Measurements of glomerular filtration in the intensive care unit are only a rough guide to renal function.

An estimate of glomerular filtration rate (GFR) is important in intensive care units (ICUs) to individualize drug dosages, and for the early detection of acute renal failure (ARF). Creatinine clearance or calculations based on plasma creatinine are the most widely used indicators of GFR, but may be very inaccurate in critically ill patients. We have assessed the accuracy of predictions of GFR by creatinine clearance and plasma creatinine in critically ill patients with variable creatinine kinetics. This is a retrospective record review set in a 24-bed multidisciplinary ICU. Eighteen patient records (9 normal renal function, 9 with ARF-250 patient days) were evaluated. Creatinine clearance measured daily over an 8-hour period, plasma creatinine, creatinine production and 8-hourly urine output were recorded for each day of ICU stay. The discrepancy between measured creatinine clearance and creatinine clearance predicted by the Cockcraft-Gault equation was determined. The coefficient of variation (CV) for each of the contributing variables and the final inherent error for the estimation of GFR were determined. The difference between measured and predicted creatinine clearance was large (predicted GFR may be from 60 ml/min above to 70 ml/min below measured creatinine clearance). The mean CV for creatinine production was 28% and for 8-hourly urine output it was 34%. The estimation of GFR from creatinine clearance can be made at +/- 52% of the calculated value, whereas the estimation of GFR by Cockcraft-Gault is accurate at +/- 37% of the calculated value. Unstable creatinine kinetics from variable urine output and creatinine production in critically ill patients make accurate estimation of GFR impossible. While prediction methods have fewer inherent sources of error, plasma creatinine may be influenced by factors other than renal function. The use of prediction equations is as inaccurate as measured clearance in unstable ICU patients.