 |
 |
 |
 |
 |
 |
 |
|
|
|
|||||||||
|
|||||||||||
Clinical Assistant Professor of Pharmacy Practice, Harrison School of Pharmacy, Auburn University, Huntsville, AL
Medical Intensive Care Clinical Pharmacist, Huntsville Hospital, Huntsville
Clinical Education Consultant, Pfizer, Inc., Helena, AL
University of Alabama-Birmingham, Huntsville Regional Medical Campus, Huntsville
Harrison School of Pharmacy, Auburn University
Reprints: Dr. Wargo, 301 Governors Dr. SW, Suite 230-N, Huntsville, AL 35801-5123, fax 256/551-4542, wargoka{at}auburn.edu
 |
Abstract |
|---|
 Top Abstract MethodsResultsDiscussionConclusionsReferences |
|---|
OBJECTIVE: To determine whether a difference exists when making antimicrobial dosage adjustments in patients with chronic kidney disease based on estimation of GFR using the MDRD and CG equations.
METHODS: We conducted an observational analysis of 409 patients with chronic kidney disease who were admitted to a tertiary care facility with an inpatient dialysis center and nephrology unit. GFR was calculated using both the 4- or 6-variable MDRD equation and the CG equation and compared using correlation and Bland-Altman methodology. Dosage discordance rates of the selected antimicrobials were determined on the basis of manufacturer renal dose recommendations.
RESULTS: Average ± SD GFR for all patients using the CG equation was 34.8 ± 12 mL/min and, using the MDRD equation, was 40.2 ± 12 mL/min (absolute mean difference 5.40; 95% CI 4.66 to 6.15; p < 0.001). The correlation coefficient between the 2 estimations, among all patients, was excellent (r = 0.80). The Bland-Altman plot yielded limits of agreement of -9.8 and 20.6; thus, the MDRD estimation may range from 9.8 mL/min below to 20.6 mL/min above the CG estimation for 95% of the cases. A discordance rate of 21-37% (p < 0.001) existed among the recommended dosing adjustments of the selected antimicrobials.
CONCLUSIONS: This analysis demonstrated statistically significant differences between the CG and MDRD equations, resulting in different dosing recommendations in 21-37% of patients. The clinical significance of these differences is uncertain in the absence of data regarding clinical outcomes that would result from the use of the discordant doses.
Key Words: Cockcroft-Gault, dosing, Modification of Diet in Renal Disease, renal
Published Online, July 11, 2006. www.theannals.com, DOI 10.1345/aph.1G635
While the GFR is the most accurate measure of kidney function, it is difficult and can be quite costly to measure directly and accurately.5 Therefore, in clinical practice it is customary to estimate the GFR based on serum creatinine concentration, using the Cockcroft-Gault (CG) equation. In March 1999, a new equation for the measurement of GFR, derived from the Modification of Diet in Renal Disease (MDRD) study, was found to provide a significantly more accurate estimation of the GFR than other commonly used equations.6 Because of this finding, questions exist as to whether this method should be used exclusively, replacing the CG equation when estimating renal function in patients with chronic kidney disease.7
Major differences exist between the MDRD and CG equations for estimating renal function.8 One difference is that the 6-variable MDRD equation takes into account 3 biochemical markers: serum creatinine (SCr), serum albumin, and blood urea nitrogen (BUN). Because not all patients routinely have these laboratory values obtained, an abbreviated version was developed to allow for simplicity in calculation, which includes only 4 variables that are more readily available.1 Both the 6- and 4-variable equations are based on body surface area (BSA); therefore, height and weight are not needed for calculation. A final difference between the MDRD and CG equations is that the MDRD accounts for ethnicity.9
One of the major responsibilities of pharmacists involves making drug dosing adjustments based on renal function. Therefore, it is increasingly important that an equation that accurately estimates renal function is used to provide the most optimal drug dosing recommendations. While the MDRD equation has many advantages, it needs further validation, particularly to determine whether statistically significant differences translate into clinically significant modifications of drug dosing in patients with chronic kidney disease. In addition, manufacturer renal dosing recommendations for medications are based on CG estimates of renal function. Therefore, it is unclear what impact estimation of GFR with the MDRD equation will have on making these dosage adjustments. Thus, the purpose of this study was to determine whether a difference exists when determining antimicrobial dosage adjustments in patients with chronic kidney disease, based on estimation of GFR using the MDRD and CG equations.
|
Methods |
|---|
|
TopAbstractMethodsResultsDiscussionConclusionsReferences |
|---|
Estimation of GFR was performed using the 6-variable MDRD equation when all variables were present: GFR = 170 x (SCr)-0.999 x (age, y)-0.176 x 0.762 (if female) x 1.18 (if black) x (BUN)-0.17 x (albumin)+0.318 and the 4-variable equation when missing one or more variables: GFR = 186.3 x (SCr)-1.154 x (age, y)-0.203 x 1.212 (if black) x 0.742 (if female). The decision to use the simplified 4 variable equation was made based on evidence presented in the National Kidney Foundation guidelines.4 The estimated GFR was then multiplied by the ratio of actual body surface area (BSA) to 1.73 m2 to determine the patient-specific GFR in mL/min. Individual patient BSA was calculated using the following equation: BSA = (weight, kg)0.425 x (height, cm)0.725 x 0.007184.10 The CG estimation of renal function was used as the comparator equation: [(140 — age) x body weight x 0.85 (if female)]/(72 x SCr). In the CG equation, the lower of actual body weight or ideal body weight (IBW) or an adjusted weight, for patients whose actual body weight exceeded their IBW by more than 30%, was used in the calculation. Adjusted body weight was determined by the equation [(actual body weight - IBW) x 0.4] + IBW.
To determine the presence of a difference when making antimicrobial dosage adjustments, dosing discordance rates among 8 commonly used antimicrobials with specific manufacturer dosing recommendations in renal impairment, based on estimation of renal function using the CG equation, were selected: cefazolin, cefepime, daptomycin, gatifloxacin, levofloxacin, meropenem, piperacillin/tazobactam, and trimethoprim/sulfamethoxazole (Table 1).11
|
STATISTICAL ANALYSIS
Data were compiled using Microsoft Access, and statistical testing was
completed using Minitab (Minitab, 1998). Using the single proportion sample
size measurement, 247 patients were needed to detect a 20% discordance rate
with a 95% CI. Comparison of continuous variables was performed by using
paired t-test, and dichotomous variables were compared using the
2 test as appropriate. ANOVA general linear model was
used to compare the absolute differences in renal function estimates using
ethnicity (African American or white) and sex as factors and age as a
covariate. Linear regression was incorporated to evaluate correlations between
continuous variables as appropriate. The Bland-Altman method was used to
assess agreement between the MDRD and CG estimations of renal
function,12,13
and 2 analysis was used to detect a difference in
dosing discordance data. Level of significance was set as p < 0.05. Data
are presented as means (range) for continuous variables, as a number for
dichotomous variables, and 95% CIs are reported as appropriate.
|
Results |
|---|
|
TopAbstractMethodsResultsDiscussionConclusionsReferences |
|---|
|
The average creatinine clearance, using the CG equation, for all patients was 34.8 ± 12 mL/min, whereas the average GFR using the MDRD equation was 40.2 ± 12 mL/min (Table 3). The absolute mean difference between the 2 estimations was 5.40 (95% CI 4.66 to 6.15; p < 0.001), and the standard deviation was 7.6. Estimation of GFR, using the 4 variable MDRD equation, was done in 25% of the patients. Factors that predicted significant differences in results included ethnicity (p < 0.001), sex (p < 0.001), and age (p < 0.001). An expected relationship was noted between age and CG estimation such that increased age was associated with lower GFR. A similar, yet weaker, relationship was found for the MDRD GFR estimation.
|
A correlation coefficient was determined for the relationship of between calculated GFR using the MDRD and CG equations among the patients evaluated (Figure 1). Excellent correlation existed (r = 0.80) among all patients (Figure 1a). When comparing the 2 estimates of GFR using the method described by Bland and Altman,12,13 the difference in values was plotted against the mean for the 2 methods to determine the variability between them. The limits of agreement were 20.6 and -9.8; thus, the MDRD estimation may be 9.8 mL/min below or 20.6 mL/min above the CG estimation for 95% of the cases (Figure 1b). For the lower limit of agreement, the confidence interval was -11.1 to -8.5 mL/min and, for the upper limit, 21.9 to 19.3 mL/min.
|
|
|
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionConclusionsReferences |
|---|
The findings presented in this analysis demonstrate the existence of a statistically significant difference between the MDRD and CG estimations of renal function. Even though a strong correlation existed when evaluating our entire cohort of patients, the Bland-Altman method for assessing agreement demonstrated a wide variation between the 2 estimations. Furthermore, based on the method chosen to estimate GFR, antimicrobials in this analysis would have been dosed differently 25% of the time (range 20-36%; p < 0.001). These findings are not only of statistical significance, but they also demonstrate the possibility of clinical importance and merit further consideration.
As stated previously, the majority of discordance existed when the manufacturer recommended a dosage adjustment according to CG estimation; however, that particular level of dosage adjustment was unnecessary, according to GFR estimation by the MDRD equation. According to this rationale, patients would have been overdosed an average of 21% (range 18-30%) of the time using the MDRD estimation, leading to the potential for adverse reactions such as seizures, arrhythmias, renal failure, gastrointestinal symptoms, neuromuscular hypersensitivity, and others. Although the potential for such adverse reactions is quite low and may not bear clinical significance, the variation between the 2 estimations was so great (20.6 to -9.8 mL/min) that a clinically significant difference may be implied. However, without actually administering an antimicrobial during the study, directly measuring GFR, comparing that measurement with our estimations, and assessing outcomes, true clinical significance can not be determined in this analysis.
Analysis contains various limitations, based on a series of assumptions. Measurement of actual GFR was not conducted on patients; instead, data presented in the MDRD study6 were used to establish that GFR can be accurately estimated using either a 6- or 4-variable equation. Thus, the major limitation of our analysis is associated with comparing 2 estimated values. Furthermore, drug concentrations were not monitored during this study due to a lack of resources.
Because the MDRD equation was chosen as the comparator estimator of renal function, it is important to control for patient demographic differences between this analysis and the Levey et al.6 study. With respect to mean weight, BUN, BSA, and race, our cohort of patients was comparable to patients in the MDRD study. Of note, the cohort of patients evaluated by Levey et al. exhibited a mean age of 50.6 ± 12.7 years, whereas our population was significantly older (73.4 ± 12.5 y). However, results from another study suggest that this difference may not be significant, as the researchers found the MDRD equation to be a more accurate predictor of GFR than the CG equation in older patients.14
A final limitation of this analysis lies within our method of selecting patients. While we were able to recruit more than a sufficient number of patients to power this analysis, we did not include those with an SCr level less than 1.3 mg/dL, with substantially decreased renal function, or those with an SCr level greater than 3 mg/dL, yet not on dialysis. Therefore, all possible patients with stages 3-5 chronic kidney disease were not captured.
|
Conclusions |
|---|
|
TopAbstractMethodsResultsDiscussionConclusionsReferences |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionConclusionsReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  P.D. Giles, P.B. Rylance, and D.C. Crothers New results from the Modification of Diet in Renal Disease study: the importance of clinical outcomes in test strategies for early chronic kidney disease QJM, February 1, 2008; 101(2): 155 - 158. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. A. Leopold Small-Molecule Glycoprotein IIb/IIIa Antagonists and Bleeding Risk in Women: Too Much of a Good Thing? Circulation, September 26, 2006; 114(13): 1344 - 1346. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
 |
 |
 |
 |
 |
 |
