Comparison of the Modification of Diet in Renal Disease and Cockcroft-Gault Equations for Dosing Antimicrobials

Study Objectives: To determine the concordance between the Modification of Diet in Renal Disease (MDRD) and Cockcroft-Gault equations for glomerular filtration rate (GFR) estimation, the impact of using each equation on antimicrobial dosing, the difference in estimated GFR in patients with acute kidney disease (AKD) versus chronic kidney disease (CKD), and the correlation between the MDRD, Cockcroft-Gault equation, and expert medical opinion for estimating GFR in patients with AKD.

Design: Retrospective cohort study.

Setting: A 689-bed academic medical center.

Patients: A total of 372 adults hospitalized with either AKD or CKD between January 1, 2007, and May 31, 2007, and who received at least one antimicrobial drug; patients with stage 1 or 2 CKD or those receiving dialysis were excluded.

Measurements And Main Results: Data were collected from electronic medical records on patient characteristics, laboratory values, antimicrobial drugs requiring dosage adjustment due to renal dysfunction, and estimated GFRs provided by the laboratory (MDRD estimation). In addition, estimated GFRs were calculated using the Cockcroft-Gault equation. For patients with AKD, a third GFR was estimated by a nephrologist. For all patients, the MDRD GFR was significantly higher than the Cockcroft-Gault GFR (p<0.001). Level of concordance for the need for dosage adjustment based on the two equations was moderate (kappa coefficient 0.57, 95% confidence interval 0.5-0.63); 99.1% of patients with discordant dose recommendations would receive a higher dose if the MDRD GFR was used. In the AKD versus CKD groups, mean MDRD GFR was significantly higher than the Cockcroft-Gault GFR in both groups (p<0.0001), but the difference was significantly greater in the CKD group (p<0.0001). In patients with AKD, the GFR estimated by expert opinion was greater than that estimated by the Cockcroft-Gault equation (p=0.04), but was similar to the MDRD equation (p=0.07).

Conclusion: The estimated GFR obtained with the MDRD equation was consistently higher than that from the Cockcroft-Gault equation in patients with AKD or CKD. In patients with AKD, the MDRD GFR more closely correlated with expert opinion than the Cockcroft-Gault, suggesting that the MDRD method may be applicable to this patient population. Moderate concordance between the two equations for the need for antimicrobial dosage adjustment due to renal dysfunction was found, but the specific dosage change was different for approximately 40% of patients, with 99% receiving higher doses when the MDRD GFR is used. These dosing differences may be significant, depending on drug safety profile, type of infection, and causative pathogen.