a comprehensive estimate of the extent of defensive medicine. For example, a multivariate study might determine that there is a difference in test ordering between physicians who have been sued and those who have not, or between physicians with higher and lower malpractice insurance premiums. It cannot, however, detect the overall level of defensive behavior that results from a generalized fear of malpractice liability among all physicians. Furthermore, even if multivariate studies succeed in finding a statistically significant association be-
A statistically significant finding is one that is unlikely to have oecurrcd solely as a result t)f chance. Through(mt (his rqxm, a finding is considered to be statistically significant if the probability that it occurred due to chance alone is no greater than five tmt of I ()@-i. e., a p value of 0.05 (w less.
tween levels of malpractice liability risk and physician behavior, the direction of causality still cannot be inferred with absolute certainty.
Case studies describe the impact of malpractice liability concerns on the use of a specific medical technology. Such studies can provide valuable detail on the role of malpractice liability in both the initial diffusion and current use of technologies. As part of this assessment, OTA commissioned a case study examining the influence of malpractice liability concerns on the diffusion of a new diagnostic technology first introduced in 1987: low osmolality contrast agents. (The findings of this case study are described in a subsequent section of this chapter. ) The primary limitation of case studies is that they typically must rely on subjective information and do not permit adequate control for the influence of factors other than defensive medicine on patterns of diffusion and use of technology.
EVIDENCE OF THE EXTENT OF DEFENSIVE MEDICINE
Thirty of the 32 studies addressed negative defensive medicine. Of these 30. eight were national surveys, nine were state-level surveys of all specialties, and 13 were state-level surveys of obstetrics providers. Figure 3-2 presents selected findings of these surveys of negative defensive medic inc. As the figure indicates, surveys were oriented toward different areas of practice and asked questions about negative defensive medicine in a variety of ways. The proportion of respondents indicating restrictions in their practices due to malpractice liability concerns ranged from 1 to 64 percent.4 A series of surveys with similar structures conducted by the American College of Obstetricians and Gynecologists between 1983 and 1992 shows an increase in the proportion of respondents reporting negative defensive medicine between 1983 and 1987 (from 31.8 to 43.7 percent). and then a slight decrease in the following years (from 41.8 percent in 1990 to 39.0 percent in 1992) (see figure 3-2). Sixteen of the 32 studies reported on positive defensive medicine. Of these, five were national surveys and 11 were state-level. Selected findings are summarized in figure 3-3. Again, a variety of different specialties were surveyed and questions were posed in a number of different ways. Across these surveys, from 20 to 81 percent of physicians indicated that malpractice liability concerns had led them to order additional tests and procedures. As the variation in question structure and responses in these surveys shows (see figures 3-2, 3-3), direct physician surveys are a highly questionable source of quantitative information about defensive medicine. In the vast majority of the studies, the respondent was made aware that the survey was about malpractice liability and changes in the malpractice climate.
SPECIFIC SURVEY QUESTION OR COMMENT: National Surveys--All Specialties
SURVEY POP./YEAR
Physicians Practice Costs and Income Survey (PPCIS)-1986: Stopped treating certain cases in the past year due to malpractice insurance costs (Rosenbach, 1986) 1
PPCIS-1986
National Surveys--Obstetrics Providers4

AAFP-1987F/GPs: Of respondents who had ever provided obstetric services, percent who discontinued or decreased obstetric services due to cost or availability of liability insurance (AAFP, 1987) ACOG 1983, 1985, 1987, 1990, 1992-OB/GYNs: Which of the following changes, If any, have you made in your personal practice, as a result of the risk of malpractice? Percent answering yes to at least one of the following: a. decreased gynecological surgical procedures b. no longer do major gynecological surgery c. no longer practice obstetrics d. decreased number of deliveries e. decreased level of high- risk obstetric care (Porter, Novelli, & Assoc, 1983; Needham, Porter, NovelIi, 1985; Opinion Research Corp., 1988, 1990, 1992) AAFP-1 987
ACOG-1983 ACOG-1985 ACOG-1987 ACOG-1990 ACOG-1992
National Surveys--Surgery
ACS-1984: Limited practice by dropping certain operations due to malpractice risks (Bligh, 1984) ACS-1984
State-Level Surveys---All Specialties Chicago-1985. Stopped performing certain high-risk procedures due to malpractice litigation or its threat (Charles et al., 1985)
Kansas-1984: Do you believe problems associated with medical malpractice have affected your practice? If yes, do you Iimit your practice to less risky procedures? (Kansas Medical Society, 1985) Maryland 1987: In the last two years, have you made any changes as a result of the current malpractice climate? Yeseliminated or cut back specific services (Weisman et al., 1989)5 New York-1989: See fewer patients or perform fewer clinical procedures today than dld ten years ago (Lawthers et al., 1992)6 Chicago-1 985
Kansas-1984
Maryland-1987
New York-1 989
Texas-1 985
Texas-1 986
Texas-1 988
Wisconsin-1987: Refer more cases due to threat of a malpractice claim (Shapiro et al., 1989)
Wisconsin-1987
SPECIFIC SURVEY QUESTION OR COMMENT: State-Level Survey---Obstetric Providers

4
SURVEY POP./YEAR
Alabama-1985-F/GPs: Of respondents who had ever practiced obstetrics, percent who quit obstetrics in last five years and listed malpractice risk/fear as a reason for doing so (Alabama Academy of Family Physicians, 1986) Georgia-1988-OB/GYNs: Had quit obstetrics in the past three years solely because of malpractice (Georgia Obstet. & Gynec. Society, 1987) 7 Illinois-1987-OB/GYNs & F/GPs: Of respondents who had ever practiced obstetrics, percent who discontinued or planned to discontinue obstetrics and cited fear of a malpractice suit as a reason for doing so (Ring, 1987) lowa-1985-F/GPs: Have you made any recent changes in your practice because of medical Iiability insurance (either its cost or availability)~ Yesstopped doing obstetrics (Iowa Medical Society, 1987) Kentucky- 1986-OB/GYNs & FIGPs: Of respondents who had practiced obstetrics any time during 1978-86, percent who had quit obstetrics and done so at least in part due to liability problems (Bonham, 1987) Louisiana 1988-OB/GYNs: Practice changes resulting from malpractice crisis-stopped obstetrics (Begneaud, 1988) Michigan- 1985-OB/GYNs: Have you changed your method of practice because of medical-legal implications? Yes--avoid care of high risk patients (Block, 1985) Michigan-1986--F/GPs: Of respondents who practiced obstetrics in 1986, percent who had quit or planned to quit and cited malpractice Iiability risk" as a reason (Smith et al., 1989) Minnesota 1984-OB/GYNs: Had quit obstetrics due to Iitigation (Meader, undated)i Rural Nevada-1985-OB/GYNs & F/GPs: Of respondents who had ever practiced obstetrics, percent that quit or had definite plans to quit and cited malpractice problem/cost/fear as a reason (Crow, 1985) Oregon- 1986-OB/GYNs & F/GPs: Of respondents who had practiced obstetrics in past two years, percent restricting their practice in ANY way who cited malpractice exposure too risky as a reason (OR Med. Assn., 1986) Washington- 1985-F/GPs: Quit or limited obstetrics practice PRIMARILY because of malpractice concerns (either increased premiums or fear of Iawsuits) (Rosenblatt and Wright, 1987) Washington- 1988-OB/GYNs, F/GPs, Nurse Midwives: Of respondents who had ever practiced obstetrics, percent who limited or discontinued obstetrics PRIMARILY because of fear of suit (Rosenblatt and Detering, 1988)
Alabama-1985
Georgia-1986
Illinois-1987
Iowa-1985
Kentucky-1986
Louisiana-1988
Michigan-1985
Michigan-1986
Minnesota-1984
Rural Nevada-1985
Oregon-1986
Washington-1985
Washington-l 986
See appendix I for full cltatlons and descr ptlons of surveys reported n this fgure 2 If the actual quest on was available t IS giver n quotatlor) marks Otherwse a bref descrptan of repOr[e[j hctla~(-~r IS pr~v ~jE)(j
3
4
Unless otherw se spec If I ed numbers are adjusted to reflect the percentage of ALL respondents who reported the md ca(ed bphav lnr F GP - family/general practce OB GYN obstetrcs gynecology 5 Maryland 1987 survey rcludecj only F GPs OB GfNs arid nternsts 6 In the La~flerS ~U~ey physlclan~ were asked to report practce changes made over the p~st ten ye~r$ for Jnv r~nsor However tbe [] Uf25tl(,r
7
was asked n the cortext o~ numerous auestlofs regardlrg malprac[ce In the 1985 Georg{a survey respordents were g vcm a ctwce between age hedfh ma@~c/mr ,ard of)f-[ ( I( [- 1] 1, re.~:(orls
SOURCE Off Ice of Technology Assessment 1994
SPECIFIC SURVEY QUESTION OR COMMENT: 2

National Surveys--All Specialties AMA, Socioeconomic Monitoring Survey-3rd quarter 1983: Percent of physicians reporting that they prescribed more tests and procedures in response to increasing professional liability risk (Reynolds et al , 1987)
AMA, Socioeconomic Monitoring Survey-4th quarter 1984: Percent of physiclans reporting that they had prescribed more tests or treatment procedures during the past 12 months in response to the growth in malpractice claims (Reynolds et al., 1987)
SURVEY POP./YEAR
PERCENT OF RESPONDENTS REPORTING 3 THE INDICATED BEHAVIOR:
National Survey s--Obstetrics/Gynecology ACOG-1983, 1985: As a result of your professional liability claim experience(s), has your practice changed the frequency with which any of the following activities are performed? Yes-Increased testing and diagnostic procedures (Porter, Novell\ & Assoc., 1983, Needham, Porter, Novelli, 1985) National Surveys--Surgery
ACS-1984:lncreased diagnostic testing as a result of the national rise in the number of malpractice suits (Bligh, 1984)
State-Level Surveys-All Specialties Chicago-1985: Due to malpractice Iitigation or its threat, ordered more diagnostic tests that that clinical iudgment deemed unnecessary (Charles et al., 1985)
Kansas-1984:Do you believe problems associated with medical malpractice have affected your practice? Yesprescribe additional diagnostic tests (Kansas Medical Society, 1985)
Maryland-1987; In the last two years, have you made any changes in your practice as a result of the current malpractice climate? Yesincreased the use of tests or monitoring procedures (Weisman et al., 1989)4
New York- 1989: Order more tests and procedures today than dld ten years ago (Lawthers et al., 1992) 5 Texas-1985:Because of the threat of malpractice suits, do you feel compelled to order more lab tests? Yes (Texas Medical Association, 1985) Texas-1986: Because of the threat of malpractice suits, do you order more lab tests? (Percent indicationg they sometimes or always order more tests) (Opinion Analysts Inc., 1986) Texas-1988:How much, if any, have you Increased [diagnostic testing] in your practice because of the threat of liability suits/clalms? (Percent indicating moderate or significant increase) (Texas Mad Assn , 1988)
See appendix I for full cltatons and descnpttons of surveys reported m thts figure 2 If the actual question was available It IS given In quotation marks Otherwse, a tmef description of reported behawor IS provided 3 Unless otherwise lndlcated nurmbers have been ad]usted to reflect percentage of ALL respondents who reported the lndlcated behavior
1
The Maryland 1987 survey Included only obstetrics gynecology, family/genera pract!tloners and internists ~ In the Lawthers survey physicians were asked to report practice changes mdde over the past ten years for ANY reason However the question was asked n the context of numerous questons regarding malpractice SOURCE Off Ice of Technology Assessment 1994
Many of the reported surveys had poor response rates. In 18 of the 32 studies, 50 percent or less of the surveyed physicians responded; in another study, the response rate was not reported (see appendix I). Low response rates raise concern about possible response biasi.e., physicians with greater concern about malpractice liability might be more likely to respond and would indicate greater levels of defensive medicine than truly exist in the study population. For example, in one study for which the response rate was 40.5 percent, respondents were more likely to have been sued (51 percent) than nonrespondents (36 percent) ( 1 23).
Results of physician surveys occasionally have been used to develop quantitative estimates of the national cost impact of defensive medicine or of the malpractice system as a wholes The most widely quoted estimate of the net national cost of the medical malpractice system was published in 1987 by Reynolds and his colleagues at the American Medical Association (AMA) ( 194). More recently, researchers at Lewin-VHI, Inc., published a range of estimates for the aggregate cost of defensive medicine based largely on the Reynolds study ( 125). Once created, estimates such as these tend to be quoted and requoted-and sometimes misquoted in the press and political debates. Consequently, OTA assessed whether the methods these researchers used provide the basis for a reliable measure of the extent of defensive medicine. The estimates are reviewed briefly here and are critiqued in greater detail in appendix J of this report.
Reynolds Estimate of the Net Costs of the Malpractice System

Reynolds and his colleagues ( 194) at the AMA sought to measure the total cost of professional liability for the health care system, not just the cost
of defensive medicine. They estimated the net impact of the medical malpractice system on the 1984 cost of physicians services. These costs included the direct costs to physicians of malpractice insurance premiums and defending against claims, and the indirect costs of practice changes made in response to increasing malpractice liability risk. Practice changes included, but were not limited to, increases in defensive medicine as defined by OTA. The authors used two separate methods of estimation: one based primarily on a survey of physicians reported behavior changes in response to malpractice risks; the other based on the statistical relationship between physicians 1984 malpractice premiums and the prices and volumes of services they reported rendering in 1984. The resulting estimates were $13.7 billion and $12.1 billion, respective] y. Although the authors acknowledged that both of our methods rely on several assumptions and are necessarily less than perfectly precise, they concluded that the similarity of the estimates increases confidence that they provide a reasonable sense of the general order of magnitude of medical [malpractice liability] costs (1 94). OTA reviewed each method for its validity as a measure of the total cost of t he malpractice system and for its ability to provide an estimate of the portion of these costs accounted for by defensive medicine. OTA concluded that the agreement between the two estimates does not increase confidence that they are reasonably accurate. The true costs of defensive medicine may be either higher or lower-and possibly substantially so-than the costs estimated by Reynolds. The first of the two methods has several sources of inaccuracy, resting as it does on the results of a direct physician survey, and therefore provides very little useful information about either the true costs of malpractice 1iabilit y or the costs of defensive medicine. (See appendix J for details. )
The second estimate is based on well-known statistical methods, but the results may be sensitive to the way the statistical model was specified and the data available to estimate it. Without reliable corroborating evidence from the first method or from other estimates, it is impossible to know how much error the statistical method may include. Finally, even if it does give a reasonable estimate of the total costs of malpractice, the statistical method does not permit one to conclude anything about the cost of defensive medicine. The results are consistent with either very high or very low frequency of defensive medicine. (See appendix J for details.)
Lewin-VHl Estimate of Defensive Medicine Costs

Lewin-VHI, Inc. (1 25) took the Reynolds estimates as a starting point for its analysis of the national cost of defensive medicine. First, it averaged together the $12.1 billion and $13.7 billion estimates and updated them to 1991 constant dollars, which yielded a total cost of $18.8 billion in physician services in 1991. It added to the $18.8 billion in physician costs an additional $6.1 billion for hospital costs (using a method described in appendix J) to arrive at a preliminary total cost of $24.9 billion in 1991. Then, because Lewin-VHI researchers believed the Reynolds number overestimated the cost of defensive medicine, 6 they reduced the $24.9 billion figure by three percentages (80, 60, and 40) to arrive at low ($5 billion), medium ($1 O billion), and high ($ 14.9 billion) final estimates of the net costs of defensive medicine to the health care system in 1991. In one respect, Lewin-VHI defined defensive medicine very restrictively compared with OTAs definition, including only those practice changes motivated solely by liability concerns. (Recall that OTAs definition allows other motivations as long as the avoidance of a malpractice suit is the
6 The adjustments were made because Lew in-V HI researchers wanted to :xclude thal pmitm of dcfensi~ e medicine not caused solely by liability ctmcems.
The reasons most frequently cited by respondents included (in decreasing order of importance): establishing a baseline, assessing prognosis, reassuring patients. and helping with treatment decisions. Minimizing risk of a malpractice suit was a relatively minor influence on test-ordering behavior (65 ).7 Evaluation and management of hypertension is not a particularly highrisk area of practice and is not associated with high litigation rates: hence, the influence of malpractice liability concerns in these clinical situations might be expected to be low (73). In a study of common diagnostic laboratory tests in a California medical training center, medical staff and residents were asked to indicate which of a 1ist of reasons for testing had influenced their decisions (256). The most common] y cited reasons were diagnosis (37 percent of all cases), monitoring (33 percent), screening (32 percent), and previous abnormal test result (12 percent). Very few physicians cited educational purposes (2 percent) or medicolegal concerns ( 1 percent) as a contributing factor (256). In another study, residents (N= 13) and faculty (N=53) in internal medicine at a university hospital and a random sample of community physicians (N=93) in the same area were asked about their perceptions of the major reasons for overutilization of diagnostic tests among their peers (258). Residents and faculty internists were asked about factors they thought influenced residents overuse of diagnostic tests. Community physicians were asked about factors causing overuse of testing by physicians in practices similar to their own. Residents cited the following as the top five of 19 reasons for test overuse: inexperience; pressure from peers or superiors: habit; confirming initial abnormal results; and correction of lab processing mistakes. delays, or duplications. Faculty internists cited the following as the top five of 19 reasons for test overuse by residents: inexperience:
habit: pressure from peers or superiors; reliance on lab results to follow daily progress: and use of laboratory rather than good history and physical exam or clinical judgment. Both residents and faculty internists ranked malpractice concerns last out of 19 factors influencing test overuse. Community physicians cited routine screening, habit, malpractice concerns, compulsion to document or explain all abnormalities, and pressure from peers or superiors as the top 5 of 19 reasons for test overuse among their peers (258).
Only one previously published study used c1inical scenarios to assess malpractice-related issues (58). OTA expanded on this approach and conducted four clinical scenario surveys in cooperation with national physician professional organizations. Finally, OTA commissioned an additional c1inical scenario survey of physicians in New Jersey. The results of all these surveys are reviewed below. The Duke Law Journal Study In a 1970 study by the Duke Law Journal (58), 827 randomly selected physicians in 10 specialties in California and North Carolina were sent specialty-specific questionnaires asking about the use of particular procedures in brief clinical scenarios. The scenarios were selected from a 1ist of practices that a group of Duke University Medical Center physicians described as meeting the following criteria: 1 ) they are frequently followed. 2) they are prompted at least in part by concern about possible malpractice litigation. and 3) they are not of sufficient medical benefit to justify the added costs and risks. Recipients were asked to indicate: 1. how often they would follow the practice (with five responses ranging from never to alWays);
2. whether the practice was of medical benefit to the patient (with five response categories ranging from useless to useful and certainly worth the cost); and 3. why they would have followed the practice described (with eight response categories, including to add to a record which might be helpful in defense of a malpractice suitsee table 3-1 ). Significantly, the survey cover letter disclosed the malpractice liability-oriented purpose of the survey, because an earlier survey not stating this purpose had a very low response rate. In three out of 17 clinical actions described in the Duke questionnaire,8 over 20 percent of respondents cited to add to a record which might be helpful in defense of a malpractice suit as the most important reason for following the specified practice (see table 3-1 ). Yet, among the procedures for which malpractice liability concerns were cited most frequently as an important motivating factor, few respondents indicated they would follow the practice. Furthermore, in all but one of the 17 scenarios, the percentages of respondents citing medical reasons (namely, either rule out undetected disease or facilitate further treatment) as the most important reason for following a practice were much larger than the percentages citing malpractice concern as most important. The estimates of defensive medicine from the Duke study are questionable for a number of reasons, and it is impossible to say whether they are too high or too low. First, because respondents were aware of the purpose of the survey and were prompted by both the cover letter and the questionnaire to think about malpractice issues, they may have exaggerated their defensive responses. Second, the wording of the question regarding reasons for choosing may have led some respon-
dents to answer it as a hypothetical question. Some physicians who indicated they would not follow the practice may have nonetheless offered reasons for doing so, thereby inflating the apparent level of defensive response. Third, other reasons listed on the Duke questionnaire (e.g., q patients peace of mind, complete chart) might indirectly reflect some degree of malpractice liability concern, and their presence in the list of reasons may have led to an underestimation of defensive response. Fourth, among physicians who cited defense of a malpractice suit as their chief reason for following the practice, many indicated they would follow the practice only some of the time. Thus, a simple frequency of citing defense of a malpractice suit as the most important reason does not translate directly into a rate of defensive practice. Finally, both clinical practice and the medic: olegal environment have changed dramatically since the Duke Study was conducted, possibly rendering the study results obsolete. OTA Clinical Scenario Surveys Goals and data collection The leadership of three medical professional societies agreed to collaborate with OTA in the conduct of clinical scenario surveys of each societys members by mail during 1993.9 The three associations were the American College of Cardiology (ACC), the American College of Obstetricians and Gynecologists (ACOG), and the American College of Surgeons (ACS). Practicing physicians were selected through stratified random sampling of each associations membership roster. ACS agreed to conduct two separate surveys: one for general surgeons; the other for neurosurgeons.
x OTA elimina[cd frtm~ its review four scenarios ((me each fr(m] derrnatoh~gy, [~bstetrlcs/gyncct)lt)g}, psychiatry, and plastic surgery) that did not meet OTAS definiti(m of defensive medwine. F(lr example, ~mc scenario read. A female nurse is present ciunng all gynecological examinations of the patient. 9
Jeremy Sugar-man, M. D., and Russell L(~ali\~, M. S., J .D., served as primary cxmsultants t{) (ITA tm the design of the survey instruments and
the survey analysls plans, respectively.
Specialty/ Hypothetical clinical situation Dermatology 1 Even though removed nevi appear clinically benign dermatologist orders a hlstopathological examination Internal medicine 1 Upon entering the hospital with a preliminary diagnosis of carcinoma of the lung the patient undergoes certain routine tests One of these iS admissions hemistries or the full battery of serum electrolytes 2 The patient IS admitted to the hospital with nonspecific abdominal complaints On the day of admission he undergoes electrocardiography 3 Same situation as in 2 above Patient undergoes an upper gastrointestinal (Gl) series 4 Same situation as in 3 above Patient undergoes a lower GI series 5 Same situation as in 4 above Patient undergoes proctoscopy Neurology 1 A student appears at campus health office with the complaint of headache for duration of three days Physician orders skull x-rays 2 In a work-up for probably Intra-cranial tumor, the patient has undergone skull x-rays cerebral arteriography, echoencephalography, and ventrlculography The neurologist orders an electroencephalogram
Percent of respondents listing defend against a possible malpractice suit as most important reason for following practice a,c 31%
Number in sample (N) 106
76
74
0 0 0
73 73 73
5 2
56 56
Obstetrics-gynecology 1 The gynecologist performs a dilatation and curettage on a 20-year-old miscarriage patient who IS otherwise healthy Orthopedics 1 After taking history and performing a physical examination the orthopedic specialist determines that the patient a 20-year-old male in otherwise good health has bruised three ribs laterally He orders x-rays to confirm his diagnosis 2 A fracture of the tibia IS reduced and cast applied The orthopedic specialist requests that the patint return the following day for a reexamination of circulation and sensation in the leg Otolaryngology 1 When the patient complains of dizziness present several months following trauma the otolaryngologist initially orders x-rays of the mastoids 2 In evaluating all forms of dizziess, the specialist initially performs audiograms Pediatrics 1 After making a preliminary diagnosis of hyperkinetic child, the pediatrician requests psychiatric consultation
112
18
107
108
11
5
71
73
99
Specialty/ Hypothetical clinical situation
Percent of respondents listing defend against a possible malpractice suit as most important reason for following practice a,c
Number in sample (N)
109 109109
SOURCE U.S. Congress Office of Technology Assessment 1994 based on data presented in Duke Law Journal A Study of Defensive Medicine Duke Law Journal 1971 939-993, 1971
The Medical Malpractice Threat
Introductory letters from both the society president and OTAs director described t he surveys as a study of c1inical decisionmaking, without mentioning malpractice or defensive medicine.
The high degree of cooperation provided by these physician associations resulted in response rates that were reasonably high for surveys of busy professionals, ranging from 56.6 to 62.3 percent. Nonetheless, these response rates leave open the possibility of response bias, Details of the survey methods are presented in appendix D and selected detailed results are presented in appendix E. The clinical scenarios were developed by expert panels selected by each of the three physician associations. Panel members were asked to identify as many clinical scenarios as they could in a two-hour brainstorming session. They were instructed to identify scenarios in which defensive medicine was likely to play a major role. These
candidate scenarios were then assessed, and two or three scenarios were selected for use in the final survey. Panel members were then asked to create a -control version of each selected scenario by adding or deleting one or more key clinical indicators (e.g., a positive result from a laboratory or radiologic test) that would substantially reduce the likelihood that malpractice concerns would be cited as the primary reason for choosing a test or procedure. OTA staff and consultants revised the final questionnaires and, with input from association staff and panel members, selected one scenario in each survey that would have both a case and control version. Box 3-1 shows (he full text of all clinical scenarios used in the surveys. Figure 3-4 reproduces the questionnaire for a sample scenario. Questionnaire format differed Slightly across the four surveys.10
ACC-1: Chest Pain Case Patient history: A 42-year-old man arrives at the emergency room complaining of chest pain The pain
IS
on the left side and
IS
worse when he changes position While it iS sore to the touch, he states
that it feels deep. The pain has persisted for one hour He has not experienced chest pain previously He jogs three times a week and does not smoke He had a normal routine physical examination a week ago Physical examination: The patient lung iS normal Additional data: A 12-lead ECG [electrocardiogram] and CXR [chest x-ray] are normal Laboratory tests including a cbc [complete blood count], electrolytes and cardiac enzymes are normal
IS
tense and anxious HiS BP [blood pressure]
IS
140/80 heart
rate 80. The anterior chest wall iS tender over the left sternal border Examination of the heart and
ACC-2: Chest Pain Control

Patient history: A 52-year-old man presents to the emergency room with retrosternal chest pressure There iS no chest soreness The pain has been recurrent for the past three weeks, it comes on with physical activity and subsides with rest He smokes two packs of cigarettes a day He had a normal routine physical examination one week ago Physical examination: The patient
IS
tense and sweating BP
IS
160/1 00, heart rate iS 95 There iS

IS
no soreness on palpitation of the chest wall Examination of the heart and lungs cluding a complete blood count, electrolytes and cardiac enzymes are normal
normal
Additional data: A 12-lead ECG shows T-wave flattening in the lateral leads Laboratory tests in-
ACC-3: Syncope (Fainting) Case:

Patient history: A 50-year-old woman collapsed in a crowded, warm church in the summer Her husband states that she was unconscious for about two minutes and recovered quickly There was no seizure activity reported and no attempt was made to see if she had a pulse or respiration at the time of the event She has never had a similar episode The patient was taken to the emergency room by ambulance for evaluation The emergency room physician refers the patient to you for care Physical examination: The patient appears well She Her exam radiation Additional data: Monitoring in the emergency room reveals isolated PVCs [premature ventlcular contractions] Complete blood count, electrolytes panel, routine blood chemistries, chest x-rays and 12-lead ECG are normal ACS-1: Breast Pain Case History of present illness: A 38 year-old woman G2P2 [gravlda 2, para 2] iS referred to you from her gynecologist for evaluation of left breast pain for one month She had her first child at age 29, and her second at age 31 She has been taking oral contraceptives subsequently Her gynecologist remarked that she has fibrocystic breast disease on annual routine examination. She has a family history of breast cancer A baseline mammogram done at age 35 showed no evidence of cancer She anticipates that her next menstrual period will begin in five days Physical examination: Slight thickening in the upper outer quadrant of her left breast with some tenderness There are no nipple changes There iS no axillary adenopathy Clinical course: Following the exam you order a mammogram A radiologists report states There is dense, dysplastic breast tissue bilaterally Vague shadows bilaterally are consistent with possible (continued)
IS IS
on no medication and was previously
healthy Her BP is 150/80 sitting and 130/70 standing Her heart rate iS 74 sitting and 85 standing remarkable only for a 11/Vl systolic murmur best heard at the left sternal border without
cysts No dominant masses or abnormal microcalcifications are present These breasts are very dense and difficult to evaluate Clinical correlation iS Indicated ACS-2: Rectal Bleeding Case History of present illness: A 35-year-oId man comes to your office complaining of bright red blood per rectum Over the past four days he has observed a few drops of blood in the toilet and on the toilet paper after having a bowel movement He denies any recent change in bowel habits and has otherwise been in good health Physical examination: Rectal examination reveals one small, external hemorrhoid which thrombosed. Otherwise the exam
IS IS
not
within normal limits
Clinical course: Anoscopy reveals non-bleeding Internal hemorrhoids A hemoglobin, hematocrit, CEA [carcinoembryonlc antigen], and flexible sigmoidoscopy are all within normal Iimits ACS-3: Rectal Bleeding Control History of present illness A 35-year-old man comes to your office complaining of bright red blood per rectum Over the past four days he has observed a few drops of blood in the toilet and on the toilet paper after having a bowel movement. He den es any recent change in bowel habits and has otherwise been in good health Physical examination: Rectal examination is normal Clinical course: Anoscopy reveals non-bleeding internal hemorrhoids A hemoccult i S positive A hemoglobin, hematocrit, CEA, and flexible slgmoidoscopy are all within normal Iimits ACS-4: Neurosurgeons Head Trauma Case History of present illness: A fifteen-year-old boy fell from his skateboard after riding over a crack in the sidewalk. He hit his head, got up and skated home Thirty minutes after the fall he told his mother about the Incident and she brings him to the ER. In the ER, the patient admits to Iight-headedness and some tenderness at the site of impact. Physical examination There
IS
an area of tenderness and swelling at left parietal area Mental status
and neurological exam are normal. ACS-5: Neurosurgeons Back Pain Case History of present illness: A 52-year-old man iS seen by you in your office, He complains of back pain and numbness of his right great toe for the past week He attributes the injury to driving over a pothole in his pick-up truck He has been able to continue to work since the Injury. Physical examination: The patient has decreased range of motion of his back There iS lumbosacral spasm Straight leg raising produces right leg discomfort at 70 degrees Ankle jerks are slightly diminished bilaterally, however, there are no other motor or sensory deficits revealed on exam There are no bowel or bladder complaints The rest of the physical examination iS normal. ACS-6: Neurosurgeons Back Pain Control History of present illness: A 52-year-old man
IS
seen by you in your office, He complains of back
pain and numbness of his right great toe for the past week He attributes the injury to driving over a pothole in his pick-up truck He has been able to continue to work since the injury Physical examination: The patient has decreased range of motion of his back There iS lumbosacral spasm He has decreased sensitivity along medial aspect of right lower leg Straight leg raising produces right leg discomfort at 70 degrees. Ankle jerks are slightly diminished bilaterally, however. there are no other motor or sensory deficits revealed on exam There are no bowel or bladder complaints The rest of the physical examination is normal (continued)
ACOG-1: Breast Lump Case History: A 31 -year-old nulliparous woman comes to your office complaining of a breast lump. Her last visit was 1 year ago At that time she had no complaints and her physical examination was normal Her last menstrual period was 3 weeks ago She family history of breast carcinoma Physical examination: There iS a 1 cm mass in the upper outer quadrant of her right breast that iS tender to palpation The nipple
IS IS
currently on oral contraceptives and has a
normal without retraction and there iS no discharge There iS n o
skin dimpling or axillary adenopathy The left breast and the remainder of the exam are normal ACOG-2: Complicated Delivery Case History: A 36-year-old primigravida presents at 39 weeks gestation after an uncomplicated pregnancy Clinical course: The patient has had 12 hours of labor, and
IS
now 3 hours into the second stage
She has been receiving oxytocin augmentation for secondary arrest of dilatation since 7 cm She iS completely dilated and effaced at +2 station, ROP [right occiput posterior position] There has been no change in the exam for over an hour Moderate variable decelerations have been present for the last 30 minutes with good beat-to-beat variability Estimated fetal weight is 75 lb and clinical pelvimetry
IS
adequate The patient
IS
fatigued and can no longer push
ACOG-3: Perimenopausal Bleeding Case History: A 51 -year-old sexually active nulliparous woman reports that her last menstrual period lasted 2 weeks It was heavier than her usual periods and there were some clots Her previous menstrual period occurred approximately 3 months ago For the prior 2 years her periods had occurred every 2 to 3 months She iS on no medications, and has not used any contraception in more than 10 years Physical examination: Vital signs are normal She otherwise normal The pelvic exam weight ACOG-4: Perimenopausal Bleeding Control History: A 51 -year-old sexually active nulliparous woman reports that her last menstrual period lasted 2 weeks It was heavier that her usual periods and there were some clots Her previous menstrual period occurred over 1 year ago For the prior 2 years her periods had occurred every 2 to 3 months She iS on no medications, and has not used any contraception in more than 10 years Physical examination: Vital signs are normal She weight
KEY ACC - Amer can College of Cardlologsts ACS - American College of Surgeons ACOG - American College of Obstetric ans ar?d Gynecologists SOURCE Off Ice of Technology Assessment 1994
IS IS IS
markedly obese The general physical exam iS
normal, but it is difficult to outline the uterus due to the patients
markedly obese The general physical exam is
otherwise normal The pelvic exam iS normal, but it iS difficult to outline the uterus due to the patients
Each survey also included an attitude questionnaire comprising three attitude scales: malpractice concern, cost consciousness, and discomfort with clinical uncertainty. 11 Finally, the surveys asked for data on selected demographic and professional characteristics of the respondents (e.g., practice setting).
Results: extent of defensive medicine OTA constructed six measures of defensive medicine based on specific patterns of reasons given for choosing selected clinical options. These six response patterns involved particular combinations of checkmarks for -malpractice concerns and other reasons (see figure 3-4). This section reports the results for the measure that most closely fit OTAs definition of positive defensive medicine: ordering additional procedures primarily, but not necessarily solely, out of fear of malpractice Iiabili y risk. The measure corresponding to this definition required the respondent to double-check malpractice concerns, but allowed single checks for any other reasons. Appendix E contains results for all six measures of defensive medicine, which span a range from nonrestrictive (requiring only a single check for malpractice concerns with single or double checks allowed for any other reasons) to highly restrictive (requiring that q malpractice concerns be the only reason checked). Table 3-2 shows the extent of defensive medicine in the case scenarios (i.e., those scenarios designed to elicit high levels of defensive medicine). The proportion of respondents citing malpractice concerns as the most important reason for choosing to perform at least one clinical action in a scenario ranged from 4.9 percent (ACS back pain scenario) to 29.0 percent (ACS head trauma scenario). The relatively high percentage in the ACS head trauma scenario is noteworthy, espe-
cially in contrast with the relatively low percentage for the back pain scenario within the same survey. Overall, these figures suggest that, if physicians actually practice as they say they would in these surveys, positive defensive medicine does exist-although not to the extent suggested by anecdotal evidence or direct physician surveys. They also suggest that defensive medicine varies considerably across clinical situations. Across the scenarios, malpractice concerns was cited considerablyess frequently than q *medical indications as the most important reason for choosing procedures. 12 Moreover, the majority of respondents who ever cited malpractice concerns as the most important reason for choosing a procedure did so for only one procedure, and very few did so for several procedures in the same scenario (data not shown). Table 3-3 further demonstrates how the citing of malpractice concerns varied across the specific clinical options given in the scenarios. Across all 54 of the *interventionist clinical actions (i.e., actions other than waiting or doing nothing), of those who would choose the action, the percentage who would do so primarily because of malpractice concerns ranged from O to 53, with a median of 8 percent. Because these scenarios were specifically designed to increase the likelihood of defensive response by physicians, they are not generallyrepresentative of all diagnostic procedures. Thus, one would expect the percentage of all diagnostic 13 procedures done consciously for defensive reasons to be less than 8 percent. Because not all physicians chose a given procedure, a smaller percentage of the clinical encounters described in the scenarios involved the performmance of a defensive medical procedure. For example, although 30 percent of surgeons who
ustxt
I I ltcT1l~ ,n t}lc ~lttltllde ScaIcs were ~dop[cd fr~)n) Previ{)uslj (77),
scales dcwclt)pxl by G(x)ld and colleagues at the University of Michigan
History:
A 31-year-old nulliparous woman comes to your office complaining of a breast lump. Her last visit was 1 year ago. At that time she had no complaints and her physical examination was normal. Her last menstrual period was 3 weeks ago. She is currently on oral contraceptives and has a family history of breast carcinoma. Physical Exam: There is a 1 cm mass in the upper outer quadrant of her right breast that is tender to palpation. The nipple is normal without retraction and there is no discharge. There is no skin dimpling or axillary adenopathy. The left breast and the remainder of the exam are normal.
Reasons for Decision Check ALL the reason (s) for your decision (check
Would you choose the following option? (Circle Yes or No)
all that apply).
If you answered NO to Question 1, go to Question 2. Otherwise go to next page. QUESTION 2. If you answered No to
Question 1 above, which actions(s) would you recommend now? Circle Yes or No for EACH Decision.
I
I I
Comments:
.
I
SOURCE Off Ice of Technology Assessment 1994
would order a computed tomography (CT) scan in the ACS back pain case would do so for defensive reasons. only 3 percent of all respondents indicated they would order the CT scan. Thus, malpractice concerns led to CT scans in only 1 percent of all responses. What do these results imply about medical practice? They support the large body of evidence (hat there is a great deal of variation in how physicians practice medicine. Furthermore, in these scenarios, beliefs about the medical appropriateness of procedures were far more influential in physicians practice choices than were concerns about malpractice liability. Case vs. control versions of scenarios In each survey, a case version of one scenario was given to a random subgroup of respondents, and a control version of that same scenario was given to the remaining respondents. The two ver-
sions were identical, except that the control version contained one or more additional clinical features designed to increase the clinical appropriateness of an intervention and hence reduce the relative importance of malpractice concerns. Higher rates of intervention were thus expected in the control scenarios, and the frequency of defensive medicine was expected to be lower. (See box 3-1 for text of case and control versions of scenarios.) OTA did find, generally, higher rates of use of tests and procedures in the control scenarios. Table 3-4 compares the percentage of physicians choosing each procedure in the case and control scenarios. Rates of use appeared to be higher in the control scenario, especially for more invasive procedures. For example, in the ACOG perimenopausal bleeding scenario, the percentage of respondents indicating they would perform an endometrial biopsy was virtually identical in the case and control versions. But much higher
Percentage of
all physicians who

chose the clinical action Scenario/ clinical action General Surgeons Breast pain (N=1 ,412) Needle biopsy Open biopsy Other Rectal bleeding (N=738) Air contrast barium enema Colonscopy Other Neurosurgeons Head trauma (N=503) Skull x-ray C-spine x-ray CT of head Other Back pain (N=252) C Lumbosacral x-ray CT MRI Other
C
Percent
95/0 confidence limits
American College of Surgeons
13,3Y0
84 145 19,2 26,2 9.7
(11 5,15 1) (7 0,9,8) (12 5,16 5) (16 2,22 2) (22.8,29.6) (7.5,1 1 .9)
2.7%
2 1 1,0
(1 .9,35)
20370
24.5 6 6
(1.3,2.9)
(O 4,1 .6) (1 3,3.3) (3 4,6.6) (0.0,07)
(14 1,26 5) (16 5,32 5) (2 8,104) (6.2,1 7.4) (1 3.0,25.0) (o 3,97)
23 5.0 0.3
11.8 19.0 2.8
337 21,1 48.8 3.9 24.4 3.4 12.6 9.4
(29 9.37.5) (17.7,24.5) (44.8,52.8) (2.3,5.5)

(19.0,29.8) (1 .2,5.6) (8.4,16.8) (5.6,13.2)
100 11.2 21.8 0.4
(74,126) (8 6,13.8) (18,4,25,2) (0.0,1 .4)
29.6j
52,9 44.7 9.3 139 298 16.0 00 (4,9,22.9) (5.5.68.0) (5.8,33.3) (0.0,14.4)
3.4
10 20 00
American College of Obstetricians and Gynecologists

Breast lump (N= 1,230) Breast sonography Mammography Needle aspiration Fine needle biopsy Open biopsy Refer to surgeon Other (1 .5,3 1) (4,2,7.0) (0.5,1 .7) (0.1 ,09) (0.0,03) (4,9,7,7) (0.0,0.3) (6.3,13.1) (9.5,15.1) (2.1 ,6.9) (2.3,14.0) (0.0,26,0) (17.0,25.8) (0.0,141) (continued)
Percentage of
all physicians who

chose the clinical action Scenario/ clinical action Complicated delivery (N= Continue pushing now Rest for 30 minutes Percent 950/o confidence limits (7 2,104) (65 97)
Percent of all respondents who chose the clinical action primarily for malpractice concerns Percent 02 02 95% confidence limits (o 00 4) (o 0,04)
Of clinical actions chosen, the percent done primarily for malpractice concerns Percent 95/0 confidence limits b .
(O 2,66)
1,230)
88 81
19 21
(o 3,72)
Perimenopausal bleeding (N=634)C 13 18 (o32 3) Hematocrlt/hemoglobin 734 (69 8,77 O) (O 8,35) 11. 1 (7 5,147) Pregnancy test (45 5,53 5) 495 55 (3 7,73) 16 19 (O62 6) Endometrial sampling (82 6,88 2) (o 9,35) 854 Pelvic ultrasound 42 (2 6,58) 76 (46, 106) 543 (50 358 3) (1 2,109) Hysteroscopy 143 (11 5,17 1) (o 01 2) 44 06 D&C 42 (2 6,58) 05 (o 01 1) 109 (2 2,289) Hysterectomy (O 0,06) (O 0,06) 02 00 00 (o 0,94.4) Other 45 (2 9,6 1) 00 (O 0,06) 00 (o 0,121 ) . . KEY C-spine = cerwcal spne CT = computed tomography D & C = dllatlon and curettage 2 DiM Mode = two dimensional and !Ime-motion mode EEG = electroencephalogram, ECG = electrocardiogram, MRI = magnetic resonance Image NSAID = nonsteroldal anti-mflammcitory drug
a Results are weighted to reflect the total population of professional society members on which the survey sample was based See appendix D for details b The confidence Intervals for the percentage of cllnlcal actions tend to be wide due to the small numbers of respondents who chose each procedure c Numbers reflect responses to case versions of the scenario only See text of chapter 3 for further explanation c~ Admit was not Ilsted In the questionnaire as an Isolated option This composite category reflects respondents who chose at least one Of the three admit oPtlOn S and dld S0 prlmarlly for malprachce reasons SOURCE Off Ice of Technology Assessment 1994 Data analyzed m collaborahon with Dr Russell Locallo of Pennsylvania State Unwerslty
Scenario/ clinical action
Percentage of physicians who indicated they would take the action Case Control
Difference [[case] - [control])
95 /0 confidence limits
American College of Cardiology

Chest pain Discharge home with NSAID Admit to hospital b Admit and observe Admit and obtain cardiac enzymes Admit and obtain ECG Stress tests Exercise ECG Stress thallium Echocardiograms 2 D/M mode Doppler Color flow doppler Transesophageal echo Angiogram (N= 162) 67 8% 271 88 215 224 502 85 188 78 84 06 06 685 400 272
408 129 123 0 6 587
(58 4, 73.6) (-77 8,-63 O)

(-85 6, -724) (-79 2, -644)
-461 * 102 -18 7*
(-55 6, -366) (-O 5, 20.9) (-26 6, -10.8)
American College of Surgeons General Surgeons

Rectal bleeding Air contrast barium enema Colonoscopy Other (N=738) 19270 262 97 -7 3* -11 1 * 3 6* (-1 1.8,-2 8) (-16 0,-6 2) (O 7,65)
(-2 1,8 1) (7 5,187) (-4 1,3 9) (-2 0,9 4) (-12.9,-4 1) (-lo 4,-4 2) (-1 0,0 4) (-o 7,3 7)
a
Results are weighted to reflect the total population of professional society members on which the survey sample was based See appendix D for details Admit was not listed m the questionnaire as an Isolated option This composite category reflects respondents who chose at least one of the three admll ophons and dld so prlmarlly for malpractice reasons q Statically slgmflcant at the p <05 level
KEY CT - computed tomography, D & C - dllatlon and curettage, 2 DIM Moje - two dimensional and Ilme-motion mode, ECG - electrocar,~lograrf, MRI - magnehc resonance image SOURCE Off Ice of Technology Assessment 1994 Data analyzed m collaborahon with Dr Russell Locallo of Pennsylvania State Unwerslty
proportions of respondents in the control scenarios said the y would perform hysteroscopy or D&C (dilatation and curettage), both of which are more invasive procedures. For the vast majority of procedures, OTA found no significant differences between case and control scenarios in the percentage of respondents who chose the procedure mainly for defensive reasons. However, the majority of procedures in the case scenarios were chosen by relatively few respondents. Therefore. the sample sizes on which to base comparisons of the frequency of defensive response were very low. The surveys were simply too small to detect such differences with adequate statistical confidence if they did exist. (Detailed results of case and control comparisons are available in a tcchnical appendix upon request to OTA. )
Open-ended vs. structured questionnaires To assess how the structure of the questionnaire
might affect responses, a supplemental sample of
concern about malpractice liability might not en-
ter as readily into their hypothetical clinical decisionmaking. Alternatively. even though the open-ended questionnaire invited physicians to cite both clinical and nonclinical reasons for their procedure choices. the respondents may have viewed the format and content of the questionnaire as being similar to a medical board examination, Such an interpretation may have reduced the likelihood of citing such nonclinical factors as malpractice concerns. Indeed, most respondents to the openended questionnaire gave detailed clinical explanations for their choices of procedures. lending support to this interpretation. These results highlight the Iimitations of surveys as a method of measuring the extent of defensive medicine. Questionnaire design can affect responses for reasons that are difficult to identify and specify.
Attitudes toward malpractice
600 general surgeons was given open-ended versions of the same c1inical scenarios used in the regular general surgeon survey. These scenarios listed the same clinical actions as in the regular survey but gave no printed "reasons from which to choose. Insted, a blank space was provided beside each clinical action in which the surgeon could write out his or her own reasons for choosing it. Open-ended responses were coded by OTA study staff into the same categories of "reasons as on the closed-ended questionnaire and were then compared with the closed-ended results. Although the percentage of physicians who chose each action did not differ significantly in the open-ended and closed-ended surveys, a substantially lower proportion of respondents to the openended questionnaire cited malpractice concerns as the primary reason for choosing a given action (see table 3-5). Two alternative explanations for this finding are possible. First, without the prompting effect of the closed-ended questionnaire, physicians
OTA examined differences in attitudes regarding malpractice concern between respondents who cited malpractice concerns as the most important reason for choosing one or more clinical actions in each scenario and those who did not. The separate items in the attitude survey that addresscd the concerns about malpract ice were combined into a composite scale. (For details, see appendix D.) OTA compared attitudes toward malpractice of respondcnts who had double-chccktxl malpractice concerns as a reason for choosing one or more c1inical actions in four selected scenarios with the attitude scores of those who had not double-checked malpractice concerns. 14 In only one scenario (ACS head trauma) did respondents who double-checked malpractice concerns have statistically significantly higher malpractice concern scale scores than those who did not double-check malpractice concerns . In two scenarios (ACS breast pain and ACOG breast
Percentage of all physicians who chose the clinical actionb Scenario/ clinical action Openended Closedended
Of clinical actions chosen, the percent done primarily for malpractice concerns Openended 6370 146 00 Closedended 20 .3% 245 66 Difference c -140 -99 -6.6 Odds ratio (OR) 0 20 0 02 0.0 95% confidence interval for ORb (0.02, O 85) (o 002, 0 07) (o 00, 1 03)
Breast pain
Needle biopsy Open biopsy Other
(N=381)
10 6% 65 126
(N=1412)
1 3.3% 84 145
Rectal bleeding
Barium enema Colonoscopy Other
a h
(N=381 )
143 250 10,2
(N=738)
192 262 9.7 3.7 40 00 11 8 190 2.8 -8.1 -150 -28 025 0 21* 00 (o 03, 1 11) (O 05, 0 60) (0. 00.6. 4)
Results are weighted to reflect the total populahon of professional society members on which the survey sample was based See appendix D for details \vAAth one ~)(~~~ii~fi @arIUiTI enert-Ia), tl-te PI upur IIUI IS of responaems cnoosmg a gwen clinical action were not statistically significantly different between open- and closed-
ended versions of the scenario Confidence intervals were constructed for the odds raho because of the small number of observations m the denominator and numerator of the calculated percentages
* = statistically significant at the p < 05 level
SOURCE Office of Technology Assessment, 1994 Data analyzed In collaboration with Dr Russell Locallo of Pennsylvania State Unlverslty
lump), malpractice attitude scores were statistically significantly lower among double-checkers compared with nondouble-checkers. 15 (Detailed results of the analysis are included in appendix E of this report). Costs of selected defensive medicine procedures Based on the results of the clinical scenario surveys, OTA estimated the potential national costs of positive defensive medicine for two scenarios for which incidence and cost data were readily available: the ACOG complicated delivery scenario and the ACS head trauma scenario. The rationale and methods for deriving these estimates, and their results, are detailed in appendix F. The aggregate incremental cost of q defensive Caesarean delivery in the 46,896 cases nationally in 1991 that were similar to the ACOG scenario16 was $8.7 million. The estimated aggregate cost of defensive diagnostic radiology of the head (skull x-ray, cervical spine x-ray, and CT scan of the head) for the roughly 530,000 minor head injuries estimated to occur annually among children and young adults aged 5 to 24 in the United States (i.e., cases similar to that described in the ACS head trauma scenario) was approximately $45 million. While these estimated costs represent only a small share of total national health care costs, they are not trivial. It is inappropriate to generalize these estimated costs beyond the specific scenarios for which they were derived. Also, the scenarios were designed to be malpractice-sensitive and thus are not representative of clinical practice generally.
Glassman Scenario Survey of New Jersey Physicians

An OTA-sponsored study by Glassman and colleagues (73) conducted a clinical scenario survey in which five of the scenarios developed for OTAs surveys were adapted for use in this study. The contractors surveyed 835 physicians covered by the Medical Insurance Exchange of New Jersey, which insures 70 percent of all New Jersey physicians. For each scenario, physicians reported the clinical actions they would take (e.g., tests, procedures, referral to other physicians). Respondents were asked to estimate on a fivepoint scale (1 = extremely influential, 5 = not at all influential) how strongly their decisions had been influenced by various factors, including the desire to reduce the possibility of malpractice litigation;" the history, physical, and lab results; the standard of patient care in their community; and patient or family expectation s. The physicians were also asked to estimate the probability that the patient had a life-threatening condition and the probability that further testing would identify the cause of the patients symptoms. The survey also queried physicians about their general attitudes regarding malpractice liability, clinical uncertainty, and cost consciousness using a set of attitude scales similar, but not identical, to those used in the OTA clinical scenario surveys. Depending on the scenario, between 2.3 and 6.4 percent of the respondents cited the desire to minimize the possibility of malpractice 1itigation as either an extremely or very influential reason for their clinical decisions and did not cite any
15 me only stat15[lca]]y si~ific~[ difference {m the other two attitude scales was in the ACC sy nctqx sccnarl(~, Where the nlean score for discomfort with clinical uncertainty was statistically significantly /ower armmg rcsp(mdents who d(mblc-chcched malpractice c(mcems compared with those who did not.
16 Womn aged so t. 39 exFnencing pro]onge~
Idx)r or
dysfunctional labor (SCC appendil F f{~r dCt:lils)
Scenario Cardiologists Syncope in 50-year-old woman Diagnostic testing Clinical management Nonspecific chest pain in 42-year-old man Diagnostic testing Clinical management
Percent of physicians who cited desire to minimize possibility of malpractice litigation a as the most influential reason for clinical decision
64-29.7%a 57-266 57-329 43-310
Internists
Syncope in 50-year-old woman Diagnostic testing Clinicall management Nonspecific chest pain in 42-year-old man Diagnostic testing Clinical management 46-305 53-295 57-315 23-275 32-241 59-422 42-289
Surgeons
Breast pain in 38-year-old woman Head trauma in 15-year-old Rectal bleeding in 35-year-old man
NOTE These numbers are based on responses to clinical scenario surveys completed by cardiologists (N- 157) internists (N- 188), and surgeons (N- 187) practicing in New Jersey Overall survey response rates were 49 percent for cardiologists 51 percent for lnternists and 59 percent for surgeons
a
In this survey respondents were not asked to rank their reasons, therefore It IS impossible to infer the primary motivation in cases where a respondent listed two reasons as equalIy Important The percentages are presented as a range The lower bound of the range includes only those respondents who cited malpractice concerns as either extremely lnfluential" or very Influenlal and cited no other reason as that Important The upper bound also includes respondents who cited malpractice concerns as either extremely influential or veryj influential and listed another reason as equally but not more important
SOURCE PA Glassman RAND Santa Monica. CA unpublished data from a study prepared under contract with the Off Ice of Technology Assessment U S Congress Washington, DC, January 1994
other reason as equally or more influential (table 3-6). However, if respondents who cited malpractice concerns as extremely or very influential but also cited mother reason as equally important are included, the defensive response across scenarios could be as high as between 24 and 42 percent (see table 3-6). 17 In contrast, medical indications were cited as the most influential factor (i.e., very or extremely
important, with no other reasons as important) by 42.8 to 60.9 percent of respondents, depending on the scenario (data not shown). The study found no statistically significant relationships between physicians tendencies to cite malpractice liability concerns as a factor in their decisions and either their malpractice attitude scale scores or their past malpractice litigation exposure (73).
17 unll~e the OTA sur~eys, G]assn)an and c()]le~guM
survey d]d
not requ Ire respmdcnts to rank rt?aw)ns. Thus, ftw CaSCS
In
Which rCsp~)n-
dents cltcxi midprac[icc I]abll ity ccmcems and medical indicati(ms as cquall y impr)rtant. II was not p(wsible to inf~r w hich was the primary motiva[ion. If (mc assumes thiit malpractu liabilit> umwms were the primary nNIII\ atl(m In tht~se casc$, h(~wwcr, the Pcrccntagc 01 rcspmden!.s displaying defcmst~c tx>h:ik lor Inwaws h) ktween 24 and 42, dqxmchng (m ttw swnam) (SW table 3-6).
Conclusions
The results of clinical scenario studies suggest that conscious positive defensive medicine does exist, although not to the extent suggested by anecdotal evidence or by some other physician surveys (see figure 3-3). Despite using somewhat different methods and measures, the three clinical scenario studies found roughly comparable levels of defensive medicine: the percentage of respondents who cited malpractice concerns as the primary reason for ordering tests or procedures ranged from zero to over 30. However, all of the studies also found that this percentage was considerably lower than the percentage of respondents who cited c1inical factors as the primary reason for choosing procedures-even though most scenarios were designed to enhance the probability y that the respondent would cite malpractice concerns. Because scenarios were also designed with the implicit assumption that conservative management was acceptable. these findings suggest that many physicians who choose to be more aggressive in diagnosis and treatment do so primarily because they believe it is medically appropriate, and not because they are conscious y concerned about liability. In the OTA clinical scenario surveys, the median defensive response across 54 interventionist clinical actions was only 8 percent. Because the scenarios were designed to be malpracticesensitive, the percentage of clinical actions arising from conscious defensive medicine is certainly lower than this figure. The estimates of defensive medicine from clinical scenario surveys are still limited in that they are based on what physicians say they would do rather than what they actually do. Furthermore, reasons such as compliance with community standards and patient expectations, although not labeled malpractice liability concerns as such, may
indirectly reflect potential liability concerns. To the extent that such reasons were listed alongside malpractice concerns as options in the questionnaires, they may have deflated the apparent influence of malpractice liability in these studies. On the other hand, the structured questionnaires may have prompted physicians to overreport true levels of defensive medicine.
Direct physician surveys and clinical scenario surveys examine the extent to which physicians report that fear of malpractice liability influences their behavior. Whether physicians actual] y do behave the way they say they do in surveys remains an open question, and the potential problems with such surveys argue for analyzing data on actual use of procedures to identify the frequency of defensive medicine. Three past studies have tried to document the existence of defensive medicine through analyses relating physicians actual exposure to malpractice claims to their actual clinical practices. As part of this assessment of defensive medicine. OTA commissioned three additional studies of this type in the areas of both positive and negative defensive medicine. The hypothesis common to such studies is that physicians with greater exposure to malpractice liability (either past personal experience or vicarious exposure through colleagues within a hospital or geographic area) will practice more defensive medicine than physicians with lower malpractice claims exposure. This section discusses the results of five studies of this type. 18 Three looked at posi tive defensive medicine: the other two examined negative defensive medicine in obstetricsnamely, the decision to withdraw from obstetrics
practice due to liability concerns. The studies used varying combinations of actual and self-reported data on malpractice claims exposure and physician practice patterns.
Studies of Positive Defensive Medicine

Caesarean deliveries in New York State, 1984 Localio and colleagues (128,129) examined the relationship bet ween malpractice 1 iabilit y risk and rates of Caesarean delivery in a sample of New York State hospitals in 1984. The study examined eight different measures of malpractice liability risk: malpractice premiums by region; physicians perceived risk of litigation as measured in a survey, by region; three measures of actual physician malpractice claims experience aggregated to the hospital level; and three measures of actual malpractice claims experience of the individual physicians ( 129). When patient severity and other factors known to affect the Caesarean rate were controlled, higher rates were associated with both higher area-level malpractice liability risk (premiums and perceived risk of litigation) and hospital-level malpractice claims risk. The estimated incremental effect of higher area- and hospital-level malpractice liability risk on the Caesarean delivery rate was quite large. For example, a patient in a hospital with a high frequency of physician obstetric malpractice claims was 32 percent more likely to undergo a Caesarean delivery than a patient in a hospital with a low claim frequency. The study did not find a statistically significant association between the physicians individual malpractice claim experience and his or her Caesarean rate (128). Analyses of patients classified at various levels of expected risk of Caesarean delivery (based on
clinical factors alone) showed that malpractice liability risk had the strongest influence in births with moderate clinical risk. For low-risk births (i.e., births in which clinical factors alone predicted a less than 5 percent chance of Caesarean), hospital- and premium-level malpractice liability risk measures were either slightly negatively or not statistically significantly associated with Caesarean delivery. For medium risk births (between 5 and 75 percent chance of Caesarean), they were positively associated with Caesarean delivery. For high-risk births (greater than 75 percent chance of Caesarean), they were also positively associated, but to a lesser degree than for medium-risk births. These findings suggest that malpractice liability risk may play a greater role in situations where clinical factors alone do not clearly point out the appropriate course of action ( 128).
Use of services in low-risk prenatal cases, Washington State, 1989 A study jointly funded by OTA and the Robert
Wood Johnson Foundation and undertaken by Baldwin and colleagues examined the association between physicians malpractice claims experience and their use of technology for low-risk obstetric patients ( 10). A stratified random sample of Washington State physicians was evaluated by linking both personal and area-level malpractice claims exposure data with data on physicians use of services for their low-risk obstetric patients. 19 Utilization measures included: ultrasound early in pregnancy (prior to 20 weeks gestation), ultrasound throughout pregnancy, type of delivery (vaginal or Caesarean), referral and consultation with specialists, and total prenatal care resource use.20
~ ~e study Sa,,lple inclu~e~ 54 urban obstetricians, ?9 - rural (Jbstc[rici .ms, 59 urban famll} physicians, and 67 rural family ph? ~icl:ins.
Patient rcc(mis were selected for up to I I h)wr-risk obstetric ptititmts pcr physician. Patients were ranch)mly sclectcd frtm~ the case rccx)rds of each physician, and those cases prescrmng with selected risk factors in thci] init]al prenatal care visit were excluded from [he anal} SIS.
20 The total prenatal care res(wrcc use ft)r a case was based {m a standardized a~cra:c charge for spccltlc prenatal serv]ccs obta]ncxt Irom Blue Cr{}ss of Washingt(m State.
Independent variables in the study included individual physicians self-reported malpractice histories and the malpractice defendant rate21 in the county in which the physician practices. These rates were obtained from Washingtons largest malpractice insurance carrier. After controlling for both patient and physician practice characteristics, the researchers found no statistically significant differences in prenatal resource use or Caesarean delivery rates between physicians with higher and those with lower malpractice claims exposure (10). Table 3-7 shows the results of the analysis that used the county malpractice defendant rate as the independent variable of interest. There were no statistically significant associations between the county defendant rate and any of the five measures of resource use.
Use of clinical services in New Jersey, 1993
An OTA contract study undertaken by Glassman and his colleagues at RAND (73) used clinical scenarios to test whether New Jersey physicians personal malpractice claims experience was associated with their reported use of resources. The study population comprised 1,540 physicians22 insured by the single largest malpractice insurance company in New Jersey. The insurance company provided data on individual physicians malpractice histories from 1977 through 1992 (both open and closed claims). The great majority of physicians surveyed had at least one claim filed against them, with some specialties as high as 93 percent. Study participants were asked to respond to two or three clinical scenarios (a total of five were used), rate their reasons for choosing among cer-
clinical choices, and answer a questionnaire on attitudes toward clinical uncertainty, malpractice, and cost consciousness.23 In relevant scenarios, physicians were asked to estimate the probability that the patient had severe disease. Physicians were blinded to the purpose of the study and were unaware that scenario results would be 1inked to their personal malpractice claims histories. The researchers found no statistically significant associations between resource use in the five clinical scenarios and the physicians own malpractice claims experience.24 The only study variables consistently correlated with resource use were physicians self-reported attitudes toward cost consciousness (negative correlate, and physicians subjective estimates of the probability of severe disease (positive correlation). Physicians self-reported attitudes toward uncertainty. cost consciousness, and malpractice were not consistently correlated with their persona] malpractice claims histories. The study did not utilize area- or hospital-level measures of malpractice claims risk.
tain
Studies of Negative Defensive Medicine

Decision to withdraw from obstetrics, New York, 1980-89 An OTA contract study conducted by Grumbach
and colleagues (81 ) examined whether New York physicians who experienced high absolute increases in malpractice insurance premiums between 1980 and 1989 were more likely than physicians with lower premium increases to withdraw from obstetrics practice during the same period. The study sample included obstetrician/gyncolo-
21 The ma/pra(fl[e defindan[ rate in a county was defined as the number of ph> s]clans In that c(mnt> who had been ln~ OIL cd in ]]~ali~r,ic [ice claims dlv idccl by [he total number of physician-years Insured In the c(mnty by Washln gtfm 1 ar:cs[ carrier. 22 A total of 835 of the 1,540 eligible physicians (54.2 pcrccnt) rcsp)ndcd tt~ the survey. ~ .%enarix for this study was rmtieleci after scenam~s dcvcl(~ped for the OT,A clinlcal sccnarlo sur~eys (see ab~~yc, .ippcnd]x [)). 24 Physicians clalms experience was measured In [w() ways I ) Categ(mca]l) (n(l cla]nls, any pasl clalnl w lth(~u[ ncg[ Igc>ncc or paJ nllnt. :m) past clalm with negligence or payrnen{, one recent claim, and more than (me recent ~la[n]). iind 2 ) OL ~r;ill phj srctan cl;itm~ ratcj coil.ipwxl (nt(l (ertilcs.
Independent variable
Mean no. of early ultrasounds per patient
Total no. of ultrasounds per patient
Obstetric Resource Use Measure Mean standardMean no. of consults or referized resource rals per patient use per patient ($)
Percent Caesarean deliveries (70)
County malpractice defendant rate

Urban obstetrician Rural obstetrician Rural family physician Urban family physician (ref.) % male Physician age HMO practice Community clinic practice Hospital practice Private practice (ref.) % high-risk patients % Medicaid patients Obstetric volume Median county household income Nursery care:b level i Level II Level Ill (ref.) Consult available Distance to tertiary hospital Physician IS residency trained Physician is board certified
-23 27* .42* 15 -04 -003 -19 -11 -07 002 .002 -001 -000005 -03 -03 05 -001 15 22
-156 15 .53
009
- - - Regression coefficients - - - - - - - - - - - - - - - - - - -79 $-1,094 02 554* 08 335 -02 158 -05 -003 .25* 04 -08 0009 0005 -0002 .00001 -11 -03 -.13* 0001 -02 -05 -118 -14 128 -161 -314 -2 3 -3 -7 -6
-02 -004 -.46* -24 -25
14 3 -1
03 352
03 -.004 12 07
196
-83 -1 -62 -14
-7
01 13 14
SOURCE L M Baldwln L G Hart M Lloyd et al Department of Family Medlclne Unwerslty of Washington, Seattle WA Malprachce Clalms Exposure and Resource Use In Low Rtsk Obstetrics prepared under contract to the Off Ice of Technology Assessment U S Congress Nov 21, 1993 unpublished data revlslons prowded 10 OTA by authors MaV 1994
gists (OB,GYNs) and family practitioners (FPs) who were active in obstetrics in 1980, The main explanatory variable was the absolute change in malpractice insurance premiums for
physicians practicing obstetrics in each specialty between 1980 and 1989 in each of New Yorks five
premium rating areas. Dependent variables included complete withdrawal from medical practice and withdrawal from obstetric practice alone during the study period. Other factors associated with withdrawal from obstetrics practice (e.g., volume of deliveries in 1980. years since 1icensure) were controlled for in the multiple regression analysis (81). Medical malpractice insurance premium increases were not associated with physician withdrawal from obstetrics practice for either OB/GYNs or FPs (81).25 Physician factors that had a statistically significant association with withdrawal from obstetrics included years since licensing (positive dissociation), volumc of deliveries in 1980 (negative association), and specialty (FPs more likely to stop than OB/GYNS) (81).26
Volume of obstetric deliveries, United States, 1987
graphic characteristics of the community in which the physician practiced. The study looked at whether OB/GYNs reported that they were practicing obstetrics at all. and also at the volume of obstetric care they reported during 1986. The study found that OB/GYNs in states with greater liability threats and who reported higher personal malpractice claims exposure were more likely to be practicing obstetrics and had higher volumes of obstetric care than their counterparts. These findings are consistent with one of the study hypotheses; namely, that obstetrics services become more concentrated among OB/GYN specialists under a worsening 1iability climate because other providers of obstetric care (e. g.. family practice physicians and nurse-midwives ) reduce their obstetric practices ( 112). This study, however, did not examine the effect of the liability climate on these other providers.
An unpublished working paper by Kington ( 112)27 examined the relationship between liability risk (measured at both the state and individual physician Ievel ) and OB/GYNs volume of obstetrics practice. The analysis used self-reported data on obstetric volume, malpractice claims history. and physician characteristics from a 1987 national survey of members of ACOG: state -level data on liability insurance premiums: and a variety of independent factors such as socioeconomic and geo -
Jacobson and Rosenquist undertook a contract case study for OTA to examine the diffusion and use of low osmolality contrast agents (LOCAs)a recently developed alternative to traditional contrast agents for radiologic imaging procedurcs ( 105 ).28 LOCAs present an opportunity to examine the relationship between legal liability and the diffusion of a new technology into medical practice. A common perception, expressed informally at professional society meetings debating the use of LOCAs, is that the widespread use of LOCAs can be explained largely as a function of
defensive medicine. The case study focused on the extent to which concerns over legal liability influenced the diffusion and use of LOCAs.
Description and Current Use of LOCAs

Radiologists and cardiologists use contrast agents to enhance a variety of radiologic imaging procedures, including angiography, intravenous urography, CT scans, and cardiac catheterization procedures. Traditional contrast agents have very high osmolality (that is, concentration of dissolved particles in solution) compared with normal body fluids, and have been associated with mild to moderate adverse reactions such as nausea and vomiting in some patients, as well as with rare but more serious adverse reactions in certain patients. The osmolality of LOCAs more closely approaches that of normal body fluids. LOCAs were first approved for the U.S. market in 1986. LOCAs and traditional contrast agents are equally effective in enhancing diagnostic images. The primary benefits of LOCAs are greater comfort for the patient due to reduced risk of mild and moderate adverse reactions and, hence, potentially better patient cooperation in the procedure. It is not clear whether LOCAs reduce the risk of more serious, but far more rare, reactions. The contractors surveyed hospitals in five regions. They found that use of LOCAs varied considerably across geographic regions and different kinds of hospitals. Some institutions reported universal use of LOCAs, while others reported using LOCAs for as few as 30 percent of patients. Some institutions had implemented selective use guidelines, although the particulars of the guidelines differed among institutions.
LOCAs were introduced in the mid-1980s (95,104). The incremental cost of using LOCAs instead of traditional contrast agents for a specific procedure may amount to $150-$200. Reimbursement for LOCAs varies widely. Hospital prospective payment systems give hospitals incentives to use less expensive alternatives on inpatients. Reimbursement for LOCAs used in outpatient diagnostic x-ray procedures varies by type of insurance coverage. Since January 1992, Medicare has reimbursed for outpatient LOCA use in selected high-risk patients.29 Private insurers have had a more liberal reimbursement policy, generally reimbursing at close to the full invoice price of the agent, depending on type of coverage. The variation in reimbursement policies for LOCAs makes it difficult to systematically compare their importance with that of malpractice concerns in explaining LOCA diffusion or use.
Legal Issues Affecting the Diffusion of LOCAs

In the absence of established legal precedent or professional consensus, it would appear that hospitals and physicians are confronted with a difficult choice in how to utilize LOCAs: how to balance the high costs of universal LOCA use with potential legal liability for improperly limiting their use. However, despite the common perception that liability fears have been driving LOCA diffusion, actual liability claims or litigation involving contrast agents are very limited. OTAs contractors were unable to identify a single court case involving the issue of whether the use of a traditional contrast agent for a low-risk patient constitutes negligence or whether the availability of LOCAs as an alternative must be disclosed to the patient. However, because LOCAs are now used almost universally for certain high-risk patients, the failure to use LOCAs for these patients might be considered negligent. At the very least, the physician would have the burden of justifying the failure to use LOCAs.
Costs of and Reimbursement for LOCAs

According to most reports and the survey information gathered for the OTA case study, LOCAs cost 10 to 20 times as much as traditional contrast agents. There has been only minimal change in the price ratio between them since
29 Medicare rein~bursemen[ policy is based (m sclcc{ive usc guidelines published by the American Ci>llege of Racli{~li)gy (3,170).
Only a few of the health professionals interviewed by OTA's contractor-s were aware of any existing litigation regarding contrast agents. Only one had been sued or had a claim filed over the use or choice of contrast agents. None of the risk managers interviewed had received any claims, and two of them asserted that there was no good risk management rationale for universal LOCA use.
Average relative rank of factora Physicians (N=29) Patient safety/comfort Reductions in adverse reactions Clinlcal indications costs Guidelines Physician preference Hospital policies Legal concerns Reimbursement policy Competitive factors Manufacturer marketing Administrators (N=17)
b
Survey Methods and Results

In an effort to gain a better understanding of physician decisionmaking regarding LOCAs, knowledgeable health care providers at a variety of different institutions in metropolitan areas in five different geographic regions of the country were interviewed about their reasons for- using LOCAs. Personal interviews were conducted with 46 indi viduals29 physicians (primarily radiologists and cardiologists) and 17 hospital administrators (including risk managers). Telephone interviews were conducted where the individual was not available in person. The trends reported are believed to reasonably reflect the current state of LOCA use. The survey included questionnaires asking respondents to indicate the importance of 11 different factors thought to influence the decision between traditional contrast agents and LOCAs. When asked to rank the factors in descending order of importance, physicians ranked legal concerns 7th out of 11 factors, and administrators ranked them 5th (table 3-8). Physicians ranked -reducing adverse reactions as the most important factor in choosing between LOCAs and traditional agents, and administrators ranked clinical indications" as the most important factor. 30 ) Cost of the agents was ranked as the 4th most important factor by physicians and as the 3rd most important factor by administrators (table 3-8). Thus, despite anecdotal information from the interviewees about the role of malpractice 1iability
1 1
3 4 5 6 7 7 9 10 11
1
3 2 3 7 5 7 5 9 10 11
J The qLlest Ion put to respondents w.)s Wbat cr terla CIId you LJ:le to make a declson on use of low vs h gtl-osmoar contrasl agent s,? Carl you rank each of the tollowlng [11] fa{ tors lr~ order of lrnportance? This columrl represents the mearl rank .Iss Ig ned for each actor Wh(>re two factors bave tbe same mean rank thc,y are ryverl the sam~ v.~ Je b I nc]Llde5 some hospital msk mwagers SOURCE P D Jacot]sor~ and C J Rosenqulst The D ffusl(jrl of 1 ow OsmolaIIty Contrast Agents lecbnolo:~c,~l Change ancj Defers ve Med clne contract report prepared for the Off I c.e o Tec hrlc, ogy Asscssrnerl t U S Congress WdStTIrglen DC Novemtmr 1 733
concerns in the decision to use LOCAs, their written responses suggest medical factors and cost considerations play a greater role than liability concerns in current decisions about the use of LOCAs. It is possible, however. that survey respondents underrated the influence of 1iability concerns because the y felt this was a more socially desirable response. While liability considerations are important to radiologists and cardiologists and might explain some of the LOCA market penetration, factors relating to general technological] advances. such as enhanced patient safety and comfort, appear to be more important in explaining LOCA use. Due to the small number of respondents and other 1imita-
tions of the case study design, however, these findings should be regarded as tentative.
CONCLUSIONS
Although direct physician surveys suggest that fear of malpractice liability is widespread among physicians and that many of them practice defensive medicine, the validity of these results is highly questionable for a number of reasonsin particular, the q *prompting of physicians to cite malpractice liability concerns and response bias due to low response rates. Consequently, the results of many of these surveys probably considerably overestimate the extent of defensive medicine. Survey-based estimates of the national cost of defensive medicine advanced by researchers at several organizations are unreliable and potentially biased. The true costs of defensive medicine may be either higher or lower than predicted by such studies. In clinical scenario surveys designed specifically to elicit a defensive response, malpractice concerns were occasionally cited as an important factor in clinical decisions; however, physicians belief that a course of action is medically indicated was the most important determinant of physicians clinical choices. These findings suggest that many physicians are more aggressive in diagnosis not because of fear of malpractice liability, but because they have come to believe that such practices are medically necessary. One large, well-designed study found a statistically significant relationship between Caesarean delivery rates and hospital- and area-level measures of malpractice liability risk (based on malpractice insurance premiums and claims) in New York State. However, to date these findings have not been replicated in other clinical situations or geographic areas. Two smaller studies commissioned by OTA failed to find similar relationships between liability risk and increased resource use in other areas of clinical practice, although limits of sample size and study design may have precluded positive findings in these studies. Neither
of the two empirical studies of negative defensive medicine found a statistically significant positive relationship between liability risk and withdrawal from obstetrics practice. A major limitation of such statistical studies is that they cannot measure the overall level of defensive medicine; they can detect only incremental differences in defensive behavior between groups of physicians with higher and lower levels of malpractice liability risk. Taken together, the findings from studies reviewed in this chapter suggest that defensive medicine is a real phenomenon that has a discernible influence in certain select clinical situations. OTA was able to document defensive practice in several isolated clinical situations, most notably the use of diagnostic radiologic examinations for young patients presenting with head injuries in emergency rooms (see table 3-3). There are probably other clinical situations not studied by OTA or others in which defensive medicine plays a major role in physicians diagnosis and treatment decisions. However, in the majority of clinical scenarios used in OTAs and other surveys, respondents did not report substantial levels of defensive medicine, even though the scenarios were specifically designed to elicit a defensive response. Based on the limited evidence available, OTA estimates that a relatively small proportion of all diagnostic procedures-certainly less than 8 percent overallis performed primarily due to conscious concern about malpractice liability risk. OTA did not attempt to make similar rough estimates of the proportion of therapeutic procedures performed for defensive reasons; in part because there was no outside information to draw on. The studies reviewed in this chapter illustrate the great difficulty of accurately measuring the true extent of defensive medicine. Although it is possible to identify particular clinical situations in which defensive medicine plays a relatively major role, it is impossible in the final analysis to draw any conclusions about the overall extent or cost of defensive medicine.
Impact of Malpractice Reform on Defensive Medicine 4

lthough it is impossible to measure with much Precision the extent of defensive medicine, the evidence summarized in Chapter 3 implies that it is neither a trivial nor a major contributor to health care costs. This chapter examines how different approaches to reforming the medical malpractice system might affect the frequency of defensive medicine. The chapter examines the potential for tort reforms (i.e., changes in the legal rules for resolving malpractice claims) to reduce defensive medicine. This is a limited policy analysis; other impacts of tort reform may be equally or more important, including:
s
Quality of care: A principle objective of medical malpractice law is to deter physicians from rendering lower-quality care, but the effect of the malpractice system on quality of care has hardly been studied. Although there is reason to believe it may have some positive effect on quality (e.g.. increased investment in risk management and quality control), the scant empirical evidence available does not support the contention that the malpractice system as it is presently configured does improve quality of care. 1 Nonetheless, tort reforms that limit physicians 1iability could adversely affect the quality of care.
I For example, in an attcmpt t{) est]nmte the deterrent effect of medical rnalpractm, researchers at Han ard LJnl/ crslty recently anal}~ed the relatitmship between the numhcr of nd pract]ce clalms pcr negligent ln]ury and the rate of negligent lnjuri~s in N~w ~orh State hospitals in 1984. They fa]leci toden~(~nstr:ite a s]gnlficant rclalionshiph ctwccn nlal practlcc claim acti~ lty and the rate of ncgllgent injury in a h(~spital (254). me anal~sls was Iirn]ted b) a small sample SIZC (Icss than 50 ht~spltal~) and a sm:lc year t~f data. Thus, the analysls may not have had $ufiiclcnt statistical power tt~ detect a dctm-mnt effect If It d]d exist.
Plaintiffs' access to the legal system: Evidence exists that the vast majority of patients injured by negligent medical care do not file a claim (130),2 and tort reforms could either make it easier or more difficult, especially for patients with 1imited financial resources; Cost of compensating victims of malpractice: Some reform proposals promise lower administrative costs (e.g., lower lawyers fees) but also would compensate a greater number of individuals. The Office of Technology Assessment (OTA) has not examined whether the overall impact of these changes would be to increase or to save costs. Physician-patient relationships: Physicians claim that their concern about malpractice liability causes their relationships with patients to suffer. Depending on its configuration, tort reform could either improve or hurt the physician-patient relationship.
most of these newer reform proposals have not been implemented, it is difficult to predict their impact on defensive medicine.
THE IMPACT OF CONVENTIONAL MALPRACTICE REFORMS ON DIRECT MALPRACTICE COSTS

Most of the traditional tort reforms retain the courts as the forum for resolvi ng malpractice suits but change certain legal rules, such as imposing limits on the time after an injury or its discovery in which a suit can be filed, or limiting the damages that can be awarded. These conventional tort reforms have been labeled pro-defendant, because they often restrict plaintiffs access to courts or limit the amounts plaintiffs can recover (254). For example, requiring a plaintiff to obtain a certificate of meritan affidavit by a physician that the claim is validprior to filing a suit can make it more difficult for low-income plaintiffs to sue (see box 4-l ) ( 166).3 Box 4-2 contains a brief description of the traditional legal reforms. In a separate background paper, OTA reviewed the results of six multistate studies that used statistical techniques to estimate the impact of specific malpractice reforms on four indicators of direct malpractice costs: 1 ) frequency of suit, 2) payment per paid claim, 3) probability of payment, and 4) insurance premiums (236). The six studies were selected because they used the most methodologically rigorous approaches to isolating the impact of malpractice reform on malpractice costs. OTA also identified several studies that either examined trends in malpractice activity in states with malpractice reforms or compared trends in such a state with those in other states without the same reforms. The results of OTAs review of the six multistate study and of the more compelling single-
More general discussions of the range of potential impacts of tort reforms are available in a number of review articles (12,2 1,37,122,208a). In this chapter OTA focuses mainly on the effects of malpractice reforms-both conventional approaches and new proposals-on defensive medic inc. Since the first malpractice insurance crisis in the mid- 1970s, almost every state has reformed one or more aspects of malpractice law (22,236). The tort reforms implemented in the states were designed primarily to reduce malpractice insurance premiums by limiting the frequency of suits, payments per paid claim, or the cost of resolving claims. Conventional tort reforms us implemented in the states have maintained the malpractice liability system while tinkering with one of more aspects of the claim resolution process. Newer reform proposals would substantially alter the process for resolving malpractice claims or would limit the physicians personal liability and substitute other quality control systems. Since
2 A rcccnt stud) t)! NCW Yfmk State h(~spttal stays rc~ealcd that apprt)xirna[cly (mc in 50 ncgl igcntly injured plaintiffs br[wght a malpr~cttce clalm ( I 30).
~ L{Iv. incf)n~c pla]ntl ffs are already Icss IILCIJ to suc than more affluent pl:iintl ffs (.? 1,230,239).
Chapter 4 Impact of Malpractice Reform on Defensive Medicine | 77
Many tort reforms explicitly Iimit the amount the plaintiff or his or her attorney can recover from a malpractice case (e g caps on damages, collateral source offsets or Iimits on attorney fees) or increase the costs of bringing a suit (e g certificates of merit) Such reforms make filing a malpractice suit less attractive for all plaintiffs. Whether these reforms disproportionately affect peoples ability to sue has not been studied As part of this study OTA was asked to examine whether Iow-income obstetric patients are more Iitigious than privately Insured patients OTA issued a background paper on this issue which found that Medicaid and Medicare patients sue physicians less often than would be expected given their relative proportion of the population (Medicaid patients) or heavy use of health services (Medicare patients) (239) OTA also commissioned a study by Morlock and Malitz to examine the impact of Marylands tort reforms on claim filings by Medicaid, Medicare and self-insured plaintiffs In July
s
1986 Maryland Implemented a package of tort reforms
a requirement that a certificate of merit be obtained within 90 days of filing a malpractice claim, a $350 000 cap on noneconomic damages, s a provision for periodic payment of damages, q a shortened statute of Iimitations for minors and s administrative reforms to Improve the pretrial screening process
q
Of these reforms the requirement that a certificate of merit be obtained within 90 days of fliling iS
most likely to pose a differential barrier based on the plaintiffs income. Obtaining such a certificate
costs $600 to $1 000 and some attorneys may require that these costs be paid by the claimant in advance of settlement or other disposition Morlock found a substantial drop in the number of claims filed by patients with no Insurance and by Medicaid patients following the Implementation of the Maryland reforms The following table shows the number of malpractice claims filed per 100000 hospital discharges in Maryland The rates are displayed by Insurance status of the Injured party A certificate of merit was required beginning in July 1986 but the Iegislation requiring the certificate was passed during the Iegislative session from January to April, 1986 Malpractice Claims Filed in the Legal System as a Result of Hospital Incidents per 100,000 Discharges in Maryland, 1979-89 Insurance Status 1979-1985 (Pre-reform) Total number of claims Claims by privately insured patients Claims by Medicare patients Claims by Medicaid patients Jan. 86 - June 86 (Transition) 599 759 51 9 671 83 July 86 - June 87 (Post-reform) 366 467 326 395 59
July 87 - Dec. 89
(Post-reform) 297 44 1 263 7.4 154
401
491 289 291
Claims by uninsured patients 5 5 2
SOURCE L L Morlock and F E Mal!tz Sho{(-~errn Effects of Tort and Adrnms[raflve Reforms on /he Clalmmg EIehawof of Prwa(e/y /nsured Mecf/care Medcad and Umnsured Paf/enfs prepared for the Ofhce of Technology Assessment U S Congress (Washington DC U S Government Prmtlng Off Ice September 1993)
Aimed at the Number of Lawsuits:
1. Attorney fee limits: Plaintiff attorneys are paid on a contingency basis, that is, they are paid a portion of the
plaintiffs damages as a fee but receive no fee when the plaintiff loses The typical contingent fee cases 2 Certificate of Merit Some states require that a plaintiff obtain an affidavit from a physician or other expert attesting that the plaintiffs malpractice claim has merit prior to filing the suit 3 Costs awardable If a plaintiff files a claim that
IS IS
33-1/3 percent of the award Some states Iimit the contingency fee percentage in large damage
subsequently judged to be without any merit, a judge may
force the plaintiff to pay the defendants court costs, and in some states the defendants legal fees
4 Pretrial screening panels: As a prerequisite to filing a suit in a court, parties may be required to submit the
malpractice claim to a hearing before a panel consisting of one or more attorneys and health care providers, and, ln certain states, a judge or Iay person. The panel wlll render a decision on Iiability and sometlmes damages The parties may choose to accept the panels findings and settle the case or file a suit in court In some states, the panels findings may be entered into a subsequent legal proceeding Some states offer panels as a voluntary option. 5 Statutes of limitations: The statute of Iimitations prescribes the time period after the injury in which a legal claim may be brought In medical malpractice this time period
IS
either measured from the date of the
negligent treatment or from the date the injury could have reasonably been discovered (the discovery rule ) Some states have shortened the time period in which a claim can be brought or Iimited the application of the discovery rule Aimed at Size of Recovery (Payment Per Paid Claim):
1 Caps on damages (noneconomic, total) Damages in medical malpractice consist of 1 ) economic damages, which are monetary awards for incurred and future costs arising from the injury (primarily medical and rehabilitative expenses and lost wages), and 2) noneconomic damages, consisting of monetary awards to compensate for the pain and suffering associated with the injury Certain states have placed Iimits (i. e , caps ) on the amount the jury can award for noneconomic damages, or for total damages ( I e , economic and noneconomic damages) 2 Collateral source offset (mandatory, discretionary,) Certain states require or permit the jury to reduce the plaintiffs malpractice award by the amount the plaintiff iS entitled to receive from collateral sources, such as health and disability insurers 3 Joint and several liability changes: Traditionally, when multiple defendants were responsible for a plaintiffs injury, the plaintiff had the right to collect from each defendant in the amount of their responsibility (jointliability) or the plaintiff could collect the entire amount from a single defendant (several Iiability), forcing that defendant to sue the other defendants for the amount that they were responsible for Some states have eliminated several Iiability, usually with respect to noneconomic damages only. 4 Periodic payments of damages (structured awards) Damages awarded to pay for future economic and noneconomic losses may be paid on a periodic basis, rather than in one lump sum Aimed at Plaintiffs Difficulty (or Costs) of Winning:
1 Expert witness requirements: Expert witnesses are used to establish the standard of care in a malpractice
trial Some states impose specific requirements on the experts qualifications for example, requiring that the physician have practiced in an area of medicine that iS related to the subject of the case
(continued)
2. Informed consent limits: Physicians must obtain informed consent from patient before performing a procedure. Some malpractice cases allege that the physician did not provide adequate information for the plaintiff to make an informed judgment The adequacy of the information provided can be judged on the basis of whether a reasonable patient would consider the Information provided adequate, or by Iooking at the practice o fother physicians The former standard is often characterlzed as pro-plaintiff, and some states restrict the use of this patient-oriented standard
I
3. Res ipsa loquitur restrictlons In medical malpractice, when the incident causing the injury was under the exclusive control of the physician and it iS obvious to an nonmedically trained person that the plaintiffs injury would not have occurred in the absence of negligence, a plaintiff will not be required to offer expert testimony of negligence Some states restrict the use of this doctrine
SOLJRCE S R Bovb]erg ~ rq~lallon on Medical M:ilpractlce Further Developments and a Prellmlnary Report Card Urr/vers/fy of Ca/I/orna DavIJ L.IW RevIe~I 22 -199-557( 1989) U S Congress Off Ice of Tectlnology Assessment Impac( of Legal Reforms on lda/prac[cc Cos(.s OTA-BP-H- 119 (VVashlngton DC Government ?rlrntlng Ofllce 1993)
state studies are summarized below. (See appendix G for a complete summary of the single-state studies ).
Multistate
Data
The six empirical studies reviewed in OTAs background paper examined the impact of a number of different reforms, but not every study examined the same set of reforms, The majority of the studies looked at the following reforms;
m m m
q q
shortening the statute of I imitations. limiting plaintiffs attorney fees, requiring or allowing pretrial screening of claims, caps on economic and noneconomic damages. amending the collateral source rule to require offsets for the portion of damages covered by health or disability insurane, and periodic payment of damages.
Across all studies, only caps on damages and amending the collateral source rule consistently reduced one or more indicators of direct malpractice costs (236). Shortening statutes of limitations and implementing pretrial screening showed inconsistent results across studies (236). Limits on attorney fees and periodic payments showed no statistical -
ance fund that paid damages exceeding $100,000, up to the $500,000 cap.4 Gronfein and Kinney found that the average payment per large paid claim was 33 and 40 percent higher in Indiana than in the neighboring states of Michigan and Ohio, respectively. This outcome probably resulted from the operation of the PCF, which gave the insurer an incentive to settle large claims when the issue of negligence was unclear, thereby shifting a portion of the liability to the PCF. On the other hand, Indiana had no payments over $500,000, whereas in Michigan and Ohio the few cases in which more than $1 million was awarded accounted for 21 and 14 percent of all malpractice payouts, respectively (79).
Therefore, overall payments for malpractice may be higher in those states despite the fact the average payment is less.
The California Studies

Supporters of malpractice reform often point to Califomia as an example of the impact tort reform can have on malpractice costs. In 1975, California passed the Medical Injury Compensation Reform Act (MICRA), which included a $250,000 capon noneconomic damages, limits on attorney fees, discretionary collateral source offsets, and periodic payments for future damages in excess of
$50,000.
Two studies concluded that MICRA significantly lowered malpractice insurance premiums or claims costs5 in California (32,34). One study found that the average malpractice insurance pre-
mium (adjusted for inflation) declined by over 60 percent from 1976 to 1991 (34), but this result in and of itself is inconclusive because 1976 marked a peak and 1991 a trough in the national cycle of malpractice premiums (236).6 More compelling is evidence that California malpractice premiums declined at a compound annual rate of 0.4 percent between 1976 and 1991 compared with a national average annual rate of increase of about 12 percent over the entire period.7 Although critics of MICRA point out that the average 1992 California malpractice premium was only slightly below the national average premium (200), Californias average malpractice premium was 65 percent above the national average as recently as 1985 (261). Not all of the relative savings can be attributed to MICRA, however, because a simple pre-post comparison does not control for other changes in the malpractice and health care markets in California over the study period. For example, physicianowned malpractice insurance companies replaced commercial malpractice insurers shortly after MICRA was passed. Also, the largest California health maintenance organization (HMO), Kaiser Foundation, with over 4 million enrollees (141), initiated arbitration for all medical malpractice cases in the early 1970s (236). California has experienced rapid growth in HMOs over the past 10 years. 8 Still, it is likely that MICRAs stringent cap did reduce California malpractice insurance premiums to some extent. The observation that malpractice insurance premiums increased more
4 Tk Indliina
cap ~ln ti~tiil d:ir]]iig~s Wiis raised to $750,000 in January of 1990 (79).
f Clwms costs Include payments made [[) plilintiffs and the insurers direct cx~sts attributable tt) the claim (fees for investigative work, expert w I[ness fees, iild IC~iil dcfcnsc ~ t)rh ). ~ Trends in ,nsllr;lnce Prelll iurll~ are ~hara~[~riled by
earn
CyC]CS. These cycles
are tied to some extenl t{) the investment Cllnlate, bCCaUSe insrJrerS
pilrt of [hclr mctmw frx~m Invcslmg prcrnlurus in inconle-prxducing assets. As the interest rilt~ txpectcd ~rorrl capital investments rises and
ii umlpetitive rate
fillls, prcrnlums arc iidjlls(~d accx)rdlngl} tt~ ilSSU1C
(If retulm to investors (2 I ()).
7 ~le ~orllpiirl~on IS b:lscd on Prcrlliurlls in current d(~llars. OTA calculated the ch~mge in Cilliformia premiums fr(m data re~)fled in a study by the C[~iilitlor] t(} Prcscrvc MICRA (34). In that study the 1976 premium (iidjtlsted for inflatitm to 1991 dollars) was $18,000 and the 1991 prernlun~ Wiis $7,000, Llsln: the c(msunwr price index-unadjusted (CPI-U) for 1976 and 1991, the 1976 premium unadjusted for inflati(m is $7,427. The niitlt~n:il cstirlliit~ IS bmed(m incrciiscs in malpractice insurimce rcpwtcd by the U.S. Health Care Financing Adnlinistrati(m (5 I F.R.
ZS77Z, 28774, 57 F,R. 55903). s Appro~ Irr]iitcl) 34,4 pcrccnt t)f the p)pulil[ion is enrolled In HMOS m (Xallf{rnia, c(~rnpared w ]th 17.3 percent nat](mwide ( 141 ).
slowly in California after MICRA is consistent with the finding that caps on noneconomic damages lower malpractice costs. California has one of the lowest caps on noneconomic damages in the country, and it has not been adjusted since 1975 (236).
Pretrial Screening Studies

Five separate studies of pretrial screening panels (three of Arizona, one of Hawaii, and one of 15 different states including Arizona) found that most plaintiffs did not appeal adverse panel decisions, which may indicate that pretrial screening led to early resolution of cases (see appendix G). Because most of the studies failed to report claim frequency before and after the screening panel was initiated, however, it is possible that pretrial screening prompted filing of more nonmeritorious claims, which were dropped after adverse panel decisions. In add it ion, almost every study found that pretrial screening panels caused significant delays in claim resolution (see appendix G). These delays may have led some plaintiffs to drop or settle cases because of the added expense of the pretrial screening process.
reduce defensive behavior. Yet, it is not known whether Localios findings for obstetricians and Caesarean delivery rates are generalizable to other procedures, other specialties, or other states. especially in light of the failure of other studies funded by OTA to find such a relationship ( see chapter 3). There are reasons to be skeptical that traditional tort reforms can reduce defensive medicine. Physicians may not react to mere reductions in malpractice risk. Instead, they may try to limit their personal risk of suit to as close to zero as possible. In the absence of any financial penalties for doing so, such an objective is a rational response to any level of malpractice risk. The long-standing concern about defensive medicine suggests that traditional tort reforms may not do much to reduce defensive medicine. In the early 1970s, when direct malpractice costs were quite low and when the malpractice signals were much weaker than they are today, there was still considerable concern about defensive medicine ( 14,20,58,243).
IMPACT OF NEWER MALPRACTICE REFORMS ON DEFENSIVE MEDICINE

Recent reform proposals either expand on traditional reforms-for example, redefining the standard of care using practice guidelines-or call for more sweeping changes, such as removing medical malpractice from the judicial system, relieving the physician of malpractice liability or eliminating the fault-based malpractice system complete1 y. These reforms all seek to make the claims resolution process more timely and less costly. Some of them would provide greater access to compensation for deserving plaintiffs. All seek to decrease the impetus for defensive medical practices. The new reform proposals fall into four categories:
s
The empirical literature discussed in chapter 3 suggests that physician behavior may be influenced in certain clinical situations by the strength of signals that the malpractice system sends about the risk of being sued. If tort reforms reduce the direct costs of malpractice, they may soften the signal and therefore also reduce defensive medicine. The best evidence for this association comes from a single study of the impact of malpractice signals on Caesarean delivery rates in New York State (129, 131 ). Localio found a strong association between the strength of the malpractice signal (i.e., high claim frequency and insurance premiums) and Caesarean delivery rates ( 129). This study supports the hypothesis that malpractice reforms that reduce claim frequency and premiums
Clinical practice guidelines as the standard of care.. At present, clinical guidelines may sometimes be entered into malpractice trials as evidence of the standard of care along with expert testimony. Several states tire developing programs in which certain clinical guidelines will be used as the definitive statement of the stan-
dard of care, replacing expert opinion when applicable. Enterprise liability: Enterprise liability would retain the current malpractice system, but the physician would no longer be a named defendant. Instead, the enterprise in which the physician practices would assume the liability for medical negligence ( 1). As originally conceived, the enterprise would be the hospital or HMO in which the physician practices(1). Under a managed competition system, liability could rest with the health insurance plan (16 1). Alternative dispute resolution: Alternative dispute resolution (ADR) removes the claim from the legal system to reduce the time and money involved in its resolution and to make the proceeding less public and adversarial. In binding ADR the dispute is heard and decided through a nonjudicial procedure, and opportunities for appeal are very limited. Because state constitutions guarantee the right to trial, binding ADR to date has been a voluntary procedure, agreed to by both parties. Selective no-fault malpractice compensation: Proposals for a selective no-fault malpractice compensation system envision a process similar to workers compensation. The leading proposal would designate certain adverse medical events that are generally avoidable as compensable under a no-fault system (221). More patients could receive compensation for medical injuries that are generally avoidable, even if there is no evidence that the injuries were caused by negligent care.
quality of care, or the potential impact on plaintiffs.
The potential impact of each of the proposed reforms on defensive medicine is examined below. OTA has not attempted to address in detail other potential benefits or limitations of these reforms, including the cost of implementing a reform compared with the present system, the impact on
See appcndi x H for a rmwe detailed discussi(m of the legal usc t)f c1 mlciil practwc guidelines, lncludlng a ret ICW of sta[c lnitia[i~cs in this
area.
Chapter 4 lmpact of Malpractice Reform on Defensive Medicine | 83
ten. In cases where the criteria in the guideline are clear, it should reduce defensive medicine. For example, there is some early evidence that adoption of the Maine guideline has substantially reduced cervical spine x-rays in emergency rooms ( 11 5). In cases where criteria for doing or not doing a procedure are less clear, the impact is more questionable. In Maine, for example, if a plaintiff proves that the guideline was not relevant given the clinical circumstances. the physician cannot use it as an affirmative defense. Because much of medical practice is subject to uncertainty, opportunities may be limited for developing guidelines explicit enough to be truly protective and to reduce defensive medicine. Physicians have also expressed concern that, if given greater weight in courts. guidelines could be used against them by patients for whom they had decided not to perform certain procedures. This concern might be particularly valid in cases where the guideline itself left considerable room for phy sician judgmentand many guidelines do. In these cases, the court would presumably defer to expert testimony to determine whether the physician exercised fair judgment. Maine addressed this concern by including a provision that specifically denies plaintiffs the right to introduce guidelines developed under the demonstration project as evidence of the standard of care. Some critics have questioned the constitutionality of this provision and the feasibility y of actually preventing plaintiffs from introducing the guidelines as evidence ( 155.1 79). In the absence of specific legislation to give guidelines more evidentiary weight. the continued development of guidelines will probably help to make practice in certain areas of medicine more uniform and hence help to clarify the legal standard of care (236). Recent evidence that guidelines are playing an increasing (though still small) role in medical malpractice litigation supports this conclusion (see appendix H ) ( 100). Howe\er. there are a number of factors that could limit their impact on medical liabi1ity and defensive medicine (see box 4-3). A major limitation is thc ability to write sufficiently explicit guidelines. Many clinical condi-
tions involve so much medical uncertainty that specific recommendations on appropriate use of technology will not be possible. For example, the National Cancer Institute ( NC I ) recommends routine mammography screening for women over 50 years of age but notes that "[e]xperts do not agree on the role of routine screening mammography for
women ages 40 to 49 ( 172). Thus. the appropriate frequency of mammography screening for women under age 50 is left to physician judgment. Indeed, the majority of clinical practice guidelines
written to date--including those developed by the federal Agency for Health Care Policy and Researchlist several diagnostic and therapeutic options for addressing specific medical conditions, leaving consider-able room for physician judgment. A guideline that leaves substantial room for physician judgment may be no more helpful in defining the proper standard of care than expert witnesses. In addition. in the absence of specific legislative changes such as those in Maine ( i e., where only certain guidelines are afforded elevated legal status), juries may choose to disregard guidelines or may be asked to make judgments about conflicting guidelines, just as they are now sometimes presented with conflicting expert testimony. Despite the limitations of guidelines, they offer several potential advantages over other malpractice reforms. Tort reforms are predicted to alter physician behavior because the> dull the tort signal and therefore allow physicians to make clini cal judgments with less anxiety about the risk of being sued. Yet. with a reduced malpractice signal, there could be a reduction in beneficial defen sive medicine as well as defensive medicine that has less clinical value. Softening the tort signal will also changc only those practices that are consciously motivated by fear of liability. Guidelines, on the other hand, can selectively target defensive medicine that does not improve the quality of care. Also. guidelincs present an opportunity for experts to reevaluate clinical practices that are performed routinely but with little evidence that they make a real difference to patient car-e. Therefore, guidelines have the potential to
Guidelines
s q
factors
q s s
Extent to which guidelines are targeted to address defensive medical practices Comprehensiveness of guidelines (i,e. , how much of medical practice iS now or can be expected m the near future to be addressed by guidelines?) Ability of guidelines to keep pace with advances in medical technology and practice Existence of multiple conflicting guidelines Criteria and process used in guidelines development (e g , medical effectiveness versus cost-effectiveness; broad consensus versus expert opinion) Source of guidelines (e g , national medical specialty society, state or federal government, Insurance
get at both conscious and unconscious defensive medicine.
ADR can take many forms, but its basic characteristic is that disputes are heard by one or more arbitrators or mediators rather than by a jury. The arbitration proceeding is often less formal, less costly, and less public than a judicial trial. In nonbinding ADR, if a party is not satisfied with the result, he or she can continue to pursue the claim through the legal system. Therefore, nonbinding ADR may not eliminate physicians anxiety about a potential malpractice trial. Binding ADR may be the most effective approach to eliminating the
physicians anxiety about a trial. The two leading binding ADR proposals are: voluntary binding arbitration under pretreatment contracts between patient and providers (or health plans), and the American Medical Association/Specialty Society Medical Liability Projects (AMA/SSMLPs) fault-based administrative system, which would remove all malpractice cases from the judicial system.
Voluntary Binding Arbitration

To implement voluntary binding arbitration, the parties must agree to waive their right to trial and instead retain one or more arbitrators to render a decision. In medical malpractice the patient and
\ ]n ~~~ltlon {)nbln~lng ADR nlay not ]ea~ K) re~ucli(ms in direct ..nm]practlcc ct~sts (i.e., the costs directly associated with rwlving a
malpractice claim) hecause of the potential for two proceedings (42.75,209).
Chapter 4 impact of Malpractice Reform on Defensive Medicine | 85
physician (or insurer) may agree to arbitrate either after an injury has occurred or before the treatment is even provided. An agreement made before treatment is rendered is called a pretreatment arbitration agreement. From the physician perspective, pretreatment arbitration agreements can provide upfront assurance that the case will be arbitrated. After an injury has occurred, the physician-patient relationship may not be conducive to negotiation of an arbitration agreement. Arbitration has several potential advantages. Arbitration replaces the lay jury with professional decisionmakers, who may have previous experience with malpractice cases. Many arbitrators are ex-judges or otherwise legally trained individuals. Though there is no good empirical evidence that jury decisions are worse than or very different from arbitration decisions, 12 physicians may perceive this to be the case. Arbitration proceedings are also less public and often may be scheduled sooner than trials. Binding arbitrat ion has not been used frequently in malpractice cases, but it is used extensively in commercial settings. Companies claim significant savings in legal costs ( 2 16). The very limitcd data available on malpractice arbitration indicates that arbitration may be less costly for resolving disputes. ] 3 Arbitration may be infrequent in medical malpractice for several reasons. Some plaintiff and defense attorneys believe that the jury is an appropriate dispute resolver, especially when factual
issues are involved ( 159). Yet the reluctance to accept arbitration may also result from a lack of experience with arbitration. 14 Attorneys familiar with arbitration also claim that arbitrators tend to reach compromise decisions in which the physician is held partially responsible (42, 158, 185). Because physicians take malpractice claims so personally, compromise decisions may not satisfy their desire to vindicate their conduct ( 159). On the other hand, arbitrators are very unlikely to award large damages, as juries sometimes do. This may be seen as a disadvantage to arbitration for plaintiffs (42, 158, 185). Pretreatment arbitration agreements also have limitations. Some states permit the patient to revoke the pretreatment agreement within a certain time after signing the contract usually 30 to 60 days) (23 1). In states without such statutory rules, the enforceability of pretreatment contracts is governed by case law. The courts often closely scrutinize such contracts, because the health care provider may have superior bar-gaining power (236). 15 For example, a health care provider could refuse to enter into a physician-patient relationship unless the patient relinquished his or her right to a trial. 16 Statutes that allow patients to revoke pretreatment agreements and court scrutiny of such contracts render pretreatment contracts of uncertain value, especially to health care providers. Whether arbitration would reduce defensive medicine depends upon the extent to which the threat of a court trial drives physicians to practice
defensive medicine. If the small risk that a suit will proceed to trial drives physicians to practice defensively, then ADR should reduce defensive medical practices. If the real driver of defensive medicine is the desire to avoid any process that judges the physicians actions, then arbitration may not affect physician behavior. It is also possible that pretreatment arbitration provisions might increase the frequency of suits, because plaintiffs may prefer arbitration over a jury trial. ] 7 Plaintiffs who would otherwise have settled their case because of the expense of trial may also decide to arbitrate. 18 The resulting increase in malpractice liability proceedings could lead to more defensive medicine.
AMA/SSMLP Administrative System

The AMA/SSMLP proposed a mandatory administrative system to replace the civil jury system for malpractice claims. The AMA/SSMLP administrative system would be part of the state medical licensing organization and would be run by a seven-member state medical board, which would include at least two physicians and possibly another health care professional. Damages awarded under this system would be limited to economic damages as determined by guidelines and reduced by collateral sources, and noneconomic changes limited to an amount equal to one-half of the average annual wage in the state multiplied by the life expectancy of the plaintiff (approximately $700,000 for a person with a
70-year life expectancy and $150,000 for someone with a 15-year life expectancy) (9). Plaintiffs would not need an attorney to file a claim. If a claim were found to have merit by a claims examiner, the plaintiff would be provided an attorney for further proceedings. If the claims examiner were to reject the claim, the claimant would have the right to appeal to one member of the medical board. If the claimant prevailed, an attorney would then be provided to him or her. If at any subsequent point in the process the claim is determined not to have merit, the plaintiff would have to obtain his or her own counsel and a certificate of merit to appeal the adverse decision. Because the proposal contemplates limiting damages, the requirements of personal counsel and a certificate of merit would discourage appeals of adverse decisions, and many cases would probably be eliminated with a single review by a claims examiner or one member of the medical bow-cl. 9 For physicians, the AMA/SSMLP proposal promises quicker claim resolution, with few claims decided in a formal proceeding resembling a trial, or even in an arbitration process. The AMA/SSMLP also proposes a number of legal changes, including: moving from the customary standard of care to a standard that accepts a physicians action if it is within a range of reasonableness; adding new requirements for expert witnesses; admitting practice guidelines and medical 1iterature without requiring that an expert witness validate its usefulness; changing informed
17
Much IS )a~e in the lalpractlce ]I[cra[ure a~)ut the
impact of the tria] On
a physician,
but
nlany
plaintiffs may also find the prospect )f a
legal batt]~ nappea]]ng, [nd~ed, this
pr,)spe~[ has been found [(J be (me factor that
disc(mrages plaintiffs fr(ml filing suits ( 14S).
18 In Mlchlo:ln N { ] Cl:ilnls ~cre fjIe~ for ~rbi[r~[ic)n and 247 (30 percent) went to an arbitrator (233). Only i O 1020 percent of llllg~td 0, claims t} pically go to tr]al (171,222,235).
I ~ C]illrlls ~roccedlng ~.yond the inltlal rc~lew would & subject to peer review by an expert retained by the tx)ard in the health providers field of c~pv-tiw. If [he first expert decided the claim had no merit, a second expert would be retained. If two independent expert reviewers detcrm]ncd that the claim did not have merit, it would be dismissed. If the claim were detemlined to have merit by a health care provider, the par-tics w(wld proceed thrtwgh a settlement prtwedure w ith the assistance of a hearing examiner (9). T() pr[)mote settlement, the systcm w(mld include financial pcrmlt]es ftw pm-ties refusing a settlement offer that a hcarin: examiner detemlines is reas(mable (9). Very few clalms w (Juld get a full hearing bcft~rc the muhcal h~ard.
consent law; and limiting noneconomic damages. The new standard of care would also be amended to take into account the resources available to the physician, a factor not explicitly considered today (9,23). Though many claims would be resolved with minimal physician involvement, the proposal would increase patients access to compensation. Thus, physicians may find themselves subject to more claims. Some experts believe, however, that claims might not increase without a consumer outreach program (23). The proposal retains the negligence standard and establishes a stronger link between malpractice claims and professional licensing. Each finding of negligence would be investigated by the medical board. This investigation might consist merely of a review of the file maintained by the medical board on that physician (e.g., previous liability determinations, settlements, disciplinary actions) to determine if a disciplinary investigation were warranted. The proposal also requires malpractice insurers to report to the medical board all cancellations, terminations, and decisions not to renew coverage (9). It is difficult to predict how physicians behavior might change in response to such an administrative system. The elimination of trials (indeed, the limits on any type of formal hearing) might reduce physicians anxieties about being sued. Physicians should also have greater confidence in the fairness of the system, because it would be run by a medical board with substantial physician representation. Yet a large increase in claims could dampen physicians enthusiasm for the proposal, and stronger links between malpractice decisions and disciplinary actions could create additional pressure to practice defensively.
In a system of enterprise liability, the physician would no longer be personall y liable for his or her malpractice. Instead, the institution in which he or she practices. or the health plan responsible for paying for the services, would assume the physician 1iability. Although some hospitals and staff-
model HMOs already assume liability for their physicians malpractice claims, few health care institutions today are fully liable for all claims originating within their organizations. Enterprise liability would eliminate the costs associated with multiple defendant suits and thereby facilitate settlement. It would promote stronger quality control within institutions and health plans while relieving physicians of some of the psychological burdens of a malpractice suit. Institutions bearing the liability risk would have a greater incentive to evaluate physicians performance. Institutional quality control programs may be a more effective deterrent to poor quality of care than the current malpractice system, because the vast majority of negligently injured plaintiffs do not sue ( 130). A model of an enterprise liability program exists today at the hospitals owned and operated by University of California. Under California law, university hospitals are 1iable for the actions of physicians practicing within their hospitals. When a claim is filed against a staff physic i tin, the general counsel office requests the plaintiff attorney to drop the physician as a party to the suit and make the Regents of the University of California the sole defendant ( 137). In virtually all cases this request has been granted. Consequently, the physician does not play as great a role in the pretrial discovery process, but if the case goes to trial the physician is the primary witness and is required to defend his or her actions (1 37). Other institutions, particularly some teaching hospitals, have similar arrangements (74), Some large teaching hospitals have an arrangement known as channeling, in which the institution and the physicians practicing in the hospital are insured under the same malpractice insurance policy. The physician pays the hospital for the insurance and is often required to agree to a joint defense. In return, the physicians receive favorable malpractice insurance rates and often high coverage limits (108, 142,197). Therefore, even without true enterprise liability, some of the administrative efficiencies of a joint defense already exist in these settings.
The impact of enterprise liability on physician practice is difficult to predict. Because enterprise liability retains the fault-based system and still calls upon physicians to defend their actions, it is unclear whether the psychological benefits of not being personally named in a claim would lead physicians to practice less defensively. To the extent that enterprise liability induces greater oversight of outcomes of care or review of malpractice claims by the enterprise, physicians may still feel pressure to practice defensively so as to avoid at all costs a poor outcome or a claim. To the extent that physicians are good judges of how to improve outcomes, this kind of defensive behavior would be beneficial to patients, though it might also be very costly. The medical profession has not seized the opportunity offered by enterprise liability to be excused as a party to malpractice suits. Some critics claim that enterprise liability threatens professional autonomy ( 148,149). Others doubt that physicians autonomy is really threatened by enterprise liability, because physicians have a great deal of influence over hospital and HMO policies, especially with respect to clinical practices (46). Yet if enterprise liability were implemented at the insurance plan level, the quality control function would be one step removed from the institution in which care is provided. The insurance plan would need to understand the quality control issues at many different institutions. Physicians might resent the suggestions or dictates of outside insurers. Finally, insurers would not be as aware of the physician abilities, skills, and other contributions to the institution, possibly leaving physicians feeling unfairly judged. Enterprise liability could increase the number of suits if patients felt more comfortable suing a corporate enterprise rather than physicians (148, 149). In return for no personal liability, physicians might therefore find themselves witnesses in a
greater number of cases and subject to greater scrutiny from the enterprise in which they provide care. It is difficult to predict the resulting impact on practice.
Some malpractice reform proponents seek to replace the fault-based system with a no-fault system, because they consider the current malpractice system ineffective in reaching its two primary goals: deterrence of poor quality care and compensation of victims of negligent injuries. Presently, very few injured patients receive compensation, and judgments about negligence can be costly and time-consuming. Certain no-fault proposals promise more equitable compensation and create other mechanisms for quality control. Other no-fault proposals address compensation issues only. Limited no-fault systems for birth-related injuries already exist in Florida and Virginia. The Virginia and Florida programs provide compensation for a limited number of obstetric injuries; they do not focus on improving the quality of care. In part this is because many injuries removed from the malpractice system by the Florida and Virginia programs may not be preventable by better quality care. A selective no-fault proposal that would cover a broader range of medical practices is in development. This proposal, which is as yet untested, would use certain adverse medical outcomes called avoidable classes of events (ACES) as a mechanism for determining liability for selected injuries. ACES could be used both to promote high-quality care and to quickly and objectively determine which patients should be compensated. When an ACE occurred, the patient could be quickly compensated through a nonjudicial insurance process, so ACES are also known as accelerated compensation events. (221).
The Virginia and Florida Birth-Related Injury Compensation Programs

Virginia and Florida have implemented an accelerated compensation program for a selected set of severe neurological birth related injuries. 20 The Virginia program was conceived out of necessity when Virginia malpractice insurers stopped writing any new obstetric policies following a Virginia Supreme Court decision upholding an $8 million obstetric award (236). Florida initiated its program shortly thereafter. Both programs came about in part because high malpractice insurance rates were thought to be responsible for a decline in the availability of obstetric services, especially for low-income people (57).21 Severe neurological injuries were chosen because the issue of causality was so muddled and malpractice insurers were frustrated by the difficulty of defending against allegations that the injury resulted from the physician's actions (or inactions) during the delivery. Many of these claims involve very large damages. Both programs stop short of being true no-fault systems. In both states, there must be evidence that the injury resulted from deprivation of oxygen or a mechanical cause during delivery (Va. Code Sec. 38.2-5008 ( 1989); Fla. Stats. Sec. 766.302 ( 1991 )).22 The Virginia and Florida programs have been operational for approximate] y 5 years. Many more claims have been brought under the system in Florida than in Virginia, probably because Florida promotes its program more aggressively ( 174, 236).23 Malpractice insurance for obstetricians is now readily available in both Virginia and Flori-
da; at least in Virginia, the program can be credited with keeping malpractice insurers in the market. The impact on malpractice insurance premiums is unclear (57,90). No studies have documented whether these programs have increased the availability of obstetric care, but the Virginia act successfully required participating physicians to work with the commissioner of health to develop a program to provide obstetric services to lowincome patients (Code of Va. Sec. 38.2-5001 (1987 )).24 Because the subset of injuries that falls under these programs is so small and the link between these injuries and physician practices so unclear, removing personal 1iability for the specified birthrelated injuries probably has very little impact on defensive medicine and may have little impact on the quality of care as well.
Accelerated Compensation Events

Under this system, medical experts would identify categories of medical injuries that are generally avoidable when a patient receives good medical care. Patients experiencing an ACE would be automatically compensated through an administrative system. Compensation would be paid either by the physicians insurer or another responsible organization. Because ACES would not account for all claims, the ACE proposal would have to operate within a larger injury compensation system, which could be the existing fault-based malpractice system or some alternative fault-based approach. Non-ACE claims could be resolved through the tort system or ADR (220).
Experts have developed 146 ACES for general surgery, orthopedic surgery, and obstetrics, but the list is still being revised.25 Examples of ACES include: complications secondary to anticoagulant therapy in preparation for surgery, consequences of misdiagnosis of breast malignancy, complications from failure to diagnose and treat hypoglycemia in a newborn, complications to infant(s) from syphilis during pregnancy that was unrecognized during prenatal care, complications to infant(s) from fetal distress (including brain damage) that was unrecognized or untreated during attended delivery, and certain complications or injuries resulting from surgical procedures, including failing to remove a foreign body from the surgical site (221). In a sample of 285 hospital obstetric claims in 24 states, the obstetric ACES accounted for 52 percent of claims, with a disproportionate number of serious injury claims and paid claims involving ACES (25). The primary benefit of ACES may be to promote predictability and consistency in the disposition of claims. ACES are developed by medical experts using epidemiologic concepts of relative avoidability on a population basis (221). In conventional malpractice cases, negligence is based on a lay jurys judgment about an individual incident. It is quite possible that the same adverse outcome will be compensated by one jury but not by another because juries will differ on whether the standard of care was met. Under a system using ACES, the primary analysis would be whether a covered adverse outcome
occurred as a result of certain clinical actions (e.g., the patient is blind following the occurrence of air embolism during a surgical procedure to remove acoustic neuroma). Compensation would be provided once a factual finding was made that certain clinical events have occurred. There would be no judging of whether an individual physicians actions were clinically acceptable or met a standard of care. Use of ACES should allow a greater number of injured patients to be compensated more quickly and for less administrative expense 26 (221). It would not be necessary to determine anew in each case the proper standard of care and to evaluate the physicians behavior against this standard. The proposal also contemplates 1imiting noneconomic damages, which are often high and sometimes inconsistent because of (he difficulty of assigning monetary values to injuries such as pain and suffering (236). Limiting these damages would decrease the open-endedness of damage awards and perhaps ease physicians anxieties about medical malpractice (see chapter 2). ACES could also have an impact on defensive medicine. ACES could relieve physicians of the psychological burden of a process that retrospectively judges their actions. Using ACES would eliminate the process of finding that the physician's actions did not meet the standard of care. Without the threat of a trial in which personal blame is assigned by a finding of negligence, there could wel1 be less motivation to practice defensive medicine in the clinical situations surrounding ACES. Because ACES are based largely on the occurrence of bad outcomes in certain clinical situations, physicians should have little incentive to perform tests or procedures that they know will not improve outcomes but merely document care
2S The unpublished I ist of research ACES were provided h) OTA have not been published previ(msly.
for review only; OTA wiis
not permitted to publish the 1]s( or any ACES tha[
b According (o (me estirnatc, $0.50 to $0.60 of every dollar spent on the nmlpriic[iccs} stem gtw~ to ailn]lnistrat]~c expenses, the majority of which are legal expenses ( 106). The cl iminati(m of a proceeding to ~stiibl ]sh laul t imd ~iiusiit]on sh( ~ulil Icad t( ) ii sign I fic;int rtduc[ I( m In iidn, in
istrative costs.
in these cases (221 ). Thus, ACES should reduce the occurrence of certain wasteful defensive medi cal procedures. ACES could also promote good defensive medicine (i.e., defensive medicine that improves outcomes). Implicit in the development of ACES is the judgment that the injury could probably have been prevented with good medical care. Thus, physicians and institutions would have incentives to change their practices and implement quality control systems to prevent the occurrence of such events. Because ACES are based on outcomes, however, they might not always provide the physician with upfront guidance on the clinical decisions necessary to avoid these outcomes. In addition, because ACES are based on statistical avoidability y, a single ACE event would not necessarily be a sign of poor care. The authors of ACES say that use of the concept would not stimulate defensive medicine, because most ACES do not involve adverse events that can be avoided by diagnostic testing (20.2 18). Indeed, one of the criteria for- designation of certain adverse medical outcomes of an ACE is that doing so will not distort medical practices or lead to unnecessary testing. Yet some ACES developed to date do involve omissions of care, including missed diagnosis. For example, complications resulting from misdiagnosis of early breast malignancy has been specified an ACE. In designating this situation tin ACE, the developers of the proposal made an explicit judgment that physicians should have strong incentives to diagnose breast cancer. even if there are many false negatives. Any determination that such an ACE occurred implies that the doctor omitted necessary procedures: thus, the physician may still feel personally responsible.27 In such situations, some physicians may feel compelled to do tests of marginal medi -
cal benefit to reduce the risk of an adverse outcome to as close to zero as possible. On the other hand, if the physician is already practicing defensively because he or she believes that any adverse outcome might lead to litigation. then having this situation removed from the fault-based liability system might reduce some of this concern. In other words. if physicians are more comfortable with an ACE compensation system than with the tort system, designation of complications from certain missed diagnosis as an ACE could relieve some anxiety about potential liability. Finally. the impact of ACES on defensive medicine might depend upon how they fit into the larger system of compensation for medical injuries. ACES will not cover all medical practices. If an ACE compensation system were layered onto the existing malpractice system, physicians might not know whether particular clinical situations could result in ACE liability or tort 1iability. More importantly, ACES might not address the c1inical situations that trigger the most defensive medicine. Since the claims that remain in the tort system might still trigger defensive medicine, the developers of ACES have suggested that an ADR system for the remaining cases would eliminate some aspects of the tort system that may drive defensive behavior+. g., adversarial proceedings, juries. or potential] y large damage awards ( 24). As discussed earlier, however, the impact of ADR on defensive medicine is not at all clear.
DEFENSIVE MEDICINE AND HEALTH CARE REFORM

Economic them-y predicts that the threat of liability will drive individuals (or organizations) to invest in activities to prevent 1iability until the cost of prevention exceeds the expected cost of 1iability (255). In a fee-for-service system, physicians
often do not bear the costs of extra tests and procedures and may sometimes get paid more money when they order them. Without counterincentives to investment in prevention of liability, extra tests or procedures would be ordered even when their marginal benefit to the patient is extremely low. As long as the investment in 1iability prevention is free or even remunerative, reducing the threat of liability might do little to change the incentive to practice defensive medicine. On the other hand, changes in health care payment that increase the cost to the clinician (or to the organization) of avoiding liability would probably reduce defensive medicine. Several current health care proposals embrace the concept of managed competition.28 Under such a system, health plans would have strong incentives to limit total expenditures on behalf of their enrollees. Plans and their physicians would weigh the cost of performing a test or procedure against the potential savings in liability costs that performing such tests can be expected to provide. Without the threat of liability, or some other effective method of quality assurance, managed competition could create too great an incentive to do less for the patient, leading to lower quality of care. Under certain health care reform proposals, physicians could find themselves in the position of not being reimbursed for delivering care they believe is appropriate. Since the legal system does not now and probably will not recognize negative reimbursement decisions as evidence of the standard of care, physicians could be caught between competing pressures of bearing the cost of procedures or bearing the risk of liability (84).
CONCLUSIONS
Conventional tort reforms that tinker with the existing process for resolving malpractice claims
while retaining the personal liability of the physician are more likely to be successful in limiting the direct costs of malpractice-claim frequency, payment per paid claim, and insurance premiums-than in altering physician behavior. Indeed, 20 years ago, when the frequency of malpractice suits, payments per paid claim, and premiums were much lower than today, physicians still claimed to practice defensive medicine frequently. Greater use of practice guidelines in malpractice proceedings may reduce defensive medicine, because practice guidelines may offer physicians specific guidance about what the courts will accept as the standard of care. Although guidelines will not be a panacea, they are likely to play an increasingly important role in malpractice proceedings. Under a payment system that seeks to reduce costs, guidelines can be used both to specify appropriate clinical actions and to shield physicians from liability for adverse outcomes occurring when the guidelines have been followed. The overall impact of guidelines on defensive medicine will probably be 1imited, however, because of the tremendous uncertainty in medical practice. Alternative dispute resolution relieves the physician of the prospect of a trial. An arbitrator may possess greater technical expertise in malpractice than a lay jury, and the process may be less adversarial and quicker. If concern about the competency of juries and the trial process is the primary motivator of defensive medicine, then this reform may have an impact on behavior. Physicians may find the process more rational and fair and therefore more readily accept the result. However, the process still involves judgments about the appropriateness of the physician clinical decision. In addition, ADR may increase the number of claims and strengthen the link between malpractice claims and professional licensing. Both of
g Managed compeif/ion in this report refers ttl a system in which each c~msumcr cht}t~ses am~mg competmg health plans that offer a standard set of benefits at different prices ( i.e., premiums). Competiti(m among plans for patients on the basis of price as well as qua] ity would presumabl y force plans to 1(NA for opp(wtunities to c1 iminate wasteful or only marginally useful services. In addition, the Admin istrati(ms prop)sal imp)ses caps (m increases in premiums. It is tnpectcd that plans w Ill exert: re:iter Influence on their participating doctors and hospitals to be more ctlst-ctmsctcws in making cllnical ciecisitms.
these factors could offset the psychological benefit of eliminating a trial. Enterprise liability removes personal liability, but the physician is still likely to be called as a witness to defend his or her clinical decision if the case goes to trial. The main advantages of this concept are reduction in administrative costs associated with multiple defendants and the prospect for better quality control systems. In addition, physicians may have less anxiety when they know they will not be named in any suit. Selective no-fault using ACES would probably limit physicians involvement in the claims process, and a payment to the plaintiff would not necessarily imply that the physician was negligent. However, the criteria used to develop ACEsi.e., generally avoidable adverse events does leave some notion of personal responsibility in the system. As for defensive medicine, it is not clear that ACES would address many of the situations in which much defensive behavior occurs. If these
situations are left in the tort system, the motivation to practice defensively may not change, Consequently, the impact of selective no-fault on defensive medicine is unpredictable. The projected impacts of these new malpractice reform proposals on physician behavior are based on logic, not experience. Missing is information about what aspects of the malpractice system drive physician behavior. If physicians mainly want to avoid jury trials, then ADR may be sufficient to reduce defensive medicine. On the other hand, if physicians are distressed about any process that questions their clinical judgment, then reforms retaining a fault-based system may not result in changes in physician behavior. Health care reform may also have an impact on defensive medicine. A different health care financing arrangement may create financial disincentives for practicing defensive medicine, making tort reform unnecessary or even unadvisable.
Appendix A: Method of Study
his assessment grew out of the debate over the role of medical malpractice in increas-
ing health care costs. Specifically, Congress was concerned that the threat of medical malpractice liability was leading physi-
cians to order many unnecessary tests and procedures. According to some estimates, these extra tests and procedures were adding $20 billion to national health care expenditures. Congressman Bill Archer, Ranking Republican Member of the House Ways and Means Committee, and Senator Orrin Hatch, member of the Office of Technology Assessments (OTAs) Technology Assessment Board, requested that OTA provide an independent estimate of the cost of defensive medicine. Additional request letters were received from Senator Edward Kennedy, Chairman of the Senate Committee on Labor and Human Resources; Senator Hatch, Member of the Senate Committee on Labor and Human Resources: Congressman John Dingell, Chairman of the Committee on Energy and Commerce; and Senators Charles Grassley and Dave Durenberger, members of OTAs Technology Assessment Board. In addition, the Congressional Sunbelt Caucus requested that OTA examine the question of whether Medicaid obstetric patients were more likely than other obstetric patients to sue their physicians.
PLANNING WORKSHOP
OTA often convenes workshops of experts in the field to assist in devising a research plan and to provide technical assistance. On November 26, 1991, before the project staff was dedicated to the assessment, OTA held a workshop to devise a method for assessing the extent of defensive medicine. The workshop included primarily academicians who had extensive knowledge of medical malpractice and defensive medicine. (Participants are listed at the end of this appendix.) This half-day workshop led OTA to a working definition of defensive medicine. The workshop
also led OTA to conclude that it would be impossible to come up with a single point estimate of the cost of defensive medicine. Instead, OTA decided to focus on a more qualitative estimate. It was also decided that physician surveys using clinical practice scenarios would not only be a feasible way to quantify defensive medicine but would also be a significant empirical contribution to research on defensive medicine.
ADVISORY PANEL
Every major OTA assessment is advised by a panel of outside experts and representatives of relevant interest groups. The role of the advisory panel is to provide guidance in project planning and to review OTAs findings. The panel is not responsible for the final contents of an OTA assessment and OTA does not attempt to get a consensus from the panel. OTA chose a 17-member advisory panel with representatives from medical and legal academia; physician organizations, including representatives of the American Medical Association; a consumer advocacy group; and a practicing plaintiffs attorney. Randall Bovbjerg, senior research associate at the Urban Institute, a Washington research organization, served as panel chair. The panel convened twice during the projectonce on August 13, 1992, to give advice about research priorities and directions for the project; and again on September 27, 1993, to review our empirical findings and to finalize the analysis plan. The panel was subsequently provided a draft of our final report for review.
willingness to cooperate, but limitations of time and resources precluded an extension of the survey to this group. Each College convened an expert panel to help devise clinical scenarios, assisted us in obtaining a sample of its member physicians, supported our survey with a letter of endorsement, helped gather the data for analysis, and generally gave freely of staff time. Without their generous efforts, OTA would not have been able to conduct the physician surveys that make up a large part of the basis for our conclusions about defensive medicine. OTA also retained the services of a clinical consultant, Dr. Jeremy Sugarman. In total, OTA surveyed 5,865 physicians; the average response rate was 60 percent. For the analysis of the data, OTA worked closely with Russell Localio of the Center for Biostatistics and Epidemiology, School of Medicine, Pennsylvania State University. An analysis plan for the surveys was discussed at the advisory panel meeting in September 1993.
ADDITIONAL EMPIRICAL RESEARCH

In addition to its clinical scenario studies, OTA commissioned several other empirical studies of defensive medicine. Initially, OTA had hoped to do a large-scale statistical analysis of the relationship between malpractice risk and use of health care services. However, after concerted efforts to identify good sources of data on malpractice claims and health care utilization, it became clear that adequate data were not avail able to conduct such analysis on a national level. OTA then considered doing a smaller analysis of this type using comprehensive hospital discharge and malpractice claims data from Floridathe only state for which such data were readily available. On June 2, 1993, OTA convened a special workshop to identify indicators of defensive medicine in a hospital setting that could be measured using discharge data abstracts. Workshop participants included seven practicing physicians with expertise in analysis of utilization data, an economist from the Center for Health Policy
CLINICAL SCENARIO SURVEYS

Having decided to use clinical scenarios to survey physicians about their medical practices and the influence of liability concerns on those practices, OTA contacted several physician professional societies for guidance. The American College of Cardiology, American College of Surgeons, and the American College of Obstetricians and Gynecologists were very willing and enthusiastic to provide assistance. In addition, the American College of Emergency Room Physicians expressed a
Appendix A: Method of Study | 97
Studies at Georgetown University, and an individual familiar with the two Florida databases. (Participants are listed at the end of this appendix.) Although the workshop produced a short list of potentially useful indicators, OTA ultimately decided not to proceed with the analysis because the data available were not adequate to control for a variety of other factors known to affect utilization of the procedures. Without those controls, the results of the analysis would have been highly equivocal. OTA was able to find several researchers with data that could be used to measure defensive medicine. OTA funded Dr. Laura-Mae Baldwin and other faculty from the Department of Family Medicine, University of Washington, to examine the impact of medical malpractice liability experience on the treatment of low-risk obstetric patients by a sample of obstetricians and family practitioners in Washington State. OTA also funded Drs. Kevin Grumbach and Harold Luft of the University of California at San Francisco to examine whether increases in malpractice premiums in New York State led obstetricians and family practitioners to drop their obstetric practice. Finally, OTA commissioned several papers on medical malpractice and defensive medicine. The major contract papers prepared under this assessment are listed at the end of this appendix. Almost all of these contract papers were sent out for external review.
practice Costs, was published in September 1993. OTA reviewed statutes and surveyed state attorneys general to document the current status of malpractice reform in the states. The paper also examined the best evidence regarding the impact of malpractice reforms on the indicators of the direct costs of the medical malpractice system malpractice insurance premiums, payments per paid claim, and frequency of claims. In addition, in response to the request from the Sunbelt caucus, OTA issued a background paper in August 1992, titled Do Medicaid and Medicare Patients Sue Physicians More Often Than Other Patients ? This paper was a review of the available literature on whether Medicaid and Medicare patients were more 1ikely to sue their physicians than patients with private health insurance or patients without insurance.
REPORT REVIEW PROCESS

Prior to completing the draft, the main contract papers were sent out for review. The 10 contract papers were reviewed by a total of 58 outside reviewers. After completing the reviews of the contract papers, a preliminary draft of OTAs report was prepared and submitted for review and critique to the advisory panel in January 1994. The advisory panel was given 10 days to review the draft for problems that were important enough to warrant attention before an outside review draft was prepared. Several panel members sent comments, but very few substantive changes were necessary before the final review draft. In February 1994, a formal draft for outside review was prepared and sent to both advisory panelists and a selected group of 80 outside reviewers. The reviewers (including the panelists) represented a wide range of expertise and interests. In all, OTA received a total of 47 sets of reviews, including those from advisory panel members. OTA rev iewed and revised the draft as appropriate in re sponse to these comments.
BACKGROUND PAPERS
As OTA began its research on defensive medicine and medical malpractice, it became apparent that there were many important issues relating to medical malpractice reform that might be of interest to Congress during the health care reform debate. OTA decided to issue a separate background paper on medical malpractice reform. The background paper, Impact of Legal Reforms on Medical Mal-
Laura -Mae Baldwin, M. D., MPH Assistant Professor Department of Family Medicine Seattle, WA Randall R. Bovbjerg, J.D. Senior Research Associate The Urban Institute Washington, DC Laura Morlock, Ph.D. Professor and Division Head Health Finance and Management Johns Hopkins University School of Public Health and Hygiene
Baltimore, MD
Richard Kravitz, M.D.

Consultant The Rand Corporation Santa Monica, CA
Russell Localio, J. D., M.P.H.

Research Associate Center for Biostatistics and Epidemiology School of Medicine Pennsylvania State University Hershey, PA
Brad Cohn, M.D.

President Physician Insurers Association of America San Francisco, CA
Lawrence R. Tancredi, M. D., J.D.

Director Health Law Program University of Texas Health Science Center Houston, TX
David Sundwall, M.D.

American Healthcare Systems Institute Washington, DC
Appendix A: Method of Study | 99
Jack Hadley, Ph.D. Center for Health Policy Studies and Department of Family and Community Medicine Georgetown University Washington, DC Richard L. Kravitz, M. D., M.S.P.H. Department of Medicine UCLA School of Medicine Los Angeles, CA Jeremy Sugarman, M. D., M.P.H. Division of Internal Medicine The Johns Hopkins University School of Medicine Arthur Garson Jr., M. D., M.P.H. Duke University Medical Center Durham, NC John Ayanian, M.D. Department of Health Care Policy Harvard Medical School Boston. MA
Jeffrey Whittle, M.D. Division of General Internal Medicine University of Pittsburgh Pittsburgh, PA
James R. Ligas, M. D., Ph.D.

Department of Surgery University of Connecticut School of Medicine Farmington, CT Mark I. Taragin, M. D., M.P.H. Division of General Internal Medicine Robert Wood Johnson Medical School University of Medicine and Dentistry of New Jersey New Brunswick, NJ James Phillips, R.R.A. Center for Health Statistics Department of Health Care Administration State of Florida Tallahassee, FL
L. M. Baldwin, M.D., M. P. H., L.G. Hart, M. D., M. Lloyd, A. R.M., M. Fordyce, M. A., and R.A. Rosenblatt, M. D., M. P.H., Department of Family Medicine, University of Washington, Seattle WA, Malpractice Claims Exposure and Resource Use in Low Risk Obstetrics, Nov. 21, 1993. P. Ehrenhaft, M. P. H., Lake Oswego, OR, Do Medicaid and Medicare Patients Sue Physicians More Often Than Other Patients? August 1992. K. Grumbach, M. D., D. Peltzman-Rennie, B. A., and H.S. Luft, Ph. D., Institute for Health Policy Studies and the Department of Family and Community Medicine, University of California, San Francisco, CA, Charges for
Obstetric Liability Insurance and Discontinuation of Obstetric Practice in New York, Dec. 7, 1993.
P.A. Glassman, M. D., M.Sc., L.P. Petersen, M.S., Bradley, M. A., B.A., J.E. Rolph, Ph. D., RAND, Santa Monica, CA,
The Effect of Malpractice Experience on Physicians Clinical Decision -Making, Dec. 1993.
M. Hall, J. D., Wake Forest University School of Law and Bowman Gray School of Medicine, The Effect of Insurance Coverage Law on Defensive Medicine, Aug. 25, 1993.
P. Jacobson, J. D., M. P. H., and C.J. Rosenquist, M. D., RAND, Santa Monica, CA, The Diffusion of Low Osmolality Contrast Agents: Technological Change and Defensive Medicine, March 1993.
E. Kinney, J. D., M. P. H., The Center for Law and Health, Indiana University School of Law, Indianapolis, IN, The
Impact of Proposed Tort Reform on the Medical Malpractice System and Physician Behavior, June 1993.
T.B. Metzloff, J. D., Duke University School of Law, Defensive Medicine and the Use of Medical Technology:
Physician Involvement in Medical Malpractice Litigation, Jan. 1994.
L. Modock, Ph. D., and F.E. Malitz, MA.S., School of Hygiene and Public Health, The Johns Hopkins University,
Baltimore MD, Short-Term Effects of Tort and Administrative Reforms on the Claiming Behavior of Privately Insured, Medicare, Medicaid and Uninsured Patients, Sept. 30, 1993.
G. Ruby, Consultant, Garret Park, MD, The Role of Medical Education in Promoting the Practice of Defensive
Medicine, Apr. 28, 1993.
Appendix B: Acknowledgments
The development of this report benefited from the advice and review of a number of people in addition to the Advisory Panel and Contractors. OTA would like to express its appreciation to the following people for their valuable assistance. Brian K. Atchinson Maine Bureau of Insurance Augusta, ME John Ayanian Harvard Medical School Boston, MA Pamela P. Bensen Oxford, ME Elizabeth Brandt California Department of Health Services Sacramento, CA Theodore Briggs Medical Mutual Insurance Co. Portland, ME Kathy Bryant American College of Obstetricians and Gynecologists Washington, DC Joel Cantor Robert Wood Johnson Foundation Princeton, NJ Rebecca Chagrasulis ME Joan Corder-Mabe Department of Health Richmond. VA Susan Cox University of Texas Dallas, TX Daniel Creasey Risk Management Foundation of the Harvard Medical Institutions Cambridge, MA Myles P, Cunningham Evanston, 11, Robert diBenedetto Baton Rouge, LA Leonard S. Dreifus Philadelphia, PA James Fabian The New York Hospital New York, NY
Casco,
David Chinoy Jacksonville. FL Mark Cohen University of California, Davis Medical Center Sacramento, CA Mark M. Connolly Chicago, IL Milton 1. Cooper Kaiser Permanente Medical Care Program Oakland, CA Philip H. Corboy ABA Special Committee on Professional Liability Chicago, IL
Don Fadjo Gold River, CA Richard M. Flowerdew Portland, ME Thomas F. Floyd Bangor. ME Larry Ganslaw American College of Cardiology Bethesda, MD Arthur Garson, Jr. Duke University Medical Center Durham, NC Lillian Gaskin American Bar Association Washington, DC Paul Geauvreau Geauvreau & Blackburn Lewis ton, ME Edward Goldman University of Michigan Ann Arbor, ME Paul Gluck Miami, FL Susan Goold University of Michigan Medical Center Ann Arbor, MI Margaret 6. Griffin Waterville, ME Jim Gross University of California, Davis Medical Center Sacramento, CA Jack Hadley Georgetown University Washington, DC
Martin Hatlie American Medical Association Chicago, IL Kenneth Heland American College of Obstetricians and Gynecologists Washington, DC Alnoor Hemani The Johns Hopkins University Baltimore, MD James R. Hines Chicago, IL Robert E. Hirshon Dummond, Woodsum, Plimpton & MacMahon Portland, ME Chris Hogan Physician Payment Review Commission Washington, DC E. Kirk Huang Potomac, MD Tricia Hunter Sacramento, CA Richard D. Judge Ann Arbor, MI Graham Kalton Westat, Inc Rockvillc, MD Mary Kaynor Risk Management Foundation of the Harvard Medical institutions Raynard Kington RAND Santa Monica, CA
Leonard 1. Kranzler Chicago, IL John A. LaCasse Medical Care Development Augusta, ME Ann Lawthers Harvard School of Public Health Boston, MA James R. Ligas University of Connecticut School of Medicine Farmington, CT Barbara A. Luke Portland, ME John Lundberg Regents of the University of California Oakland, CA Robert Markowitz Federation of Jewish Philanthropies Service Corp. New York. NY William C. McPeck Augusta, ME Irving Meeker Portland, ME Daniel Mendelson LewinVHI, Inc. Fairfax, VA Daniel L, Meyer Augusta, ME Fabrizio Michelassi Chicago, IL Jean M. Milligan Augusta, ME
Appendix B: Acknowledgments | 103
Don Harper Mills Professional Risk Management
David L. Roseman Chicago, IL Robert Rubin Lewin-VHl, Inc. Fairfax, VA Kevin A. Schulman
GeorgetownUniversityMedical
Corp.
Long Beach, CA Penny S, Mills American College of Cardiology Bethesda, MD David Nagey Baltimore, MD Joseph P. Newhouse Harvard University Boston, MA Neil Newton Augusta, ME Michael A. Nocero, Jr. Orlando, FL Paul Nora American College of Surgeons Chicago, IL Jeffrey OConnell University of Virginia Law School Charlottesville, VA DonaId C. Overy Bloomfield, MI James Phillips State of Florida Department of Health Care Administration Tallahassee, FL Stuart M. Poticha Chicago. IL O. Howard Reichman Maywood, IL John Rizzo Yale University School of Medicine Ncw Havce, CT
Mark 1. Taragin Robert Wood Johnson Medical School New Brunswick, NJ George O. Thomasson COPIC Insurance Company Englewood, CO Ronald J. Trahan Augusta, ME Sandy Ulsaker St. Paul Companies, Inc. St. Paul, MN Harvey F. Wachsman Pegalis and Wachsman Great Neck, NY Walter Wadlington University of Virginia Charlottesville, VA Kate Wallace Consumer Product Safety Commission Washington, DC Sylvan L. Weinberg Dayton. OH Jeffrey Whittle University of Pittsburgh Pittsburgh, PA Debra Williams Prospective Payment Assessment Commmission Washington, DC Sidney Wolfe Public Citizen Health Research Group Washington, DC Stephen Zuckerman The Urban Institute Washington, DC
Center Washington, DC Ruth Schwartz Rochester, NY David Shapiro Physician Payment Review Commission Washington, DC Tonya Sharp Chapel Hill, NC Rebecca Shaw Des Moines, IA Frank Sloan Duke University Durham, NC Robert Smith San Francisco, CA Robert Stolt Lipman & Katz Augusta, ME Robin Stombler American College of Surgeons Washington. DC James L. Stone Cook County Hospital Chicago, IL Selma Taffel National Center for Health Statistics Washington, DC
Appendix C: The Impact of Nonclinical Factors on Physicians Use of Resources
lthough clinical factors are still the most important determinants of physicians clinical decisions (61 ), research suggests that a number of nonclinical factors also influence physicians diagnosis and treatment choices, among them malpractice liability concerns. The influence of malpractice risk on physician behavior is discussed at length in chapters 2 and 3 of this report. This appendix briefly reviews some evidence on the influence of other nonclinical factors in physicians decisions about resource use.
FINANCIAL INCENTIVES
Several studies have found that diagnostic testing and other service use is lower in prepaid and salaried practice settings than in fee-for-service systems (64,92, 136, 140,208). Other types of financial incentives have also been shown to have an effect on use. For example, a study of physicians in a forprofit chain of ambulatory care centers found that use of laboratory tests and x-rays increased substantially (23 and 16 percent, respectively) after physicians were offered bonuses for increasing patient care revenues (91 ). Other studies have shown that physicians respond to reduced fees by increasing the volume of services they perform ( 189,195,205). Finally, physician ownership of testing and treatment facilities has been associated with increased resource use (93,2 14,245).
AWARENESS OF AND SENSITIVITY TO TEST COSTS

A number of studies have suggested that physic cians are sensitive to costs when ordering tests and prescribing treatments (1 1,65,97,1 33,225). For example, one study found that physicians who were given information on test costs ordered 14 percent fewer tests per patient than physicians who are not given cost information (225). In a study of test use for hypertensive patients, cost to patient was cited as an important reason for not ordering electrocardiograms (65). An OTAsponsored clinical scenario study found that physicians with greater levels of cost-consciousness (measured by using attitude scales) reported they would use fewer resources than physicians with lower levels of cost-consciousness (73).
INSURANCE COVERAGE
Insurance status of patients has also been associated with willingness to use resources. This may reflect physicians sensitivity to both their own and patients financial concerns. Research has consistently shown that hospitalized patients with private insurance coverage stay in the hospital longer and receive more procedures (especially more discretionary and high-cost procedures)
Appendix C: The Impact of Nonclinical Factors on Physicians Use of Resources | 105
than patients with Medicaid coverage or patients who lack health insurance (238). For example, a recent study of low-income pregnant women in Massachusetts (82) found that public health insurance coverage increased their likelihood of undergoing a Caesarean section. Service-specific financial incentives did not play a role, as the public insurance program paid a global fee regardless of type of delivery. Another study of patients with ischemic heart disease in California hospitals found that, after controlling for demographic, clinical, and hospital characteristics, the frequency of coronary revascularization procedures (coronary artery bypass surgery and coronary angioplasty) was almost two times higher in fee-for-serv ice patients than in health maintenance organization (HMO) and Medicaid patients (121). The same study also found that the rate of coronary revascularization increased more quickly in fee-for-service and HMO patients than in Medicaid patients between 1983 and 1985 (121 ).
lities were three times as likely as patients in hospitals without those facilities to undergo coronary angiography. After adjusting for clinical factors, the existence of onsite catheterization facilities was the strongest predictor of use of coronary angiography (66). A similar study in New York corroborated these results, finding that AMI patients admitted to facilities offering cardiac catheterization, bypass surgery, and angioplasty services were two to six times as likely as patients in facilities not offering them to receive these services (18). Another study of physician practice patterns suggested that some of the otherwise unexplained variation may be influenced by differences in physicians enthusiasm for using certain interventions (39). This enthusiasm may be a byproduct of other related issues, such as greater familiarity with the technique, a role in its pioneering, or availability of technology.
OTHER FACTORS PROXIMITY OF TECHNOLOGY

Some studies have shown that the availability of technologies influences their use. For example, a recent study of acute myocardial infarction (AMI) patients in Seattle found that patients admitted to hospitals with onsite cardiac catheterization faciOther factors associated with physicians use of tests and procedures include physician specialty and training (62, 123,126, 175,257,259), practice setting (e.g., managed care versus unrestricted private practice) (135, 136) and patient expectations (144).
1 For example. one study found that internists and family practitioners ordered rm~re diagm~stic tests than general practititmers (62).
Appendix D: Methods Used in the OTA Clinical Scenario Surveys

his appendix summarizes the methods used to develop and analyze surveys of three physician professional societies. The Office of Technology Assessment (OTA) cooperated with three physician associations to conduct clinical scenario surveys of association members by mail from February through August of 1993. 1 The three physician associations, listed in the order in which they were surveyed, were:
editing. For the other two surveys, these tasks were shared between OTA and the respective association. OTA performed all final data editing, processing, and analysis. Strict rules protecting respondent confidentiality were observed by all participating organizations.
SURVEY INSTRUMENT CONTENT AND FORMAT

The main goal of each survey was to ascertain, as unobtrusively as possible, the extent to which physicians would choose malpractice concerns from among several reasons for selecting or rejecting specific diagnostic or therapeutic procedures in treating specific hypothetical cases. Respondents were presented two or three specific clinical scenarios appropriate to their respective specialties. Introductory letters from both the physician association and OTA described the purpose of the survey in general terms, without mentioning malpractice or defensive medicine. Two separate instruction pages, including an example scenario, explained how the questionnaire should be
The ACS component actually involved two separate surveys: one for general surgeons and the other for neurosurgeons. Thus, four distinct surveys were actually conducted. The questionnaire for each survey was developed jointly between OTA and the respective association. ACC maintains an ongoing practice panel sample of its practicing members and conducted its own mailout, data entry, and initial data
] Dr. Russell Loealio of Pennsylvania State University and Dr. Jeremy Sugannan of Duke University were consultants to OTA on the design
of the survey instruments and statistical analysis. Dr. Loealio designed the sampling plan and data analysis components of the surveys and participated extensively in the analysis and interpretation of the survey results. Dr. Sugarman consulted on the development of the fomlat and content of the clinical scenarios used in the surveys.
Appendix D: Methods Used in the OTA Clinical Scenario Surveys | 107
all survey instruments are appendix available from
presented between 3 and 13 diagnostic or therapeutic procedures, including the option of essentially doing nothing; and presented four reasons for choosing or rejecting
Scenario Format and Content

The clinical scenarios in each of the four surveys were developed by an expert panel containing from seven to 10 members of the relevant physician association (selected by association leadership in cooperation with OTA project staff and consultants). During a one-day meeting at the association headquarters, the panel members were asked to brainstorm at least 20 clinical scenarios in which concerns about liability would be expected to strongly influence clinical actions. Then the panel was asked to select from these candidates three or four scenarios that would be expected to elicit the strongest defensive medicine responses for inclusion in the survey. Panel members were also asked to create a q control version of each selected case by adding or deleting one or more key clinical indicators (e.g., a result from a laboratory or radiologic test) that would, in the opinion of the panelists, greatly reduce the likelihood that malpractice concerns would be cited as the primary reason for choosing any action. OTA staff and consultants then selected and refined the final scenarios, with input from association leaders and panel members. Each questionnaire was pretested on a small sample of association members who were excluded from the final survey. Each clinical scenario:
q
described the patients demographic characteristics, symptoms, vital signs, and initial diagnostic test results;
~ In place of other, the ACC survey used institutional pr(~t(}ct)ls/prtJfessit~ nal guidel incs as the fifth rcas{m, Although c~thcr was listed as a procedure on the ACC qucsti(mnaire. the assoeiati(m did not c(xle the presence or absence of a w rltten rcsp(mse In that box. C(msequently, OTA was unable u) Include t~ther pr(wdure in its analysis of the ACC data.
expected to be frequent); the other received the control scenario and one or two common scenarios, The specific combination of scenarios presented to each group of respondents is summarized in table D- 1. Special analytical problems posed by this case-control design are discussed later in this appendix.
Open-Ended Version of the ACS General Surgeon Survey

A supplemental sample of general surgeons was sent an open-ended version of each ACS clinical scenario used in the main survey of general surgeons (case versions onlysee previous section). The open-ended questionnaire offered no specific reasons for choosing procedures. Instead, a blank space was provided beside each procedure, in which respondents could fill in their own reasons, in their own words, for choosing the
Association American College of Cardiology American College of Surgeons General surgeons
Group Group 1 (case) Group 2 (control)
Scenario 1 (case/control) Chest pain case Chest pain control
Scenario 2 (common) Syncope Syncope
Group 1 (case) Group 2 (control) Group 1 (case) Group 2 (control) Group 1 (case)
Rectal bleeding case Rectal bleeding control Back pain case Back pain control Perimenopausal bleeding case
Breast pain Breast pain Head injury Head injury Breast lump Complicated delivery Breast lump Complicated delivery
Neurosurgeons American College of Obstetricians and Gynecologists
Group 2 (control) SOURCE Off Ice of Technology Assessment, 1994
Perimenopausal bleeding control
3 These characteristics were jointly selected by staff members of OTA and the relevant physician association, c(msidenng not only differences among the specialties, but also the unavailability of some characteristics in each associations membership database (also see the section on sampling, below). Most importantly, the following measures were not available: in the ACC survey, the number of years in practice; in the ACS survey, geographic region; and in the ACOG survey, whether the respondent held an academic appointment. Also, the categories of the respondents usual practice setting differed SI ightly from survey to survey, reflecting the different categories used by the associations themselves. Finally, as measures of the number of years in practice, ACS used years since board certification, whereas A COG used years of membership m the association. These unavoidable variations in measurement reduced the comparability of results from the f(mr surveys.
s q s
Malpractice Concern, Cost Consciousness, and Discomfort with Clinical Uncertainty.
SAMPLING
OTA and its consultant, Russell Localio, developed a sampling plan for each survey, with input from association staff. Sampling fractions were based on statistical power calculations for twosample comparisons, with rough assumptions about the survey response rate and the number of respondents who would choose clinical procedures primarily because of malpractice concerns. Sampling fractions varied across sampling strata to ensure adequate numbers of respondents in each subclass of physicians. Each physician association then drew a sample from its membership database according to detailed instructions provided by OTA. Population sizes, sample sizes, numbers of respondents, and response rates for each survey are displayed in table D-2. All four surveys targeted only association members who, according to the membership database:
m
Additional items regarding satisfaction with medical practice were developed by OTA and Dr. Goold to serve as decoy items in the surveys. Each attitude item offered five response categories, scored as 1 through 5 (respectively): strongly agree, agree, unsure, disagree, and strongly disagree. The Malpractice Concern scale contained five items, the Cost Consciousness scale contained six items, and the Discomfort with Clinical Uncertainty scale originally contained three items. However, OTA did not use the entire Uncertainty scale for the ACOG survey (only one Uncertainty item was included in that survey), after receiving written comments from ACS respondents regarding how similarly worded the items were. Each respondents scores (1 through 5) on all the items in a given scale were summed to obtain a total scale score.4 To make a 5 represent agreement rather than disagreement (so that the summed scores would measure agreement), the item scores were reversed by subtracting them from 6, except where an item was worded negatively (e.g., where agreement represented low malpractice concern). The scores for the five-item Malpractice Concern scale thus ranged from 5 (minimal malpractice concern) to 25 (maximal malpractice concern), whereas the six-item Cost Consciousness scale ranged from 6 (minimal cost consciousness) to 30 (maximal cost consciousness). The three-item Uncertainty scale, which ranged from 3 (minimal discomfort with clinical uncertainty) to 15 (maximal discomfort with clinical uncertain y), was computed on] y for ACC and ACS respondents because the ACOG survey contained only one Uncertainty item (see above).
s s s
had earned the degree of either Medical Doctor (MD) or Doctor of Osteopathy (DO). were not in residency training, were not retired, were board certified in the relevant specialty, and were currently practicing in the United States.
All four samples were drawn from the associations membership database through systematic stratified random sampling. However, due to 1imitations of the membership databases and special association concerns, the stratification factors differed somewhat from survey to survey. These and other features of the four samples are summarized in table D-3. Other differences also existed among the four samples:
q
ACC used its existing Professional Practice Panel, a standing sample of about 1,500 practicing members who are occasional] y surveyed
Survey
American College of Cardiology b American College of Surgeons General surgeons
Group Total Case Control
Population 11,541
Sample 622 311 311 3,004 2,401 1,196 1,205 603 859 427 432 1,983 1,002 981
Respondents a 352 184 168 1,793 1,412 739 673 381 503 252 251 1,230 634 596
Response rate 566 591 540 597 588 618 559 63.2 586 59.0 581 623 633 608
Total 12,972 Closed-ended Case Control Open-ended Total Case Control Total Case Control 1,384
Neurosurgeons
American College of Obstetricians and Gynecologlstsc
20,832
a The numbers of respondents shown In this table may differ silghtly from the scenario-specific numbers of respondents shown in text tables
3-2 through 3-5 in chapter 3 because a few respondents completed one scenarlo but not the other b The American College of Cardiology sample included only adult cardiologists c The American College of Obstetrlclans and Gynecologists sample excluded gynecological oncologists and reproductive endocrinologists SOURCE Office of Technology Assessment, 1994
ing each respondents sampling stratum is described in the next section. The ACS survey included physicians practicing in U.S. territories (Puerto Rico, Guam, etc.), whereas the ACC and ACOG surveys did not. The ACC and ACS surveys contained government-employed physicians, including military doctors (except those practicing overseas,), whereas the ACOG sample excluded military physicians.
In the ACS and ACOG surveys, the numbers of case and control respondents were not equal, for two reasons. First, for ease of data processing, random assignment of respondents to the case or control group (every other respondent) was performed within each sampling stratum rather than throughout the entire sample. In the ACC survey, the overall numbers of case and control respondents were equal; however, the case respondents were selected by taking a simple random subsample of the overall sample, without regard to the stratification variable of geographic region. Second. response rates differed slightly between the
Feature Stratification factors
American College of Cardiology a Census region
American College of Surgeons Academic appointment yes, no Year of first board certification post-1981, 1972-81, pre-1972 Practice setting solo, group, medical school, hospital, other
American College of Obstetricians and Gynecologistsb Geographic region (4 regions) Years in ACOG < 6, 6-10, 11-20, >20 Gender 32
Number of strata
30, plus two additional, one for some missing data, the other for all missing data None
Special exclusionsC
U S trust territories
U S trust territories, military, Public Health Service May 271 1993 June 30, 1993
First mailing Second mailing
Feb. 4, 1993 Feb. 23, 1993
March 4, 1993 . None
a The ACC survey Included only adult cardiologists b The ACOG survey excluded gynecological oncologists and reproductive endocrinologists c For general exclusion criteria see text
case and control groups. The numbers of case and
control respondents therefore differed within each region by as much as 11 percent. Differences in response rates were corrected by reweighting the respondents according to case/control group and sampling stratification factors (e.g., region ).
DATA PROCESSING
ACC conducted its own mailouts, data entry, and initial data editing. Individual respondents were tracked, and initial nonrespondents were sent another copy of the questionnaire. In the ACS and A COG surveys, the general procedure was as follows: The association providcd OTA with mailing labels for sampled members. OTA produced the questionnaires and mailed them with a prepaid return envelope addressed to the association's Washington. DC. office. Upon receiving the responses, the association photocopied them and shipped the originals to OTA for processing.
There were several variations on this basic process between the ACS and ACOG surveys. The identity of individual ACOG respondents was tracked by ACOG personnel by means of a relatively unobtrusive identification number printed on the first page of the questionnaire as well as on the mailout label and the postage-paid return envelope. As noted earlier, a second mailing of the ACOG questionnaire was sent to initial nonrespondents. Five such respondents apparently returned both questionnaires, for they had duplicate ID numbers. We allowed one of each pair of data records for these duplicate respondents to be randomly discarded through a computer sorting and matching routine (see the next section). ACS, on the other hand, preferred not to track individual respondents; thus, no followup mailing of the questionnaire to initial nonrespondents was possible. To track the sampling stratum to which the respondent belonged, OTA devised a method of unobtrusively tracking the respondents sampling stratum by varying the features of the return mailing label.
Eighty-nine respondents did not use the return envelope provided but instead sent the questionnaire back in an irregular envelope (i e., without the tailored mailing label). For 61 of these respondents (68.5 percent), ACS was able to use the return address or postmark on that envelope to identify the sampling stratum to which the respondent belonged. ACS kept the individual identity of these 89 respondents confidential. OTA made no attempt to identify any individual respondents and analyzed all data separately from any identifying materials.
DATA EDITING AND ENTRY

The major rules used to edit the data in all four surveys are summarized in a technical appendix available from OTA upon request. OTA and the associations made concerted efforts to refine the questionnaire instructions based on responses to the three pretests. Despite these precautions, respondents in all four surveys sometimes provided answers that were inconsistent with the instructions; these responses required editing. The most frequent q *error was failure to circle no for unselected clinical options or failure to check the reasons for circling no for such options. That is, many respondents circled *yes only for selected options and checked reasons for choosing only those options. Fortunately, this kind of error did not substantially affect the analysis, which focused on respondents who chose yes for a given option (see the next section). Another very infrequent error (on the order of 0.1 to 0.6 percent of all responses) that would affect the analysis was failure to check reasons for c1inical options where yes was circled. These respondents (who circled yes for an option but failed to check any reasons for doing so) were included in the denominator when the percentage of choosers (see below) was calculatedimplying that, if the respondent had cited a reason, it would
A p)ssible exceptitm here is the clinical opti(m of refer to surgeon, which appeared in the ACOG breast lump scenario. Physicians who chose this option had possibly decided not [o intervene themselves (depending on whether they chose (o Perfornl other procedures I isted in the scenario), and thus may have been engaging in negative defensive medicine. On the other hand, referral to a surgeon can imply an expectati(m that relatiwly aggressive and p~tentially c(~stly mterventi{m will he undertaken, and may thus reflect p~sitive defensive medicine.
hence chose no for the do nothing option). Thus, for each procedure. the denominator was the group of respondents who chose yes for that procedure. Excluded from this denominator were not only respondents who explicitly chose no, but also those who chose neither yes nor no (i.e., those who had left that entire row of the questionnaire blank). Respondents who did not respond at all to a given scenario, but who responded to other parts of the questionnaire, were excluded only from the analysis of that particular scenario. Of this denominator (respondents who chose yes for a given procedure), the numerator of greatest interest was the group of respondents who checked malpractice concerns as a reason for choosing that procedure (with either a single- or double-check). However, the malpractice responses could not be analyzed in isolation, because another reason (usually medical indications) was often cited along with "malpractice concerns by the same respondents. This meant that these respondents were selecting procedures not only on the basis of malpractice concerns, but also in part because they felt that the procedures were at least somewhat medically indicated. These combinations of responses suggested that differing degrees or levels of defensive motivation were being expressed in these surveys. each of which required a separate measure. Tables showing the distribution of responses by clinical procedure and reason for procedure choice are presented in a technical appendix available from OTA upon request.
definition of defensive medicine to the least restrictive definition. The measures are cumulative, i.e., the least restrictive measure (measure 6) includes respondents meeting measures 1 through 5. Measure 1: DOUBLE check for malpractice concerns AND NO check at all for ANY other reason. Measue 2: Measure 1 PLUS a DOUBLE check for malpractice concerns AND NO check for medical indications (single checks for other reasons are allowed). Measure 3: Measure 2 PLUS a DOUBLE check for malpractice concerns AND a SINGLE check for medical indications (single checks for other reasons are allowed). Meusure 4: Measure 3 PLUS a SINGLE check for malpractice concerns AND NO check for medical indications (single or double checks for other reasons are allowed). Measure 5: Measure 4 PLUS a SINGLE check for malpractice concerns AND a SINGLE check for medical indications (single or double checks for other reasons are allowed). Measure 6: Measure 5 PLUS a SINGLE check for malpractice concerns AND a DOUBLE check for medical indications (single checks for other reasons are allowed}. The rationale underlying these measures is as follows. Defensive medicine is most strongly indicated when the respondent cites only malpractice
To gauge the extent of defensive medicine expressed in these surveys, we constructed six measures of defensive medicine based on specific patterns of reasons given for choosing a given diagnostic or therapeutic procedure. These response patterns involved particular combinations of check marks for malpractice concern s, medical indication s, and other reasons. The six measures are 1isted in order below from the most restrictive
concerns and no other reason (measure 1). Even though there are no medical indications or patient expectations for performing the procedure, the physician would perform it anyway, solely out of fear of malpractice litigation. This response should be infrequent, since it is arguably a violation of medical ethics. Citing other reasons, particularly medical indications, dilutes the degree of defensive medicine indicated. Moreover, a single check for *malpractice concerns represents a weaker level of defensive medicine than does a double check. These six measures of defensive medicine were computed on the basis of two different denominators, thereby creating two separate measures that provide two different interpretations of the results for a given procedure in a given scenario: Percentage of choosers: Here the denominator was the number of respondents who would
choose the procedure (i.e., circled *yes). The measure of defensive medicine was thus the percentage of respondents choosing the procedure who cited malpractice concerns as a reason for doing so.
tion of positive defensive medicine: physicians performing procedures primarily, but not necessarily solely, out of fear of malpractice litigation (see chapter 2). Tables showing the distribution of responses on all six measures of defensive medicine are presented in appendix E.
Percentage of scenario respondents: Here the

denominator was the total number of respondents to the overall scenario. The measure of defensive medicine was thus the percentage of all respondents who, when presented with the scenario, would choose the procedure for defensive reasons. This percentage was much smaller than the percentage of choosers and represents the frequency with which concerns about malpractice would be expected to enter clinical decisions in situations of this type.
With six separate measures of defensive medicine, the number of comparisons between the percentages for various groups of respondents (case versus control, academic versus nonacademic, etc.) would have been unmanageable. Consequently, for such comparisons we used only measure 3 (double-check for malpractice concerns, with single checks allowed for any other reasons, including q medical indications). This measure most closely approximated OTAs working defini-
All data were treated as coming from a sample survey with unequal probability of selection in a stratified (cross-classified) population (114,117, 124). Compared with simple random sampling, the effect of weighting the data to compensate for unequal probability of selection is generally to increase the variance of estimators, while the effect of stratification is generally to reduce that variance. Data from the surveys supported our reliance on this general experience. Test analyses using methods for 1 ) unweighed simple random samples, 2) weighted simple random samples, 3) unweighed stratified samples, and 4) weighted stratified samples demonstrated that the effects of stratification and weighting in fact did offset each other to a considerable degree. Variances were not increased markedly owing to the use of unequal weights in this sampling design. Rates (or proportions) of respondents who would choose a clinical procedure, and of those who did so primarily because of malpractice concerns (see above), were calculated using sampling weights that compensated for nonresponse as well as unequal probability of selection across the sampling strata. Wherever possible, variance estimates and confidence intervals for these point estimates used methods that are common in survey analysis and assumed both stratification and sampling without replacement (i.e., use of the finite population correction). Where possible, comparisons among subclasses of respondents were made by differences in rates (or proportions), and calculations of the variance of those differences took into consideration the sampling design. In several instances we departed from the use of rate differences in
rate differences.6 This approach allowed us to take advantage of the stratified sampling design, where the numbers of respondents were sufficient, and alternative methods where the numbers of respondents were too small to justify large-sample techniques. Tests for rate differences and odds ratios are comparable for these data.
the independent variable was the respondents group (case or control, labeled group in the model); and the control variable was whether or not the re-
Case-Control Comparisons
Comparisons of responses to the case and control scenarios presented special problems. First, the design of the surveys did not permit withinphysician comparison of case and control responses, because the same respondents could not be given both the case and control scenarios without possibly revealing our purpose. The case and control responses were thus independent, thereby reducing the efficiency of the case-control comparisons (greater variances for the same sample size). Second, although the case and control groups were each stratified random samples, they could differ in systematic waysmost importantly, in their propensity to cite malpractice concerns. As a proxy for this control variable, we examined whether or not the respondent double-checked "malpractice concerns for one or more procedures in the common scenario for each survey (the scenario received by every respondent in a given surveysee table D-l). This adjustment was computed as follows. Where the numbers of respondents were adequate (again, at least 10 in each category), we used sample-weighted logistic regression, as implemented in the PROC LOGISTIC procedure in SUDAAN ( 193), to perform the equivalent of stratified 2-by-2 contingency table analysis in which:
s
the dependent variable was whether or not the respondent double-checked *malpractice concerns in the case-control scenario (labeled response in the model shown below);
6 EKcept where m~ted, the calculations arc evict iKMs ratios and the]r acc(m)pany ]ng c~act 95-percent c(mfldcncc intcrk als and p-values, cxmlpukxl accxmilng t{) [hc mc[h(ds of Mehta, Gray, and Pak?l ( 156).
the PROC FREQ procedure and Cochran-Mantel Haenszel statistics on the normalized weighted data in SAS (203) (see table D- 1 ).7 The DIFFVAR option in PROC DESCRIPT in SUDAAN (193) was used to test the significance of difference in
mean attitude scale scores between respondents who double-checked malpractice concerns in the common scenario for each survey (see table D-1 ) and those who did not.
The cxmlrmm sccn;iric}s were used f~~r this analysis SC) thai it w(mld be based on all respondents in a g]ven survey.
Appendix E: Detailed Results of the OTA Clinical Scenario Surveys
T
q
I
s
he main features of the results of the Office of Technology Assessment (OTA) clinical scenario surveys 1 are highlighted in chapter 3. This appendix contains: for each clinical option in each case scenario, weighted frequencies and percentages of responses using six different definitions of defensive medicine (tables E1 through E8); and a comparison of attitude scale scores between respondents who cited malpractice concerns as the primary reason for choosing procedures and those who did not (table E-9).
The following additional results are presented in a technical appendix available from OTA upon request: unweighed frequencies and percentages of respondents who single-checked or double
s
These results were ctm]piled in collatxwati{m with Dr. Russell Lt~ali{J of Pennsylvania State University.
1118
Appendix E:
Detailed Results of the OTA Clinical Senario Surveys | 119
Of clinical actions chosen, percent done for malpractice concerns Most restrictive definition Scenario b / clinical action Syncope(N=346) Admit Exercise ECG Stress thallium 2 D/M mode Doppler Color flow doppler Transesophageal echo Hotter monitor Tilt table Carotid doppler EEG Brain MRI Chest pain (N=162) Discharge home w/NSAID Admit and observe Admit/obtain enzymes Admit and obtain ECG Exercise ECG Stress thallium 2 D/M mode Doppler Color flow doppler Transesophageal echo Angioqram
a
Least restrictive definition Definition 3 Definition 4 12 5% 80 23 11 22 32 00 42 06 162 163 289 Definition 5 Definition 6 57 2% 278 310 249 216 192 299 272 94 39.8 489 53.0 13.0 556 62.3 624 477 307 391 346 24.1 00 1000
% of respondents who chose the clinical action
Definition 1
Definition 2
10 8%
71 23 11 22 32 00 33 00 137 14.9 203
Results are weighted to reflect the total population of professional society members on which the survey sample was based (see append!x D for detads) b Results shown for case versions of scenarios only (see appendix D for explanation) KEY 2 D/M = 2 dimensional/time-motion mode, ECG = electrocardiogram, EEG = electroencephalogram, NSAID = nonsteroldalantl-mf lammatory drug NOTE Starting with defmltlon 1, the data are cumulatwe. q Defmlhon 1 Malpractice Concerns double checked with no checks for any other reason q Defmltlon 2 defmltlon 1 phxs Malpractice Concerns double-checked no checks for Medical Indlcatlons, but single checks for other reasons allowed s Defmltion 3 defmltlon 2 phx Malpractice Concerns double-checked, a single check for Medical Indlcatlons, and single checks for other reasons allowed q Defmltlon 4 deflnltlon 3 phs Malpractice Concerns single-checked, no checks for Medical Indlcatlons, but single or double checks for other reasons allowed s Defmltion 5 defmmon 4 phs Malpractice Concerns single-checked, Medical I ndlcahons single-checked, and single or double checks allowed for other reasons = Defmltlon 6 defmmon 5 phs Malpractice Concerns single-checked, Medical Indlcahons double-checked, and single checks for other reasons allowed SOURCE Office of Technology Assessment, 1994 Data compiled m collaborahon with Dr Russell Locaho of Pennsylvania State Uruversify
NOTE Starting with Dehmtlon 1, the data are cumulatwe q Deflnltlon 1 Malpractice Concerns double-checked with no checks for any other reason s Deflmtlon 2 deflndlon 1 PM Malpractice Concerns double-checked, no checks for Medical Indlcahons, but single checks for other reasons allowed s Defln!tlon 3 defmmon 2 phx Malpractice Concerns double-checked, a single check for Medical Indlcatlons, and single checks for other reasons allowed q Deflnltlon 4 defmltlon 3 PIUS Malprachce Concerns single-checked, no checks for Medical Indlcatlons, but single or double checks for other reasons allowed s Defmltion 5 defmltlon 4PM Malpractice Concerns single-checked Medical Indlcahons single-checked and single or double checks allowed for other reasons s Deflnltion 6 definition 5 PIUS Malpractice Concerns single-checked, Medical Indlcatlons double-checked and single checks for other reasons allowed SOURCE Off Ice of Technology Assessment, 1994 Data compiled m collaboration with Dr Russell Locaho of Pennsylvama State Unlvers!ty
SOURCE Office of Technology Assessment, 1994 Data compiled m collaborahon wth Dr Russell Locallo of Pennsylvania State Unwerslty
Appendix E:
Of clinical actions chosen, percent done for malpractice concerns Most restrictive definition Scenario b / clinical action /0 of respondents who chose the clinical action
0
Least restrictive definition Definition 3 Definition 4 Definition 5 Definition 6
Definition 1
Definition 2
Back pain (N=252) Lumbosacral x-ray CT MRI Other

a
244 3,4 12,6 93
1.2 00 57 00
24 00 57 0.0
139 298 16.0 0.0
169 36.2 16.0 00
204
36.2 337 00
503
51 1 520 00
Results are weighted to reflect the total population of professional society members on which the survey sample was based (see appendix D for details) b Results shown for case versions of scenarios only (see appendix D for explanation) KEY CT = computed tomography, MRI = magnetic resonance image
NOTE Starting with Defmihon 1, the data are cumulatwe q gef:n:t:~n ; ~v~a~p:a~;ice C~~~erfi~ d~ubie-~f-lecked WI[h rIO ch=-ks for any olner reason ~ Definition 2 deflrmtlon 1 PIUS Malprachce Concerns double-checked, no checks for Medical Indications, but single checks for other reasons allowed s Defirutlon 3 defmltton 2 phs Malpractice Concerns double-checked, a single check for Medtcal Indications, and single checks for other reasons allowed = Defmihon 4 deflnltlon 3 phs Malpractice Concerns single-checked, no checks for Medical Indlcatlons, but single or double checks for other reasons allowed s Definition 5 definition 4 phs Malpractice Concerns smgie-checked, Mechcal Indlcatlons single-checked, and single or double checks allowed for other reasons q Definihon 6 defmtton 5p/us Malpractice Concerns single-checked, Medical Indlcahons double-checked, and single checks for other reasons allowed SOURCE. Off Ice of Technology Assessment, 1994 Data compkd m collaboration wth Dr Russell Locaho of Pennsylvania State Unwerslfy
Appendix E:
@.J-oJ-30 0 0 0
Results are weighted to reflect the total population of professional society members on which the survey sample was based (see appendix D for details) b Results shown for case verslms of scenarios only (see appendix D for explanation) KEY D & C = dllatlon and curettage NOTE Starting with Defrutlon 1, the data are cumulatwe 9 Defmmon 1 Malpractice Concerns double-checked with no checks for any other reason q Defmltlon 2 defmthon 1 PIUS Malpractice Concerns double-checked, no checks for Medical Indlcatlons, but single checks for other reasons allowed s Defmltlon 3 defmltlon 2 plus Malpractice Concerns double-checked, a single check for Medical Indlcahons, and single checks for other reasons allowed q Defmmon 4 defmlhon 3 phx Malpractice Concerns single-checked, no checks for Medical Indlcatlons, but single or double checks for other reasons allowed q Defmltlon 5 defmtlon 4 ph.fs Malpractice Concerns single-checked, Medical Indicahons single-checked, and single or double checks allowed for other reasons s Deflnmon 6 defmlhon 5 ph Malpractice Concerns single-checked, Medical Indlcahons double-checked, and single checks for other reasons allowed SOURCE Office of Technology Assessment, 1994 Data compiled m collaborahon with Dr Russell Locallo of Pennsylvama State Unwersl!y
Appendix E:
Mean attitude scale scores Respondents citing malpractice concerns as primary reason for choosing one or-more clinical actionsa
Attitude
scale/scenario
All other respondents
950/o confidence Difference
Malpractice concern (5 items, range 5-25) ACC syncope (N-339) ACS breast pain (N-1 377) ACS head trauma (N-492) ACOG breast lump (N-1 192) Cost consciousness (6 items, range 6-30): ACC syncope (N-340) ACS breast pain (N -1 369) ACS head trauma (N - 488) ACOG breast lump (N-1 185) Discomfort with clinical uncertainty (3 items, range 3-15) ACC syncope (N-330) ACS breast pain (N - 1,368) ACS head trauma (N-486)
15.55 1442 1774 1403
1841 1874 2191 1842
1890 1886 2263 1846
-049 -012 -072 -004
794 770
955
* Statlstlcally slgnlflcanl at the p ~ 05 level a Excludes respondents who dld not complete the attitude questionnaire b Because the ACOG survey Included only one Item on discomfort with cllnlcal uncertainty ratner lharl three (see appendix D), ACOG attitude scale scores for discomfort with cllnlcal uncertainty are not Included In the comparlsorl
KEY ACC = American College of Cardlologsts ACOG = American College of Obstetrlclans and Gync~coloq Ists ACS American College of Surgeons
SOURCE Off Ice of Technology Assessment 1994 Data analyzed In collaboration with Dr Russell Local Io of Pennsylvania State Unwerslty
Appendix F: Estimates of the Costs of Selected Defensive Medical Procedures
survey data is not possible, for two reasons. First, the OTA surveys covered only 13 clinical scenarios, nine of which were deliberate] y designed to increase the likelihood of a defensive response (see chapter 3 and appendix D). (The other four were control scenarios, in which concern about liability was expected to be much less important.) Second, reliable incidence and cost data could not be readily obtained for most of the procedures listed in the OTA scenarios. OTA was able to estimate the annual cost of defensive medicine associated with procedures selected in two scenarios: a complicated obstetrical delivery (American College of Obstetricians and Gynecologists (ACOG) survey) and head injury in a 15-year-old (American College of Surgeons (ACS) neurosurgeons survey). These two scenarios were chosen because they exhibited a high frequency of defensive practice and because national incidence and cost data were available.
rejecting the overall cost of defensive medicine based on the Office of Technology Assessment (OTA) clinical scenario
APPROACH
OTAs basic approach was. first, to obtain national data on the incidence of the clinical condition described in the chosen scenario. Such data are not available for patients who match each and every demographic and clinical characteristic of the simulated patient. OTA applied the results to patients in a similar age range who fit the broader diagnoses into which the simulated patient might be classified. Second, the estimated incidence of the clinical case was multiplied by the percentage of OTA survey respondents who chose the selected procedure primarily due to malpractice concerns (see table 3-3 in chapter 3), resulting in a national estimate of the annual frequency with which the procedure was performed primarily because of malpractice concerns in similar situations. Finally, OTA obtained estimates of the average cost of performing the procedure and multiplied this per-service cost by the estimated number of defensively performed procedures to arrive at an estimated aggregate annual cost of defensive
Appendix F:
Estimates of the Costs of Selected Defensive Medical Procedures | 29
Number of live births complicated by prolonged labor or dysfunctional labor among women aged 30 to 39 in 1991 a
45,126
Incremental cost of Caesarean section over and above normal delivery in 1991c
X $3,106
performance of the procedure. These calculations, discussed in further detail in the following two sections, are displayed in tables F-1 (Caesarean section in a complicated delivery) and F-2 (diagnostic radiology for head injury in young people). These estimates do not necessarily represent any savings in health care costs that might accrue from elimination of defensive medical practices. Ordering or performing a procedure defensivelycould save health care costs in the future if poor outcomes are avoided or the patient condition is managed better. OTA assumed that such savings would be negligible in the scenarios used here.
CAESAREAN DELlVERY IN A COMPLICATED LABOR
Annual number of head injuriesa Proportion of head injuries that are apparently minorb Annual number of apparently minor head injuries Proportionof emergency room visits for head injury in persons aged 5 to 24 in 1992C Annual number of apparently minor head injuries in persons aged 5 to 24
1,975,000 x 070 -1,382,500 X O 3837168 -530,488
Cervical spine x-ray:

Annual number of apparently minor head injuries among persons aged 5 to 24 (see above) Proportion of ACS neurosurgeon respondents who chose cervical spine x-ray primarily because of malpractice concerns in the head trauma scenariod Annual number of cervical spine x-rays performed primarily because of malpractice concerns, for apparently minor head injury in persons aged 5 to 24 Estimated private Insurance reimbursemente for cervical spine x-rayg in 1992 2. Aggregate cost of defensive cervical spree x-ray for apparently minor head injury in persons aged 5 to 24 in 1992 530,488 x o 112
-
59,415 x $72
-$4,277,880 530488 x 0218

- 115,646
Computed tomography (CT) scan of head:

Annual number of apparently minor head Injuries among persons aged 5 to 24 (see above) Proportion of ACS neurosurgeon respondents who chose CT scan of head primarily because of malpractice concerns in the head trauma scenariod
Annual number of CT scans of the head performed primarily because of malpractice concerns, for apparently minor head inlury in persons aged 5 to 24 Estimated private Insurance reimbursement e for CT scan of the head in 1992
h
x $315 3 Aggregate cost of defensive CT scan for apparently minor head injury in persons aged 5 to 24 in 1992 -$36,428,490 Total annual cost of defensive radiology for apparently minor head injury in persons aged 5 to 24, 1992 (sum of aggregate costs for: 1) skull x-ray, 2) cervical spine x-ray, and 3) CT scan of head, shown above)
a
= s 44,791,143
J F Kraus, Epldemlology of Head injury Heacf/r?/uV, 3rd Ed Cooper, P R (ed ) (Balhmore Wlhams & Wilkins 1993), data from 1985-87 National Health interview Survey b M Ellastam, E Rose, H Jones, et al Utlllzatlon of Dlagnostlc Radlolog[c Examlnatons In the Emergency Department of a Teaching Hospital, The Journal of Trauma 2061-66 1980 c Consumer Product Safety Commmon. Nahonal Electronic Inlury Surveillance System, unpu blmhed data obtained from Kathryn Wallace Congressional Relatlons Specialist ~J S Consumer Product Safety Commlss[on, Jan 3, 1994 Data are for all head mlunes presenting In an emergency room, for all Ievelsof severity and all causes associated with all consumer products (excluding motor vehicles and publlc transportation) The pro~rtton was calculated by summtng the number of vlslts for ages 5 to 14 and 15 to 24 and dwldmg this s~m by the total number of vrslts d See table 3-3 m chapter 3 e Private insurance costs were estimated using Medicare data For outpatient hosp(tals, the average Medlcaw reimbursement was dlwded by O 542, obtained by dlwdmg the payment-to-cost ratio computed from Medicare data (O 90) by that from a private multlple-insurer database (MEDSTAT) for 1991 (1 66) (Prospective Payment Assessment Commmslon unpublished data for 1990 but using 1992 reimbursement rules, supplled by Deborah Wllllams, Semor Policy Analyst, Jan 21, 1994 and Feb 3, 1994 ) For physicians offices (and free-standing Imaging centers), the average Medicare reimbursement (Physlclan Payment Rewew Commlsslon, unpublished data for 1992 supplled by Chris Hogan, Prlnclpal Pollcy Analyst, Jan 19, 1994) was dlwded byO 70, the ratio of Medicare to private Insurance fees for phystclan Imagmg serwces (M E Mtller, S Zuckerman, and M Gates How Do Medtcare Physlclan Fees Compare with Private Payers~ Hea/lh Care Fmancmg Rewew 1425-39 1993) The resultlng prwate Insurance reimbursement estlmatesfor outpatient hospital; and physicians offices were averaged weighted bythe proportion of Medicare procedures performed In each setting (private Insurance data on this were not available) f lde~tlfled by Codes 70250 and 70260 m American Medical Assoclatlon Current %ocecfwal Terrnmo/ogy qth Ed (Chcago 1993) The re~mbursement figures for these two codes were averaged weighted by the number of procedures performed for each ~ ldentlfled by Codes 72040, 72050, and 72052 (n Arner(Ca.fi Medical Assoc[atlon, Cur{en/ Procecfufa/ Terfrrmo/ogy 4th Ed (Chicago, 1993) The reimbursement figures for these three codes were averaged, weighted by the number of procedures performed for each h Identified by code 70450 m American Medical Association Current Procedura/ Terrnmo/ogy, 4th Ed (Chicago, 1993) This code IS for CT scan of head or bran without contrast material whch IS used to detect tumors rather than blood The reimbursement figures for thm code for outpatient hosplfals and physicians offices were averaged, weighted by the numbers of procedures performed In each setting SOURCE Office of Technology Assessment, 1994
Appendix F:
Estimates of the Costs of Selected Defensive Medical Procedures | 131
good beat-to-beat variability. Estimated fetal weight is 7.5 lbs. and clinical pelvimetry is adequate. The patient is fatigued and can no longer push.
DIAGNOSTIC RADIOLOGY FOR HEAD INJURY IN YOUNG PEOPLE
National incidence data for women aged 30 through 39 for calendar year 1991 were obtained from birth certificate data compiled by the National Center for Health Statistics (250). Two kinds of delivery complications that most closely fit the simulated patient were prolonged labor and dysfunctional labor. OTA divided the number of live births in the selected age category (30 to 39) involving these complications by the total number of live births for which the nature of any birth complications was known (250). This gave the rate of each complication in births to women in the selected age range. OTA then multiplied this rate by the total number of live births to women in the selected age range to obtain the total number of live births with the selected complications. This number was then multiplied by the percentage of ACOG survey respondents who chose Caesarean delivery primarily due to malpractice concerns (see table 3-3 in chapter 3), giving a national annual estimate of the number of times that a Caesarean de] i very was performed primarily because of malpractice concerns in situations similar to the ACOG scenario. National estimates of the incremental cost of Caesarean delivery over and above those of a normal delivery for calendar year 1991 were obtained from the Health Insurance Association of America (89). OTA multiplied this cost estimate by the estimated number of Caesarean deliveries performed primarily due to malpractice concerns in situations similar to the ACOG scenario. This gave the final aggregate estimate of the national annual cost of defensive Caesarean delivery in complicated deliveries involving prolonged or dysfunctional labor.
History of present illness: A 15-year-old boy fell from his skateboard after riding over a crack in the sidewalk. He hit his head, got up and skated home. Thirty minutes after the fall he told his mother about the incident and she brings him to the ER. In the ER, the patient admits to lightheadedness and some tenderness at the site ofimpact. Physical examination: There is an area of tenderness and swelling at left purietal area. Mental status and neurological exam are normal.
OTA used an estimate of the annual total number of head injuries per year (11 8), obtained from the National Health Interview Survey for 1985-87. OTA then estimated the proportion of all head injuries that are apparently minor. Discussions with clinicians indicated that the clinical features of a head injury (e.g., loss of consciousness, neurological deficit) are more important than its cause (e.g., fall from a skateboard) in determining severity. OTA therefore broadened the basis for this cost projection beyond the cause-specific ACS c1inical scenario to reflect all minor head injuries in young people. A conservative estimate of the proportion of all head injuries that appear to be minor upon clinical examination in the emergency room is available from a study by Eliastam and colleagues (63). In that study, the researchers reported the proportion of all head injuries presenting to the emergency room of a suburban teaching hospital for which diagnostic x-rays were ordered. but that were classified immediately prior to the x-ray as not meeting specified criteria for likely skull fracture. This estimate is conservative because it excludes all head injuries for which x-rays were not or-
1 Althtwgh Ellastam and ct)llcagues (63) used [he tern] medIto/e,gal t~~ characterize such injuries, they did not attempt I(J detu-mlnc w hclhcr
the x-ra}s pcrfomml (m those patwm c(mstitutd de fcnslve medicine.
dered. This proportion was applied to the National Health Interview Survey data to generate an annual estimate of the frequency of apparently minor head injuries. National data on the age distribution of minor head injuries, or even all head injuries, do not exist. However, OTA obtained national data by age group on the number of head injuries (regardless of severity) caused by consumer products (excluding motor vehicles and public transportation) and treated in emergency rooms from the National Electronic Injury Surveillance System (242). The available age categories nearest age 15 (the age of the patient in the ACS head trauma scenario) were 5 to 14 and 15 to 24, which OTA combined into a single category of 5 to 24. Multiplying the estimated number of apparently minor head injuries by the percentage of consumer product-related emergency room visits for head injury among persons aged 5 to 24 gave the estimated number of apparently minor head injuries among persons aged 5 to 24. This number was then multiplied by the percentage of ACS survey respondents (neurosurgeons) who chose each radiologic procedure (skull x-ray, cervical spine x-ray, or computed tomography (CT) scan) primarily due to malpractice concerns in the ACS head trauma scenario (see table 3-3 in chapter 3). This gave a national annual estimate of the number of times that each procedure was performed primarily due to malpractice concerns in clinical situations similar to the ACS scenario. National estimates of the cost of performing each radiologic procedure under Medicare (the only readily available and reliable national data) were obtained from the Physician Payment Re-
view Commission (PPRC) and the Prospective Payment Assessment Commission (ProPAC). Data on average per-service Medicare reimbursement rates for each procedure performed in physicians offices and free-standing imaging centers during calendar year 1992 were obtained from PPRC (187). To estimate the average private insurance reimbursement rate for each procedure, OTA divided these Medicare rates by 0.707, the ratio of Medicare to private insurance fees for physician imaging services found in a recent study by Miller and colleagues (162). Data on average per-service Medicare reimbursement rates for each procedure performed in hospital outpatient departments during calendar year 1990 (but using 1992 reimbursement rules) were obtained from ProPAC ( 192). To estimate the average private insurance reimbursement rate for each procedure, OTA divided these Medicare rates by 0.542, the ratio of Medicare to private insurance fees for all nonfee-schedule outpatient hospital services (1 92). 2 OTA averaged these per-service private insurance cost estimates for radiology services in physicians offices and outpatient hospitals, weighted by the number of Medicare services performed in each setting (private insurance data by setting were not available). This estimated average private insurance reimbursement rate was then multiplied by the estimated number of times that each procedure was performed primarily due to malpractice concerns in situations similar to the ACS scenario. This gave the final aggregate estimate of the national cost of defensive radiologic procedures for apparently minor head injuries among persons aged 5 to 24.
z This ratio was obtained by dividing the payment-to-cost ratio computed from Medicare data (0.90) by tha[ from a private mult]ple-]nsurer database (MEDSTAT) ftw 1991 ( 1.66).
Appendix G: Summary of State Studies on Tort Reforms
Comments ing Office, Medical Malpractice: SIX State Case Studies Show Claims and Insurance Costs Still Rise Despite Reforms, HRD-87-21 (Washington, DC U S Government Printing Office, December 1986) W.P. Gronfein, and E. Kinney, Controlling Large Malpractice Claims The Unexpected Impact of Damage Caps, Journal of Health Politics, Policy and Law 16(3) 441-483, 1991
Insurance premiums, and the cost of resolving
claims in Arkansas, California, Florida Indiana New York and North Carolina from 1980 to 1986 Method: Comparison of trends among states
The study was unable to determine whether tort reforms had slowed the growth in claim frequency, payment per paid claim, or insurance premiums because no data were collected on trends prior to the reforms The methodology did not control for other factors that might affect malpractice claim activity
Method: Statistical regression analysis to determine whether Indiana's $500,000 cap on total malpractice damages lowered the average payment per paid claim for large claims The analysis controlled for the effects of plaintiff's age and sex, year
There was no pre-reform and post-reform comparison of payment levels for malpractice claims The higher mean and median payment per claim may be a result of the operation of lndiana's Patient Compensation Fund, which was passed at the same point as the cap on damages and not the result of the cap on damages Although the average payment per paid claim was higher in Indiana the study could not determine whether Indianas tort reforms resulted in an overall savings in malpractice claims payments , . According to data gathered by the U S Health Care Financing Administration national average premiums increased at a compound annual rate of approximately 12 percent between 1976 and 1985 (51 F R 28772, 28774 57 F R 5903) Therefore California claims costs (a proxy for premiums) Increased at a slower rate after MICRA than national malpractice insurance premiums The reductions in claim costs may be unrelated to MICRA especially since MICRA was not upheld by the courts until 1985, which may have Iimited its impact There may be alternative explanations for the findings for exam,ple
of settlement, severity of injury, and allegations of negligence (e g diagnosis, anesthesia surgery medication patient monitoring, etc. )
California Medical Association, Actuarial Study of Professional LIability Insurance prepared by Future Cost Analysts Newport Beach CA May 31 1985
after 1975 most commercial Insurers were replaced by physlclan-owned companies
Comments Patient Protection, The . Coalition to Preserve MICRA, MICRA Informa- , tion, January 1 1993
s
Physician feesAmerican Medical Association survey Malpractice premiums in CaliforniaPhyslclan Insurance Association of America
The magnitude of the decline may have been overstated by comparing a peak in premium levels (1 976) to a relative trough in premiums (1991) a In addition comparisons of singleyear premiums can be misleading because premiums are based on expected revenue
Malpractice premiums in New York Florida MichiganMedical Liability Monitor National Malpractice PremiumsTlllinghast
s
needs and are often adjusted upward or downward when better Information IS available 1992 average malpractice Insurance premiums were lower in California than in New York, Florida, or Michigan
q
Method: Comparison of trends in California with those in other states and the nation to assess the impact of MICRA reforms Data: Harvey Rosenfeld, California MICRA Profile National per capita health care spending data of a Failed Experiment in U S Health Care Financing Administration and Tort Law Restrictions , the Center for National Health Statistics U S PubVoter Revolt, Los AnIiC Health Service geles CA (no date) .
Estimate of Californias personal health care expendituresCalifornia Almanac (5th Ed 1991)
The study did not control for any other factors in California that may have led to lower insurance premiums or physician fees e g changes in the malpractice insurance market or health care delivery market In 1985 Californias average premium was 65 percent above the national average, therefore, the decline to less than the national average IS noteworthy b The study did not control for other factors that contribute to changes in malpractice and health costs therefore, one cannot conclude that MICRA was solely responsible for lower premiums or moderate growth in health care costs
In 1990 the average California malpractice insurance premium was $7,741 as compared with a national average premium cost of $8,327
Incurred malpractice Insurance losses as a
Average medical consumer price index from Los Angeles, San Francisco, and San Diego Malpractice Insurance premiums, profits, and lossesNational Association of Insurance Commissioners
percent of health care costs declined in California between 1987 and 1990 at a greater rate than in the nation
Methods: Comparison of trends in the measures listed above from 1975 to 1991, and comparison of these measures among states in various years
Study
Academic Task Force for Review of the insurance and Tort Systems, Preliminary FactFinding Report on Medical Malpractice, Gainesville, FL, August 14, 1987. #
Major reported findings The rate of closed claims per 100 physicians
remained stable from 1975 to 1986
s
Comments The study did not do a pre-post reform comparison of trends The 1985-86 reforms were unlikely to have had an effect on the data analyzed because most claims were closed prior to implementation of reforms. The study looked at gross trends in malpractice cost indicators, but made no attempt to assess the individual impact of particular reforms on those Indicators
The
a v e r a g e p a y m e n t per paid claim increased 14.8% per year from 1975 to 1986
Claims with million dollar plus awards accounted for 4.9% of total paid claims in 1981 but 29 1% in 1986
The average cost of defending a claim increased at an annual rate of 17% from 1975 to
1986,
Increases in payment per paid claim were the primary factor driving Increases in premiums in Florida same or similar locality.
Data: Various statistics on the operations of 15 pretrial screening panels in Arizona (Maricopa County), Delaware, Hawaii. Indiana Louisiana, Massachusetts, Montana, Nevada, New Jersey, New Mexico, New York, Pennsylvania, Tennessee, Virginia, and Wisconsin Method:
q s q
There were no comparisons of clalm dlsposltlon prior to the implementation of the panel Because pretrial panels offer plaintiffs a reiatwely Inexpenswe mechanism for screening the merits of a case, their existence may have e n c o u r a g e d pialntlffs with nonmentorlous suits to file This could explain the high rate of decisions for defendants and the low rate of plalntlff appeals
Analysis of data Review of the empirical literature Interviews with pretrial panel administrators and members of state medical societies and state bar associations
q
The long delays In panel hearings may lead some plaintiffs to drop clalms or settle after moceedlng thrrwgh the pre!r!a! ~~reenl ng process
Major reported findings J.K. Mardfin, Medical Malpractice in the State of Hawaii, Department of Commerce and Consumer Affairs, Honolulu Hl, January 1986 Data: 453 pretrial screening panel decisions between 1979 and 1984 in Hawaii Method: Comparison of disposition of pretrial screening panel decision and subsequent disposi tion of claim
s q
The majority of claims were settled or dropped after a panel hearing
In the 328 cases in which no Iiability was found, 3% settled without filing suit and 221 claimants (67%) apparently took no . further action
The majority of claimants took no further action following the pretrial screening panel hearing This indicates that the panel promoted early settlement However, the researchers were not completely confident about the status of the cases they reported as taking no further action They did not know whether plaintiffs were still considering a suit or engaged in settlement negotiations The relatively large number of no-liability panel decisions that resulted in payment to the plain tiff raises a question about the accuracy of the panels decisions
Comments
A majority of plaintiffs who filed suit after a panel decision of no-liability received a payment
s
Data was available on 71 suits filed folIowing a panel finding of no-liability Only 51 were closed by the time the study was completed In 28 cases (55%), plaintiffs received a payment In 10 of these cases, the amount paid to the plain , tiff exceeded $100,000
55%
The average time from filing a claim to the
panels decision was 7/2 months, with claims beinq settled within 1 month Howard, D.A An Evaluation of Medical Liability Review Panels in Arizona State Courts Journal 519-25, 1981 from primary malpractice Insurers in Arizona, 1975 to 1979
s q
Of
Court data:
q
The percentage of malpractice cases that went to trial dropped from 15/0 in 1975 to 6A in 1978 The percentage of stipulated dismissals (indi-
The data set only Included 1 year of data for claims filed prior to the enactment of pretrial screening, and 3 years of claims data postpanel The use of only a slngle year of prepanel
data
IS
inadequate for comparison of trends
Insurance claim data for Arizona, 1975 to 1979

s
cating settlement prior to trial) Increased after 1975 Median time for resolution of claims Increased after panels were Instituted Cases that went through the panel process were slowest q There were significant delays in convening panels and scheduling hearings. Insurance claims data:
q q
Interviews with judges and attorneys in Arizona (circa 1980) Method: Analysis of trends before and after implemnentation of pretrial screening panels in 1976
The decline in the number of trials may result from delay in claim resolution, 27% of claims filed in 1977 and 56% of those filed in 1978 had not been closed by the time the study was completed in May 1980 Changes in patterns of disposition of claims may be a result of changes in the malpractice Insurance market A major shift from commercial to physician-owned Insurance companies occurred at the same time panels were implemented
Probability of payment remained stable Average payment per paid claim similar for screened and nonscreened claims Average cost to the insurer to defend a claim e Increas d Average time to resolve a claim Increased Claim frequency increased after the implementation of the panel (1 978 1979)
q q
Study S. Shmanske, and T. Stevens, The Performance of Medical Malpractice Review Panels, Journal of Health Politics, Policy and Law 11 (3) 525-535, 1986
I
Data and methodology Data: Claims data from two Insurance companies in Arizona prior to (1 972-75) and after (1 976-79) pretrial screening panels were implemented The data set Included only claims that closed within 2 years of filing and claims that were filed within 1 year of the incident Method: Pre-post comparison of differences in
Major reported findings Claim frequency Increased Claims took longer to resolve Probability of payment remained the same There was no overall Increase in average indemnity payment, but claims that closed quickly had higher average payment
q
Comments
There were no controls for other factors that may have led to changes in malpractice claim activity for example, the change from commercial insurer to a physician-owned mutual company, changes in demographics, and na-
tional trends in malpractice claims activity
Participants tended to believe that pretrial screening panels did not promote settlement Pretrial screening Increased the cost of litigation
General dissatisfaction with the operation of the pretrial screening panel system About one-third of plaintiff attorneys said there was no reason to enter settlement negotiations prior to the panel decision arbitration studies Data: 1,353 malpractice claims brought between 1966 and 1975 against Southern California hospitals One group of 8 hospitals had Implemented an arbitration project in which patients were presented with an arbitration agreement upon entering the hospital (the arbitration hospitals) The other group of 8 hospitals did not promote arbitration (the nonarbitration hospitals) Method: Comparison of claims experience in arbitration and nonarbitration hospitals before and after implementation of the arbitration program in 1970
Fewer claims were filed in arbitration hospitals as compared with nonarbitration hospitals
q
Thus, there was potential for response bias in results
U.S. Department of Health, Education and Welfare, Public Health Service, Health Resources Administration, National Center for Health Services Research, An Analysis of the Southern California Arbitration Project, January 1966 Through June 1975, prepared by D H Heintz, HHEW Pub 77-3159 (Washington DC: U.S. Government Printing Office, 1975)
The amount paid per closed claim was lower in arbitration hospitals There was a statistically significant decline in the defense cost per claim in the arbitration hospitals over the period of the study The average length of time to resolve a claim was shorter For arbitration hospitals the time period was measured from the filing of the claim Prior to the initiation of the arbitration project the arbitration hospitals had taken
longer to resolve a claim than the nonarbitratlon hospitals
m
Hypotheses were stated in terms of differences between arbitration hospitals and nonarbitration hospitals in the levels of certain variables (e. g. , the number of malpractice claims) but the test statilstic measures the difference between the two groups of hospitals in the rates of change in those variables A number of hypotheses were tested using a test statistic that appears to be Incorrectly specified. Consequently, the statistical significance-though not necessarily the direction-of the findings must be questioned There was evidence that arbitration hospitals were using more intensive efforts to resolve claims earner in the process
Appendix H: Clinical Practice Guidelines and Malpractice Liability

linical practice guidelines have been hailed as tools that can help reduce defensive medicine, improve the quality of care, and protect health care providers from unpredictable liability by clarifying the legal standard of care (59,101 ,188). Medical professional societies have been developing clinical practice guidelines for some years now. In 1989, Congress established the federal Agency for Health Care Policy and Research (AHCPR), which is charged with conducting medical effectiveness research and developing and disseminating national clinical practice guidelines (249). Despite high hopes in Congress and the Administration and continuing enthusiasm among academics for the clinical practice guidelines movement (30,59), a number of factors are likely to limit the impact of guidelines on medical liability and physician behavior. This appendix examines the potential impact of clinical practice guidelines on medical liability. First, it describes the existing legal standard of care and the current
role of clinical practice guidelines in helping to determine it. Second, it discusses limitations of guidelines as legal standards of care. Third, it describes some state initiatives to promote the use of guidelines in litigation. Finally, it comments on the potential role of guidelines in bringing about more cost-effective medical care as our health care system struggles to contain costs.
CURRENT USE OF GUIDELINES AS LEGAL STANDARDS

Because they are more or less concise statements of what the profession deems to be appropriate care, clinical practice guidelines developed by groups of physicians are clearly relevant evidence of the legal standard of care, which is based on customary practice. In fact, the development and acceptance of national guidelines for hospital care provided impetus for abandoning the strictly local standard of care for hospitals in some jurisdictions. 2 However, factors inherent in both the legal
1 In this appendix, gude/ine refers to a clinical practice guideline itself, and srundardrefers to the legal standard of care. In general practice, as well as in certain places in this appendix, these terms as well as others (e. g., parameter and prorocol) are used interchangeably. z In Cornje/df ~. ibngen, 262 N.W. 2d 684 (Minn. 1!)77), the appeals court detemlined that [he trial court had erred in not admitting Joint C(mmlissi(m on the Accreditation of Hospitals as evidence of the legal standard of care. See also Darling v. Charleston Communi~ Hospifai,
33 ]]1. 2d 326,2 I I N,E. 2d 253 (Ill. 1965) (55).
Appendix H: Clinical Practice Guidelines and Malpractice Liability | 141
system and in guidelines themselves limit the role guidelines currently play in the litigation process.
The Legal Standard of Care

To prove that a medical practitioner committed medical malpractice, a plaintiff must establish: 1) that the provider owed a duty of care to the patient, 2) that the provider breached this duty by failing to provide care that met the applicable standard of care for that practitioner under the specific circumstances, 3) that the patient sustained compensable damages, and 4) that the physicians breach of duty was the proximal cause of those damages. It is in establishing the second element, negligent conduct, that clinical practice guidelines have a potential role. The applicable standard of care in a given case is established through expert testimony. Both the plaintiff and defense counsel call to the stand expert witnesses who testify as to what constituted an appropriate level of care in the patients case and whether or not the defendant physician breached this standard. Expert testimony is based on the experience of the witnesses themselves as well as their knowledge of the literature (which may include textbooks, journal articles, or clinical practice guidelines); hence, the courts defer to the medical profession rather than to some objective or lay standard in determining the scope of a physicians duty to a patient. 3 After testimony has been delivered, it is up to the jury to decide whether or not the physician has breached the standard of care, although in extreme cases the court may
take this decision away from the jury by directing a verdict. Until relatively recently, the legal standard of care was articulated as a strictly local standard:
A physician is bound to bestow such reasonable and ordinary care, skill, and diligence as physicians and surgeons in good standing in the same neighborhood, in the same general line of practice, ordinarily have and exercise in like cases (1 90).
Today, most jurisdictions apply a national standard for medical specialists that allows plaintiffs and defendants access to expert witnesses from outside their locality.4 The specific standard varies from state to state. In some jurisdictions, the standard recognizes situational resource constraints--e.g., a practitioner would not be held liable for failing to perform a magnetic resonance imaging study if no facilities were available (86). Additional safe harbors under the customary standard are the respectable minority rule, which allows practices that deviate from the professional norm as long as they are followed by a respected minority of practitioners;5 and the error in judgment rule, which protects a physician who chooses between two or more legitimate courses of treatment (109).
How Guidelines Are Admitted as Evidence

Courts generally bar written guidelines from being admitted as evidence under the hearsay rule, which prohibits the introduction of out-of-court statements as evidence (150). In these cases, guidelines can only color the evidence to the extent that expert witness testimony reflects their contents. Certain guidelines, however, may be ad-
The prt)fcssltmally detem~med standard was challenged successfully in Helling J. Carey, 83 Wash. 2d 514, 519 P. 2d 98 I (Wash. 1974). in wh]ch the ctwrt rejected the professional standard for glaucoma screening in favor of its own higher standard. The precedent set by this case, which sparked c(mskkrahle c(mccm [n [he pr(}~lderc[)nln]uni[~f, has since heen restricted to apply (rely to situations itf obvious negligence (83 ).
4 M(NI jurisdicti(ms apply a national standard of care for board-certified specialists, but a significant number still apply a local standard ft)r general practiti{mcrs. The most con]rmm f(mrnulatmn of the skmdard cumently is a n]{xiificd locality rule, which requires physicians to meet the standard of physicians practicing In the same or similar l(~alit]es (9). ~ LTee ,.! ~ ~, C}lldnlbier ~. ,Mccilirc, sOS F. Xi 489 (~th Clr. 1974).
mitted into evidence as learned treatises, a class of statements that are granted exception from the hearsay rule in many jurisdictions (1 13). Federal Rules of Evidence, which have been adopted in a similar form by most states, define the learned treatise exception as follows: . . . statements contained in published treatises, periodicals, or pamphlets on a subject of history, medicine, or other science or art, established as a
reliable authority by the testimony or admission of the witness or by other expert testimony or by judicial notice (150).
There is no hard and fast rule as to which guidelines have reliable authority. Guidelines reflecting comprehensive analysis of scientific evidence and broad consensus among members of the profession are likely candidates, but courts themselves are likely to defer to expert opinion regarding the scientific validity of a guideline rather than make such judgments themselves (113).6
I
guideline might not be persuasive if expert witnesses testify that most physicians do not follow it. In spite of extensive and focused guidelines development in some areas of practice, physicians are sometimes slow to incorporate them (1 32). Additional incentives and dissemination tactics may. be needed to change physician behavior m accordance with guidelines. A recent study suggests that guidelines currently play only a small role in litigation but that this role may be increasing ( 100). The authors studied guideline use from the three different perspectives in order to assess their use in the various phases of medical malpractice litigation.
m
Use of Guidelines in Establishing the Legal Standard of Care

Once admitted as evidence of the legal standard of care, guidelines do not carry greater legal weight than any other expert testimonyi.e., they are not regarded as definitive statements of the standard of care. Once all testimony has been heard, it is left to the jury to decide the applicable legal standard of care. Even when a guideline is quite explicit and straightforward, it is not clear how much weight it will be accorded by the jury. OTA knows of no studies that have examined the reactions of juries to the use of guidelines as evidence. Under the current customary standard of care, clinical practice guidelines can only influence the standard to the extent that they are adopted into common medical practice. The existence of a
A national review of all published court opinions between 1980 and 1993 found only 32 cases in which the opinion indicated that guidelines had been used as evidence of the standard of care. A review of a sample of 259 claimsboth open and closedfrom two malpractice insurance companies found that only 17 involved the use of guidelines. In a random sample survey of medical malprac tice plaintiff and defense attorneys, 36 percent of attorneys reported that they had at least one case per year where guidelines played an important role. Moreover. 30 percent of attorneys reported they felt the use of guidelines in litigation was increasing ( 1 00).
The study identified more claims involving the use of guidelines by plaintiffs than claims involving the use of guidelines by defendants. In many cases, attempts to use guidelines as proof or rebuttal of negligence or nonnegligence were unsuccessful. The most frequently cited guidelines were those published by the American College of Obstetricians and Gynecologists ( 100).
b A recent us+ supreme ct~urf decision, ~aub~rf t: J4errel/ DOW PhiJrwlattl(ti~~]/s, I 13 S. Ct. 2786, 125 L. Ed.2d 469 ( 1993 ), gives Judges
greater responsibility for making independent judgments of the scientific validity {~f evidence before it is admitted m ctmt. It is unclear how this decision wilt affect [he admissibility of cl inical practice guidelines as evidence of the professi(mal standard of care, bu[ it dtws herald a shift away from relying solely on expert opinitm [o mahe such judgments.
Appendix H:
Clinical Practice Guidelines and Malpractice Liability | 143
BARRIERS TO THE USE OF GUIDELINES AS LEGAL STANDARDS

One factor limiting the impact of guidelines in litigation is that their language and form are often not amenable to use as legal standards. Some guidelines offer several treatment options, while others offer a single option but do not hold it forward as the only acceptable one. A typical guideline frequently includes allowances for deviation based on professional judgment. Many medical societies consciously avoid the use of words such as always and never when drafting guidelines and avoid referring to their guidelines as standards for fear of potential adverse legal consequences (232). AHCPR has also been concerned with potential legal consequences of guidelines development and has sought immunity from civil liability for the members of its guidelines panels (2.54). The American Medical Association (AMA) shares these concerns about the legal implications of guidelines. Although it encourages the devel opment and dissemination of practice guidelines as a means of improving and further standardizing the practice of medicine, the AMA resists the use of guidelines as an absolute legal standard of care:
ters will vary depending upon the origins and content of the parameter and the circumstances of the case. As a policy matter, this result seems entirely appropriate. Rules of law, like parame-
ters, must
added) (6)
maintain sufficient flexibility to adjust
to the needs of
the
particular case.
(emphasis
The AMA endorses and encourages building flexibility into guidelines in order to avoid cookbook medicine (6). Such flexibility may be warranted: however, it may limit the usefulness of guidelines in a legal context. The vastness and complexity of medical knowledge pose additional barriers to the courts
ability to depend on practice guidelines. While it may be possible to develop explicit criteria for diagnosis and treatment of certain pathologies, the current state of medical knowledge is insufficient to support the development of explicit criteria for the majority ofclinicalsituations(101 ). One study estimated that there could be over 10 billion possible pathways for diagnosing common medical problems (56). Adding treatment algorithms would increase the number even further. Even if good evidence were available on which to base guidelines for a subset of medical conditions. its complexity could be daunting in a court of law. Court decisions could be complicated further in cases where conflicting guidelines were introduced into evidence. In a 1992 survey, a random sample of state trial and appellate judges ranked clinical practice guidelines third among 30 scientific topics on which they felt a need for greater information (262). To satisfy this need, a major project is currently under way to publish desk books that will give judges guidance on the evaluation of scientific evidence. However, because the medical community is still debating the relative merits of different types of evidence on the effectiveness of medical treatments,7 it maybe some time before judges have the tools necessary to evaluate clinical practice guidelines from an evidentiary standpoint. Finally, the continuing evolution of medical practice presents a challenge for efforts to keep guidelines current. Some critics argue that the adoption of rigid guidelines as legal standards of care could hinder the development and adoption of new medical technologies in the future.
INITIATIVES TO PROMOTE LEGAL USE OF GUIDELINES

Today, clinical practice guidelines carry limited evidentiary weight in medical malpractice litigation. To enhance the role of guidelines in the
courts, two different approaches could be taken. One approach would be to give greater evidentiary weight to certain guidelines in the litigation process (e.g., by authorizing judges to exercise more discretion with respect to admissibility of guidelines or by adopting certain guidelines under administrative law). A mere passive approach would be to continue current efforts in guidelines development at the national level in the expectation that, over time, guidelines would figure increasingly in medical malpractice litigation. The first approach requires legislative action. In fact, such action was taken in the early 1970s as a part of the Medicare Program. A provision of the Medicare Act8 grants immunity from civil liability to practitioners who exercise due care in complying with treatment criteria developed by Medicare peer review organizations (PROS). Although this provision has been on the books for over two decades, it has never been invoked, probably because the criteria developed are not explicit enough to be of much use in a legal context (85, 116). Even if sufficiently explicit criteria were available, legal scholars dispute how much additional protection the provision would confer because of a lack of clarity in the legislative language (17, 116, 169). Another likely explanation for the disuse of the Medicare provision is its link to the PRO program, which has itself been the subject of considerable controversy and change since the adoption of the immunity provision (85). In recent years, however, several states have passed legislation that may allow for greater use of guidelines in determining the legal standard of care. Four statesMaine, Florida, Minnesota, and Vermontrecently passed legislation that accords greater weight to certain guidelines in medical malpractice litigation. Maines 5-year Medical Liability Demonstration Project, begun in 1991, makes state-developed guidelines admissible as a defense in medical malpractice proceedings (24 M.R.S. Sees.
2971 et. seq. (1993)). The projects goals include reducing malpractice suit rates and insurance premiums; reducing defensive medicine; reducing variation in practice patterns; and containing overall health care costs. Guidelines for selected areas of practice in obstetrics/gynecology, emergency medicine, radiology, and anesthesia were developed by four medical specialty advisory committees appointed by the Maine Board of Registration in Medicine (see box H-l). Guidelines were developed in areas of practice where defensive medicine was believed to be extensive. The statute permits physicians electing to participate in the demonstration to use these guidelines as an affirmative defense in medical malpractice proceedings. Under the affirmative defense provision, use of guidelines as evidence is no longer a matter of the judges discretion. If a physician introduces the guideline as a defense, he or she must prove only that the guideline was followed. In order to deny a physician this affirmative defense, the plaintiff must either: 1 ) prove that the physician did not follow the guideline, or 2) prove, through expert testimony, that the guideline is not applicable to the given case. If the plaintiff is unable to do this and the physician proves that he or she complied, the physician is cleared of liability. Another provision of the Maine Statute prohibits plaintiffs from introducing a state guideline into evidence in an effort to prove that the physicians performance was substandard (24 M. R. IS. Sec. 2975 ( 1993)). This provision was included to allay fears on the part of physicians that the guidelines, instead of protecting them from liability, would be used against them (212). Some critics, however, claim that this provision may be subject to challenge on state or federal constitutional grounds because it selectively denies plaintiffs the use of evidence that may be critical to proving malpractice (215). A hearing on such a constitutional challenge would probably not occur for sev -
~ 42 U.s.c.
sec. 1 32&-6(c)
Appendix H:
Clinical Practice Guidelines and Malpractice Liability | 145
Emergency Medicine
q
Criteria for performing cervical spine x-rays on asymptomatic trauma patients in the emergency room Checklist for criteria to be met in accordance with federal statute before affecting a patient transfer
Obstetrics and Gynecology

Caesarean delivery for failure to progress Assessment of fetal maturity prior to repeat cesarean or elective induction of labor Management of singleton breech presentation Management of Intrapartum fetal distress Antepartum management of prolonged pregnancy Hysterectomy for diagnosis of abnormal uterine bleeding in women of reproductive age or diagnosis of Ieiomyomata Tocolysis Diagnosis and management of ectopic pregnancy Management of perinatal herpes simplex virus infection
Anesthesiology
q q
Preoperative testing Preoperative, interoperative, and postoperative monitoring
Radiology
q q q q
Screening mammography Antepartum ultrasound Outpatient angiography Adult barium enema examination
SOURCE State of Maine Board of Reglstratlon In Medlcme Department of Professional and Fmancml Regulation, Rule02-373 chs 20 22 24 26 Medical Llablllty Demonstration ProjectSpecialty Practice Parameters and Risk Management Protocols
eral years. As of May 1994, the states largest medical malpractice insurance carrier had only received one claim for which the adopted guidelines were potentially relevant (29). Florida legislation in 1993 authorized a 4-year demonstration project similar to that in Maine. Outcomes data on hospital patients collected through a statewide mandatory reporting system will be used to help develop practice parameters for inpatient care. These parameters, as well as parameters for selected outpatient services, will be developed by the Florida Agency for Health Care Administration in conjunction with relevant state
health professional associations and boards. Once adopted under state rulemaking procedures. these parameters will be admissible as an affirmative defense in medical malpractice proceedings (Fla. Stat. Sec. 408.02 (1993)). Unlike Maine, however, the Florida legislation does not bar plaintiffs from trying to use the parameters to prove that a physicians care was substandard. A plaintiff might be able to introduce the parameter as evidence, but the parameter would not be accorded greater weight than any other expert testimony. Minnesota recently passed legislation that allows guidelines developed or adopted by a special
state commission to be used as an absolute defense in malpractice litigation (164).9 Like the Maine statute, Minnesotas law also bars the plaintiff from introducing the guideline as evidence that the physician failed to meet the standard of care. As of May 1994, the first round of guidelines had yet to be developed (72). Vermonts approach is more moderate, amounting to a change in the rules of evidence that would allow a wider variety of guidelines--e. g., guidelines developed by health care professional groups, the federal government, or health care institutionsto be directly admitted as evidence of the standard of care by either the plaintiff or the defendant in future mandatory medical malpractice arbitration proceedings (18 V. S. A., part 9, chapter 21, Sec. 1 ( 1992)). This provision would make it easier to introduce guidelines as evidence but would not give them legal weight any greater than other expert testimony. Maryland, in a departure from the strategies adopted by other states, recently adopted legislation that mandates the development of state guidelines but explicitly prohibits them from being introduced as evidence by any party in a malpractice suit (Maryland, State House of Representatives, House Bill 1359, enacted Apr. 13, 1993.) A few other states have passed legislation authorizing the development of guidelines and encouraging consideration of their use in the future as legal standards of care. Some patient rights advocates may oppose the approach taken by Maine and Minnesota because it offers no safeguard against bad guidelines i.e., the plaintiff cannot contest the reasonableness of the guidelines themselves ( 179). Some critics contend that the use of guidelines as rigid legal standards may be problematic due to the continual evolution of medical practice and the inability of written guidelines to reflect changes in a timely manner (94).
State guidelines initiatives raise the potential for conflict between national, state, and even institutional guidelines. For example, most of Maines guidelines were based on nationally recognized guidelines, but others were developed de novo by Maine physicians (53) and could be construed as setting a precedent for reconversion to a more local standard of care. Guidelines developers in Minnesota anticipate using national guidelines as models and amending them if necessary to conform to the realities of health care delivery in the state (72). In Vermont, the statutory description of guidelines could be interpreted as including even written hospital protocols. It will be some time before evidence of the effects of these state efforts is available. Some early reports suggest that the Maine initiative has reduced defensive practices in selected areas (e.g., the use of cervical spine x-rays in the emergency room) ( 115). Given the modest nature of the changes and the limited number of guidelines adopted, however, it is unlikely that these programs will have much of an impact overall on the practice of medicine. The extent to which Maine and Minnesotas programs will streamline the litigation process is also questionable. In both states, expert testimony will still be required to establish whether the guidelines are relevant to the case and, because of the complicated nature of medical practice, whether they were in fact followed. In cases where several different guidelines can be introduced as evidence, expert testimony may also be necessary to determine which, if any, represents the legal standard of care.
PRACTICE GUIDELINES IN AN ERA OF COST CONTAINMENT

Increasing concern over the costs of medical care has sparked the introduction of cost as a factor in medical decisionmaking (204). Costs as well as
II IS unclear exactly how Minnesf~tas &.\o/uIe dejcnse provlst(m differs fr{)m Maines a(~irn~arr]e d(:tensc. The legal meaning may he essentially the same+. c., the plaintlff must pr{~ve that the ph) slcian chcin t ft)llt~u the guidcl inc or that the guldct Ine IS not applicable tt~ the specific case in order to deny [he physician this al cnuc of defense. H{)wc\cr, unt II there h:it e hem test ~iis~s in~folving the gu idel incs, it renm~ns unclear how exactly h{)w judges w ill Interpret the st:itutes (83).
Appendix H: Clinical Practice Guidelines and Malpractice Liability | 147
plications following coronary artery bypass surgery. The patients primary physician had requested an 8-day extension, but the Medicaid program authorized only 4 days. The patient was discharged after a 4-day extension and suffered post-discharge complications that ultimately resulted in a leg amputation. The court concluded that the state Medicaid program was not liable for Wicklines injury because the decision of when to discharge was the responsibility of the treating physician. The primary physician testified that he felt that Medi-Cal had the power to tell him, as a treating doctor, when a patient must be discharged from the hospital.13 However, all three physicians involved in the patients care testified that the decision to discharge after the 4-day extension was consistent with customary practice. 14 The court stated that, although: . . . cost consciousness has become a permanent feature of the health care system, it is essential that cost limitation programs not be permitted to corrupt medical judgment. We have concluded, from the facts in issue here, that in this case it did n{~[.I5.16 Some legal scholars have argued that, as cost concerns enter increasingly into physicians treatment decisions, the customary standard will come to reflect these concerns either implicitly or explicitly (85,1 99), as suggested in Wickline. Practice guidelines, to the extent that they reflect cost considerations and are given evidentiary weight in court, are clearly one of the more systematic ve-
I o See, ~.g.. .srnlr/f ~, }Ij/ic, 194 ,A. 2d 167 (P:~. 1963), ITlarh I. Lnifcd state\. 402 F. 2d 950 (Clr. DC. 1968), Wi/Lrn!mr i. Ve.$e), 295 A. 2d
hicles that might be used to bring about such a change. There is still considerable argument regarding the incorporation of cost concerns into practice guidelines (33,1 88). The AMA does not include cost as one of its criteria for guidelines development (8) and maintains that practice guidelines should be developed independent of considerations of cost (227). An entire area of law is under development that may expose payers to liability for negligent utilization review and payment decisions that result in harm to patients (84).
It remains to be seen whether courts will come to accept economic factors as determinants of the legal standard of care for physicians. Resolution of these difficult questions maybe central to effective health care reform. If they can be used to protect physicians from liability, clinical practice guidelines may be a potential means for reconciling broader social goals (e.g., health care cost containment) with a more individual-oriented legal standard of medical care.
Appendix I: Description of 32 Direct Physician Surveys of Defensive Medicine Reviewed by OTA
| 149
Author, year of release

Porter, Novelli &
Survey year
1983
Sample population location National
Specialty Obstetrician/ Gynecologists (Ob\Gyn)
Survey characteristics
Response rate (percent) 50.1%.
Associates, 1983a
Survey of random sample of American College of Obstetricians and Gynecologists (ACOG) members regarding medical Iiability Insurance premiums, claims experience, and practice changes in response to malpractice risks Data from the 3rd quarter 1983 and 4th quarter 1984 American Medical Association (AMA) Socioeconomic Monitoring Surveys on practice changes made in response to Iiability risk Survey of members regarding medical Iiability Insurance premiums, claims experience, and practice changes in response to medical Iiability Survey of all members for data and opinions on the medical professional Iiability environment Survey of random sample of ACOG members regarding medical liability Insurance premiums, malpractice claims experience, and practice changes in response to malpractice risks Survey regarding professional Iiability and defensive medicine Survey of physicians to assess the personal and professional impact of malpractice Iitigation Survey of all members regarding obstetric practice
Reynolds et al 1987b
1983/1 984
National
All
630
Bligh, American College of Surgeons, 1984C
1984
National
Surgeons
36
Kansas Medical Society, 1985d
1984
Kansas
All
50
Needham, Porter, Novelli, 1985e 1985
National
Ob\Gyn
397
Texas Medical Association, 1985f Charles, Wilbert, & Frankel 1985g Alabama Academy of Family Physicians 1986h
1985
Texas
All
232
1985
Chicago
All
366
1985
Alabama
Family and General Practitioners (F\GP) F\GP
84
Iowa Family Physician Survey 1985 Michigan State Medical Society, 19851
1985
lowa
Survey on medical Iiability
1985
Michigan
Ob\Gyn
Survey to measure the potential impact of the professional Iiability Insurance problem
Phone survey of rural doctors regarding obstetrical
Ob\Gyn and F\GP
care and malpractice concerns
-.
Author, year of release
Survey year 1985 1985 1986
Sample population location Oregon Washington National
Specialty Ob\Gyn and F\GP F\GP All
Survey characteristics Survey to assess the impact of professional Iiability issues on access to obstetrical care Survey to assess the impact of rising malpractice Insurance premiums on the practice of obstetrics Interview survey regarding costs and availability of malpractice Insurance and their impact on physician practice Survey to assess impact of cost and availability of liability Insurance on the practice of obstetrics Survey to measure the impact of professional Iiability Insurance rates on the medical profession Survey of how malpractice liability affects obstetric care Survey regarding professional liability
Response rate (percent) 81 1 803 742 0
1986 1986 1986 1986 1986
National Texas Georgia Kentucky Michigan
F\GP All Ob\Gyn Ob\Gyn and F\GP F\GP
337 355 61 42 o
Survey to describe the characteristics of family physicians who practice obstetrics and identify factors prompting them to discontinue practice Survey to describe the impact of rapidly rising malpractice premiums on obstetric practice and to assess the impact of tort reform on professional liability costs
815
1986
Washington
Ob\Gyn, F\GP, and midwives Ob\Gyn
635
1987
National
Survey of random sample of ACOG members regarding 484 medical liability Insurance premiums, claims experience and practice changes in response to malpractice risks Survey to assess the impact of malpractice litigation on the doctor-patient relationship and to collect data that might suggest effective tort reform Survey on changes in availability of obstetrical services Telephone survey regarding practice changes as a result of the current malpractice Iiability cilmate 427
1987
Wisconsin
All
1987 1987
Illinois Maryland
Ob\Gyn and F\GP Ob\Gyn F\GP
256 65
and Internal Medicine
(continued)
Author, year of release Texas Medical Association, 1988Y
Survey year 1988
Sample population location Texas Louisiana New York
Specialty All Ob\Gyn All
Survey characteristics Survey to assess impact of malpractice Insurance premiums cost and Iiability risk on physician practice
Response rate (percent) 41 384 405
Louisiana Section of ACOG, 1988Z 1988 Lawthers et al , 1992aa 1989
Survey on professional liability Survey of physicians perceptions of the risk of being sued and their impact on physician practice Survey of random sample of ACOG members regarding medical liability Insurance premiums, claims experience, and practice changes in response to malpractice risks Survey of random sample of ACOG members regarding medical liability insurance premiums, claims experience, and practice changes in response to malpractice risks General survey regarding income and malpractice Insurance cost concerns Survey regarding professional Iiability Insurance problems facing physicians
Opinion Research Corp 1990 bb 1 9 9 0
National
Ob\Gyn
540
Opinion Research Corp ,1992 CC 1 9 9 2
National
Ob\Gyn
51
Minnesota Ob\Gyn Survey (Meader, no date)dd West Virglnia State Medical Association, no dateee
a
no date no date
Minnesota West Virginia
Ob\Gyn All
Not provided 50
Porter, Novellt &Associates, Professional Liability Insurance and Its Effects Report of a Survey of ACOGs Membership, prepared for the American College of Obstetrlclans and Gynecologists, Washington, DC, August 31, 1983 bR A Reynolds, J A Rl=o, and M I- Gonzalez, The cost of Medical Professional Llablllty Journa/ofthe Arnerlcan Med/ca/Assoc/at/on 257(20) 2776-2781, May 22/29 1987
T J Bhgh, American College of Surgeons Professional Liability Survey Report, 1984 , Executwe Services Department for the Regents Ad Hoc Committee on Professional Llablllty, American College of Surgeons, Washington, DC, 1984 dKanSaS MedlCal Society, professlona/ Ltablllty Survey, Kansas Med;c/ne P 43 February 1985 Needham, Porter, Novell!, Professional Llabll Ky Insurance and Its Effect Report of a Survey of ACOGs Membership, prepared for the American College of Obstetnclans and Gynecologists, Washington, DC, November 1985 f Texas Medical Association, Texas Medical Assoclatlons 1985 Professional LlabllKy Survey (unpubhshed), Austin, TX September 1985 9 S C Charles, J R Wllbert and K J Franke, sued and NonSued physicians Self- R e p o r t e d R e a c t I o n s t o Ma[practlce I-ltigatlon, Amerjcan Journa/ of PsYchlat~ 142(2) 437-440, April 1985 hAlabama Academy of Faml[y physlclans, A Survey of Family Physlclans Provldmg Obstetrical Care A Prellmmary Report, Alabama Academy of Family Physlclans Mont-
gomery, AL, February, 1986 I Iowa Medical Society, Iowa Family Physlclan Survey Fmdmgs (unpublished),1 987 I M Block, Professional Liablllty Insurance and Obstetrical Practice, commissioned by Mlchlgan State Medical Society, July 1985 kH E Crow Ljnlverslty of Nevada School of Medlclne, Off Ice of Rural Health, Survey of Rural Doctors Regarding Their parttclpatlon (Or not) In Obstetrics, Off Ice of Rural Health
Unwerslty of Nevada School of Medicine, Mar 11, 1985 The Oregon Medical Assoclatlon, Ad Hoc 06 Task Force on Professional Llabtllty, The Impact of Professional Llablllty Issues on Access to Obstetrical Care In Oregon, Oregon Medical Assoclahon, March 1986 m R A Rosen blatt and C L Wright, Rising Malpractice Premiums and Obstetric practice patterns The lrn~a~t nn F~rrilly PhYSICl~nS !n Wach tnrmtnn --! !.! l,j .. ~tate ! ,UL , ?~e we~~~~,rj J~~,~,y~/ of A4ed;cme 146(2) 246-248, February 1987 n M L Rosenbach and A G Stone Malpractice Insurance Costs and Physlclan Practice, 1983 -1986, Hea/fh Atfalrs 9(4) 176-185, 1990
(continued)
) Amerlcan Academy of Family Physicians Commitee on Professional Liablility and Division of Research and Information Services Family Physicians and Obstetrics A Professional Liability Study 1987 p Opinion Analysts Inc The Texas Medical Association Professional Liability Insurance Survey prepared for the Texas Medical Association September 1986 q Georgia Obstetrical and Gynecological Society GOGS 1987 Survey Results Atlanta GA 1987 r G S Bonham Survey of Kentucky Obstetric Practice Jolurnal of fhe Kentucky Medical Assocliation 349 353, June 1987 s M A Smith L A Green and T L Schwenk "Family Practice Obstetrics In Michigan Factors Affecting Physician Participation on The Journal of Family Practice 28(4) 433 437 1989 t R A Rosenblatt and B Deterlng Changing Patterns of Obstetric Practice in Washington State The Impact of Tort Reform Famlily Medicine 20(2) 101 107, March/April 1988 u Opinion Research Corp , Professional Liability and Its Effects Report of a 1987 Survey of ACOG's Membership prepared for the American College of Obstetricians and Gynecologists Washington, DC March 1988 v R S Shapiro, D E Simpson, S L Lawrence et al "A Survey of Sued and Nonsued Physicians and Suing Patients Archives of Internal Medicine 1492190 2196 October 1989 w M C Ring, Draft Report Changes in Availability of Obstetrical Services in in Illinois" Division of Local Health Administration, Illinois Department of Public Health 1987 x C S Weisman, L L Morlock, M A Teitelbaum et al , Practice Changes in Response to the Malpractice Litigation Climate Medical Care 27(1) 16 24 January 1989 Y Texas Medical Association, Texas Medical Assocition 1988 Professional Liability Survey summer 1988 z W P Begneaud, Obstetric and Gynecologic Malpractice in Louisiana Incidence and Impact prepared for the Louisiana Section of the American College of Obstetrics and Gynecology, Lafayette, IA 1988 aaA.G. Lawthers, A R Localio and N M Laird, Physicians Perceptions of the Risk of Being Sued Journal of Health Politics, Policy and Law 17(3) 463-482, 1992 bbOpinion Research Corporation, professional Liability and Its Effecfs: Report of a 1990I Survey Of ACOGs Membership, prepared for the American College of Obstetricians and Gynecologists, Washington, DC, September 1990 ccOpinion Research Corporation, "Professional Liability and Its Effects Report of a 1992 Survey of ACOG's Membership, prepared for the American College of Obstetricians
and Gynecologists, Washington DC, October 1992

ddE.C. Meader, Jr., , Minnesota Obstetrics and
Gynecology Practice Survey Summary, prepared
for the Minnesota Section of the American College Of Obstetrics and Gynecolo-
gy, no date eeWest Virglnia State Medical Association, West Virglnia State Medical Associations Physician Survey (unpublished), undated
SOURCE Off Ice of Technology Assessment, 1994
Appendix J: Detailed Critique of Reynolds et al. and Lewin-VHI Estimates

n chapter 3 of this report, the Office of Technology Assessment (OTA) reviewed two wide] y publicized estimates of the costs of defensive medicine and the medical malpractice system-one published in 1987 by Reynolds and colleagues at the American Medical Association (194) and the other published in 1993 by Lewin-VHI, Inc. (1 25). This appendix provides a detailed critique of the data, methods, and assumptions that underlie those estimates.
THE REYNOLDS ESTIMATES Method 1: Survey of Physicians

Reynolds and colleagues tried to estimate the full impact of the malpractice system on physician costs, including:
m
q
malpractice insurance premiums; the time lost in defending against malpractice claims and lawyers fees not covered by malpractice insurance; and practice changes, including increased recordkeeping, use of more tests or treatment procedures, increased time spent with patients. and increased followup visits.
Of all the practice changes, only two-increases in tests or treatment procedures and followup vis-itsfall within OTAs definition of defensive medicine. Though some observers would claim that more time spent with patients or in documenting medical records is defensive medicine, OTA excluded these practices because it is extremely difficult to measure their frequency and magnitude and because the positive impact of these practices on the quality of care is less equivocal. In contrast, procedures and followup visits are documented in utilization data, offering an empirical check. Estimation of malpractice insurance premiums was based on the American Medical Association (AMA) Socioeconomic Monitoring System (SMS) survey, which asks physicians to report their malpractice insurance premiums and other practice costs. The SMS also gives information on days lost from work to defend against malpractice claims and the amount paid for outside attorneys. These data items, though subject to the usual problems of recall bias, are sufficiently accurate for the purposes at hand. (They are also subject to verification with objective premium data and other survey data. ) The main problem comes in esti-
I 154
Appendix J: Detailed Critique of Reynolds et al. and Lewin-VHl Estimates | 155
mating the net costs of practice changes resulting from malpractice liability. In its fourth quarter 1984 survey, the AMA
asked a series of questions about whether physicians were maintaining mm-e detailed records, prescribing more diagnostic tests and treatment
procedures, spending more time with patients. and having more followup visits with patients in the last 12 months in response to their malpractice risks ( 194). If physicians answered in the affirmative to any of these items, they were asked to quantify the change over the past 12 months in percentage terms. Table J-1 summarizes the results of the survey. The physicians reported that in 1984 they increased tests and procedures by 3.2 percent and followup visits by 2.6 percent in response to changes in the frequency of malpractice claims. These two practice changes fall within OTAs definition of defensive medicine. The other practice changes, such as increasing recordkeeping and time spent with the patient, may result from the same desire to avoid a malpractice suit, but these practice changes lead to increases in the cost per visit or procedure. Such cost increases would be passed on to consumers in the form of higher fees rather than additional procedures or visits. Reynolds estimated the cost of all of the 1984 practice changes except the cost of extra tests and procedures, which was excluded because the researchers could not find a good way to estimate the average cost of such a diverse array of services.
The average cost per physician of the remaining practice changes was $4.600. of which $1,900 was the cost of reported changes in followup visits. The authors computed the ratio of the 1984cost of practice changes ($4,600) to the 1984 increase in malpractice insurance premiums ($ 1,300), and applied this ratio (3.53) to the average 1984 malpractice premium ($8,400) to arrive at a per-physician cost of practices done in response to the malpractice system: $29,700. or 14percent of average physician revenues. In the aggregate, this cost corresponds to $10.6 billion in 1984. To summarize, under method 1. Reynolds total estimate of the cost of the malpractice system for physicians$ 13.7 billion in 1984---comprises the following elements:
s s
premiums-$3.O billion. other costs of incurring malpractice claims$0.1 billion, and practice changes-$ 10.6 billion.
Of the $13.7 billion in total cost, about $4.3 billion, or 30 percent, represents defensive medicine under OTAs definition. The estimate of the cost of practice changes has several potential sources of bias. On the one hand, there is reason to believe that Reynolds estimate of the malpractice systems impact on health care costs is too low because Reynolds and colleagues excluded the reported 1984 cost impact of increased tests and treatment procedures. The importance of this exclusion is unknown. but it rep-
Activity
Increased recordkeeping Prescription of more test or treatment procedures Increased time spent with patients Increased followup visits Percent of physicians with at least 1 listed practice charge
Percent of physicians making change in 1984 31.0% 200 170 170 41 8
Average percent change in 1984a 2 9% 32 24 26
Calculations Include zeros for physclans who did not make practice change
SOURCE American Medical Assoc Iaf IoP Socloeconomlc Mon torlng System sJrVf?y as rep@rtw r] R A R~yr301ds J A RIZZO and M L Gonzalez The Cost of Medical Professional Ltabllty Journa/ o/ Amer/can Medlca/ Assocalm 257(20) 2776-2781 May 2229 1987 Copyright 1987 American Medical Assoclatlon
resents the essence of OTAs definition of defensive medicine and means that the Reynolds estimate probably does not capture the greatest part of defensive medicine. On the other hand, there is reason to believe that Reynolds estimate is too high, because the survey may have prompted physicians, who regularly articulate negative feelings about malpractice liability, to overestimate the impact of rising malpractice claims on their practices. Data from the National Ambulatory Medical Care Survey (NAMCS) show no change between 1981 and 1985 in the per-capita number of followup visits; they also show an annualized rate of increase of less than 1 percent in total per-capita physician office visits over the period (70). Barring some dramatic factor at work between 1983 and 1984 to otherwise reduce the frequency of followup visits by as much as 2.3 percent, physicians responses to the AMA survey appear to exaggerate their actual change in behavior. 1 If physicians overestimated the malpractice systems impact on follow up visits, they may also have done so with the other practice changes. Finally, Reynolds approach involved an arbitrary assumption with unknown effects on the validity of the estimate. Reynolds assumed that the ratio of the change in practices (in response to
malpractice risk) to the change in premiums can predict the ratio of the level of such activities to the level of premiums in 1984. The authors had no empirical evidence for this assumption, and there is reason to believe that it may be inaccurate.2 As a consequence of these issues, OTA concluded that Reynolds first method does not offer a sufficiently reliable estimate of the full cost impacts of malpractice liability and does not offer a basis for estimating the costs of defensive medicine.
The researchers examined the relationship between the level of malpractice liability risk, as measured by the 1984 malpractice premium reported by each physician responding to the AMA survey, and the physicians fees and volume of selected services reported in the same survey. Regression of utilization and fees on premiums3 and other demographic variables (e.g., physicians per 1,000 population, years in practice, board certification, etc. ) gave estimates of the impact of each $1 of premium on the utilization or fee for a given procedure. Doctors with higher premiums were found to have higher fees, but they had lower lev-
I II is theoretically feasible that physicians responding to the AMA suwey were able to differentiate between extra followup visits they would like to have provided and extra visits that they actually realized, after other independent impacts on visits were taken into account. If; for example, the demand for visits declined over the period, physicians might have ordered more follow up visits for defensive reasons but nevertheless actually provided fewer net visits overall. To accept this possibility y, one would have to believe that physicians responding to surveys could accurately estimate the partial impact of their defensive behavior on the volume of visits. 2 me assunlptlon in)p]les a Ilnem re]at;onship between the frequency of the cited practices ~d the level of malpractice insur~ce Premiums, with the graph of the line intersecting the y-axis at the origin. Because ordering extra tests, procedures, and visits does not cost physicians money
and is often financially remunerative, there is no reason to believe that as malpractice premiums decline, the motive to practice defensively declines in a linear fashion to the origin. Indeed, one would expect that physicians in 1984 were practicing on the flat of the curve where they were already as defensive as they knew how to be. Thus, to the extent that their reported 1984 behavior changes reflect reality, the linearity assumption would understate theamount of defensive medicine. On theother hand, practice changes that take up more time (such as increiised time with the patient) would increase the physicians costs and presumably be more directly responsive to increases in premiums. Whether the relationship is linear or not is unknown.
3 The malpractice premium used in the regression analysis was an estimated value based on a first-stage regression of premiums on demographic characteristics, the status of various malpractice reforms in the physician state, and the malpractice claim frequency in the state. This two-stage method t)f estimation is referred to as the in.$mumenfa/tariab/e technique. The rationale for such an approach is to make the instrumental variable (premiums m this case) a better measure of the actual variable (malpractice risk in this case) than it would be were the actual value used in the regressi(m.
els of use of the most important services studied. Table J-2 summarizes the results for each service. Reynolds took the findings presented in table J-2 as the basis for estimating what utilization and fees would have been if malpractice insurance premiums (and, presumably, malpractice liability risk) had been zero in 1984. These rates were compared with actual reported utilization and fees to obtain an estimate of the impact of premiums on physician revenues. The eight services chosen for the analysis represented about 70 percent of the average revenues of self-employed physicians in 1984. Without any malpractice insurance premiums, these revenues would have been reduced (according to the regression estimates) by 11.2 percent of average reve-
nues. In the aggregate, a reduction of 11.2 percent in average physician revenues represents an $8.4 billion saving in expenditures if there were no malpractice insurance premiums (and presumably no malpractice liability system). If the services constituting the 30 percent of average revenues not studied by Reynolds were influenced by premiums to the same extent as the eight studied, the physician revenues saved by no malpractice liability would amount to $12.1 billion in 1984. The most striking feature of this analysis is that virtually all of the impact on cost comes through increased fees, not through increases in utilization of procedures. In fact, utilization of most of the procedures studied appeared to be reduced by higher malpractice insurance premiums. Any pos-
Procedure
Coefficient
Standard Error
/0 change in fee or utilization per a /0 change in premiums
Fees
Established patient office visit New patient office visit Followup hospital visit Electrocardiogram Obstetric care, normal delivery Hysterectomy Hernia repair Cholecystectomy O 85 1 16 1 18 148 2224 2538 311 -238 -6641 -1381 -4515 606 146 -049 -051 070 0 17b 0 .37b 0 .22b 0 .46b 4 .53b 5 .74b 566 860 28 .97b 7 .33c 20 .84b 3499 1 31 063 1 12 095 O 272 0212 0340 O 205 O 427 0349 0069 -0033
Monthly utilization
Established patient office visit New patient office visit Followup hospital visit Electrocardiogram Obstetric care, normal delivery Hysterectomy Hernia repair Cholecystectomy
a
The premium levels used In the computation are the averages for the specialties used (n estimating the premium effect for each procedure For
-0171 -0209 -0297 0073 0168 -0276 -0224 0217
patient wsls, these include all Speclalttes except radiology, psychiatry, pathology and anesthesiology for electrocardiograms general family pracllce and Internal medlclne for obstetric care and hysterectomies, obstetrics-gynecology, and for hernia repairs and cholecystectomles, general surgery b Indicates regression Coefftclent IS dlferent from O at the 01 519nlflcance level
c
Indtcates regression coeff Iclent IS dttferent from O al the 10 sign lflcance level
SOURCE R A Reynolds J A RIZZO and M L Gonzalez The Cost of Medical Professional Llablhfy, The ~ourna/ of Arrwrtcan Mecflca/Assoclaflon 257(20) 2776-2781, May 22/29 1987 table 2 Copyright 1987, American Medical Association
itive effects of malpractice risk on defensive medicine are apparently overshadowed by the negative effect of malpractice risk on demand that results from the higher fees that physicians with higher malpractice risk charge their patients. Thus, if the statistical analysis is correct, high malpractice risk depresses the demand for services as much as or more than it increases defensive medicine. The method underlying the estimates is based on a standard econometric technique, but as with all econometric analyses, the results might be sensitive to the specification of the statistical model and the ability to measure the relevant variables.4 Just how sensitive they might be is impossible to tell without more analysis of the quality of the premium measure of malpractice risk or corroborating evidence from other analyses. To turn the results of the statistical analysis into an estimate of the net costs of the malpractice system, the authors assumed that the relationship between malpractice insurance premiums and practice fees and volumes is linear throughout the range of potential premiums. The assumption that defensive medicine or other practice changes decline in lock-step linear fashion with declines in premiums all the way to the point of zero premiums is unlikely to be accurate, for reasons discussed above. Thus, OTA is unable to verify the accuracy of the estimates derived from the second method. Even if the total cost estimates are accurate, they do not allow any inferences about the extent or cost of defensive medicine, whose practice is embedded in a larger set of utilization changes re-
sulting from the malpractice system. High or low rates of defensive medicine are equally consistent with the results of the statistical model.
LEWIN-VHI ESTIMATES
Lewin-VHI began with the Reynolds estimates of the cost of the malpractice system (an average $18.8 million in 1991 constant dollars) and added another $6.1 billion for extra costs incurred in hospitals. Lewin-VHI obtained this hospital cost estimate by assuming that the cost of hospital professional liability in excess of hospital malpractice insurance premiums ($2.7 per dollar of premium) was the same as the ratio of physicians costs to physicians premiums estimated in the Reynolds study. s The preliminary total cost of malpractice$24.9 billion in 199lwas then reduced by three percentages (80, 60, and 40). This produced "low,($5 billion) medium ($10 billion) and *high ($1 4.9 billion) final estimates of the net costs of defensive medicine to the health care system in 1991. The adjustments were made because Lewin-VHI researchers wanted to exclude that portion of defensive medicine not caused solely by liability concerns. To help justify their estimates, Lewin-VHI researchers described three technologies whose utilization may be influenced by malpractice risk: electronic fetal monitoring in labor and delivery, skull x-rays in emergency rooms, and preoperative laboratory testing .6 Lewin-VHI researchers concluded that the low estimate of defensive medicine costs ($5 billion) represents a reasonable lower bound on defensive medicine costs based on a brief review of the literature on unneces-
For example, the asserti(m tha[ individual physicians prcmiums are a g(Nti measure of Il:ibil it] nsh using the instrumental vanablcs tt:ch nique cannot be assessed with the inforn]ation presented in the paper or its unpublished te~hni~iil iippcntli x Rccen[ resciirth suggests that If :in instrumental \ ariablc is not a g(NKI (me, it can lead to misleading and b]ascd results ( 173,213). The auth(~rf had a ]))tii~urt {)f ~liilni frequency available tt) therm which they might also have used as a direct measure of malpractice risk. Whe[hcr [hcsc ftict(m w~~uld ~hiing~ [he rcsulls is impassible m know with(mt carrying out such analyses. f Lcwin-\H1 obtained th]s ratio (2.7) from AMA res~ar~hers, It IS lower than the ratio publlshtxl In the Rcj n(~lds study .of prcopera[iic tests that cla]rntxl about $ 2 . 7 b]]] i!m eitra IS s~nl cii~h > car
(3,2).
~f}r
~ For ~.:llllp]e, the au[h(~rs cited (~n~ stud~
unnCCCssary
prtx)perat]w lestin~ ( 13S). Because dwt[~rs t} pi~ally d{) not gain finan~iall~ trtml (mltmn~ $u~h t~$[~, [h~ LCW ]n-YHl ;iu[hors c(}ncludcd that iin appreciable pmmm of these costs results fr(ml fear of malpractice I]abil it} ( 125).
sary use of these three procedures. Lewin-VHI offered no justification for the upper bound of the range. Although the Lewin-VHI researchers acknowledged the great uncertainty surrounding any estimate of defensive medicine, the objective basis for their specific adjustments from the Reynolds estimate is weak. The evidence presented in the three clinical examples used for the lower bound estimate does not necessarily reflect the percentage of unnecessary procedures motivated solely (or even primarily) by fear of malpractice liability.
Also, the estimates of the number of unnecessary procedures in the studies cited by Lewin-VHI were based on small and sometimes subjective assessments. Finally, they represent only three relatively narrow areas of medicine. To summarize, Lewin-VHI began with the estimates by Reynolds and colleagues, whose accuracy is unknown and unverifiable, and then made downward adjustments using a fragile base of evidence. Consequently, the Lewin-VHI estimate is not a reliable gauge of the possible range of defensive medicine costs.
Appendix K: Glossary
Accelerated compensation events (ACE)
A set of medical injuries deemed to be statistically avoidable with good medical care which would be compensated under a limited no-fault claims resolution system. Affirmative defense A response by the defendant in a legal suit that, if true, constitutes a complete defense to the plaintiffs complaint. Alternative dispute resolution (ADR) A process outside the judicial system for resolving legal claims. Decisions are made by dispute resolution professionals. ADR can be binding or nonbinding (see arbitration). American Medical Association/Specialty Society Malpractice Liability Project (AMA/SSMLP) Administrative System A proposed alternative to the malpractice system in which the medical licensing boards in each state would decide medical malpractice cases based on fault (negligence), using an administrative process designed to be more abbreviated and less costly than the current malpractice system. Arbitration A form of ADR in which the parties agree to have one or more trained arbitrators hear the evidence of the case and make a determination on liability
or damages. The rules of evidence and other procedural matters may often be specified by the parties. There are two types of arbitration: binding and nonbinding. In binding arbitration the arbitration decision is subject to very limited judicial review. If arbitration is nonbinding, the parties may proceed to trial if they are not satisfied with the outcome of the arbitration. Some states require parties to submit a claim to nonbinding arbitration before trial (see also pretrial screening).
Attorney fee limits
Legislation that either limits a plaintiff attorney fees to a set percentage of the award or allows for court review of the proposed fee and approval of what it considers to be a reasonable fee. Awarding costs, expenses, and fees Statutes that provide that the losing party in a friv-O1OUS suit may be required to pay the other partys reasonable attorney and expert witness fees and court costs. These provisions are designed to deter the pursuit of frivolous medical injury claims. Caps on damages Legislative limits on the amount of money that can be awarded to the plaintiff for economic or noneconomic damages in a personal injury claim. such as medical malpractice. The limit is imposed regardless of the actual amount of economic and noneconomic damages.
Appendix K: Glossary | 161
Certificate of merit
As a prerequisite to filing suit, some states require that a plaintiff obtain a written affidavit from an independent physic i an attesting that the plaintiff suit has merit. This provision is designed to limit nonmeritorious suits.
Claim frequency
Damages See economic damages and noneconomic dam-
ages.
Defensive medicine
A rate expressing the frequency with which physicians are named in malpractice claims. It is usually expressed as the number of malpractice claims per 100 physicians per year.
Collateral source rule
The ordering of extra tests, procedures, and visits or the avoidance of high-risk patients or procedures primarily (but not necessarily solely) to reduce their risk of malpractice 1iability. The performance of extra procedures for defensive purposes is positive defensive medicine. Avoidance of high-risk patients or procedures is negative defensive medicine.
Difference-of-means test
A rule of evidence that prohibits the introduction at trial of an y evidence that a patient has been compensated or reimbursed for the injury from any source (e.g., health or disability insurer). Legislation modifying the collateral source rule has taken two basic approaches: 1 ) permitting the jury to consider the compensation or payments received from some or all collateral sources and decide whether to reduce the award by the amount of collateral sources; or 2) requiring a mandatory offset against any award in the amount of some or all collateral source payments received by the plaintiff.
Confidence interval
A test of the statistical significance of the difference between two groups in their mean scores on a single variable.
Direct malpractice costs The net costs of compensating injuries through the
medical malpractice system, including costs borne by malpractice insurers, defendants, and plaintiffs.
Discovery
An interval that contains, with certain probability, the true value of a statistic. The mean is a typical statistic. The true mean lies within the bounds of the 95-percent confidence interval in 95-percent of all samples.
Correlation A statistic that gauges the strength of association
Pretrial tools for obtaining information in preparation for trial. The tools include written and oral questioning of relevant parties, requests for documents, and physical examination of evidence and physical premises. The process of discovery is governed by federal and state rules of civil procedure.
Economic damages
between two variables. The value of a correlation coefficient usually ranges from a minimum of zero (no association at all between the two variables) to a maximum of one (perfect association between the two variables). Some correlation coefficients also have a sign indicating the direction of association between the two variables: a positive sign indicates direct association (as one variable increases in value. the other also increases); and a negative sign indicates inverse association (as one variable increases in value, the other decreases).
Monetary damages that compensate the plaintiff for his or her actual economic lossesi.e., past and future medical expenses, lost wages, rehabilitation expenses, and other tangible losses,
Enterprise liability
A system under which a health care institution or health insurance plan assumes full legal liability for the actions of physicians acting as their agents, and individual physicians cannot be named as defendants.
Error in judgment rule
An exception to the general requirement that the physician must meet the prevailing standard of care provided by his or her profession. A physi-
cians conduct will not be judged to fall below the standard of care if the physician chooses between two or more legitimate choices of treatment, even though a better result might have been obtained with a different treatment.
Guidelines
formed decision. The required elements of disclosure differ from state to state.
Joint and several liability
Generally referring to clinical practice guidelines, which are defined by the Institute of Medicine as systematically developed statements to assist practitioner and patient decisions about appropriate health care for specific clinical circumstances. However, q guidelines in some cases refers to clinical practice guidelines developed with additional goals explicitly in mind, such as cost containment or reduction of defensive medicine.
Health maintenance organization (HMO)
A rule under which each of the defendants in a tort suit can be held liable for the total amount of damages, regardless of his or her individual responsibility. In other words, even if a defendant was only 20 percent responsible, he or she could be held liable for 100 percent of the damages if other defendants are unable to pay. Several states have eliminated joint and several liability for medical malpractice so that physicians are liable only in proportion to their responsibility.
Low osmolality contrast agent (LOCA) A contrast agent is a substance that is used to im-
A health care organization that, in return for prospective per capita payments (cavitation), acts as both insurer and provider of comprehensive but specific health care services. A defined set of physicians (and often other health care providers such as physician assistants and nurse midwives) provide services to a voluntarily enrolled population. Prepaid group practices and individual practice associations, as well as q staff models, are types of HMOs.
Iatrogenic injury
prove the visibility of structures during radiologic imaging-e. g., angiography, intravenous urography, or computerized tomography (CT) scans. A low osmolality contrast agent has an osmolality (i.e., concentration of dissolved particles in solution) that is closer to the osmolality of body fluids than the osmolality of traditional contrast agents.
Malpractice cost indicators
Unintended, detrimental effects on a patients health as a result of medical care. The term is commonly applied to secondary infections, adverse drug reactions, injuries, or other complications that may follow treatment.
Indirect malpractice costs A cost of the malpractice system that is not direct-
Factors that reflect direct costs of the medical malpractice system, such as claim frequency, payment per paid claim, and malpractice insurance premiums (see direct malpractice costs).
Multivariate analysis
ly associated with the compensation of persons injured by medical malpractice. Defensive medicine is an example of an indirect cost of the malpractice system (see defensive medicine, compare direct malpractice costs).
Informed consent
Statistical analysis of three or more variables simultaneously. The most widely used form of multivariate analysis is multiple regression analysis, in which a single dependent variable (the presumed effect) is analyzed as a function of two or more independent variables (presumed causes).
Negligence
In medical malpractice, conduct that falls below the prevailing standard of care in the medical profession (see standard of care).
No-fault compensation program
As applied to clinical care, a patients agreement to allow a medical procedure based on full disclosure of the material facts needed to make an in-
A malpractice reform under which certain medical injuries would be compensated regardless of whether they are caused by negligence. This reform
Appendix K: Glossary | 163
would be administered in a manner analogous to workers compensation programs in the states.

Noneconomic damages
Pretrial screening
Monetary damages that compensate the plaintiff for pain and suffering, which includes:
q
tangible physiologic] pain suffered by a victim at the time of injury and during recuperation, the anguish and terror felt in the face of impending death or injury, emotional distress and long-term loss of love and companionship resulting from injury or death of a close family member, and loss of enjoyment of life by the plaintiff who is denied pleasures of a normal person because of physical impairment.
An alternative dispute resolution procedure that parties use prior to filing a legal suit. The pretrial screening panel usually comprises health care professionals, legal experts, and sometimes, consumers. The panel hears the evidence, including expert testimony, and makes a finding on liability and, in certain cases. on damages. Pretrial screening may be voluntary or mandatory, as specified by legislation. The panel decision is not binding on the parties, so parties may continue to pursue claims through the legal system.
Punitive damages
Normal distribution
A bell-shaped frequency distribution of the values of a variable, so that most of the values fall in the middle of the distribution and few of them fall at the extremes.
Odds ratio
Monetary damages awarded when the defendant conduct is found to be intentional, malicious, or outrageous, with a disregard for the plaintiffs well-being. (Punitive damages are rarely awarded in malpractice suits. ) Reliability The reproducibility of a measure. A measure is reliable if it yields similar results each time it is used on similar samples, or if its components yield similar results for the same or similar samples (compare validity).
Res ipsa Ioquitur A legal doctrine that allows plaintiffs with certain
The ratio of the odds of an event occurring under one set of circumstances to the odds of the event occurring under mother set of circumstances.
Patient compensation fund (PCF)
A go~ernment-operated mechanism that pays the portion of any judgment or settlement against a health care providcr in excess of a statutorily designated amount. A PCF may pay the remainder of the award or it may have a statutory maximum (e.g.. $1 million).
Payment per paid claim
types of injuries to prevail without having to introduce expert testimony of negligence. (Literall y, *the thing speaks for itself.) A plaintiff must establish that the procedure or incident causing the injury was under the exclusive control of the physician and that such injuries do not occur in the absence of negligence.
Respectable minority rule
The average dollar amount awarded to plaintiffs for claims that result in payment.
Periodic payments Payments to the plaintiff for future damages made
over the actual lifetime of the plaintiff or for the actual period of disabi1ity rather than in a prospective lump sum.
Point estimate
An exception to the general rule that a physician must meet the prevailing standard of care provided in his or her profession. A physician is shielded from liability when his or her clinical decision is consistent with the practices of a minority of physicians in good standing.
Right of subrogation
A sample-based estimate of the true population value of a statistic-e. g., the mean of a variable (see also confidence interval).
A provision typically found in health and disability insurance contracts that requires a plaintiff to reimburse the insurance company for any pay-
ments received from the tort system that were for services reimbursed by the insurer. Scale A composite statistical measure comprising several variables. Schedule of damages A set of guidelines for juries to use in deciding appropriate awards for noneconomic damages in malpractice cases. Standard of care A legal standard defined as the level of care provided by the majority of physicians in a particular clinical situation. In a malpractice action, a physicians actions are judged against the prevailing standard of care. Negligence is defined as failure to meet the standard of care. Statistical significance A statistically significant finding is one that is unlikely to have occurred solely as a result of chance. Throughout this report, a finding is considered to be statistically significant if the probability that it occurred by chance alone is no greater than five out of 100i.e., a p value of 0.05 or less. Statute of limitations A legal rule that determines how long after an injury one can bring a lawsuit-e. g., t wo years after the injury. In many states, the clock does not start until discovery of the injury. The discovery rule states that the date of injury, from which the statutory time period is measured, is the date that it was reasonable for the plaintiff to have discovered the injury rather than the actual date of injury. Injuries may be discovered years after the treatment was provided, so the time period for filing action may be uncertain. Stratified random sampling A method of drawing a random sample from a population that has been grouped by population characteristics.
Tort law
A body of law that provides citizens a private, judicially enforced, remedy for injuries caused by another person. Legal actions based in tort have three elements: existence of a legal duty from defendant to plaintiff, breach of that duty, and injury to the plaintiff as a result of that breach.
Tort reform
A legal reform that changes the way tort claims are handled in the legal system or removes claims from the civil judicial system.
Tort signal
Direct or indirect signals from the malpractice system that apprise physicians of their liability risk (e.g., litigation exposure of self or peers, malpractice insurance rates, professional literature and popular media). Unweighed results Statistical results based on a disproportionate stratified sample (see stratified random sampling) without applying sampling weights (see weight).
Validity
Broadly, the extent to which an observed situation reflects the true situation. Internal validity is a measure of the extent to which study results reflect the true relationship of an intervention to the outcome of interest in the study subjects. External validity is the extent to which the results of a study may be generalized beyond the subjects of the study to other settings, providers, procedures, diagnostics, etc. (compare reliability).
Weight
A multiplier applied to each element of a given stratum of a sample (see stratified random sampling) so that the sample accurately represents the population from which the sample was drawn. A weight can be thought of as the number of members of the population represented by each respondent.
Weighted results Results to which sampling weights have been ap-
plied (see weight).
References
1. Abraham, K. S., and Weiler, P. C., Organizational Liability for Medical Malpractice: An Alternative to Individual Health Care Provider Liability for Hospital Related Malpractice, unpublished paper, Rich-
2.
mond, VA, December 1992. + AHCPR Exploring Purchase of Liability Insurance for Panel Members, Health News Daily 4(1 40): 1-2, July 21, 1992.
3. American College of Radiology, Commitb tee on Drugs and Contrast Media, Report of the Current Criteria for the Use of Water Soluble Contrast Agents for Intravenous Injections, monograph, Reston, VA, 1990. 4. American Healthcare Systems, Quality and Risk Management: A Mode] Program, 1986/ 1987, San Diego, CA, 1987. 5. American Healthcare Systems, and Johnson & Higgins, Inc., Physician Office Practice Risk Management Manual, monograph, Washington, DC, 1989. 6. American Medical Association. Legal Implications of Practice Parameters (Chicago, IL: 1990). 7. American Medical Association, Socioeconomic Characteristics of Medical Practice (Chicago, IL: Center for Health Policy Research, 1992).
8. American Medical Association, Attributes To Guide the Development of Practice Parameters (Chicago, IL: 1990), 9. American Medical Association/Specialty Society Medical Liability Project, A Proposed Alternative to the Civil Justice System for Resolving Medical Liability Disputes: A Fault-Based, Administrative System, monograph, Chicago, IL, January 1988. 10. Baldwin, L. M., Hart, L. G., Lloyd, M., et al., Department of Family Medicine, University of Washington, Seattle, WA, Malpractice Claims Exposure and Resource Use in Low Risk Obstetrics, prepared under contract to the Office of Technology Assessment, U.S. Congress, Nov. 21, 1993. 11. Becker, E. R., and Sloan, F. A., Utilization of Hospital Services: The Roles of Teaching, Case Mix, and Reimbursement, Inquiry 20:248-57, 1983. 12. Bell, P. A., Legislative Intrusions into the Common Law of Medical Malpractice: Thoughts About the Deterrent Effect of Tort Liability, Syracuse Law Review 35(3): 939-993, 1984. 13. Bernzweig, E. P., Defensive Medicine, Appendix: Report of the Secretarys Com-
I 165
14.
15.
16.
17.
18.
19.
20.
21.
mission on Medical Malpractice, U.S. Department of Health, Education and Welfare, Secretarys Commission on Medical Malpractice, DHEW Pub. No. 73-89 (Washington, DC: U.S. Government Printing Office 1973). Bernweig, E. P., Special Assistant for Malpractice Research and Prevention, Office of the Director of Community Health, U.S. Department of Health Education and Welfare, Washington, DC, * Statement, Medical Malpractice: The Patient Versus the Physician, testimony presented at hearing before the Subcommittee on Executive Reorganization, Committee on Government Operations, Senate, U.S. Congress, Washington, DC, Nov. 20, 1969. Binder, D. A., On the Variances of Asymptomatically Normal Estimators from Complex Survey s, International Statistical Review) 51 (3):279-292, 1983. Binder, D. A., Gratton, M., Hidiroglou, M. A., et al., Analysis of Categorical Data from Surveys with Complex Designs: Some Canadian Experiences, Survey Methodology 10: 141-156, December 1984. Blumstein, J., Rationing Medical Resources: A Constitutional, Legal, and Policy Analysis, Texas Law Review 59: 13451397, 1981. Blustein, J., High-Technology Cardiac Procedures: The Impact of Service Availability on Service Use in New York State, Journal of the American Medical Association 270(3):344-349, July 21, 1993. Borzo, G., Quality Assurance Prenatal Systems Reduce Risk for OBs, American Medical News 36(8):31, Feb. 22, 1993. Bovbjerg, R. R., Senior Research Associate, The Urban Institute, Washington, DC, personal communications, Oct. 20, 1993, Feb. 14, 1994, March 23, 1994, and April 1994. Bovbjerg, R. R., and Havighurst, C. C., Medical Malpractice: Can the Private Sector Find Relief? Law and Contemporary Problems 49(2): 1-321, spring 1986.
Bovbjerg, R. R., Legislation on Medical Malpractice: Further Developments and a Preliminary Report Card, University of Davis Law Review 22:499-556, 1989. 23. Bovbjerg, R. R., Reforming A Proposed Tort Reform: Improving on the American Medical Associations Proposed Administrative Tribunal for Medical Malpractice, Courts, Health Science & The Law 1(1): 19-28, Ju]y 1990. 23a. Bovbjerg, R. R., Sloan, F. A., Blumstein, J. F., Valuing Life and Limb in Tort: Scheduling Pain and Suffering, Northwestern University Law Review 83(4):908-76, 1989. 24. Bovbjerg, R. R., and Tancredi, L. R., " Reform of the Medical Malpractice System Should Go Beyond ADR and Tort Law, World Arbitration and Mediation Report 3(3): 75-77 (London, England: BNA International Inc., 1992). 25. Bovbjerg, R. R., Tancredi, L. R., and Gaylin, D. S., Obstetrics and Malpractice: Evidence on the Performance of a Selective NoFault System, Journal of the American Medical Association 265(21):2836-2843, June 5, 1991. 26. Brennan, T. A., Professor of Law and Public Health, Harvard School of Public Health, Boston MA, personal communication, Jan. 25, 1994. 27. Brenner, R. J., Medicolegal Aspects of Breast Imaging: Variable Standards of Care Relating to Different Types of Practice, American Journal of Radiology 156:719723, April 1991. 28. Brenner, R., and Sickles, E., Acceptability of Periodic Follow-up as an Alternative to Biopsy for Mammographically Detected Lesions Interpreted as Probably Benign, Radiology 171 :645-646, 1989. 29. Briggs, T., Chief Operation Officer, Medical Mutual Insurance Company of Maine, Portland, ME, personal communication, May 3, 1994. 30. Brook, R., q Practice Guidelines and Practicing Medicine: Are They Compatible?
22.
References | 167
Journal of the American Medical Association 262(21):3027-3030, Dec. 1, 1989. 31.

Burstin, H. R., Johnson, W. G., Lipsitz, S. R.. and Brennan, T. A., DO the Poor Sue More?
41.
32.
33.
A Case Study of Malpractice Claims and Socioeconomic Status, Journal of the American Medical Association 270( 14): 1697-1701, 1993. California Medical Association. Actuarial Study of Professional Liability Insurance, Newport Beach, CA, May 31, 1985. California Public Employees Retirement System, Health Benefits Advisory Council,
"Designing a Fair and Reasonable Basic Benefit Plan Using Clinical Guidelines:
42.
43.
34.
35.
36.
37.
38.
39.
40.
Summary of a Conference Sponsored by CalPERS, CalPERS, Sacramento, CA. 1992. Californians Allied for Patient Protection, Micra Information Booklet, Sacramento, CA, Jan. 1, 1993. "Campaign 92: Transcript of the First Presidential De bate, The Washington Post. Oct. 12, 1992, p. Al 6. Campion, F. X., The Risk Management Foundation of the Harvard Medical institutions, Inc., and American Medical Association, Grand Rounds on Medical Malprac tice (Chicago. IL: American Medical Association, 1990). Charles, S. C., Wilbert, J. R., and Franke, K. J., "Sued and Nonsued Physicians SelfReported Reactions to Malpractice Litigat ion , American Journa[ of Psychiatry 142(4):437-440, April 1985. Charles. S. C., Wilbert, J. R., and Kennedy, E. C., Physicians Self-Reports of Reactions to Malpractice Litigation, American Journal of Psychiatry 141(4):563-565, April 1984. Chassin, M. R., Explaining Geographic Variations: The Enthusiasm Hypothesis, Medical Care 31(5): YS37-YS44, supplement 1993, Cheney, F.. Posner, K., Caplan, R. A., et al.. Standard of Care and Anesthesia Liabil-
44.
45.
46.
47.
48.
49. 50.
51.
Journal of the American Medical Association 261 (11 ): 1599-1603, 1989. Cohen, M., Hospital Risk Manager, University of California at Davis Medical Center, Sacramento, CA. personal communications, Nov. 15 and 25, 1993. Cooper, M. I., Senior Counsel, Medical Legal Section, Regional Legal Department, Northern California, Kaiser Permanence Medical Care Program. Oakland, CA, personal communications, Oct. 25 and Nov. 30, 1993, COPIC Insurance Company, Handbook on COPIC Participatory Risk Management Program, Englewood, CO, 1993. Corder-Mabe. J., Nurse Consultant, Division of Woman and Infants, Department of Health, Commonwealth of Virginia, Richmond, VA, personal communication, Nov, 30, 1993. Crane, M., Could a Malpractice Suit Wipe Out Your Assets? Medic.al Economics 69(1 3): 146-153, July 6, 1992. Creasy, D., President. Risk Management Foundation of Harvard Medical Institutions, Inc., Cambridge, MA, personal communication. August 1993. Crothers, L. S.. Professional Standards Review and the Limitation of Health Services, Boston University Law Review 54(5):931-945 ( 1974). Cm-ran, W. J., and Moseley, G. B., . The Malpractice Experience of Health Maintenance Organizations, Northwestern University Law Review 70(1 ):69-89, 1975. Cytel Software, StatXact-Turbo (Cambridge, MA: 1992). Daly, L., Simple SAS Macros for the Calculation of Exact Binomial and Poisson Confidence Limits, Computer in Biology and Medicine 22:351-361, 1992. Dasse, P. S., Vice President of Loss and Prevention, Risk Management Foundation of the Harvard Medical Institutions, Inc., Boston, MA, personal communication, Feb. 25, 1994.
ity,"
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
Dasse, P. S., Commentary: House Staff Risk Management Issues, Forum 14(4): 1-2, September 1993. David, E., Chairman, Maine Board of Registration in Medicine, Bangor, ME, personal communication, Dec. 15, 1992. Development of Clinical Practice Guidelines Outside AHCPR Panels, Health News Daily 4(189):5-6, Sept. 29, 1992. Dornette, W. H. L., The Legal Impact of Voluntary Standards in Civil Actions Against the Health Care Provider, New York Law School Law Review 22:925-42, 1977. Dubois, R. W., and Brook, R. H., Assessing Clinical Decision Making: Is the Ideal System Feasible? Inquiry 25:59-64, spring 1988. Duff, D. G., q Compensation for Neurologically Impaired Infants: Medical No-Fault in Virginia, Harvard Journal of Legislation 27:391-451, 1990. Duke Law Journal, The Medical Malpractice Threat: A Study of Defensive Medicine, Duke Law Journal 1971:939-993, 1971. Eddy, D. M., Clinical Decisionmaking: from Theory to Practice, pts. 1-4, Journal of the American Medical Association 263(2): 287-290; 263(3):441-443; 263(6): 877-880; 263(9):1265-1275; 263(22):30773084, 1990. Eichorn, J. H., Cooper, J. B., Cullen, D. J., et al., Standards for Patient Monitoring During Anesthesia at Harvard Medical School, Journal of the American Medical Association 256(8): 1017-1020, Aug. 22-29, 1986. Eisenberg, J. M., Physician Utilization: The State of Research About Physicians Practice Patterns, Medical Care 23(5):461483, May 1985. Eisenberg, J. M., and Nicklin, D., q Use of Diagnostic Services by Physicians in Community Practice, Medical Care 19:297, 1981.
63.
64.
65.
66.
67.
68.
69. 70.
71.
Eliastam, M., Rose, E., Jones, H., et al., Utilization of Diagnostic Radiologic Examinations in the Emergency Department of a Teaching Hospital, Journal of Trauma 20(1):61-6, 1980. Epstein, A. M., Begg, C. B., and McNeil, B. J., The Use of Ambulatory Testing in Prepaid and Fee-for-Service Group Practices: Relation to Perceived Profitability, New England Journal of Medicine 314: 1089-1094, 1986. Epstein, A. M., and McNeil, B.J., Relationship of Beliefs and Behavior in Test Ordering, American Journal of Medicine 80:865-870, May 1986. Every, N. R., Larson, E. B., Litwin, P. E., et al, The Association Between On-Site Cardiac Catheterization Facilities and the Use of Coronary Angiography After Acute Myocardial Infarction, The New England Journal of Medicine 329(8):546-551, Aug. 19, 1993. Failure To Diagnose Claims Head List of Allegations, St. Paul Finds, Medical Liability Monitor 18(8):5-6, Aug. 16, 1993 Farber, H. S., and White, M.J., Medical Malpractice: An Empirical Examination of the Litigation Process, working paper, National Bureau of Economics, Cambridge, MA, September 1990. FOX, R. B., The Sociology of Medicine (Englewood Cliffs, NJ: Prentice Hall, 1986). Gagnon, R., Chief, Ambulatory Care Statistics Branch, Division of Health Statistics, Office of Vital and Health Statistics, National Center of Health Statistics, Public Health Statistics, U.S. Department of Health and Human Services, Hyattsville, MD, personal communication, Dec. 17, 1993. Garg, M. L., Gliebe, W. A., and Elkhatib, M. B., The Extent of Defensive Medicine: Some Empirical Evidence, Legal Aspects of Medical Practice, February 1978, pp. 25-29.
References | 169
72.
Gifford, G., Research Analysis Specialist, Minnesota Department of Health, Minneapolis, MN, personal communications, July 12, 1992, Nov. 1993 and May 18, 1994. 73. Glassman, P. A., Petersen, L. P., Bradley, M. A., et al., The Effect of Malpractice Experience on Physicians Clinical DecisionMaking, paper prepared under contract to the Office Technology Assessment, U.S. Congress, Washington, DC, 1993. 74. Goldman, E., Director Risk Management, University of Michigan, Arm Arbor, Michigan, MI, personal communication, Feb. 17, 1994 75. Goldschmidt, P. G., Can Practice Guidelines Reduce Malpractice Claims? (comment), JournaI of the American Medical Association 267(19):2602-2603, May 20, 1992. 76. Goold, S. D., Lecturer, Department of Internal Medicine, University of Michigan, Ann Arbor MI, personal communication, Oct. 1, 1993. 77. Goold, S. D., Hofer, S., Zimmerman, M. A., et al., Measuring Physician Attitudes Postulated To Influence Practice Style, Abstracts in Clinical Research 40:613A, 1992. 78. Goyert, G. L., Bottoms, S. F., Treadwell, M. C., et al., The Physician Factor in Caesarean Birth Rates, New England Journal of Medicine 320(11):706-709, Mar. 16, 1989. 79. Gronfein, W. P., and Kinney, E. D., *Controlling Large Malpractice Claims: The Unexpected Impact of Damage Caps, Journal of Health Politics, Policy and Law 16(3): 441-483, fall 1991. 80. Gross, J., Manager, Clinical Quality Improvement, University of California at Davis Medical Center, Sacramento, CA, personal communication, Nov. 16, 1993. 81. Grumbach, K., Peltzman-Rennie, D., and Luft, H. S., Charges for Obstetric Liability Insurance and Discontinuation of Obstetric Practice in New York, prepared under contract for the Office of Technology Assess-
ment, U.S. Congress, Washington, DC, Dec. 7, 1993. 82. Haas, J. S., Udvarhelyi, S., and Epstein, A. M., The Effect of Health Coverage for Uninsured Pregnant Women on Maternal Health and the Use of Caesarean Section, Journal of the American Medical Association 270( 1 ):61 -64, July 7, 1993. 83. Hall, M. A., Professor of Law and Public Health, Wake Forest University School of Law, Winston-Salem, NC, personal communication, Feb. 24, and Mar. 17, 1994. 84. Hall, M. A., The Effect of Insurance Coverage Law on Defensive Medicine, prepared under contract to the Office of Technology Assessment (Springfield VA: National Technical Information Service, Aug. 25, 1 993). 85. Hall, M. A., *The Defensive Effect of Medical Malpractice Policies in Malpractice Litigation, Law and Contemporary Problems 54(2): 119-145, spring 1991. 86. Hall, M. A., The Malpractice Standard Under Health Care Cost Containment, Law, Medicine, & Health Care 17:347, 1989. 87. Hall, F. M., Storella, J. M., Silverston, D., et al., Nonpalpable Breast Lesions: Recommendations for Biopsy Based on Suspicion of Carcinoma at Mammography, Radiology 167:353-358, 1988. 88. Harvard Law Review, Rethinking Medical Malpractice Law in Light of Medicare CostCutting (note), Harvard Law Review 94: 1004-1022, March 1985. 89. Health Insurance Association of America, Source Book of Health Insurance Data: 1992 (Washington, DC: Health Insurance Association of America, 1992). 90. Heland, K., Associate Director, Department of Professional Liability, American College of Obstetricians and Gynecologists, Washington DC, personal communication, Nov. 29, 1993. 91. Hemenway, D., Killen, A., Cashman, S. B., et al., Physicians Responses to Financial Incentives: Evidence from a For-Profit Am-
bulatory Care Center, New England Journal of Medicine 322(1 5): 1059-1063, 1990. 92. Hickson, G. B., Altemeier, W. A., and Perrin, J. M., Physician Reimbursement by Salary or Fee-for-Service: Effect on Physician Practice Behavior in a Randomized Prospective Study, Pediatrics 80:344-350, 1987. 93. Hillman, B.J., Joseph, C. A., Mabry, M. R., et al., Frequency and Costs of Diagnostic Imaging in Office PracticeA Comparison of Self-Referring and Radiologist-Referring Physic ians, The New England Journal of Medicine 323(23): 1604-1608, Dec. 6, 1990. 94. Hirshfeld, E. B., q *Should Practice Parameters Be the Standard of Care in Malpractice Litigation? Journal of the American Medical Association 266(20):2886-2891, Nov. 27, 1991. 95. Hirshfeld, Jr., J. W., LOW Osmolality Contrast AgentsWho Needs Them? New England Journal of Medicine 326(7):48284, Feb. 13, 1991. 96. Hodson, J. D., Medical Malpractice: Loss of Chance Causality, American Law Reports, 4th Ed., vol. 54, sees. 2-6 (Rochester, NY: Lawyers Cooperative Publishing, 1987). 97. Hoey, J., Eisenberg, J. M., Spitzer, W. O., et al., Physician Sensitivity to the Price of Diagnostic Tests: A U.S.-Canadian Analysis, Medical Care 20:302-07, March 1982. 98. Holder, A. R., Medical Malpractice Law, 2nd Ed. (New York: John Wiley and Sons, 1978). 99. Holzer, J. F., The Advent of Clinical Standards for Professional Liability, Quality Review Bulletin 16(2):71 -79, February 1990. 100. Hyams, A., Brandenburg, J., Lipsitz, S., et al., Practice Guidelines and Malpractice Litigation, report prepared for the Physician Payment Review Commission, Washington, DC, Jan. 25, 1994. 101. Institute of Medicine, Guidelines for Clinical Practice: From Development to Use
(Washington, DC: National Academy Press, 1992). 102. Institute of Medicine, Medical Professional Liability and the Delivery of Obstetrical Care (Washington, DC: National Academy Press, 1989). 103. Jacobson, P. D., Who Decides Who Gets Low-Osmolar Contrast, Diagnostic Imaging 13(4):77-84, April 1991. 104. Jacobson, P. D., and Rosenquist, C.J., The Introduction of Low Osmolar Contrast Agents in Radiology: Medical, Economic, Legal, and Public Policy Issues, Journal of the American Medical Association 260: 1586-1592, Sept. 6, 1988. 105. Jacobson, P. D., and Rosenquist, C. J., The Diffusion of Low Osmolality Contrast Agents: Technological Change and Defensive Medicine, prepared under contract for the Office of Technology Assessment, U.S. Congress, Washington, DC, November 1993. 106. Kakalik, J., and Pace, N., Costs and Compensation Paid in Tort Litigation (Santa Monica, CA: Rand Corp., 1986). 107. Kalton, G., and Kasprzyk, D., The Treatment of Missing Survey Data, Survey Methodology 12: 1-16, 1986. 108. Kaynor, M., Underwriting Manager, Harvard Risk Management Foundation, Cambridge, MA, personal communication, Aug. 8, 1993. 109. King, J. H., The Law of Medical Malpractice, 2nd Ed. (St. Paul, MN: West Publishing Company, 1986). 110. King, J. H., Causation, Valuation, and Chance in Personal Injury Torts Involving Preexisting Conditions and Future Consequences, The Yale Law Journal 90: 13531397, 1981. 111. King, J. H., Jr., In Search of a Standard of Care for the Medical Profession: The Accepted Practice Formula, Vanderbilt Law Review 28(6): 1213-1276, 1975. 112. Kington, R. S., School of Medicine, University of California, Los Angeles, CA, Li-
References I 171
ability and the Practice Patterns of Obstetricians and Gynecologists, draft working paper provided to the Office of Technology Assessment, March 1994. 113. Kinney, E., and Wilder, P., Medical Standard Setting in the Current Malpractice Environment: Problems and Possibilities, University of California/Davis Law Review 22(2):421-450, 1989. 114. Kish, L., Survey Sampling (New York, NY: Wiley, 1965). 115. Kladiva, S., Assistant Director for Health Financing Issues, Human Resources Division, General Accounting Office, U.S. Congress, Washington, DC, personal communication, Sept. 30, 1993. 116. Kleinman, K., PSRO: Malpractice Liability and the Impact of the Civil Immunity Clause, Georgetown Law Journal 62: 1499-1513, 1974. 117. Kern, E. L., and Graubard, B. I., Epidemiologic Studies Utilizing Surveys: Accounting for the Sampling Design, American Journal of Public Health 81: 1166-1173, 1991. 118. Kraus, J.F. Epidemiology of Head Injury, Head Injury, 3rd Ed., P.R. Cooper (cd.) (Baltimore, MD: Williams & Wilkins, 1993). 119. Kravitz, R. L., Rolph, J. E., and McGuigan, K., Malpractice Claims Data as a Quality Improvement Tool: I. Epidemiology of Error in Four Specialities, Journal of the American Medical Association 266(15): 2087-2092, Oct. 16, 1991. 120. LaCava, F. W., *The Roles of Legal Counsel in Hospital Risk Management, Quality Review Bulletin 11(1) :20-24, January 1985. 121. Langa, K. M., and Sussman, E. J., "The Effect of Cost-Containment Policies on Rates of Coronary Revascularization in California, New England Journal of Medicine 329(24): 1784-1789, Dec. 9.1993. 122. Law, S., A Consumer Perspective on Medical Malpractice, Law and Contemporary Problems 49(2):305-320, spring 1986.
123. Lawthers, A., Localio, A., Laird, N., et al., Physicians Perceptions of the Risk of Being Sued, Journa[ of Health Politics, Policy, and Law 17(3):463-482, 1992. 124. Levy, P. S., and Lemeshow, S., Sampling of Populations (New York, NY: Wiley, 1991 ). 125. Lewin-VHI, Inc. Estimating the Costs of Defensive Medicine, report prepared for MMI Companies, Inc., Fairfax, VA, Jan. 27, 1993 126. Linn, L. S., Yager, J., Leake, B. D., et al., Differences in the Numbers and Costs of Tests Ordered by Internists, Family Physicians, and Psychiatrists, inquiry 21(3 ): 266-275, fall 1984. 127. Litan, R.E. (cd.), Verdict Assessing the Civil Jury System (Washington, DC: The Brookings Institution. 1 993). 128. Localio. A. R., Lawthers, A. G., Bengtson, J. M., et al., Relationship Between Malpractice Claims and Caesarean Delivery, Journal of the American Medica[ Association 269(3):366-373, Jan. 20, 1993. 129. Localio, A. R., Lawthers, A. G., Bengston, J. M., et al.. The Relationship Between Malpractice Claims Risk and Caesarean Delivery, report prepared for the Agency for Health Care Policy and Research, Public Health Service, U.S. Department of Health and Human Services, Grant No. R03 HS 07070-01, Rockville, MD, Sept. 12.1991. 130. Localio, A. R., Lawthers, A. G., Brennan, T. A., et al.. "Relation Between Malpractice Claims and Adverse Events Due to Negligence: Results of the Harvard Medical Practice Study III. New England Journal of Medicine 325:245-25 1, July 25, 1991. 131. Localio. A. R., Weaver, S. L.. Landis. J. R., et
al.. Clinical Decision-Making on Adverse Events in Medical Care: Measuring Agreement and Implications for Professional Li-
ability. Final Report for the Agency for Health Care Policy and Research, U.S. Department of Health and Human Services, Grant No. ROl HS07067-01, Washington, DC, July 30, 1993.
172
Defensive Medicine and Medical Malpractice
132 Lomas, J., Anderson, G. M., DomnickPierre, K., et al., Do Practice Guidelines Guide Practice? New England Journal of Medicine 321 (19): 1306-1311, Nov. 9,1989. 133. Long, M.J., Cummings, K. M., and Frisof, K. B., The Role of Perceived Price in Physicians Demand for Diagnostic Tests, Medical Care 21:243, 1983. 134. Louisell, D. W., William, H., Medical Malpractice, vol. 1, sec. 9.05 (New York, NY: Mathew Bender& Co. Inc., 1993). 135. Luft, H. S., Variations in Clinical Practice Patterns, Archives of Internal Medicine 143(10): 1861-1862, October 1983. 136. Luft, H. S., How Do Health Maintenance Organizations Achieve Their Savings: Rhetoric and Evidence, New England Journal of Medicine 298: 1336-1343, June 15, 1978. 137. Lundberg, J., Deputy General Counsel, Regents of the University of California, Oakland, CA, personal communication, October 1993 and Dec. 21, 1993. 138. Macario, A., Roizen. M. F., et al., A Tale of Three Cities: Has Reassessment of Preoperative Laboratory Testing Changed the Test-Ordering Patterns of Physicians? Journal of Surgical Gynecology and Obstetrics (forthcoming) as cited in Lewin-VHI, Inc., Estimating the Costs of Defensive Medicine, report prepared for MMI Companies, Inc., Fairfax, VA, Jan. 27, 1993. 139. MacCoun, R.J., Lind, E. L., Hensler, D. R., et al., Alternative Adjudication: An Evaluation of the New Jersey Automobile Arbitration Program (Santa Monica, CA: RAND Corp., 1988). 140. Manning, W. G., Leibowitz, A., Goldberg, G. A., et al., A Controlled Trial of the Effect of a Prepaid Group Practice on Use of Services, New England Journal of Medicine 310: 1505-1510, Jan. 7, 1984. 141. Marion Merrell Dow, Inc., Managed Care Digest: HMO Edition, monograph, Kansas City, MO, 1993.
142. Markowitz, R., President, Federation of Jewish Philanthropies Service Corporation, personal communication, Feb. 7, 1994. 143. Marsh, F. H., Health Care Cost Containment and the Duty To Treat , The Journal of Legal Medicine 6(2): 157-190, 1985. 144. Marton, K. I., Sex, Jr., H. C., Wasson, J., et al., The Clinical Value of the Upper Gastrointestinal Tract Roentgenogram Series, Archives of Internal Medicine 140:191-5, 1980. 145. May, M. L., and Stengel, D. B., Who Sues Their Doctors? How Patients Handle Medical Grievances, Law and Society Review 24(1): 105-119, 1990. 146. McClure, J. D., The Distress of Internship: Causes and Prevention, New England Journal of Medicine 312(7):449-452, February 1985. 147. McCormick, B., Public Access To Data Bank Dropped, Likely To Resurface, American Medical News 36(43): 10, Nov. 15, 1993. 148. McCormick, B., Reformers Willing, Critics Wary on Alternative Tort System, American Medical News 36:63-67, Apr. 26, 1993. 149. McCormick, B., In Face of Doctor Onslaught: Enterprise Liability Backers Stand Firm, American Medical News 36(l): 35-37, June 21, 1993. 150. McCormick, C. T., McCormick on Evidence, 3rd Ed., (Lawyers Edition), E. Cleary (cd.), (St. Paul, MN: West Publishing Co., 1984). 151. McMahon, M. J., Medical Malpractice: Measure and Elements of Damages in Actions Based on Loss of Chance American Law Reports, 4th Ed., vol. 81, sees. 1-20 (Rochester, NY: Lawyers Cooperative Publishing, 1990). 152. Medical Board of California, Medical Board 92: MBC Annual Report 91/92 Fiscal Year, Sacramento, CA, 1992.
References | 173
153. Medical Board of California, "MBC Complaint Process, information sheet, Sacramento, CA, April 1991. 154. Medical Mutual Insurance Company of Maine, Participatory Risk Management Program, Portland, ME, 1993. 155. Mehlman, M. J., Assuring the Quality of Medical Care: The Impact of Outcome Measurement and Practice Standards, Law, Medicine, & Health Care 18(4):368-384, winter 1990. 156. Mehta, C. R., Patel, N. R., and Gray, R., On Computing Exact Confidence Intervals for the Common Odds Ratio in Several 2 x 2 Contingency Table s, Journal of the Amerian Statistical Association 80:969-973, 1985. 157. Meierhoefer, B. S., Court-Annexed Arbitration in Ten District Courts (Washington, DC: The Federal Judicial Center, 1990). 158, Metzloff, T. B., Defensive Medicine and the Use of Medical Technology: Physician Involvement in Medical Malpractice Litigation, prepared under contract for the Office of Technology Assessment, U.S. Congress, Washington, DC, January 1994. 159. Metzloff, T. B., Alternative Dispute Resolution Strategies in Medical Malpractice, Alaska Law Review 9(2):429-457, 1992. 160. Meyer, J. E., Eberlein, T., Stomper, P., et al., Biopsy of Occult Breast Lesions: Analysis of 1261 Abnormalities, Journal of the American Medical Association 263(17): 2341-2343, May 2, 1990. 161. Miller, F. H., and Harrison, A., Malpractice Liability and Physician Autonomy, The Lancet 342:973-975, 1993. 162. Miller, M. E., Zuckerman, S., Gates, M., q How Do Medicare Physician Fees Compare with Private Payers? Health Care Financing Review 14:25-39, 1993. 163. Mills, D. H., Medical Director, Professional Risk Management Corporation, Long Beach, CA, personal communications, Oct. 2 and 29, and Nov. 5 and 9, 1993.
164. Minnesota, State of, Minnesota Health Rights Act: Conference Committee Report on HF No. 200, Conference Report (St. Paul, MN: 1992). 165. Montague, J., Breaking the Bank? MDs Say NPDB Needs Screening, Hospitals & Health Networks 67(20):51, Oct. 20, 1993. 166. Morlock, L. L., and Malitz, F. E., ShortTerm Effects of Tort and Administrative Reforms on the Claiming of Behavior of Privately Insured, Medicare, Medicaid and Uninsured Patients, prepared under contract for the Office of Technology Assessment, U.S. Congress, Washington, DC, Sept. 30, 1993. 167. Morlock, L. L., and Malitz. F. E., Do Hospital Risk Management Programs Make a Difference?: Relationships Between Risk Management Program Activities and Hospital Malpractice Claims Experience, Law and Contemporary Problems 54(2): 1-22, spring 1991. 168. Morreim, E. H., Cost Constraints as a Malpractice Defense, Hastings Center Report 18(1):5-10, February-March 1988. 169. Morreim, E. H., Cost Containment and the Standard of Medical Care, California Law Review 75:17 19-1763, 1987. 170. Morse, W., Health Insurance Specialist, Division of Medical Services Payment, Bureau of Policy Development, Health Care Financing Agency, U.S. Department of Health and Human Services, Washington DC, personal communication, Aug. 4, 1993. 171. National Association of Insurance Commissioners, Malpractice Claims: Medical Malpractice Closed Claims, 1975-1978 (Brookfield, WI: National Association of Insurance Commissioners, 1980). 172. National Cancer Institute, National Institutes of Health, U.S. Department of Health and Human Services, Bethesda, MD, Statement: Breast Cancer Screening, Dec. 3.1993.
173. Nelson, C. R., and Startz, R., The Distribution of the Instrumental Variables Estimator and Its t-Ratio When the Instrument Is a Poor One, Journal of Business Review 63( 1), pt. 2 S 125-S 140( 1990). 174. No-Fault Malpractice Reform: An Unproven Rx for Medical Liability, State Health Notes 15(177):4-5, April 4, 1994. 175. Noren, J., Frazier, T., Altman, I., et al., Ambulatory Medical Care: A Comparison of Internists and Family Practitioners, New England Journul of Medicine 302(1):11 -16, Jan. 3, 1980. 176. Nye, D. J., Gifford, D. G., Webb, B. L., et al., The Causes of the Medical Malpractice Crisis: An Analysis of Claims Data and Insurance Company Finances, Georgetown Law Journal 76: 1495-1561, April 1988. 177. Oh, H. L., and Scheuren, F. J., *Weighting Adjustment for Unit Nonresponse, Incomplete Data in Sample Surveys: Theory and Bibliographies, W.G. Madow, 1. Olkin, and D.B. Rubin (eds. ) (New York, NY: Academic Press, 1983). 178. Ohio Law Journal, causation in Medical Malpractice: A Modified Valuation Approach (note), Ohio Law Journal 50:469485, 1989. 179. Peters, J. D., Attorney, Charfoos and Christensen. Detroit, MI, personal communication, Mar. 3, 1993. 180. Peters, J. D., Nerd, S. K., and Woodson, R. D., An Empirical Analysis of the Medical and Legal Professions Experiences and Perceptions of Medical and Legal Malpractice, University of Michigan Journal of Law Reform 19(3):601-636, spring 1986. 181. Physician Insurers Association of America, Medication Error Study, monograph, Lawrenceville. NJ, June 1993. 182. Physician Insurers Association of America, Lung Cancer Study, monograph, Lawrenceville, NJ, January 1992. 183. Physician Insurers Association of America, Colon Cancer Study, monograph, Lawrenceville, NJ, May 1991.
184. Physician Insurers Association of America, Breast Cancer Study, monograph, Lawrenceville, NJ, March 1990. 185. Physician Insurers Association of America, A Comprehensive Review of the Alternatives to the Present System of Resolving Medical Liability Claims, Lawrenceville, NJ, 1989. 186. Physician Insurers Association of America, Data Sharing Reports, Cumulative Reports, January 1, 1985-June 30, 1989, monograph, Lawrenceville, NJ, 1989. 187. Physician Payment Review Commission, unpublished data on average per-service Medicare reimbursement rates supplied by Chris Hogan, Principal Policy Analyst, personal communication, Washington, DC, 1994. 188. Physician Payment Review Commission, Annual Report 1992 (Washington, DC: 1992). 189. Physician Payment Review Commission, Annual Report 1989 (Washington, DC: Physician Payment Review Commission, 1989). 190. q Physicians, Surgeons, etc.Standard of Care (See 205), American Jurisprudence 2nd. Ed., vol. 61 (Minneapolis, MN: West Publishing Company, 1981) 191. Posner, J. R., q Trends in Medical Malpractice Insurance, 1970- 1985, Law and Contemporary Problems 49(2):37-56, spring 1986. 192. Prospective Payment Assessment Commission, unpublished data supplied by Deborah Williams, Senior Policy Analyst, personal communications, Jan. 21 and Feb. 3, 1994. 193. Research Triangle Institute, SUDAAN, Release 5.50 (Research Triangle Park, NC: 1992). 194. Reynolds, R. A., Rizzo, J. A., and Gonzalez, M. L., The Cost of Medical Professional Liability, Journal of the American Medical Association 257(20):2776-2781 , May 22/29, 1987. 195, Rice, T. H., The Impact of Changing Medicare Reimbursement Rates on Phvsician-In-
References | 175
duced Demand, Medical Care 21 :803-815,

1983. 196. Risk Management Foundation of the Har-
vard Medical Institutions, Inc., Forum (published bimonthly). 197. Risk Management Foundation of the Harvard Medical Institutions, Inc., Malpractice Insurance Program: Information Booklet, 1993 -1994, Cambridge MA, 1993. 198. Rock, S. M., Malpractice Premiums and Primary Caesarean Section Rates in New York and Illinois, Public Health Reports 103(5):459-463, September-October 1988. 199. Rosenblatt, R., b Rationing Normal Health Care: The Hidden Legal Issues, Texas Law Review 59: 1401-1418, 1981.
200. Rosenfield, H., Voter Revolt, California MICRA: Profile of a Failed Experiment in Tort Law Restrictions, Los Angeles, CA (no date). 201. Rubsamen, D., Who Benefits When Mal-
practice Case Outcome Ends in a Hung Jury? Physician Financial News 10(21 ): 29, Dec. 15, 1992.
202. Saks, M., Do We Really Know Anything About the Behavior of the Tort Litigation Systemand Why Not? University of
Pennsylvania Law Review 4(140):11471292, 1992. 203. SAS Institute, Inc., SAS, Release 6.03 (Cary, NC: 1988). 204. Schwartz, J. S., Ball, J. R., and Moser, R. H., Safety, Efficacy, and Effectiveness of Clinical Practice: A New Initiative, Annals of Internal Medicine 96:246-47, February 1982. 205. Schwartz, M., Martin, S. G., Cooper, D. D., et al., The Effect of a Thirty Percent Reduction in Physician Fees on Medicaid Surgery Rates in Massachusetts, American Journal of Public Health 71 :370-375, 1981. 206. Schwartz, W. B., and Mendelson, D. N., Physicians Who Have Lost Their Malpractice Insurance, Journal of the American Medical Association 262( 10): 1335-1341, Sept. 8, 1989.
207. Schwartz, W. B., and Mendelson, D. N., The Role of Physician-Owned Insurance Companies in the Detection and Deterrence of Negligence, Journal of the American Medical Association 262(10): 1342-1346, Sept. 8, 1989. 208. Scitovsky, A. A., The Use of Medical Services Under Prepaid and Fee-for-Service Group Practice, Social Science and Medicine 15: 107-116, 1981. 208a. Shapiro, R. S., et al., A Survey of Sued and Nonsued Physicians and Suing Patients, Archives of Internal Medicine 149:21 90-96, October 1989. 209. Shmanske, S., and Stevens, T., q The Performance of Medical Malpractice Review Panels, Journal of Health Politics, Policy and Law 11 (3):525-535, fall 1986. 210. Sloan, F. A., Bovbjerg, R. R., and Githens, P. B., Insuring Medical Malpractice (New York, NY: Oxford University Press, 1991 ). 211. Smarr, L., Executive Director, Physicians Insurance Association of America, Washington, DC, personal communication, Mar. 3, 1994. 212. Smith, G., Maine Medical Association, Manchester, ME, personal communication, Nov. 23, 1993. 213. Staiger, D., Asymptotic for Instrumental Variables Regressions with Weakly Correlated Instruments, Working Paper (Boston MA: National Bureau of Economic Research, July 1993). 214. State of Florida Health Care Cost Containment Board, Joint Ventures Among Health Care Providers in Florida, vol. II (Tallahassee, FL: 1991 ). 215. Stolt, R., Attorney, Lipman & Katz, Augusta, ME, personal communication, June 24, 1992. 216. Stuller, J., Settling for Bearable Unhappiness, Across the Board 1993, pp. 17-22. 217. Tan, M. W., and Bierman, E., Shoulder Dystocia/Erbs Palsy Claims, F o r u m 14(5): 13-15, November 1993.
218. Tancredi, L. R., M. D., J. D., Private Consultant, New York, N. Y., personal communication, April 1994. 219. Tancredi, L.R. Identifying Avoidable Events in Medicine, Medical Care 12:935-943, 1974. 220. Tancredi, L. R., and Bovbjerg, R. R., q Rethinking Responsibility for Patient Injury: Accelerated-Compensation Events, A Malpractice and Quality Reform Ripe for a Test, Law and Contemporary Problems 54(2): 147-178, spring 1991. 221. Tancredi, L. R., and Bovbjerg, R. R., Creating Outcomes-Based Systems for Quality and Malpractice Reform: Methodology of Accelerated-Compensation Events, Milbank Quarterly 70: 183-216, 1992. 222. Taragin, M. I., Willett, L. R., Wilczek, A. P., et al., The Influence of Standard of Care and Severity of Injury on the Resolution of Malpractice Claims, Annals of Internal Medicine 117(9):780-784, Nov. 1, 1992. 223. Thomasson, G., Vice-President, Medical Risk Management, COPIC, personal communication, Feb. 18, 1994. 224. Thorton, L., Supervising Investigator, Central Complaint and Investigation Control Unit, Medical Board of California, Sacramento, CA, personal communication, Dec. 28, 1993. 225. Tierney, W. M., Miller, M. E., and McDonald, C. J., The Effect on Test Ordering of Informing Physicians of the Charges for Outpatient Diagnostic Tests, The New England Journal of Medicine 322(21):14991504, May 24, 1990.
229.
230.
231.
232.
233.
234.
226. Todd, J. S., Reform of the Health Care System and Professional Liability, New England Journal of Medicine 329(23):17331734, Dec. 2, 1993. 227. Todd, J. S.. Only Parameters Will Give MDs Needed Flexibility (1988 Top Stories; The Year-in review) (interview),* American Medical News 32(1): 15, col. 1, Jan. 6, 1989. 228. Ulsaker, S., Counsel and Graham, T., V.P. Actuarial, Division of Medical Services, The St. Paul Fire and Marine Insurance
235.
236.
Company, St. Paul, MN, personal communication, Apr. 1, 1994. U.S. Congress, General Accounting Office, Medical Malpractice: Maines Use of Practice Guidelines To Reduce Costs, Report No. GAO/HRD-94-8 (Washington, DC: U.S. Government Printing Office, October 1993). U.S. Congress, General Accounting Office, Medical Malpractice: MedicarelMedicaid Beneficiaries Account for a Relatively Small Percentage of Malpractice Losses, GAO/ HRD-93-126 (Washington, DC: U.S. Government Printing Office, August 1993). U.S. Congress, General Accounting Office, Medical Malpractice: Alternatives to Litigation, GAO/HRD-92-28 (Washington, DC: U.S. Government Printing Office, January 1992). U.S. Congress, General Accounting Office, Practice Guidelines: The Experience of Medical Specialty Societies Report No. GAO/PEMD-91-l 1 (Washington, DC: U.S. Government Printing Office, February 1991). U.S. Congress, General Accounting Office, Medical Malpractice: Few Claims Resolved Through Michigans Voluntary Arbitration Program, GAO/HRD-91-38 (Washington, DC: U.S. Government Printing Office, December 1 990). U.S. Congress, General Accounting Office, Medical Malpractice: A Framework for Action, GAO/HRD-87-73 (Washington, DC: U.S. Government Printing Office, May 1987). U.S. Congress, General Accounting Office, Medical Malpractice: Characteristics of Claims Closed in 1984, GAO/HRD-87-55 (Washington, DC: U.S. Government Printing Office, April 1987). U.S. Congress, Office of Technology Assessment, Impact of Legal Reforms on Medical Malpractice Costs-Background Paper, background paper for OTAs Project on Defensive Medicine and Medical Malpractice, OTA-BP-H-119 (Washington, DC:
References | 177
U.S. Government Printing Office, October I 993). 237. U.S. Congress, Office of Technology Assessment, personal communications with J. Crist, Vice President, American Health Care Systems Insurance Management Services, San Diego, CA; R, Mayagi, Director of Legal Medical Services and Legal Counsel, Kaiser Permanence Northwest Region, Portland, OR; S. Fillimore, Director of the Department of Quality Management, University Hospital & Clinics, Portland. OR; B. Wheatley, Risk Manager, Good Samaritan Hospital, Portland, OR; and P. Yeats, Assistant Administrator, Willamette Falls Hospital, Oregon City, OR, January through July 1993. 238. U.S. Congress. Office of Technology Assessment, Does Health Insurance Make a Difference-Background Paper, background paper for OTAs Project on Technology, Insurance, and the Health Care System, OTA-BP-H-99 (Washington, DC: U.S. Government Printing Office, September 1992). 239. U.S. Congress, Office of Technology Assessment, Do Medicaid and Medicure Patients Sue More Often Than Other Patients?-Background Paper (Washington, DC: August 1992) 240. U.S. Congress, Office of Technology Assessment, Evaluation of the Oregon Medicaid Proposal (Washington, DC: U.S. Government Printing Office. May 1992). 241. U.S. Congress, Senate, Committee on Labor and Human Resources, Agency for Health Cure Policy and Research Reauthorization Act of 1992, Committee Report, Serial No. 102-426 (Washington, DC: Government Printing Office, 1992). 242. U.S. Consumer Product Safety Commission, unpublished data on head injuries caused by consumer products (excluding motor vehicles and public transportation) that were treated in emergency rooms, supplied by Kathryn Wallace, Congressional
Relations Specialist, Bethesda, MD, personal communication, 1994. 243. U.S. Department of Health, Education and Welfare, Secretarys Commission on Medical Malpractice, Repot-t of the Secretarys Commission on Medical Malpractice, DHEW Publ. No. (OS) 73-88 (Washington, DC: U.S. Government Printing Office, 1973). 244. U.S. Department of Health, Education and Welfare, Secretarys Commission on Medical Malpractice, Appendix: Report of the Secretarys Commission on Medical Malpractice, Reports. Studies, and Analysis, DHEW Publ. No. (OS) 73-89 (Washington, DC: U.S. Government Printing Office, 1973). 245. U.S. Department of Health and Human Services, Office of Inspector General, Financial Arrangements Between Physicians and Health Care Businesses, prepared by Richard P. Kusserow, Inspector General, OAI12-88-01410 (Washington, DC: U.S. Government Printing Office, 1989). 246. U.S. Department of Health and Human Services, Public Health Service, Health Resources and Services Administration, Office of Communications, National Practitioner Data Bank Announces New Practitioner Statement Feature, HRSA Note, Mar. 31, 1994. 247. U.S. Department of Health and Human Services, Public Health Service, Health Resources and Services Administration, Bureau of Health Professions, Division of Medicine, Council on Graduate Medical Education 4th Report to Congress and the Department of Health and Human Services Secretary (Rockville, MD: January 1994). 248. U.S. Department of Health and Human Services, Public Health Service, National Institutes of Health, National Cancer Institute, Cancer Facts, information sheet, Bethesda, MD, October 1993. 249. U.S. Department of Health and Human Services, Public Health Service, Agency for
Health Care Policy and Research, AHCPR: Purpose and Programs (Rockville, MD: September 1990). 250. U.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, National Center for Health Statistics, Division of Vital Statistics, Natal, Marriage, and Divorce Statistics Branch, Advance Report of Maternal and Infant Health Data from the Birth Certificate, 1991 , Monthly Vital Statistics Report, forthcoming 1994, supplied by Selma M. Taffel, Statistician, October 1993. 251. Wadlington, W., Waltz, J. R., and Dworkin, R. B., Law and Medicine (Mineola, NY: The Foundation Press, Inc., 1980). 252. Wagner, L., Defensive Medicine: Is Legal Protection the Only Motive? Modern Healthcare 22(14):26-28,30, Sept. 10, 1990. 253. Weiler, P. C., Medical Malpractice on Trial (Cambridge, MA: Harvard University Press, 1991). 254. Weiler, P. C., Hiatt, H. H., Newhouse, J. P., et al., A Measure of Malpractice (Cambridge, MA: Harvard University Press, 1993). 255. Weinrib, E.J., Understanding Tort Law, Valparaiso University Law Review 23(3): 485-526, 1989. 256. Wertman, B. G., Sostrin, S. V., Pavlova, Z., et al., Why Do Physicians Order Laboratory Tests? A Study of Laboratory Test Re-
quest and Use Patterns, Journal of the American Medical Association 243(20): 2080-2082, May 23-30, 1980. 257. Wilensky, G. R., and Rossiter, L. F., The Relative Importance of Physician-Induced Demand for Medical Care, Milbank Quarterly 61 (2):252-277, spring 1983. 258. Williams, S. V., Eisenberg, J. M., and Pascale, L. A., *Physicians Perceptions About Unnecessary Diagnostic Testing, Inquiry 19(4):363-70, winter 1982. 259. Young, M. J., Lisbeth, F. S., Eisenberg, J., et al., Do Cardiologists Have Higher Thresholds for Recommending Coronary Arteriography Than Family Physicians? Health Services Research 22:623-35, 1987. 260. Zuckerman, S., q Medical Malpractice: Claims, Legal Costs, and the Practice of Defensive Medicine, Health Affairs 3(3): 128133, fall 1984. 261. Zuckerman, S., Norton, S., and Wadler, B., A State-Based Survey of Malpractice Premiums: Implications for Medicare Physicians Payment Policy, Report 6090-02 (Washington DC: The Urban Institute, March 1993). 262. Zweig, F. M., and Witte, H. A., Assisting Judges in Screening Medical Malpractice Guidelines for Health Care Litigation, Journal on Quality Improvement 19(8):342354, August 1993.
I ndex
A ACC. See American College of Cardiology Accelerated compensation events, 15, 18, 19,88, 89,90-91 ACES. See Accelerated compensation events; Avoidable classes of events ACOG. See American College of Obstetricians and Gynecologists ACS. See American College of Surgeons Acute myocardial infarction, 105 ADR. See Alternative dispute resolution Agency for Health Care Policy and Research, 18, 83, 142. 145, 149 Alternative dispute resolution, 13-14, 82, 84-87, 89, 90-93 AMA. .See American Medical Association AMA, SSMLP. See American Medical Association/ Specialty Society Medical Liability Project American College of Cardiology. 5-6, 8, 50, 58, 96, 106-117 American College of Emergency Room Physicians, 96 American College of Obstetricians and Gynecologist, 5-6, 8, 43, 50, 56, 58, 63, 65, 71, 96, 106-117, 144 American College of Surgeons, 5-6, 8, 50, 56, 58, 63,65, 96, 106-117 American Health Care Systems. Inc.. 32-33 American Medical Association, 30, 145, 47-48, 150, 156-160 American Medical Association/Specialty Society Medical Liability Project, 14, 84,86-87 AM I. See Acute myocardial infarction Arbitration. See Alternative dispute resolution Archer, Bill, 2, 95 Arizona pretrial screening studies, 81 Avoidable classes of events. See accelerated compensation events
B
Baldwin, Laura Mae, 9, 68-69.70, 97 Birth-related injuries, 14-15, 88, 89 Bovbjcrg, Randall, 96 Breast biopsy, 24-25 Brigham and Womens Hospital, 24-25 Bush, George, 2
c
Caesarean delivery, 2,5,8, 11,68,81, 105, 129, 131 California, 28-29, 49.80-81, 87, 105, 149-150 Cancer, 9.24-25, 31-32 Cardiologists. See American College of Cardiology. Case studies methodology, 43 use of low osmolality contrast agents. 1 (), 71-74 Channeling arrangements, 87 Clinical practice guidelincs, 2, 12-13, 17-18.81-84, 87,92, 142-150 Clinical scenario surveys Duke Law Journal study. 49-50, 51-52 Classman survey of New Jersey physicians. 9. 65-66 methodology, 41-42 OTA surveys. 5-618, 50, 52-65 Congressional Sunbelt Caucus. 95 Conventional malpractice reforms compensation guidelines, 11-12 description, 2, 11-12, 78-79.92 direct malpractice costs impact, 81 low-income plaintiffs and, 76, 77 multistate data. 79, 133:141 policy option, 16-17 pretrial screening studies, 81, 133-141 pm-defendant bias, 76 single-state studies, 79-81, 133-141 small multistate studies, 79-81. 133-141 Cost Consciousness scale. 109
Cost of defensive medicine Caesarean delivery in a complicated labor example, 129, 131 cost containment and practice guidelines, 148-149 customary practice standard, 149 estimate surveys, 128-132, 156-161 head injury example, 5, 131-132 Lewin-VHI, Inc. estimates, 48, 160-161 Reynolds and colleagues estimates, 47-48, 156-160 Customary practice standard, 149 D Definitions of defensive medicine benefit or harm to the patient and, 22-25,36 categories of defensive medicine, 23-24 examples, 24-25 conscious versus unconscious practice, 2, 22, 36 definitions other than OTAs, 23 Lewin-VHI, Inc. definition, 48 OTA definition, 1,3,21-22,95-96 primary versus sole motivation, 22,36 probability of disease and medical consequences, 25-26 Delayed diagnosis breast malignancy claims, 24-25 Diagnostic x-rays - see x-rays Dingell, John D., 2,95 Direct physician surveys methodology, 4, 41, 43 findings, 4,43-46 poor response rates, 47 Discomfort with Clinical Uncertainty scale, 109 Duke Law Journal Project findings, 50 methodology, 5, 41-42 structure, 49-50 Durenbergcr, Sen. Dave, 2,95
learned treatise exception, 144 Fee-for-service system health care reform and, 2, 15,91-92 lower diagnostic testing use in, 104 Financial consequences of malpractice suits. See also Cost of defensive medicine income loss, 27-28 malpractice premiums and, 29, 159 malpractice reporting systems and, 10,28-29 misperceptions about, 28 Florida, 14-15,29, 82,88,89,96-97, 147 FPs. See Family practitioners
G
Glassman, P., 4,9,65-66,69 Goold, Susan, 108-109 Graduate medical education, 33-36 Grassley, Sen. Charles E., 2 Gronfein, and Kinney, 79-80 Grumbach and Lueft, 69,71,97 Guidelines. See Clinical practice guidelines
H
Harvard Medical Institutions, 33 Hatch, Sen. Orrin, 2,95 Hawaii, 81 Head injuries, 5, 130, 131-132 Health care reform, 2, 15-16,91-92,93 Health Insurance Association of America, 131 Health maintenance organizations, 15,31,87, 105 HMOs. See Health maintenance organizations Hospitals, 32-34
J E
Economic issues. See Cost of defensive medicine; Financial consequences of malpractice suits Eliastam, 131 Enterprise liability, 13, 18,82,87-88,93 Epstein, A. and McNeil, B., 48-49 Erbs palsy study, 32 "Error in judgment rule, 143 Expert witnesses, 30,83, 143 Jacobson, P. and Rosenquist, C. 10,71-74 Joint Commission on Accreditation of Health Care Organizations, 32
K
Kaiser Foundation, 80 Kennedy, Sen. Edward M., 2,95 Kington, R., 71 Kinney. See Gronfein and Kinney
F
Failure-to-diagnose claims, 30-31 Family practitioners, 5,9,29,69,71, Federal Rules of Evidence
L
Learned treatise exception, 143-144 Legal standard of care, 30-32, 142-145 Lewin-VHI, Inc., 48, 160-161
Index | 181
Localio R, 2,5,8, 11,68,81 LOCAs. See Low osmolality contrast agents "Loss of chance doctrine, 31-32 Low osmolality contrast agents, 10,72-74
o
OB/GYNs. See Obstetricians/gynecologists Obstetric claims. See also Caesarean delivery, 4,8, 68-69,90 Obstetricians/gynecologists. See also American College of Obstetricians and Gynecologists, 5,9, 29,69,71, 125-126 OTA clinical scenario surveys, 5-6,50,52-65,67, 106-111, 113-114, 118-127, 130-132
M
Maine, 12,82-84, 109, 146-147, 148 Malpractice reform. See Reforms Mammograms, 24-25,83 Managed competition, 15,92 Maryland, 148 Massachusetts, 105 Medical Injury Compensation Reform Act, 80-81 Medical Insurance Exchange of Ncw Jersey, 65-66 Medical Liability Demonstration Project, 12,82-84, 146-147, 148 Medicare Act, 146 Medicare reimbursement rates, 132 Methodology of studies. See also Study evidence behavioral model of physician test ordering, 39.40 case studies, 43 clinical scenario surveys, 5-6, 8, 41-42 direct physician surveys, 41 prompting issue, 41,63,74 statistical analyses, 42-43 for this report, 95-100 Meyer, J., 24-25 MICRA. See Medical Injury Compensation Reform Act Minnesota, 82, 147-148 Multistate studies of malpractice reform, 79-81, 133-141
P
Patient Compensation Funds, 79-80 PCFs. See Patient Compensation Funds Physician Payment Review Commission, 132 Physician test ordering surveys, 48-49 Physicians attitudes, 2,9-10,26-32,37, 104-105, 108-109, 127 Physicians Insurance Association of America, 24-25 Policy options, 16-19 Positive defensive medicine studies, 2, 5, 8-9, 11, 68-69,81 Pretreatment arbitration agreements. See Voluntary binding arbitration Pretrial screening studies, 81, 133-141 Project structure advisory panel, 96 background papers, 97 clinical scenario surveys, 96 contract papers, 97, 100 empirical research in addition to clinical scenario surveys, 96-97 planning workshop, 95-96 report review process, 97 workshop participants, 98-99 Prompting issue, 41,63,74 Prospective Payment Assessment Commission, 132 Prostate specific antigen test use, 9 Psychological consequences of malpractice suits, 29
N
National Ambulatory Medical Care Survey, 158 National Cancer Institute, 83 National Center for Health Statistics, 131 National Electronic Injury Surveillance System, 132 National Health Interview Survey, 131, 132 National Practitioner Data Bank, 10,28, 29 Negative defensive medicine, 3,5,9,69,71 Neurological injuries. See also Head injuries, 88, 89 Neurosurgeons, 123-124 New Jersey, 9, 4, 9,65-66,69 New York, 2,5, 8, 11, 28,68-69,71,81, 105 No-fault malpractice reform proposals, 14-15, 18-19,82, 88-91,93 Nonclinical factors in physicians resource USC , 104-105 NPDB. See National Practitioner Data Bank
Q
Quality assurance influence on defensive medicine, 32-33
R
Reforms alternative dispute resolution, 13-14, 82, 84-87, 89,90,91,92-93 clinical practice guidelines, 12-13, 81, 82-84,92, 142-150 conventional, 11-12, 76-81
enterprise liability, 13, 18, 82, 87-88,93 health care reform considerations, 15-16,91-92 newer reforms, 81-91 no-fault compensation, 14-15, I8-19, 82, 88-91, 93 Relative avoidability concept, 90 Residency training. See Graduate medical education "Respectable minority rule, 143 Reynolds R., 47-48, 156-160 Risk management, 32-33 Risk Management Foundation, 32 Robert Wood Johnson Foundation, 9,68-69,70 Rosenquist. See Jacobson and Rosenquist
physicians reasons for ordering tests and procedures, 48-49 specific measures, 113-114 statistical analyses, 67-71 survey-based estimate of cost, 47-48 Study summaries conclusions, 74 methodology, 41-43 study evidence, 43-74 SUDAAN software, 115-117 Surgeons. See also American College of Surgeons, 121-122 Survival rates, 31-32
s
Secretarys Commission on Medical Malpractice, 23 Shoulder dystocia study, 32 Single state studies of malpractice reform, 79-81,
133-141
T
Tort reform. See Reforms Traditional reforms. See Conventional malpractice reforms
SMS survey. See Socioeconomic Monitoring System survey Socioeconomic Monitoring System survey, 156-157 Sources of defensive medicine, 26-36 St. Pauls Fire and Marine Insurance Company, 30 Stanford University Medical Center Emergency Department, 25 Statistical analyses common hypothesis, 67 methodology, 4, 42-43 multivariate analyses, 42 negative defensive medicine studies, 9, 69, 71 OTA clinical scenario surveys, 114-115 positive defensive medicine studies, 68-69 StatXact-Turbo software, 115-116 Study evidence. See also Methodology of studies case study of LOCAs, 71-74 clinical scenario surveys, 5-6, 8, 49-67 direct physician surveys, 4,43-47
u University of California, 87
v
Vermont, 82, 148 Virginia, 14-15,88-89 Voluntary binding arbitration, 13-14,84-86
w
Washington State, 4,8,68-69, 105
Wickline v. State of California, 149-150
x
X-rays criteria for when not to obtain cervical spine x-ray, 2, 5, 25, 82-83, 130-132
Y Youngberg v.. Romeo, 149
0 U S GOVERNMENT PRINTING OFFICE 1994-301 -804/1 7414

July 1994: Defensive Medicine and Medical Malpractice

Uploaded by

Copyright:

Available Formats

July 1994: Defensive Medicine and Medical Malpractice

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

July 1994: Defensive Medicine and Medical Malpractice

Uploaded by

Copyright:

Available Formats

Defensive Medicine and Medical Malpractice July 1994

OTA-H-602 NTIS order #PB94-193257 GPO stock #052-003-01377-1

Panel Chair Senior Research Associate The Urban Institute Washington, DC

2 Defensive Medicine: Definition and Causes 21

3 Summary of the Evidence on Defensive Medicine 39

4 Impact of Malpractice Reform on Defensive Medicine 75

REFERENCES 165 INDEX 179

Findings and Policy Options

2 | Defensive Medicine and Medical Malpractice

Chapter 1 Findings and Policy Options | 13

DEFINING DEFENSIVE MEDICINE

THE EXTENT OF DEFENSIVE MEDICINE

4 | Defensive Medicine and Medical Malpractice

Chapter 1 Findings and Policy Options | 5

6 | Defensive Medicine and Medical Malpractice

Percent of respondents Percent of respondents

Of clinical actions chosen,

I I 3.6 3.4 1.4

Lumbosacral x-ray I L CT scan MRI

139 2 1 9 6 123 2 2 11 1 109 1 4 5 . 8

-- Mammography Refer to surgeon

Complicated delivery Perimenopausal bleeding

Caesarean delivery Pregnancy test D&C 4 . 2

8 | Defensive Medicine and Medical Malpractice

Procedure Utilization Studies

Chapter 1 Findings and Policy Options | 9

RECENT FACTORS AFFECTING THE AMOUNT OF DEFENSIVE MEDICINE

NOT found to be more likely than phy-

10 | Defensive Medicine and Medical Malpractice

THE IMPACT OF MALPRACTICE REFORM ON DEFENSIVE MEDICINE

Chapter 1 Findings and Policy Options | 11

12 | Defensive Medicine and Medical Malpractice

Clinical Practice Guidelines

Chapter 1 Findings and Policy Options | 13

Alternative Dispute Resolution

14 | Defensive Medicine and Medical Malpractice

The American Medical Association/ Specialty Society Medical Liability Project

Chapter 1 Findings and Policy Options | 15

DEFENSIVE MEDICINE IN AN ERA OF HEALTH CARE REFORM

16 | Defensive Medicine and Medical Malpractice

Chapter 1 Findings and Policy Options | 17

18 | Defensive Medicine and Medical Malpractice

Chapter 1 Findings and Policy Options | 19

Defensive Medicine: Definition and Causes

DEFINING DEFENSIVE MEDICINE

22 | Defensive Medicine and Medical Malpractice

Conscious vs. Unconscious Defensive Medicine

Defensive Medicine: Good, Bad, or Both?

Chapter 2 Defensive Medicine: Definition and Causes | 23

Effect on health care costs

Cost reducing) quality raising

Cost raising quality raising

Cost reducing/ quality reducing

Cost raising j quality reducing

24 | Defensive Medicine and Medical Malpractice

Number of Percent biopsies malignant a

other hospitals and officesb Lobular carcinomas considered benign

Example #1: Referrals for Breast Biopsy After Screening Mammography

Chapter 2 Defensive Medicine: Definition and Causes | 25