Making the case for precision dosing: visualizing the variability of cefepime exposures in critically ill adults

To investigate and describe the variability in cefepime exposures among 'real-world', critically ill patients by using population pharmacokinetic modelling and simulations, and with translation of these findings to visualizations.

A cohort of adult medical ICU patients who received cefepime with therapeutic drug monitoring was studied. Two compartment models were developed to estimate cefepime clearance (Model 1) and simulate cefepime exposures among 1000 patients, each with identical creatinine clearance of 60 mL/min and receiving a regimen of cefepime 1 gram IV over 30 minutes, every 8 hours (Model 2). Variability in the relationship between cefepime clearance and creatinine clearance (CrCL) was visualized, and a random, representative sample of 10 simulated patients was utilized to illustrate variability in cefepime exposures.

A total of 75 adult medical ICU patients (52% female) and 98 serum cefepime samples were included in the study. Population parameter estimates for cefepime displayed a wide range of variation in Model 1 (CV: 45% to 95%), with low bias at the individual level at 0.226 mg/L but high bias in the population model 10.6 mg/L. Model 2 displayed similar fits, demonstrating that correcting for individual patient creatinine clearance slightly improves the bias of the population model (bias = 4.31 mg/L). Among 10 simulated patients that a clinician would deem similar from a dosing perspective (i.e. equivalent creatinine clearance), maximum concentrations after three simulated doses varied more than 8-fold from 41.2 to 339 mg/L at the 5th and 95th percentiles, and clearance profiles were highly different.

Creatinine clearance estimates alone are inadequate for predicting cefepime exposures. Wide variations in cefepime exposure exist among ICU patients, even for those with similar kidney function estimates. Current population adjustment schemes based solely on creatinine clearance will result in unintended high and low exposures leading to safety and efficacy concerns, respectively.