Pitfalls of using estimations of glomerular filtration rate in an intensive care population

Background: Accurate knowledge of the glomerular filtration rate (GFR) is imperative in the intensive care unit (ICU) as renal status is important for medical decisions, including drug dosing.

Aims: Recently, an estimation of GFR (eGFR) was suggested as a method of estimating GFR. How well this formula predicts GFR in unwell patients with normal initial serum creatinine concentrations has not been examined.

Methods: The accuracy of the eGFR (before and after adjustment for actual body surface area (BSA)) was compared with measured and with estimated creatinine clearance using the Cockcroft Gault (CG) formula adjusted for total and lean body weight.

Results: A total of 237 observations was recorded in 47 subjects. These were initially analysed independently, and then using the first observation only. Overall the mean difference between measured creatinine clearance and eGFR was -12 mL/min (95% confidence interval (CI) -20 to -3), between measured creatinine clearance and CG +17 mL/min (95% CI 9-24), between measured creatinine clearance and CG adjusted for ideal body weight +12 mL/min (95% CI 4-21) and between measured creatinine clearance and eGFR 'unadjusted' for BSA 5 mL/min (95% CI -2-13).

Conclusions: Using either eGFR or CG formulae to estimate renal function in ICU subjects with normal serum creatinine concentrations is inaccurate. Although correcting for BSA improves the eGFR, this requirement to measure height and weight removes a major attraction for its use. We suggest that eGFR should not be automatically calculated in the ICU setting.