Predictive Performance of Estimated Creatinine Clearance in HIV-Infected Individuals

Abstract

Objective:

To evaluate the predictive performance of published creatinine clearance (Cl_cr) predictor equations in individuals infected with HIV.

Methods:

A retrospective review of medical records of patients with a 24-hour urine collection to determine Cl_cr was conducted. Cl_cr was estimated with each of six published equations. Estimated Cl_cr was compared with actual values, and bias and precision were calculated.

Results:

Cl_cr determined using the 24-hour urine collections were only moderately related to various patient factors such as weight and serum creatinine. The Cockcroft—Gault method significantly overestimated Cl_cr in patients with HIV (ANOVA; p = 0.0087) and demonstrated greatest bias (mean percentage prediction error, 52; 95% CI 18 to 87) and least precision (percentage root mean-squared error, 76; 95% CI −26 to 177) of the estimator methods.

Conclusions:

None of the six equations evaluated were precise predictors of Cl_cr in individuals with HIV. Using these equations, particularly the Cockcroft—Gault method, to predict renal function in patients with HIV could lead to overdosing renally eliminated medications in this population. Accurate assessment of renal function in these patients warrants a 24-hour urine collection or isotopic clearance measurement.