Temporal effects of errors in estimating kidney function after critical illness on drug dosing categories

Aims This study aimed to clarify how the trajectories of the Cockcroft–Gault (CG), Modification of Diet in Renal Disease (MDRD) and Chronic Kidney Disease Epidemiology Collaboration (CKD‐EPI) equations differ from measured creatinine clearance (Clcr) and to determine their impact on drug dosing categories in critically ill patients. Methods We used data from 5355 days following critical illness to estimate the trajectories of the above‐mentioned kidney function equations using a linear mixed model. In addition, we determined whether the CG, MDRD and CKD‐EPI equations would change drug dosing categories (<15, 15–29, 30–59, 60–89, 90–119, 120–149, or ≥150 mL/min) over time compared with that observed using measured Clcr. Results The rate of change relative to baseline in measured Clcr remained almost unchanged during the 28 days of observation (−0.02 [95% confidence interval, CI: −0.19–0.14] mL/min/1.73 m 2 /day). In contrast, all indirect kidney function estimating equations showed a time‐dependent increase in the rate of change, which was largest for the CG equation, followed by the MDRD and CKD‐EPI equations (+2.00 [95% CI: 1.89–2.12], +1.50 [95% CI: 1.34–1.67] and +0.77 [95% CI: 0.61–0.93] mL/min/1.73 m 2 /day, respectively). More than half of the CG, MDRD and CKD‐EPI equations showed discrepancies with the measured Clcr and drug dosing categories, with underestimation in the early phases and overestimation in the later phases being more common. Conclusion Creatinine‐based indirect kidney function estimation equations increased over time, indicating erroneous drug dosing categories on over half of the critical care days.