Population pharmacokinetic modeling of piperacillin in critically ill adult patients: consideration of sex-based differences during model development

Piperacillin population pharmacokinetic models reportedly perform poorly for critically ill females compared to males. We aimed to explore potential methods that may better adjust for female data during model development. Total piperacillin concentrations were used from a prospective observational study in NONMEM v7.5.1. Two models were developed following different approaches: classic stepwise approach and sex-specific approach. Relationship between covariates and estimated parameters were explored by statistically and graphically assessing their performance on males and females separately. Dosing regimen simulations were also performed separately by sex. A one-compartment model based on data from 70 critically ill patients (49/21 males/females) with 233 concentrations best fit the data with both approaches. Creatinine clearance was the most significant covariate for the classic approach model, while creatinine clearance was best for male patients and estimated glomerular filtration rate was best for female patients with the sex-specific approach. Dosing recommendations were different between male and female patients with the sex-specific model. This study is the first to consider sex-specific covariates during the modeling process for piperacillin in critically ill patients. This approach may help reduce differences in model predictions between males and females in model-informed precision dosing strategies.