Biostatistics and Methods

of Epidemiology

DIRECTIONS: Each question below contains five suggested responses. Select the
one best response to each question.

1. Epidemiology can be defined as 3. A measure of the amount of varia¬

the study of tion of a set of values about the mean
(A) the etiology of disease in humans is the
(B) the frequency of causes of death (A) regression coefficient
in humans (B) standard error of the mean
(C) the determinants of frequency of (C) standard deviation
disease in humans (D) range
(D) the distribution and determinants (E) correlation coefficient
of frequency of disease in humans
(E) the patterns of organization and 4. The association between low birth
financing of health care weight and maternal smoking during
pregnancy can be studied by obtaining
2. A sample of 1,000 people includes smoking histories from women at the
120 who are hearing-impaired and 50 time of the prenatal visit and then sub¬
who are diabetic. If the number who sequently correlating birth weight with
are both diabetic and hearing-impaired smoking histories. What type of study
is 6, then which of the following is is this?
true? (A) Clinical trial
(A) Diabetes and hearing impairment (B) Cross-sectional
appear to be independent (C) Cohort (prospective)
characteristics (D) Case-control (retrospective)
(B) Diabetics appear to be protected (E) None of the above
from hearing impairment
(C) Diabetics appear to be at greater
risk of hearing impairment
(D) There is an interaction between
diabetes and hearing impairment
(E) There is insufficient information
to make any of the statements
above

1
2 Preventive Medicine and Public Health

5. An investigator wishes to perform 7. A study is undertaken to deter¬

a randomized clinical trial to evaluate mine whether use of Newman and
a new beta blocker as a treatment for Browner’s PreTest Self-Assessment and
hypertension. To be eligible for the Review reduces sexual dysfunction
study, subjects must have a resting related to anxiety before board exami¬
diastolic blood pressure of at least 90 nations. One group of students study¬
mmHg. One hundred patients seen at ing for their board examinations was
the screening clinic with this level of given the book to read; the other was
hypertension are recruited for the study not. The results are as follows:
and make appointments with the study
nurse. When the nurse obtains their Got Did Not
blood pressure 2 weeks later, only 65 Outcome Book Get Book
of them have diastolic blood pressures
Sexual
of 90 mmHg or more. The most likely
dysfunction 3 34
explanation for this is
No sexual
(A) spontaneous resolution dysfunction 61 / 32
(B) regression toward the mean j
(C) baseline drift All the following statements are true
(D) measurement error EXCEPT
(E) Hawthorne effect
(A) these data could be analyzed using
the chi-square test
6. Reye syndrome is an acute en¬
(B) about half of the control group
cephalopathy of childhood, accompa¬
experienced sexual dysfunction
nied by fatty infiltration of the liver. It
(C) unless one knows on what basis
has been linked to use of aspirin during
group assignment was made (i.e.,
infections with chickenpox or
randomized or not), the results are
influenza. Children commonly pro¬
difficult to interpret
gress through a sequence of stages that
(D) the difference is probably due to
are used to grade severity and progno¬
chance
sis. If 30 percent of children progress
(E) a t test is inappropriate to analyze
only to stage I, 40 percent only to
the data because the variables are
stage II, 20 percent only to stage ID,
categorical rather than continuous
and 10 percent to stage IV, what is the
probability that a child will progress to
8. Which of the following measures
stage IV, once that child has already
is used frequently as a denominator to
reached stage II?
calculate the incidence rate of a
(A) 10 percent disease?
(B) 14 percent
(A) Number of cases observed
(C) 30 percent
(B) Number of new cases observed
(D) 33 percent
(C) Number of asymptomatic cases
(E) None of the above
(D) Person-years of observation
(E) Persons lost to follow-up
 Biostatistics and Methods of Epidemiology

9. If the chance of acquiring sero¬ Q In nine families surveyed, the

logic evidence of human immunodefi¬ numbers of children per family were 4,
ciency virus (HIV) from an infected 6,2,2, C'3,2, 1, and 7. The mean,
sexual partner is a constant 30 percent mg^Ji^n, and mode numbers of children
per year, what is the approximate per family are, respectively,
v
chance of becoming infected within 3 (A) 3.4,2, 3
years? (B) 3,3.4,2/x‘
(A) 10 percent (C) 3,3,2
(B) 30 percent (D) 2,3.5,3
(C) 66 percent (E) none of the above li
(D) 90 percent
(E) None of the above W/ 7 >> 3 z
V
10. In 1971, the crude birth rate in the 1 2
United States was approximately 17
per 1,000 population; the death rate
was 10 per 1,000; and the net in-
migration rate was 2 per 1,000. What
was the net growth rate per 1,000?
u
£
(A) 5 %
(B) 7
(C) 9
(D) 15
(E) 25
,1--

l\4
4 Preventive Medicine and Public Health

Questions 12-14

The results of a study of the incidence of pulmonary tuberculosis in a village in

India are given in the table below. All persons in the village are examined during
two surveys made 2 years apart, and the number of new cases was used to
determine the incidence rate.

Category of Household
at First Survey Number of Persons Number of New Cases

With culture-positive case 500 10

Without culture-positive case 10,000 10

12. What is the incidence of new cases 14. What is the relative risk of acquir
per 1,000 person-years in households ing tuberculosis in households with a
that had a culture-positive case during culture-positive case compared with
the first survey? households without tuberculosis?
(A) 0.02 (A) 0.05
(B) 0.01 (B) 0.5
,60'
(C) 1.0 (C) 2.0
(D) 10 (D) 10
(E) 20 (E) 20

13. What is the incidence of new cases

per 1,000 person-years in households
that did not have a culture-positive
case: during the first survey?
(A) 0.001
(B) 0.1
(C) 0.5
(D) 1.0 p0
r.* $

(E) 5.0
 Biostatistics and Methods of Epidemiology 5

15. During the investigation of an outbreak of food poisoning at a summer

camp, food histories were obtained from all campers as indicated in the table
below. Which of the food items was probably responsible for the outbreak?

Proportion Ill (Percent)

Campers Who Ate Campers Who Did Not Eat
Food Specified Food Specified Food

(A) Hamburger 61 48
(B) Potatoes 70 35
(C) Ice cream 40 50
(D) Chicken 73 10
(E) Lemonade 20 45

16. Which of the following statements 17. In the study of the cause of a dis¬
concerning statistical inference is ease, the essential difference between
correct? an experimental study and an observa¬
tional study is that in the experimental
(A) If the p value = 0.05, then there is
investigation
a 5 percent probability that the
null hypothesis is true (A) the study is prospective
(B) The null hypothesis generally (B) the study is retrospective
states that there is a difference (C) the study and control groups are
between the groups of equal size
(C) If the p value is greater than (D) the study and control groups are
alpha, the null hypothesis is not selected on the basis of history of
rejected exposure to the suspected causal
(D) Knowledge of the sampling factor
method is not important in deter¬ (E) the investigators determine who is
mining statistical significance and who is not exposed to the sus¬
(E) None of the above pected causal factor
6 Preventive Medicine and Public Health

Questions 18-19

About 1 percent of boys are bom with undescended testes. To determine

whether prehataTexposure to tobacco smoke is a cause of undescended testes in
newborns, the mothers of 100 newborns with undescended testes and 100 new¬
borns whose testes had descended were questioned about smoking habits during
pregnancy. The study revealed an odds ratio of 2.6 associated with exposure to
smoke, with 95 percent confidence intervals from 1.1 to 5.3.

18. Which of the following statements 19. Which of the following statements
is true? is true?
(A) The odds ratio could be falsely (A) The results provide no evidence
elevated by the inclusion of infants that maternal cigarette smoking is
whose testes were descended (but associated with undescended
retractile) in the case group (mis- testes in the offspring
classification bias) (B) If the study results are accurate,
(B) The odds ratio would be falsely they suggest that a baby boy
reduced if parents of affected whose mother smoked is about
infants were more likely to 2.6 times as likely to be bom with
remember or report their smoking testes undescended as a baby boy
(recall bias) whose mother did not smoke
(C) Because the cases are newborns, (C) The fact that the confidence inter¬
but the exposure data came from val excludes 1.0 indicates that
their mothers, this is not a true p > 0.05
case-control study (D) The 90 percent confidence in¬
(D) Since the study was not blinded, it terval for these results would
is impossible to rule out a placebo probably include 1.0
effect (E) None of the above
(E) None of the above

20. The probability of being bom with 21. All the following are important
condition A is 0.10 and the probability steps in decision analysis EXCEPT
of being bom with condition B is 0.50.
(A) construction of a decision tree
If conditions A and B are independent,
(B) estimation of probabilities at
what is the probability of being bom
chance nodes
with either condition A or condition B
(C) assignment of utilities
(or both)?
(D) estimation of sample size
(A) 0.05 (E) determination of how changes in
(B) 0.40 probability estimates affect the
(C) 0.50 decision
(D) 0.55
(E) 0.60
 Biostatistics and Methods of Epidemiology 7

22. In order to determine the relation¬ 23. All the following statements
ship between serum levels of sodium regarding the normal (Gaussian) dis¬
and antidiuretic hormone (ADH) in tribution are true EXCEPT ,
patients who have meningitis, the most
(A) the mean = median = nfode
appropriate study design would be
(B) about 95 percent of observations
(A) repeated measurement of sodium fall within 2 standard deviations
and ADH in a patient with of the mean
meningitis (C) approximately 68 percent of
(B) measurement of both sodium and observations fall within 1 standard
ADH in a set of patients with deviation of the mean
meningitis (D) the number of observations be¬
(C) measurements of sodium in one tween 0 and 1 standard deviation
set of patients and of ADH in a from the mean is the same as the
different set of patients number between 1 and 2 standard
(D) measurements of ADH in a set of deviations from the mean
patients with meningitis and a set (E) the shape of the curve does not
of controls with other illnesses .depend on the value of the mean
(E) none of the above
8 Preventive Medicine and Public Health

Questions 24-26 25. Dr. Stewells next performs the test

on 1,000 patients admitted to the hos¬
Lou Stewells, a pioneer in the study
pital with profuse diarrhea. The results
of diarrheal disease, has developed a
are as follows:
new diagnostic test for cholera. When
his agent is added to the stool, the
Cholera No Cholera
organisms develop a characteristic ring
around them. (He calls it the “Ring- RAC test + 312 79
Around-the-Cholera” [RAC] test.) He RAC test — 31 578
performs the test on 100 patients
Totals 343 657
known to have cholera and 100
patients known not to have cholera
Which of the following statements is
with the following results:
correct?

Cholera No Cholera (A) The predictive value of a positive

result is 312/343
RAC test + 91 12 (B) The predictive value of a positive
RAC test — 9 8§ result is 79/312
Totals 100 100 ■(C) The predictive value of a negative
result is 578/(578 + 31)
(D) The incidence rate of cholera in
24. All the following statements about this population is 343/1,000
the RAC test are correct EXCEPT (E) None of the above
(A) the sensitivity of the test was
about 91 percent
(B) the specificity of the test was
about 12 percent
(C) the false negative rate was about 9
percent
(D) the predictive value of a positive
result cannot be determined from
the information above
(E) the predictive value of a negative
result cannot be determined from
the information above
 Biostatistics and Methods of Epidemiology 9

26. After many years and dozens of 28. In comparing the difference
publications, the RAC test achieves between two means, the value of p is
widespread acceptance, and Dr. found to be 0.60. The correct interpre¬
Stewells’ position in the Diarrhea Hall tation of this result is
of Fame seems assured. However, with (A) the null hypothesis is rejected
improvements in hygiene, the preva¬ (B) the difference is statistically
lence of cholera gradually falls from significant
the high of about 35 percent of hospi¬ (C) the difference occurred by chance
talized diarrhea patients to only 5 per¬ (D) the difference is compatible with
cent. Which of the following state¬ the null hypothesis
ments about the effect of this fall in (E) sampling variation is an unlikely
prevalence is true? explanation of the difference
(A) The change in prevalence will
reduce the predictive value of a 29. In a study of the cause of lung
negative result cancer, patients who had the disease
(B) The predictive value of a positive were matched with controls by age,
result will decline sex, place of residence, and social
(C) The specificity of the test is likely class. The frequency of cigarette smok¬
to decline ing was then compared in the two
(D) The specificity of the test will groups. What type of study was this?
increase at the expense of its (A) Cohort (prospective)
sensitivity (B) Case-control (retrospective)
(E) None of the above (C) Clinical trial
(D) Historical cohort
(E) None of the above
27. Randomization is a procedure
used for assignment or allocation of
subjects to treatment and control
groups in experimental studies. Ran¬
domization ensures
(A) that assignment occurs by chance
(B) that treatment and control groups
are alike in all respects except
treatment
(C) that bias in observations is
eliminated
(D) that placebo effects are eliminated
(E) none of the above
10 Preventive Medicine and Public Health

30. The relationship between urinary Questions 31-33

tract infection (UTI) and lack of cir¬
Smoking and alcohol use are both
cumcision may be causal. Studies have
risk factors for esophageal cancer. The
also suggested that the relative risk for
hypothetical table below shows how
UTI in uncircumcised male infants is
the incidence of esophageal cancer
about 10. Which of the following is a
varies with either smoking or alcohol
valid implication of this observation?
use (but not both):
(A) In male infants with fever, the
yield of a urine culture is likely to Incidence
be higher if the boy is circumcised (cases/10,000 person-yr)
(B) In male infants in whom a UTI is
Nonsmoker Smoker
documented but no abnormalities
are detected on physical examina¬ Nondrinker 10 50
tion, the yield of urinary tract Drinker X
30 '
imaging to look for congenital
urologic malformations (which
predispose to UTI) is likely to be 31. All the following statements are
much higher if the infant is true EXCEPT
uncircumcised (A) in nonsmokers, the excess risk of
(C) All male infants should be cir¬ esophageal cancer from drinking is
cumcised to prevent UTI 20 cases per 10,000 person-years
(D) If the association is causal, about (B) in nondrinkers, the relative risk of
90 percent of UTIs in uncircum¬ esophageal cancer from smoking
cised male infants are due to the is 5.0
fact that these infants are (C) the overall relative risk for smok¬
uncircumcised ing cannot be determined without
(E) None of the above knowledge of X and the propor¬
tion of the population that drinks
(D) the table shows that smoking
causes more cases of esophageal
cancer in this population than does
drinking
(E) in nondrinkers, the excess risk of
esophageal cancer from smoking
is 40 cases per 10,000 person-
years

if o 4-5 o 4- )c
30
30+ 50 =
/0"X + ? S'V
0 * 5o ^ ~X ^<£c>'/c>

« x ~ \ 'X ^ +0,
 Biostatistics and Methods of Epidemiology 11

32. If smoking and alcohol act inde¬ 33. If smoking and alcohol act inde¬
pendently to cause esophageal cancer pendently according to a multiplicative
under an additive model, what would model, what would be the expected
be the expected value of X? value of X?
(A) 60 (A) 70
(B) 70 (B) 80
(C) 80 (C) 120
(D) 150 (D) 1500
(E) None of the above (E) None of the above

34. In the table below, data are presented on the number of children suffering
from acute leukemia who were admitted to a hospital between 1970 and 1984.
Correct conclusions about the data include which of the following?

Number of Children Admitted in Interval

Age (years) 1970-74 1975-79 1980-84

0-4 12 23 31
5-9 8 17 36
10-14 10 7 4

Total 30 47 71

(A) The incidence rate of leukemia decreased in children 10 to 14 years old "
(B) The prevalence rate of leukemia increased in children between 1970 and
1984
(C) The incidence rate of leukemia increased in children 5 to 9 years old -
(D) The number of children 9 years old and under admitted to the study hos¬
pital because of acute leukemia increased between 1970 and 1984
(E) None of the above
12 Preventive Medicine and Public Health

35. The Coronary Drug Project was a 36. The attributable risk of a disease
randomized trial to evaluate the effi¬ estimates the maximum proportion of
cacy of several lipid-lowering drugs. the disease in the population attribut¬
The 5-year mortality of the men who able to a particular risk factor. For
adhered to the regimen of clofibrate example, in the U.S. the attributable
(i.e., took 80 percent of their medicine risk of lung cancer from active smok¬
or more) was 15 percent; among those ing is about 85 percent; this means that
assigned to the clofibrate group who active smoking may explain up to 85
were less compliant, it was 24.6 per¬ percent of the cases of lung cancer that
cent. The result was highly statistically occur in the U.S. For a particular dis¬
significant (p = 0.0001). From this one ease and risk factor, the attributable
can conclude risk depends on
(A) clofibrate was very beneficial to (A) the size of the population of
the patients who took it reliably interest
(B) clofibrate is not effective unless (B) the prevalence of the disease
patients take at least 80 percent of (C) the relative risk (risk ratio) of the
the recommended doses disease associated with the risk
(C) either clofibrate or something factor
associated with taking it reliably is (D) the duration of the disease
strongly associated with reduced (E) none of the above
total mortality
(D) there was a problem with blinding 37. A randomized, double-blinded
in this study trial finds that oral corticosteroids are
(E) none of the above superior to placebo in hastening the
resolution of otitis media with effusion.
Possible reasons why this study might
have given a falsely positive result
include
(A) the sample size may have been
too small
(B) the apparent effect might be a
result of chance
(C) it may be difficult to determine
accurately which effusions have
resolved, which will lead to errors
in determining the outcome of the
study
(D) lax inclusion criteria may have led
to inclusion of some subjects in
the study who did not really have
otitis media with effusion
(E) none of the above
 Biostatistics and Methods of Epidemiology 13

38. Correct statements concerning sta¬ 39. An investigator is designing a ran¬

tistical inference include which one of domized clinical trial to see whether
the following? vitamin E will prevept cancer in
(A) If the p value is very low, the dif¬ smokers. Which of the following is the
ference between the groups must LEAST important consideration in
be very large planning the sample size for the study?
(B) The standard error of the mean is (A) The expected incidence of cancer
used to estimate how closely the in the placebo group
mean of a sample approximates (B) The frequency with which subjects
the true population mean are likely to be lost to follow-up
(C) If the sample size is large enough, or die from noncancer causes over
it is easy to achieve statistical sig¬ the duration of the study
nificance at the 0.05 level, even (C) The prevalence of smoking in the
when there is no difference population
between the groups (D) The values for alpha and beta, the
(D) Use of one-tailed rather than two- type I and type 2 error rates
tailed tests of statistical signifi¬ (E) The magnitude of the preventive
cance will less often lead to reject¬ effect that the investigator wishes
ing the null hypothesis when it is to be able to detect
true
(E) None of the above
14 Preventive Medicine and Public Health

Questions 40-41

The following questions refer to the receiver operating characteristic (ROC)

curves for three different diagnostic tests (labeled 1, 2, and 3) shown below.

40. Which of the following statements about the figure is true?

(A) Test 2 is more sensitive than test 1
(B) Test 3 is totally useless
(C) If it is particularly important to avoid a false positive result, test 2 is prefera¬
ble to test 1
(D) Test 3 performs perfectly
(E) None of the above

41. Which of the following statements about points A and B is correct?

(A) A and B are two different patients with disease on whom test 1 was
performed
(B) A and B are two different patients without disease on whom test 1 was
performed
(C) A and B represent varying thresholds for considering the result of test 1 to be
indicative of disease
(D) If false negative results are to be avoided, point A represents the best choice
(E) None of the above
 Biostatistics and Methods of Epidemiology 15

42. In a study of level of blood lead as 44. All the following statements about
a predictor of IQ in children, a simple life expectancy in the U.S. are true
linear regression analysis is performed. EXCEPT
The regression coefficient for lead is
(A) less than half of all deaths occur
—0.1 IQ points per ng Pb/dL blood.
by age 75 years
The coefficient is highly significant
(B) a 65-year-old woman can expect
statistically (p < 0.001). Correct con¬
to live about 19 more years
clusions include which of the
(C) on average, women live about 7
following?
years longer than men
(A) All else being equal, the expected (D) most of the increase in life expec¬
difference in IQ between two chil¬ tancy since 1900 has been a result
dren whose lead levels differ by of decreasing rates of coronary
50 Mg/dL is about 5 points heart disease
(B) The relationship between blood (E) if mortality does not change in his
lead and IQ is linear or her lifetime, a child bom today
(C) Lead causes a decrease in the IQ will live an average of 75 years
in children of about 0.1 point for
each Mg/dL of blood lead 45. True statements concerning cohort
(D) The study provides convincing studies include all the following
evidence that lead and IQ are EXCEPT
independently associated (A) cohort studies are longitudinal in
(E) None of the above design
(B) subjects are selected on the basis
43. The prevalence of HLA-B27 of characteristics present before
among North American whites is 7 the onset of the condition being
percent; among white American studied
patients with ankylosing spondylitis, (C) subjects are observed over time to
the prevalence of HLA-B27 is 90 per¬ determine the frequency of occur¬
cent. This indicates that rence of the condition under study
(A) the prevalence of ankylosing (D) cohort studies can be retrospective
spondylitis is 0.7 percent (E) cohort studies are necessary to
(B) the prevalence of ankylosing estimate the prevalence of diseases
spondylitis is 6.3 percent
(C) the relative prevalence of ankylos¬
ing spondylitis associated with
HLA-B27 is greater than 100
(D) the relative prevalence of ankylos¬
ing spondylitis associated with
HLA-B27 is about 13
(E) the relative prevalence of ankylos¬
ing spondylitis associated with
HLA-B27 cannot be estimated
16 Preventive Medicine and Public Health

Questions 46-49

A research team wishes to investigate a possible association between smoke¬

less tobacco use and oral lesions among professional baseball players. At spring
training camp, they ask each baseball player about current and past use of
smokeless tobacco, cigarettes, and alcohol, and a dentist notes the type and
extent of the lesions in the mouth.

46. What type of study is this? 48. All the following statements about
(A) Case-control the association between smokeless
(B) Cross-sectional tobacco use and oral lesions in this
(C) Cohort study are true EXCEPT
(D) Clinical trial (A) the odds ratio is (80 X 34)/(2 X
(E) None of the above 30) = 45.3
(B) smokeless tobacco users are about
47. After the players have been ques¬ 45.3 times as likely as nonusers to
tioned about smokeless tobacco use have oral lesions
and examined for lesions of the mouth, (C) the prevalence of smokeless
the data on the 146 players are tabu¬ tobacco use in this sample is
lated as follows: 110/146
(D) the results are likely to be statisti¬
Mouth No cally significant
Lesions Lesions Total (E) the statistical significance could
be calculated using a chi-square
User 80 30 110
test
Nonuser 2 34 36
Total 82 64 146
49. If the association between smoke¬
less tobacco use and oral lesions in the
In this study, what is the incidence rate
last question is causal, approximately
of mouth lesions among the baseball
what proportion of the mouth lesions
players who use smokeless tobacco?
in the smokers are due to smokeless
(A) 80/30 tobacco use?
(B) 80/110
(A) 27 percent
(C) 82/146
(B) 50 percent
(D) 30/110
(C) 73 percent
(E) Cannot be calculated from the
(D) 92 percent
data given
(E) 99 percent
 Biostatistics and Methods of Epidemiology 17
t
50. For many diagnostic or screening 53. A randomized trial shows that a
tests, there is a tradeoff between sensi¬ new thrombolytic agent reduces total
tivity and specificity. True statements mortality in the first 30 days after a
include which of the following? suspected myocardial infarction by 30
(A) Sensitivity would be extremely . percent compared with a placebo (p <
important when testing for amyo¬ 0.002). Which of the following ques¬
trophic lateral sclerosis (ALS) tions would be the most'important to
because there is no good treat¬ have answered?
ment for it (A) Was the trial blinded?
(B) Because hypothyroidism in (B) What was the power of the study?
infancy is devastating if missed, a (C) What happened to surviving
screening test for it should be patients in the next year?
highly specific (D) What percentage of patients in
(C) Specificity is more important than each group actually had a
sensitivity for screening tests myocardial infarction?
(D) In evaluating the potential use¬ (E) What was the effect on mortality
fulness of a screening test, the from coronary heart disease?
effectiveness of treatment for the
disease screened for is important 54. A radiologist develops a new
(E) None of the above screening test for cancer of the pan¬
creas. The sensitivity and specificity of
51. The occurrence of a group of the test are said to be “very high, at
illnesses of similar nature at a rate least 98 percent.” You are asked to
above the expected number is called determine whether it would be useful
to screen everyone in your primary
(A) hyperendemic
(B) epidemic care practice with this test in the next
year. To answer, you would need to
(C) endemic
know all the following EXCEPT
(D) enzootic
(E) pandemic (A) the incidence of asymptomatic
pancreatic cancer
52. The time interval between entry of (B) the prevalence of asymptomatic
an infectious agent into a host and the pancreatic cancer
onset of symptoms is called (C) the exact values of the sensitivity
and specificity of the tests
(A) the communicable period
(B) the incubation period (D) the cost of the test
(E) the survival rate from pancreatic
(C) the preinfectious period
cancer cases detected by the new
(D) the noncontagious period
test
(E) the decubation period
18 Preventive Medicine and Public Health

Questions 55-57

In a study of the effectiveness of pertussis vaccine in preventing pertussis

(whooping cough), the following data were collected by studying siblings of
children who had the disease.

Immunization Status Number of Siblings Number of Cases

of Sibling Contact Exposed to Case among Siblings

Complete 4,000 400

None 1,000 400

55. What was the secondary attack 57. What was the relative risk of
rate of pertussis in fully immunized contacting whooping cough in the
household contacts? unimmunized children compared with
(A) 0 percent the fully immunized children?
(B) 10 percent (A) 0.25
(C) 25 percent (B) 0.5
(D) 40 percent (C) 1.0
(E) 75 percent (D) 2.0
(E) 4.0
56. What was the protective efficacy
of whooping cough vaccine?
(A) 25 percent /
(B) 40 percent / 10 f )/r
HtfW. ; 10 % 2>*'V

(C) 75 percent
4^ "to ,0
(D) 90 percent
Tcr tf ' '
(E) None of the above
 Biostatistics and Methods of Epidemiology 19

Questions 58-63

Data from an investigation of an epidemic of German measles in a remote vil¬

lage in Brazil are given in the table below:

Number Not III

Age but with
Group Number in Number III Antibody Rise Number Percent
(years) Population (symptomatic) (asymptomatic) Uninfected Infected
0-9 204 110 74 20 90
10-19 129 70 46 13 90
20-39 161 88 57 16 90
40-59 78 42 28 8 90
60+ 42 2 2 38 10

Totals 614 312 207 95

58. Which expression represents the 59. Which expression represents the
calculation to determine the incidence calculation to determine the percent¬
of illness for all age groups (as a age of infection that is asymptomatic
percentage)? (subclinical)?
312 312
(A) -X 100% = 50.8% (A) X 100% = 50.8%
614 614
207 207
(B) -X 100% = 33.7% (B) X 100% = 33.7%
614 614
95 95
(C) —X 100%= 18.3% (C) X 100% = 18.3%
519
519 519
(D) -X 100% = 84.5% (D) X 100% = 84.5%
614 614
-
207 207
(E) -X 100% = 39.9% (E)
519
X 100% = 39.9% , sTf/
519
60. Based on the age-specific infection
rates, when did German measles pre¬
viously occur in this village—in rela¬
tion to the current epidemic?
(A) 1 to 9 years ago
(B) 10 to 19 years ago
(C) 20 to 39 years ago
(D) 40 to 59 years ago
(E) 60 years ago
20 Preventive Medicine and Public Health

61. The authors of a study state that 63. Current techniques for estimating
careful autopsies show that 60 percent the number of cancers caused by low-
of all persons who die have evidence level radiation are based primarily on
of recent or previous pulmonary embo- (A) data from women who have
lism and conclude that pulmonary undergone mammography
embolism is the leading cause of death (B) data from atomic bomb victims
in the U.S. Possible reasons for dis¬ (C) the Ames model
agreeing with the conclusion include (D) cohort studies of x-ray technicians
all the following EXCEPT (E) none of the above
(A) confounding
(B) selection bias 64. In a previously well 35-year-old
(C) random error man with 3 h of severe, crushing sub-
(D) lack of definition of pulmonary sternal chest pain and 5 mm of acute
embolism ST elevation in leads V2 to V4, appli¬
(E) lead-time bias cation of Bayesian reasoning would
indicate that
62. “The absence of valvular calcifica¬
(A) he is likely to have a nonathero-
tion in an adult suggests that severe
sclerotic form of coronary heart
valvular aortic stenosis is not present.”
disease
This statement means that
(B) he is likely to have a serum
(A) valvular calcification is a sensitive cholesterol above 9 mmol/L
test for severe valvular aortic (326 mg/dL)
stenosis (C) he is likely to be having an acute
(B) valvular calcification is an insensi¬ myocardial infarction
tive test for severe valvular aortic (D) an elevated level of a creatine
stenosis kinase MB isozyme is likely to be
(C) valvular calcification is a specific a false positive
test for severe valvular aortic (E) measurement of creatine kinase
stenosis MB isozymes would not be a very
(D) valvular calcification is a nonspe¬ sensitive test
cific test for severe valvular aortic
stenosis
(E) the positive predictive value of
valvular calcification is high
 Biostatistics and Methods of Epidemiology 21

65. Decision analyses often include a 66. A decision analysis of the appro¬
patient’s utilitiesin the determination priate treatment of suspected herpes
of the best decision. These utilities encephalitis shows that the decision
measure whether to treat with acyclovir is
(A) whether a patient favors one deci¬ insensitive to the likelihood of severe
sion over another complications of brain biopsy needed
(B) whether a physician favors one to make the diagnosis. This implies
decision over another that
(C) the difference between a patient’s (A) a brain biopsy has a low sensitiv¬
decision and the physician’s ity for the diagnosis of herpes
decision simplex encephalitis
(D) the relative value a patient places (B) a brain biopsy has a low specific¬
on a particular outcome ity for the diagnosis of herpes
(E) the relative likelihood of a particu¬ simplex encephalitis
lar outcome (C) a brain biopsy has a high specific¬
ity for the diagnosis of herpes
simplex encephalitis
(D) the exact frequency of severe
complications following brain
biopsy is not important
(E) acyclovir therapy is unlikely to be
successful in a patient who has
had a brain biopsy
22 Preventive Medicine and Public Health

DIRECTIONS: Each group of questions below consists of lettered headings fol¬

lowed by a set of numbered items. For each numbered item select the one lettered
heading with which it is most closely associated. Each lettered heading may be used
once, more than once, or not at all.

Questions 67-70 Questions 71-74

For each of the studies below, For each of the following descrip¬
choose the most appropriate statistical tions, pick the appropriate epidemio¬
test to analyze the data. logic term.
(A) Chi-square analysis (A) Confounding
(B) Student t test (B) Effect modification
(C) Analysis of variance (interaction)
(D) Paired t test (C) Misclassification bias
(E) Linear regression (D) Lead-time bias
(E) None of the above
67. Comparison of systolic blood pres¬
sures in independent samples of preg¬ 71. Elevated bilirubin levels in neo¬
nant and nonpregnant women nates are associated with brain damage
only in babies who also have infections
68. Comparison of the prevalence of or severe hemolytic disease
hepatitis B surface antigen (HBsAg) in
medical and dental students 72. People who drink coffee tend to
smoke more, and for this reason coffee
69. Comparison of the level of blood drinkers have a higher risk of lung
glucose in male and female rats fol¬ cancer
lowing administration of three differ¬
ent drugs 73. Higher lead levels in hyperactive
children may be due to increased con¬
70. Comparison of serum cholesterol sumption of paint in children who
before and after ingestion of ham¬ were already hyperactive
burgers in a sample of fast-food
patrons 74. The association between blood
lead levels and IQ may be a result of
the fact that both are associated with
socioeconomic status
 Biostatistics and Methods of Epidemiology 23

Questions 75-78 Questions 79-83

In each statement below, data are Match the examples below with the
presented based on a cohort study of appropriate epidemiologic terms.
coronary heart disease. Choose the
(A) Lead-time bias
parameter that best describes each of
(B) Surveillance bias
these statements.
^t(C) Recall bias
(A) Point prevalence (D) Type 1 error
(B) Incidence rate (E) Power
(C) Standardized morbidity ratio
(D) Relative risk 79. Medical students who fail a physi¬
(E) None of the above ology examination are more likely to
report missing two or more physiology
75. At the initial examination,-17 per¬ lectures than those who fail a neuro¬
sons per 1,000 had evidence of coro¬ anatomy examination
nary heart disease (CHD)
80. The chance of discovering the
76. Among heavy smokers, the truth that twice as many of your
observed frequency of angina pectoris friends are at the movies as are study¬
was 1.6 times as great as the expected ing for their board examinations
frequency during the first 12 years of
the study 81. In a class of 150 medical students,
there will likely be a few who can
77. During the first 8 years of the answer this question correctly without
study, 45 persons developed coronary understanding the material
heart disease per 1,000 persons who
entered the study free of disease 82. The likelihood of finding a lost
biochemistry notebook in your apart¬
78. The risk of coronary heart disease ment is higher in the month of June
was 23 percent higher in relatives of than in the month of March
patients with CHD than in the general
population 83. Medical students enrolled in a
first-year anatomy class are more
likely to remain at their same
addresses for the next 2 years than
medical students enrolled in fourth-
year clerkships
24 Preventive Medicine and Public Health

Questions 84-87 Questions 88-91

Choose the rate that best describes For each of the descriptions of statis¬
each statement below. tical procedures below, choose the sta¬
tistical error.
(A) Secondary attack rate
(B) Case-fatality rate (A) Observations are not
(C) Morbidity rate independent
(D) Age-adjusted mortality (B) Variable is not normally
(E) Crude mortality distributed
(C) Unequal group sizes or
84. Death occurs in 10 percent of unequal variances
cases of meningococcal meningitis (D) Not enough degrees of
freedom
85. Approximately 9 people die each (E) None of the above
year in the United States for every
1,000 estimated to be alive 88. The frequencies of infection of the
urinary tract are compared in 24 chil¬
86. Eighty percent of susceptible dren treated with intermittent catheter¬
household contacts of a child with ization and 9 children treated with uri¬
chickenpox develop this disease nary diversion. In the catheterization
group, 85/231 urine cultures were posi¬
87. Children between the ages of 1 tive, compared with 34/55 in the diver¬
and 5 have an average of eight colds sion group. The difference between
per year these two proportions was statistically
significant (x2 = 11 -4, degrees of free¬
dom = 1.0, p < 0.001)

89. The mean lengths of hospital stay

were compared in 20 premature infants
treated surgically and 20 premature
infants treated medically for patent
ductus arteriosus. The mean lengths of
stay in the two groups were 50 ± 30
and 30 ± 30 days (mean ± standard
deviation), respectively. The t test was
not significant (t = 1.5, p = 0.07)
 Biostatistics and Methods of Epidemiology 25

90. The parity of 100 women with 91. An investigator wishes to examine
breast cancer was compared with that the efficacy of varicella vaccine. For
of 200 controls. Among the cases, 40 convenience, he randomizes four
percent were nulliparoijs,'30 percent classes of third graders for the study:
had one child, and 30 percent had more two classes receive the vaccine and two
than one child; among the controls the do not. Over the next year, only 2/57
numbers were 20, 30, and 50 percent, children in the vaccine group, but
respectively. The mean (± SD) parity 25/60 children in the control group get
in the cases was 1.6 ± 2.1; in the con¬ chickenpox (x2 = 23, degrees of free¬
trols it was 2.2 ± 2.3. The difference dom = 1.0, p < 0.001). The investiga¬
was statistically significant using the t tor concludes that the vaccine is highly
test (t — —2.2; p = 0.04) effective

Questions 92-95 94. Acute anxiety neurosis was diag¬

nosed among 250 patients and follow¬
For each of the studies described
up data were available on 80 percent
below, select the critical statement that
of these patients 10 years later. The
best explains why the conclusion is
mortality experience of this cohort was
misleading or false.
no different -thafr tHat of the general
(A) Lack of a control group population. The authors concluded that
(B) Lack of proper follow-up the diagnosis of acute anxiety neurosis
(C) Lack of adjustment for age was not associated with a decrease in
(D) Lack of denominators longevity
(E) Lack of adjustment for race
95. Of 143 patients who died of bac¬
92. Of 250 consecutive, unselected terial endocarditis and on whom
women in whom acute cholecystitis autopsies were performed, 2 percent
was diagnosed, 75 were under age 50 were less than 10 years of age. The
and 175 were over age 50. The inves¬ authors concluded that bacterial endo¬
tigator concluded that older women carditis is rare in childhood
are at greater risk of acute cholecysti¬
tis than are younger women

93. In a review of 3,000 patients in

whom adult-onset diabetes was diag¬
nosed, 2,000 of these patients were
obese at the time of diagnosis. The
investigator concluded that there was
an association between diabetes and
obesity
26 Preventive Medicine and Public Health

Questions 96-99 Questions 100-103

Choose the term that best fits the The following two-by-two table
description. represents the findings of a 5-year
cohort study in which the incidence of
(A) Matching
suicide in veterans who served in Viet¬
(B) Stratification
nam was compared with that of veter¬
(C) Age adjustment
ans who served elsewhere. Match the
(D) Multivariate statistical
analysis name of the parameter below with the
(E) Survival analysis appropriate formula.

96. In a cohort study of hypertensive No

men, the proportions of subjects with Suicide Suicide
high and low renin levels who survived Served in Vietnam a b
for 5 years are compared separately Served elsewhere c d
among those aged 40 to 49, those aged
50 to 59, and those aged 60 to 69 at (A) ad/bc
entry (B) (a + b)/(a + b + c + d)
(C) (a + c)/(a + b + c + d)
97. A sampling strategy is used to (D) [a/(a + b)]/[c/(c + d)]
achieve comparability of the groups (H) a/(a + b) — c/(c + d)
being studied
100. The odds ratio
98. A technique that takes into ac¬
count variable length of follow-up is 101. The relative risk
used
102. The excess risk of suicide in
99. Six different risk ratios are calcu¬ Vietnam veterans
lated: one for each sex at each of three
social class levels 103. The overall incidence (per 5
years) of suicide in the study
 Biostatistics and Methods of Epidemiology 27

Questions 104-107 Questions 108-112 I

Match each description of a sam¬ A new test for chlamydial infections
pling procedure with the correct term. of the cervix is introduced. Half offne
(A) Systematic sampling women who are tested have a positive
(B) Paired sampling test. Compared with the gold standard
(C) Simple random sampling of careful cultures, 45 percent of those
(D) Stratified sampling with a positive test are infected with
(E) Cluster sampling chlamydia, and 95 percent of those
with a negative test are free of the
104. Each individual of the total infection. Match the epidemiologic
group has an equal chance of being terms below with the correct
selected percentage.
(A) 5 percent
105. Households are selected at ran¬ (B) 25 percent
dom, and every person in each house¬ (C) 80 percent
hold is included in the sample (D) 90 percent
(E) None of the above
106. Individuals are initially as¬
sembled according to some order in a 108. Sensitivity of the test
group and then individuals are selected
according to some constant determi¬ 109. Specificity of the test
nant; e.g., every fourth subject is
selected 110. Prevalence of chlamydial infec¬
tion in that community
107. Individuals are divided into sub¬
groups on the basis of specified char¬ 111. Predictive value of a positive test
acteristics and then random samples
are selected from each subgroup 112. Predictive value of a negative
test
28 Preventive Medicine and Public Health

Questions 113-116

Select the letter corresponding to the figure that best fits each description.

113. Distributions 1 and 2 have the same median and mean, but 2 has the
greater variance

114. Distributions 1 and 2 have the same variance, but different means

115. Compared with curve 2, curve 1 is skewed to the right

116. This figure represents a tasty dessert

 Biostatistics and Methods of Epidemiology 29

Questions 117-120 Questions 121-125

For each result or conclusion de¬ For each variable described below,
scribed below, select the choice that choose the type of measurement scale.
might best explain it.
(A) Dichotomous scale
(A) Ecologic fallacy (B) Nominal scale
(B) Type 1 error (C) Ordinal scale
(C) Type 2 error (D) Interval scale
(D) Selection bias (E) Ratio scale
(E) Misclassification bias
121. Survival of a particular patient
117. A randomized blinded trial of for at least 5 years
aspirin to prevent myocardial infarc¬
tion fails to find a difference between 122. Frequency of somnolence during
aspirin and placebo groups after 5 biochemistry lectures: never, some¬
years (N = 500 per group; p = 0.11) times, usually, or always

118. A study of patterns of contracep¬ 123. Birth weight

tive use finds that counties with the
highest per capita use of condoms also 124. Type of medical specialty
have the highest pregnancy rates (N =
100,000; p < 0.001) and concludes 125. Year of birth
that condoms are ineffective as
contraceptives

119. An investigator analyzes data

from the National Health Interview
Survey and finds that there is a positive
association between consumption of
turkey and degenerative joint disease
in black women 50 to 59 years old
(N = 50; p < 0.05)

120. In a case-control study of lung

cancer, cases’ spouses are chosen as
controls. The odds ratio for smoking is
3.0, which does not quite reach statis¬
tical significance (N = 30 per group;
p = 0.07)
30 Preventive Medicine and Public Health

Questions 126-130 126. “I found that the sensitivity of the

Blues test was...”
Dr. Vera Blues, a noted psychiatric
epidemiologist, is interested in the
127. “The specificity of the Blues test
diagnosis of depression. She develops a
was...”
new test for its diagnosis, which she
calls the Blues test. According to the
128. “The likelihood that someone
gold standard, which involves meeting
with depression would have a positive
the DSM III criteria, about 1 percent of
Blues test was...”
adults in the U.S. are depressed. Dr.
Blues studies her new test in 100 per¬
129. “The likelihood that someone in
sons diagnosed as being depressed by
the population with a positive Blues
the gold standard; 80 have a positive
test would be depressed was...”
Blues test. She finds 400 persons who
are not depressed; again, 80 have a
130. “The likelihood that someone in
positive test. She reports her findings in
the population with a negative Blues
the Journal of the Society of Academic
test would be depressed was...”
Psychiatrists (JSAP). Match the state¬
ments that Dr. Blues made in her arti¬
cle with the appropriate percentage.
(A) 80 percent
(B) 50 percent
(C) 20 percent
(D) < 1 percent
(E) None of the above
aa»JI VJrfv -
$-£> & O hn>-

•£uv
\ ^
3
<ko