Predictability of creatinine clearance estimates in critically ill patients

Abstract

a) To evaluate the predictive ability of different creatinine clearance methods as compared with the criterion standard, inulin clearance; and b) to determine which of the predictive methods yields the most accurate estimation of creatinine clearance. Prospective study. Medical intensive care unit (ICU) of a university-affiliated tertiary care hospital. Glomerular filtration rate was measured by the criterion standard, inulin clearance. Twenty mechanically ventilated adults. Renal function was assessed by the following procedures: inulin clearance using a standard protocol, 30-min creatinine clearance, 24-hr creatinine clearance, and creatinine clearance estimates by the Cockcroft-Gault equation. Ideal body weight, total body weight or lean body mass with actual serum creatinine or serum creatinine concentration corrected to 1 mg/dL (85 mumol/L) in cachectic patients were sequentially incorporated into the Cockcroft-Gault equation. The Cockcroft-Gault equation, using ideal body weight and the corrected serum creatinine concentration, was the best predictor of inulin clearance with the smallest bias (9.7 +/- 8.6, 95% confidence interval 5.7 to 13.8). The bias encountered with the 30-min creatinine clearance was not different from that value with the 24-hr creatinine clearance (21.6 +/- 33.0, 95% confidence interval 6.2 to 37.1 vs. 25.4 +/- 28.3, 95% confidence interval 11.8 to 42.9). Good correlations existed between inulin clearance and the Cockcroft-Gault equation, using ideal body weight and the corrected serum creatinine concentration (r2 = .81; p = .0001), as well as between inulin clearance and the Cockcroft-Gault equation, using the lower of ideal or total body weight and the higher of the actual serum creatinine concentration or corrected serum creatinine (r2 = .75; p = .0001). The 30-min creatinine clearance and the 24-hr creatinine clearance had poorer agreement with inulin clearance. The incorporation of a corrected serum creatinine value into the Cockcroft-Gault equation consistently led to better predictions and higher correlation coefficients. The utilization of the Cockcroft-Gault equation as used clinically (the lower of ideal or total body weight and the higher of actual serum creatinine or corrected serum creatinine concentration to 1 mg/dL [85 mumol/L]) results in more accurate predictions of glomerular filtration rate in the medical, critically ill patient than urine creatinine clearance measures. If creatinine clearance measures are used, the 30-min collection provided results not different from those results obtained with 24-hr urinary collections.