Evaluation of Assumptions in the Clinical Use of the Cockcroft-Gault Equation
KT. Baron, AK. Birnbaum, JC. Cloyd, J. St. Peter, and RC. Brundage
Department of Experimental and Clinical Pharmacology, College of Pharmacy, University of Minnesota Twin Cities
Objectives: The Cockcroft-Gault (CG) equation is the most widely used method in US clinical settings for estimating creatinine clearance (CLCR)1. Clinicians commonly make adjustments to CG to improve estimation, such as using an alternate body size metric (e.g. ideal body weight) and rounding serum creatinine (Scr) up to 1 mg% in elderly subjects. A 15% reduction in CLCR is also assumed for female subjects. This study aims to evaluate the appropriateness of these adjustments.
Methods: A database of measured 24-hour CLCRs was assembled consisting of 748 measurements in 319 ambulatory subjects. The CG model was re-estimated on the current data using WinBUGS. Predictive performance measures2 for the original CG equation were calculated after plugging in actual (WGT), ideal (IBW), lean (LW), or dosing (DW=IBW+0.4*(WGT-IBW)) weight into the equation. The distribution of percent prediction error (%pe) conditioned on different covariates was summarized. Confidence intervals (alpha=0.05) for %pe were constructed using non-parametric bootstrap methods. Models were constructed to estimate a data-driven body size metric for the original CG equation.
Results: WGT-normalized 24-hour excretion rate (EXCR) was regressed on AGE with a proportional reduction for female subjects. The estimates revealed a slower decline in EXCR with AGE (-0.13 [-0.16,-0.11] mg/kg/24hour/year in males) compared with the CG estimate (-0.2 [-0.22,-0.17]). On average, females had 11% [7-15%] lower CLCR compared with males, conditional on the other published CG parameters. Overall, using WGT in CG gave the least biased estimates of CLCR (median %pe: -5.8 [-8.8,-3.3]). Using IBW, LW, or DW in CG resulted in persistent under-prediction that increased with WGT. For subjects greater than 100 kg, median %pe was +2% when using WGT in CG, but -19.4% with DW, and -35% when using IBW. Models estimating a modified dosing weight for use with CG always reduced to WGT as the body size metric. Plots of %pe versus AGE and Scr indicate that rounding Scr to 1 mg% in the elderly gives estimates that are more negatively biased than those obtained with actual Scr values.
Conclusions: A 15% reduction in CLCR for females is a reasonable assumption. Using WGT in CG resulted in the least-biased estimates of CLCR. Using IBW, LW, or DW in CG resulted in poorer rather than improved estimates. Rounding Scr up to 1 mg% in elderly subjects also diminished the predictive performance of the CG equation.
References:
[1] Cockcroft, D.W. and Gault, M.H. (1976). Prediction of creatinine clearance from serum creatinine. Nephron. 16(1): 31-41.
[2] Sheiner, L.B. and Beal, S.L. (1981). Some suggestions for measuring predictive performance. Journal of Pharmacokinetics and Biopharmaceutics. 9(4): 503-512.
