Estimating creatinine clearance: a meta-analysis

To determine which body weight descriptor most accurately predicts measured creatinine clearance (Cl(cr)) and whether rounding serum creatinine concentration (S(cr)) to 1 mg/dl when it is less than 1 mg/dl accurately predicts measured Cl(cr).

Meta-analysis of 13 English-language trials comparing 24-hour measured Cl(cr) with Cockcroft-Gault estimated Cl(cr) by using various body weights or rounded S(cr) values.

A total of 1197 patients (mean age 53.3 yrs, 48.7% male) were included in the meta-analysis.

A thorough literature search of the PubMed, MEDLINE, and the Cochrane Library databases was performed (1976-June 2010) to identify relevant clinical trials. Patient population, number of subjects, age, and 24-hour measured and estimated Cl(cr) were extracted independently by two investigators by using standardized data collection forms. Mean difference (MD) between estimated and measured Cl(cr) was assessed. Thirteen studies met all selection criteria. A random-effects model was applied secondary to a high level of heterogeneity among the studies (I(2)>50%). Total body weight in the Cockcroft-Gault equation overestimated measured Cl(cr) (MD 15.91 ml/min, 95% confidence interval [CI] 7.17-24.65 ml/min). Ideal body weight underestimated Cl(cr) (MD-5.15 ml/min, 95% CI -9.92 to -0.38 ml/min). No body weight (i.e., assumes body weight is 72 kg, thus removing the factor of 72 from the denominator of the equation) closely estimated Cl(cr) (MD 0.43 ml/min, 95% CI -5.42-6.27 ml/min). Adjusted body weight with correction factors of 0.3 or 0.4 also closely estimated Cl(cr) (MD 4.55 ml/min, 95% CI -11.41-20.50 ml/min, and MD 19.94 ml/min, 95% CI -9.6-49.49 ml/min, respectively). Total body weight with a rounded S(cr) value closely estimated measured Cl(cr) (MD 3.51 ml/min, 95% CI -17.18-24.20 ml/min). Ideal body weight with a rounded S(cr) value underestimated Cl(cr) (MD -29.45 ml/min, 95% CI -48.46 to -10.43).

Using the Cockcroft-Gault equation with no body weight and actual S(cr) value most closely estimated measured Cl(cr). In obese patients, it may be reasonable to use actual body weight with a correction factor of 0.3 or 0.4 and actual S(cr) value in the Cockcroft-Gault equation. Based on this analysis, the use of total body weight, ideal body weight, and a rounded S(cr) value cannot be recommended.