Individualized Pharmacokinetic Monitoring Results in

Less Aminoglycoside-Associated Nephrotoxicity and

Fewer Associated Costs
Daniel S. Streetman, Pharm.D., Anne N. Nafziger, M.D., M.H.S.,
Christopher J. Destache, Pharm.D., and Joseph S. Bertino, Jr., Pharm.D.

Study Objective. To examine the impact of individualized pharmacokinetic

monitoring (IPM) on the development of aminoglycoside-associated
nephrotoxicity (AAN).
Design. Retrospective case-control study.
Setting. Two teaching hospitals.
Subjects. Two thousand four hundred five patients who received
aminoglycosides.
Intervention. Aminoglycoside therapy dosed by either IPM or physicians’
directions.
Measurements and Main Results. Patients receiving IPM were significantly
less likely to develop AAN by both univariate (7.9% vs 13.2%, p=0.02) and
multivariate methods (odds ratio 0.42, p=0.002). Female sex was
protective against AAN. Age 50 years and above, high initial
aminoglycoside trough, long duration of therapy, and concurrent
piperacillin, clindamycin, or vancomycin increased risk of AAN. We
estimated that IPM decreased AAN costs by $90,995/100 patients.
Conclusion. Individualized pharmacokinetic monitoring significantly
decreased the frequency of AAN and its associated economic costs.
(Pharmacotherapy 2001;21(4):443–451)

Nephrotoxicity is the primary toxicity associated nephrotoxicity (AAN) is the result of

associated with aminoglycoside therapy. 1 Its aminoglycoside accumulation in the renal
reported frequency varies widely, primarily due to cortex. 2, 4–6 The polycationic aminoglycoside
different definitions, but in most reviews it was molecule binds to the anionic brush border of
5–15%.2, 3 It is believed that aminoglycoside- proximal tubular renal cortical cells, leading to
From the Clinical Pharmacology Research Center (Drs.
cellular uptake of the agent. Intracellularly, the
Streetman, Nafziger, and Bertino), and the Departments of aminoglycoside interferes with normal
Medicine (Drs. Nafziger and Bertino) and Pharmacy phospholipid transport, but it is unclear if this
Services (Dr. Bertino), Bassett Healthcare, Cooperstown, process is responsible for cortical cell death.5, 7
New York; and the Department of Pharmacy Practice, Individualized pharmacokinetic monitoring
Creighton University School of Pharmacy and Allied Health
Professions, Omaha, Nebraska (Dr. Destache).
(IPM) is used by clinicians in an attempt to
Presented in part at the 38th Interscience Conference on optimize aminoglycoside pharmacodynamics.8
Antimicrobial Agents and Chemotherapy, San Diego, Patient-specific pharmacokinetic parameters are
California, September 24–27, 1998. examined to design a dosage that provides
Address reprint requests to Daniel S. Streetman, specific peak (and target maximum
Pharm.D., UHB2D301/Box 0008, University of Michigan
Health System, 1500 East Medical Center Drive, Ann Arbor,
concentration:minimum inhibitory concentration
MI 48109-0008. ratios) and trough concentrations to achieve
444 PHARMACOTHERAPY Volume 21, Number 4, 2001

maximum therapeutic effect, 8–10 reduce Clinical Pharmacology Service (CPS) of BHC or
development of resistance, 11 and minimize by the Clinical Pharmacokinetic Service (CPK) of
toxicity.12–15 Although many studies reported that SJH. In both IPM and non-IPM groups, initial
IPM enhances aminoglycoside efficacy, 8–10 few doses were determined by physicians and not
examined its impact on AAN. Most that did altered by the CPS or CPK unless they would
reported less AAN in patients receiving IPM,9, 12, result in significant underdosing or overdosing.
13
whereas others failed to find significant benefit After the first dose, patients who were random-
of IPM.16, 17 Unfortunately, most of these studies ized to receive IPM for the duration of amino-
lacked appropriate control groups and compared glycoside therapy were in the IPM group. The
their results only with historical controls or non-IPM group was composed of patients who
reported rates of AAN in the general population. were randomized to receive no IPM monitoring
Despite AAN’s well-known multifactorial in a previous study at SJH.22, 23
nature,18–20 these studies either did not account Pharmacokinetic dosing in the IPM group was
for this or were not designed to control for it. based on one predose concentration and one or
Much work to identify risk factors for AAN has two postdose concentrations. The nursing staff
been done in patients receiving traditional dosing was instructed to infuse aminoglycoside doses
every 8 hours 19, 20 and in those who received over 30 minutes, and the actual duration of each
IPM18; however, IPM itself was never included in infusion was recorded. Predose concentrations
analyses as a potential risk or protective factor. were scheduled to be drawn 30 minutes before
In addition, considering the substantial economic the next dose. Postdose concentrations were
costs of AAN, 21 IPM could have significant scheduled to be obtained 30 minutes or longer
economic benefits by reducing the disorder’s after the end of the infusion. One or two addi-
frequency. Only a few studies have examined tional postdose concentrations were drawn at the
these potential economic benefits,9, 16, 22, 23 and all beginning of therapy in an attempt to sample
reported shorter hospital stays and significant over one estimated half-life. For first-dose
cost savings. studies, two serum concentrations were obtained
The primary goal of this study was to examine over one estimated half-life. Actual times all
the impact of IPM on the development of AAN samples were obtained were recorded and used in
by comparing the frequency of AAN in patients all calculations. Reported times were verified by
receiving IPM with that in patients not receiving checking drug administration records and/or
IPM. A secondary goal was to examine the nursing flowsheets. An earlier internal audit
significance of IPM as a risk factor for AAN using showed that compliance with scheduled infusion
multiple logistic regression. We also evaluated and blood-drawing times was approximately
economic costs and benefits associated with IPM 90%.
compared with traditional dosing, and All samples were analyzed within 6 hours of
determined the significance of other potential collection. For subjects receiving concurrent
risk factors for the development of AAN using carbenicillin sodium, piperacillin sodium, or
both univariate and multivariate statistical ticarcillin disodium, samples were immediately
procedures. spun and frozen at -20°C until analysis; all other
samples were spun and refrigerated at 4°C until
analysis.
Methods
Analyses were done using an enzyme-
Adult (age ≥ 18 yrs) inpatients receiving multiplied immunoassay technique (Syva, Palo
gentamicin or tobramycin for 48 hours or longer Alto, CA) from May 1982–December 1982 and a
for treatment of a suspected or documented fluorescence polarization immunoassay
infection were eligible for the study. All subjects technique (TDX; Abbott Laboratories, Abbott
received treatment between May 1982 and May Park, IL) from January 1983–May 1997. Serum
1997 at Bassett Healthcare (BHC; Cooperstown, concentration data were analyzed by a Bayesian
NY) or between January 1988 and June 1988 at model24, 25 (SJH) or by the method of Sawchuck
Saint Joseph Hospital (SJH; Omaha, NE). and Zaske 26 (BHC), fitting the data to a one-
Patients with cystic fibrosis and those receiving compartment intravenous infusion model. These
peritoneal dialysis or hemodialysis were two methods are equally effective at achieving
excluded. Some patients at BHC were included optimum aminoglycoside dosages. 27 Dosages
in earlier studies of risk factors for AAN.12, 18 were adjusted to maintain particular target 1-
All data were collected prospectively by the hour (extrapolated) peak (C max ) and trough
 INDIVIDUALIZED PHARMACOKINETIC MONITORING REDUCES AAN Streetman et al 445

concentrations (C min) below 2 µg/ml. Target causes, and patients with nephrotoxicity who did
peak concentrations were 6–10 µg/ml for not have another definable cause were classified
pneumonia; 5–6 µg/ml for sepsis, neutropenic as having AAN. Outcomes of AAN were
fever, intraabdominal sepsis, and skin and soft determined for 118 patients for whom such
tissue infections; 4–5 µg/ml for urinary tract follow-up data were available. Information on
infections; and 3–4 µg/ml for synergy against outcome (recovered, died, lost to follow-up),
gram-positive organisms. initial Scr, maximum Scr, and time to changes in
After the first dose, all dosage adjustments in Scr was recorded for each of these patients.
the IPM group were based on recommendations All analyses were performed using the SAS
made by the CPS or CPK. In the non-IPM group, software system version 6.08 (SAS Institute, Cary,
changes in dosage were made by physicians NC). Univariate analyses comparing IPM and
without input from the CPS or CPK. In the IPM non-IPM groups and AAN and non-AAN groups
group, follow-up serum aminoglycoside were performed with two-tailed Student’s t tests
concentrations (one predose, one postdose) were for continuous variables and x2 for categoric
measured within 24–72 hours of the first variables. A graphic analysis of the effect of age
adjustment, and further adjustment was made if on the frequency of AAN showed a significant
necessary. Serum aminoglycoside trough increase in risk after age 50 years; therefore, age
concentrations were assessed approximately was included in univariate and multivariate
every 72 hours unless pharmacokinetic values analyses as a categoric variable as younger or
were unstable or the patient’s condition had not older than 50 years. Variables identified as
improved. In those instances, more detailed significant predictors of AAN from univariate
(two- or three-sample) and/or more frequent analyses were entered into a multivariate
pharmacokinetic studies were performed based stepwise logistic regression model. Interaction
on clinical judgment of the CPS or CPK. terms were tested for all significant variables. A
p value of 0.05 or below was considered
Nephrotoxicity Analysis statistically significant.
Data on demographics, organism culture,
Pharmacoeconomic Analysis
organism sensitivities, chemistry, hematology,
urinalysis, clinical course, outcome, amino- Data regarding cost of AAN were derived from
glycoside dosage, and pharmacokinetics were a subset of 1165 patients at BHC who received
collected for all patients. Serum creatinine (Scr) IPM and for whom total cost of care, outcome,
concentrations routinely were measured at least and total length of stay were available. Total cost
every other day. For patients in the IPM group of care represents the total cost to the system and
without adequate documentation of Scr, a request includes all costs. Total length of stay represents
was made by the CPS or CPK for the physician to the time from hospital admission to discharge.
order the test. Creatinine clearance (Clcr) was Cost data were adjusted by the medical inflation
calculated by a computerized Bayesian dosing rate from the years they were obtained until 1999
program (SJH) or by the method of Cockcroft using data from the U.S. Bureau of Labor
and Gault28 (BHC). Ideal body weight was used Statistics (http://stats.bls.gov). They are reported
when calculating Clcr in obese subjects.29 in 1999 dollars.
Nephrotoxicity was defined as a rise in Scr of The cost of AAN in the IPM and non-IPM
0.5 mg/dl or more (≥ 44.2 µmol/L) if the baseline groups was calculated using the frequency of
value was below 3.0 mg/dl (265 µmol/L), or as a AAN reported herein for each group.
rise of 1.0 mg/dl or more (88.4 µmol/L) if the Cost/patient with AAN was defined as the
baseline was 3.0 mg/dl or more (265 µmol/L).21 difference in total cost between these patients
Change in Scr (∆Scr) was calculated as follows: and those without AAN. This figure multiplied
∆Scr = (highest Scr during aminoglycoside therapy by the frequency of AAN in each group
or within 3–5 days after end of therapy) - represents the cost of AAN/100 patients.
(baseline S cr). If the S cr began to rise during Differences in cost of length of stay for patients
therapy but had not peaked by the end of with AAN versus those without AAN were
therapy, it was followed until it peaked. Peak Scr compared with Student’s t test.
was used to determine nephrotoxicity. All Cost of providing IPM is included in total cost
subjects meeting the criteria for nephrotoxicity of care as described; however, a separate estimate
were evaluated for the presence of other possible of the cost was calculated and included costs of
446 PHARMACOTHERAPY Volume 21, Number 4, 2001
Table 1. Demographic and Clinical Variables for IPM and Non-IPM Patients
Variable Non-IPM Patients IPM Patients p Valuea
No. (%) 152 (6.3) 2253 (93.7) —
Age (yrs) 61.4 ± 21.5 61.2 ± 17.7 0.90
No. (%) women 87 (57.2) 1046 (46.3) 0.009
No. (%) men 65 (42.8) 1211 (53.7)
Total body weight (kg) 67.9 ± 18.8 72.4 ± 19.8 <0.008
Initial Clcr (ml/min/1.73m2) 71.4 ± 37.3 66.4 ± 34.1 0.08
Aminoglycoside, no. (%)
Gentamicin 104 (68.4) 1678 (74.5) 0.10
Tobramycin 48 (31.6) 575 (25.5)
Concurrent antimicrobial, no. (%)
Cephalosporin 73 (48.0) 853 (37.8) 0.012
Piperacillin 13 (8.6) 419 (18.6) 0.002
Clindamycin 16 (10.5) 567 (25.1) <0.001
Vancomycin 10 (6.6) 252 (11.2) 0.08
By x2 analysis or unpaired t test where appropriate.
a

Table 2. Site of Infection of IPM and Non-IPM Patients

Non-IPM Patients IPM Patients
Type of Infection no. (%) no. (%) p Valuea
Pneumonia 61 (40.1) 781 (34.6) 0.17
Intraabdominal 17 (11.2) 515 (22.8) 0.001
Urinary tract 26 (17.1) 385 (17.1) 0.99
Sepsis 32 (21.1) 283 (12.5) 0.003
Fever of unknown origin 0 (0) 309 (13.7) <0.001
Wound 12 (7.9) 176 (7.8) 0.97
Neutropenic fever 0 (0) 142 (6.3) <0.001
Skin, soft tissue 13 (8.6) 99 (4.4) 0.018
Bone and joint 2 (1.3) 57 (2.5) 0.58
Endocarditis 2 (1.3) 28 (1.2) 0.71
Central nervous system 2 (1.3) 10 (0.4) 0.17
Prophylaxis 0 (0) 3 (0.1) 1.00
Comparisons by x2 analysis.
a

salaries, serum concentration monitoring, and Results

supplies. Time spent on IPM was defined as 90
During the study period 2446 patients
minutes for an initial work-up, 10 minutes/day receiving gentamicin or tobramycin were
for follow-up, and an additional 20 minutes of evaluated; most of them (93.7%) received IPM.
follow-up on days when a repeat pharmaco- Patients in the IPM and non-IPM groups were of
kinetic study was performed. Salary cost was similar age but differed with regard to several
estimated using a clinical pharmacist salary of other characteristics (Table 1). Patients receiving
$60,000/year and benefits of $16,800/year. IPM were significantly more likely to be male,
Cost/minute was estimated based on 48 working had higher initial total body weight, were less
weeks/year and 40 hours of work/week. The cost likely to receive concurrent cephalosporins, and
of determining an aminoglycoside serum were more likely to receive concurrent
concentration was conservatively estimated to be piperacillin or clindamycin. The IPM group also
$10/sample, as that reflects costs over much of had a nonsignificant trend toward lower initial
the period during which data were collected.30 Cl cr, more gentamicin therapy, and increased
The cost of aminoglycoside serum concentrations, concurrent vancomycin therapy.
however, decreased at BHC from $10 during Types of infections for which patients were
1990–1993 to $6 in 1996.31 receiving gentamicin or tobramycin are shown in
Data are presented as mean ± SD or as number Table 2. Significantly more patients being treated
(%) unless otherwise noted. for an intraabdominal infection, neutropenic
 INDIVIDUALIZED PHARMACOKINETIC MONITORING REDUCES AAN Streetman et al 447

fever, or fever of unknown origin received IPM, Table 3. Reasons for Increases in Serum Creatinine other
and significantly fewer patients being treated for than Aminoglycoside Therapy
sepsis or infection of the skin or soft tissue Reason No. of Patients
received IPM. Acute renal failure in progressa 29
Acute respiratory distress syndrome 3
Two hundred forty (9.8%) patients met criteria Acute nephrosclerosis 3
for nephrotoxicity. Of these, 41 (17.1%) had Interstitial nephritis 3
other determinable causes (Table 3) of changes in Myeloma 2
Scr and were excluded from further analysis. This Myoglobinuria 1
a
left 199 (8.1%) patients who met criteria for AAN Evidenced by an acute rise in serum creatinine before start of
aminoglycoside therapy.
and comprised the AAN group for all statistical
analyses.
By x2 analysis a significantly higher frequency
of nephrotoxicity was seen in patients not After including the 12 risk factors into a
receiving IPM than in those receiving IPM stepwise logistic regression model, initial Clcr,
(13.2% and 7.9%, respectively, p=0.024). therapy with tobramycin, and initial Cmax were
Univariate analyses also identified 11 other no longer predictive risk factors (Table 5).
statistically significant risk factors for AAN Positive risk factors were initial Cmin, duration of
(Table 4). Ten positive risk factors included age therapy, concurrent piperacillin, age 50 years or
50 years or older; initial Clcr; tobramycin therapy; older, concurrent clindamycin, concurrent
initial C max; initial C min; duration of therapy; vancomycin, and presence of pneumonia. Again,
concurrent piperacillin, vancomycin, or IPM and female sex were protective risk factors.
clindamycin; and presence of pneumonia. Use of Outcome of AAN was evaluated for 118
IPM and female sex were significant protective (59.3%) patients defined as having AAN (Table
risk factors. 6). Approximately 43% were followed to

Table 4. Risk Factors for AAN Identified by Univariate Analyses (t Test or x2, where appropriate)
Risk Factor AAN No AAN p Value
Age ≥ 50 years, no. (%) 174 (87.4) 1638 (74.3) <0.0001
Initial Clcr (ml/min/1.73m2) 57.3 ± 29.1 67.6 ± 34.7 0.0001
Tobramycin therapy, no. (%) 64 (32.2) 559 (25.2) 0.035
Initial Cmax (µg/ml) 5.5 ± 2.8 5.0 ± 2.5 0.034
Initial Cmin (µg/ml) 1.2 ± 1.2 0.8 ± 0.9 0.0001
Duration of therapy (days) 9.7 ± 5.6 7.5 ± 5.0 0.0001
Male sex, no. (%) 119 (59.8) 1156 (52.4) 0.045
Concurrent piperacillin, no. (%) 65 (32.7) 367 (16.6) <0.001
Concurrent vancomycin, no. (%) 40 (20.1) 222 (10.1) <0.001
Concurrent clindamycin, no. (%) 62 (31.2) 518 (23.5) 0.015
Pneumonia, no. (%) 93 (46.7) 748 (33.9) <0.001
Non-IPM, no. (%) 20 (13.2) 179 (7.9) 0.024
Cmax = maximum (peak) concentration; Cmin = minimum (trough) concentration.

Table 5. Predictors of AAN Identified by Stepwise Multiple Logistic Regression

Parameter Odds 95% Confidence
Variable Estimate (± SE) Ratio Interval p Value
Initial Cmin (µg/ml) 0.2672 ± 0.0670 1.31 1.15–1.49 0.0001
Duration of therapy (days) 0.0570 ± 0.0129 1.06 1.03–1.09 0.0001
Concurrent piperacillin 0.8087 ± 0.1786 2.25 1.58–3.19 0.0001
Concurrent clindamycin 0.7538 ± 0.1763 2.13 1.50–3.00 0.0001
Concurrent vancomycin 0.8145 ± 0.2086 2.26 1.50–3.40 0.0001
IPM -0.8606 ± 0.2804 0.42 0.24–0.73 0.0021
Age ≥ 50 years 0.6316 ± 0.2513 1.88 1.15–3.08 0.0119
Pneumonia 0.3963 ± 0.1644 1.49 1.08–2.05 0.0159
Female sex -0.3345 ± 0.1592 0.72 0.52–0.98 0.0356
Cmin = minimum (trough) concentration.
448 PHARMACOTHERAPY Volume 21, Number 4, 2001
Table 6. Outcome and Serum Creatinine Data for 118 Patients with AAN
Initial Scr Scr-max Days to Days to Days to
Outcome (mg/dl) (mg/dl) Rising Scr Scr-max Baseline Scr
Recovered (n=51) 1.1 ± 0.4 2.4 ± 1.3 10.4 ± 6.0 12.5 ± 6.7 21.9 ± 21.8
Died (n=41) 1.2 ± 0.5 3.6 ± 2.1 8.0 ± 4.8 12.8 ± 7.1 NA
Lost to follow-up (n=26) 1.4 ± 0.8 3.4 ± 2.2 11.5 ± 6.0 15.7 ± 6.8 NA
NA = data not available.

recovery. Nearly 35% died during the same in a multivariate analysis, IPM was a significant
hospitalization as the episode of AAN (none negative risk factor, with a relative risk of 0.42.
considered secondary to AAN complications); This represents a 58% reduction in risk for
22% were lost to follow-up. Of the 118 patients, development of AAN. These results support the
only 6 required hemodialysis (one session each). findings of several earlier studies that found less
Of 1165 patients evaluated to determine the AAN with IPM 9, 12, 13, 16 but conflict with the
cost of IPM and its impact on the cost of AAN, 96 findings of another study that reported no benefit
(8.2%) met criteria for AAN. Patients without of IPM.17
AAN had a mean total hospitalization A secondary goal of this study was to examine
cost/patient of $32,416 ± 37,921 with a mean the significance of other potential risk factors for
length of stay (LOS) of 23 ± 41 days. Patients AAN. Most identified risk factors were similar to
with AAN had a mean total hospitalization those in other large, similarly designed
cost/patient of $49,585 ± 45,158 (p<0.01 vs studies18–20 with a few notable differences. Our
cost/patient without AAN) with a mean LOS of finding that tobramycin therapy was a significant
32 ± 24 days (p<0.01 vs LOS for patients without risk factor by univariate analysis is in contrast
AAN). The mean increase in total hospital cost with other studies that reported increased AAN
for patients with AAN was $17,169/patient. with gentamicin compared with tobramycin.6, 32
Patients with AAN had an increased average LOS In the multivariate analysis no difference between
of 9 days. Using AAN frequency in the non-IPM aminoglycosides was found, supporting
group of 13.2%, the total cost difference of conclusions of several reviews claiming no
AAN/100 patients not receiving IPM would be significant difference between these agents with
$226,630. In the IPM group, using a frequency respect to nephrotoxicity. 3, 33 Despite earlier
of 7.9%, the cost of AAN would be $135,635/100 findings that concurrent therapy with
patients. Subtracting this from the total cost of cephalosporins is a risk factor for AAN,18, 34 this
AAN in the non-IPM group ($226,630) gives the was not a significant risk factor in our study
added cost of $90,995/100 patients in the non- (p=0.256). We also found female sex to be a
IPM group. significant negative risk factor for AAN, which is
The cost to the institution of providing IPM in agreement with some authors18 but not with
was estimated to be $186/patient. The total time others 19, 20 It is important to note that some
required to provide IPM was calculated to be reported risk factors (liver disease, ascites, serum
approximately 188 minutes/patient. Salary and albumin, shock, leukemia) were not evaluated in
benefit cost for a clinical pharmacist to perform this study.
IPM was estimated to be $0.667/minute. Total In two published studies12, 18 the frequency of
salary cost was estimated to be $125/patient. AAN in patients receiving IPM was compared
Based on an average of six concentrations/patient with that in historical controls not receiving IPM.
receiving IPM, mean cost of serum concentrations
Subjects were 890–1490 patients who had
was $60/patient. The cost of additional supplies
received IPM by a clinical pharmacology service
(paper, photocopying, etc.) was estimated to be
using the Sawchuck-Zaske method. The reported
$1/patient.
frequency of AAN in these patients was only
7.2–7.7%, which was significantly lower than
Discussion that of historical controls and the 17.5% reported
In 2446 subjects receiving therapy with by another group20 (p<0.0001, x2 analysis). The
gentamicin or tobramycin for 72 hours or longer, authors of these studies concluded that IPM can
those receiving IPM had significantly (p=0.024) significantly reduce the frequency of AAN and
less AAN than patients not receiving IPM. Also, should be used to dose patients receiving
 INDIVIDUALIZED PHARMACOKINETIC MONITORING REDUCES AAN Streetman et al 449

aminoglycosides since AAN cannot be predicted the disorder’s multifactorial nature.

accurately. In addition to finding that IPM is a negative
In one study, the risk of AAN was decreased risk factor for AAN, we noted that by reducing
significantly in 105 patients receiving IPM using the frequency of AAN, IPM was associated with a
Bayesian methods compared with 127 controls substantial cost savings. In our analysis of a
(13.4% vs 2.9%, p<0.01).9 The trend was toward subgroup of 1165 patients who received IPM, we
overall decreased mortality in the IPM group found a mean cost excess of $17,169 and a 9-day
(8.6% vs 14.2%, p=0.26). Unfortunately, that increase in LOS/episode of AAN. The cost of
study compared only the frequency of AAN in providing IPM was approximately $186/patient,
the two groups and did not include IPM in any which is between two previously reported
multivariate models to account for potentially estimates.16, 22, 23 Using the AAN frequency for
confounding variables. both IPM (7.9%) and non-IPM (13.2%) groups,
The frequency of AAN was evaluated in 175 we estimated an overall cost savings associated
patients receiving IPM by the Sawchuck-Zaske with IPM of more than $900/patient.
method. 13 Nephrotoxicity was defined as an Our economic analysis is merely a comparison
increase in Scr of 30% or greater compared with of total costs of patients with AAN versus
baseline. Only 5 (4%) of 125 patients receiving patients without AAN, all of whom received IPM.
gentamicin and 1 (2.6%) of 39 receiving In 242 patients (106 with AAN), mean cost
tobramycin met the definition of AAN. All excess was $19,722 and mean increase in
patients who developed AAN were older than 50 LOS/episode of AAN was 13.1 days. 21 After
years (mean age 67 yrs). Although the authors regression analyses to control for potential
did not attempt to make statistical comparisons confounding variables, the authors concluded
with non-IPM groups, they did note that the that AAN resulted in an increased cost of
frequency of AAN in patients receiving IPM was $2501/episode of AAN.
less than that reported in existing literature. In another study, both LOS and total costs were
In another study, the impact of IPM by a significantly lower in the group receiving
Bayesian pharmacokinetic program on the individualized monitoring than in the control
frequency of AAN was studied in 147 patients group.9 Length of stay decreased from 26.3 ± 2.9
treated with an aminoglycoside, of whom 72 days in the control group to 20.0 ± 1.4 days in
were randomized to receive IPM and 75 the group receiving individualized monitoring
(controls) dosing according to physician choice.16 (p=0.045). Total costs were decreased by almost
Nephrotoxicity was defined as a rise in Scr of 0.5 $1900/patient in the latter group ($8431 ± 8861
mg/dl or more if baseline Scr was 1.5 mg/dl or less vs $6563 ± 4634, respectively, p=0.027). Based
or a 30% rise from baseline. There was less AAN on the observed frequency of AAN and using an
in the IPM group than in the control group (5.6% inflation-adjusted cost of $3220/episode, the
vs 9.3%, respectively), but the difference was not authors hypothesized that individualized
statistically significant. The authors stated that monitoring could result in an overall cost savings
the lack of significant difference may be due in of $33,810/100 patient courses.
part to monitoring by physicians of serum The economics of IPM by Bayesian
aminoglycoside concentrations in some members pharmacokinetic methods showed a cost
of the control group. avoidance of IPM of $1311.45/patient
The frequency of AAN was compared in monitored.16 The actual cost of providing IPM
primarily older veterans with several medical was $297.23/patient, resulting in an average of
illnesses: 80 received pharmacist-directed IPM $4.09 saved for every $1.00 invested in IPM. The
by a Bayesian pharmacokinetic program, 90 impact of IPM on LOS was determined using
received only pharmacist-assisted dosing, and 73 multivariate analysis with LOS as the dependent
acted as controls.17 Nephrotoxicity was defined variable. Only IPM and length of therapy were
as a 100% increase in Scr from baseline (at least significant predictors of LOS, and IPM was
0.5-mg/dl increase). The overall frequency of associated with a mean decrease of 4.3 days in
AAN was 20.6% and was not significantly total LOS.
different among the three groups. As a result, the In another assessment of IPM, the 75 patients
authors concluded that IPM with a Bayesian for whom pharmacokinetic recommendations
pharmacokinetic program did not decrease the were always accepted had lower mean direct
risk of AAN. Unfortunately, they examined IPM costs ($7103) than either 35 patients for whom
only as a single risk factor and did not control for recommendations were followed less than 100%
450 PHARMACOTHERAPY Volume 21, Number 4, 2001

of the time ($19,630, p=0.002) or 70 patients in decrease in total cost, even when the cost of
the control group ($13,759, p<0.05).22, 23 The providing the service is included.
calculated cost of IPM was $85/patient. Patients
receiving IPM 100% of the time also had a References
decreased LOS compared with both the group 1. Chan GLC. Alternative dosing strategy for aminoglycosides:
receiving IPM less than 100% of the time (15.7 impact on efficacy, nephrotoxicity, and ototoxicity. Drug Intell
days less, p=0.001) and controls (5.0 days less, Clin Pharm 1989;23:788–94.
2. Bennett WM. Mechanisms of aminoglycoside nephrotoxicity.
p=0.087). Clin Exp Pharmacol Physiol 1989;16:1–6.
A few limitations of our study should be 3. Meyer RD. Risk factors and comparisons of clinical
addressed. An important one is that all non-IPM nephrotoxicity of aminoglycosides. Am J Med 1986;80(suppl
6B):119–25.
patients were from one site (SJH). Based on our 4. Kaloyanides GJ, Pastoriza-Munoz E. Aminoglycoside
experiences at these institutions, we believe that nephrotoxicity. Kidney Int 1980;18:571–82.
physicians at SJH dosed patients receiving 5. Mingeot-Leclercq M-P, Tulkens PM. Aminoglycosides:
nephrotoxicity. Antimicrob Agents Chemother
aminoglycosides in the same manner as 1999;43:1003–12.
physicians at BHC; however, it would have been 6. Schentag JJ, Plaut ME, Cerra FB. Comparative nephrotoxicity
preferable to have non-IPM patients at both sites of gentamicin and tobramycin: pharmacokinetic and clinical
studies in 201 patients. Antimicrob Agents Chemother
to ensure internal validity. Unfortunately, it was 1981;19:859–66.
not possible to obtain data concerning non-IPM 7. Swan SK. Aminoglycoside nephrotoxicity. Semin Nephrol
patients from BHC as virtually all such patients 1997;17:27–33.
8. Kashuba ADM, Bertino JS Jr, Nafziger AN. Dosing of
receive IPM while being treated with amino- aminoglycosides to rapidly attain pharmacodynamic goals and
glycosides. Also, as a result of the retrospective hasten therapeutic response by using individualized
design, patients were not randomized to IPM or pharmacokinetic monitoring of patients with pneumonia
caused by gram-negative organisms. Antimicrob Agents
non-IPM and data regarding certain variables Chemother 1998;42:1842–4.
therefore were not available. The two groups had 9. van Lent-Evers NAEM, Mathot RAA, Geus WP, van Hout BA,
some differences at baseline that we believe Vinks AATMM. Impact of goal-oriented and model-based
clinical pharmacokinetic dosing of aminoglycosides on clinical
illustrate that physicians are likely to request IPM outcome: a cost-effectiveness analysis. Ther Drug Monit
for patients who are more ill and/or suspected to 1999;21:63–73.
have unusual pharmacokinetics (e.g., fluid- 10. Zaske DE, Bootman JL, Solem LB, Strate RG. Increased burn
patient survival with individualized dosages of gentamicin.
loaded patients). Based on these differences, our Surgery 1982;91:142–9.
results actually may understate the benefits of 11. Blaser J, Stone BB, Groner MC, Zinner SH. Comparative study
IPM; unfortunately, no indexes of overall disease with enoxacin and netilmicin in a pharmacodynamic model to
determine importance of ratio of antibiotic peak concentration
severity were available for most of these patients, to MIC for bactericidal activity and emergence of resistance.
prohibiting us from evaluating this further. Antimicrob Agents Chemother 1987;31:1054–60.
Although some may view the extended period of 12. Bertino JS Jr, Timm EG, Nafziger AN. Individualized
pharmacokinetic dosing of aminoglycosides: impact of
data collection as a limitation, pharmacokinetic monitoring by a clinical pharmacy service on the frequency of
monitoring procedures have changed minimally aminoglycoside nephrotoxicity and its associated costs [abstr].
at each institution during this time period. Clin Pharmacol Ther 1991;49:150.
13. Pancorbo S, Compty C, Heissler J. Comparison of gentamicin
Another potential limitation is the use of only and tobramycin nephrotoxicity in patients receiving
total cost in the economic analysis. As evident individualized-pharmacokinetic dosing regimens. Biopharm
by an earlier evaluation, 21 accounting for Drug Dispos 1982;3:83–8.
14. Dahlgren JG, Anderson ET, Hewitt WL. Gentamicin blood
confounding variables, such as severity of illness levels: a guide to nephrotoxicity. Antimicrob Agents Chemother
and age, can have a significant impact on total 1975;8:58–62.
cost of AAN. Sufficient data regarding 15. Zaske DE. Aminoglycosides. In: Evans WF, Schentag JJ, Jusko
WJ, eds. Applied pharmacokinetics: principles of therapeutic
comorbidities were not available to perform a drug monitoring, 2nd ed. Spokane, WA: Applied Therapeutics,
similar regression cost analysis. However, data 1986.
on total excess cost from other studies are 16. Burton ME, Ash CL, Hill DP, Handy T, Shepherd MD, Vasko
MR. A controlled trial of the cost benefit of computerized
included so that appropriate comparisons may be Bayesian aminoglycoside administration. Clin Pharmacol Ther
made. 1991;49:685–94.
In conclusion, IPM was associated with a 17. Leehey DJ, Braun BI, Tholl DA, et al. Can pharmacokinetic
dosing decrease nephrotoxicity associated with aminoglycoside
significant overall decrease in the frequency of therapy? J Am Soc Nephrol 1993;4:81–90.
AAN in patients receiving gentamicin or 18. Bertino JS Jr, Booker LA, Franck PA, Jenkins PL, Franck KR,
tobramycin for 48 hours or longer. It was a Nafziger AN. Incidence of and significant risk factors for
aminoglycoside-associated nephrotoxicity in patients dosed by
significant protective risk factor against using individualized pharmacokinetic monitoring. J Infect Dis
development of AAN. By reducing the frequency 1993;167:173–9.
of AAN, IPM is associated with a significant 19. Moore RD, Smith CR, Lipsky JJ, Mellits ED, Lietman PS. Risk
 INDIVIDUALIZED PHARMACOKINETIC MONITORING REDUCES AAN Streetman et al 451
factors for nephrotoxicity in patients treated with which utilize multiple intravenous infusions: gentamicin in
aminoglycosides. Ann Intern Med 1984;100:352–7. burn patients. J Pharmacokinet Biopharm 1976;4:183–95.
20. Sawyers CL, Moore RD, Lerner SA, Smith CR. A model for 27. Bottger H-C, Oellerich M, Sybrecht GW. Use of amino-
predicting nephrotoxicity in patients treated with glycosides in critically ill patients: individualization of dosage
aminoglycosides. J Infect Dis 1986;153:1062–8. using Bayesian statistics and pharmacokinetic principles. Ther
21. Eisenberg JM, Koffer H, Glick HA, et al. What is the cost of Drug Monit 1988;10:280–6.
nephrotoxicity associated with the aminoglycosides? Ann 28. Cockcroft DW, Gault MH. Prediction of creatinine clearance
Intern Med 1987;107:900–9. from serum creatinine. Nephron 1976;16:31–41.
22. Destache CJ, Meyer SK, Rowley KM. Does accepting 29. Devine BJ. Gentamicin pharmacokinetics. Drug Intell Clin
pharmacokinetic recommendations impact hospitalization? A Pharm 1974;8:650–5.
cost-benefit analysis. Ther Drug Monit 1990;12:427–33. 30. Bertino JS Jr, Rodvold KA, Destache CJ. Cost considerations in
23. Destache CJ, Meyer SK, Bittner MJ, Hermann KG. Impact of therapeutic drug monitoring of aminoglycosides. Clin
clinical pharmacokinetic service on patients treated with Pharmacokinet 1994;26:71–81.
aminoglycosides: a cost-benefit analysis. Ther Drug Monit 31. Bertino JS Jr, Rotschafer JC. Editorial response: single daily
1990;12:419–26. dosing of aminoglycosides—a concept whose time has not yet
24. Jellife RW. Clinical applications of pharmacokinetics and come. Clin Infect Dis 1997;24:820–3.
control therapy: planning, monitoring, and adjusting 32. Smith CR, Lipsky JJ, Laskin OL, et al. Double-blind
dosage regimens of aminoglycosides, lidocaine, digitoxin, comparison of the nephrotoxicity and auditory toxicity of
a n d d i g o x i n . I n : M a ro n d e R , e d . To p i c s o f c l i n i c a l gentamicin and tobramycin. N Engl J Med 1980;302:1106–9.
pharmacology and therapeutics. New York: Springer Verlag, 33. Kahlmeter G, Dahlager JI. Aminoglycoside toxicity—a review
1986:26–82. of clinical studies published between 1975 and 1982. J
25. Sheiner LB, Rosenberg B, Melmon K. Modeling of Antimicrob Chemother 1984;13(suppl A):9–22.
individualized pharmacokinetics for computer-aided drug 34. Wade JC, Smith CR, Petty BG, et al. Cephalothin plus an
dosage. Comput Biomed Res 1972;5:441–59. aminoglycoside is more nephrotoxic than methicillin plus an
26. Sawchuck RJ, Zaske DE. Pharmacokinetics of dosing regimens aminoglycoside. Lancet 1978;2:604–6.