Validation of a predictive risk score of aggressive PK/PD target non-attainment with continuous infusion piperacillin/tazobactam or meropenem in critically ill patients having documented Gram-negative infections

Objective: To validate a predictive risk score of early aggressive pharmacokinetic/pharmacodynamic (PK/PD) target non-attainment with continuous infusion (CI) piperacillin/tazobactam or meropenem in a retrospective cohort of critically ill patients having documented Gram-negative infections. Methods: Critically ill adult patients receiving treatment of documented Gram-negative infections with CI piperacillin-tazobactam or meropenem and undergoing first real-time beta-lactam therapeutic drug monitoring (TDM) instance within 72 h from starting standard dosing regimens were retrospectively included. A receiving operating characteristic (ROC) curve analysis was performed by using the proposed predictive score as the test variable and early aggressive PK/PD target non-attainment as the state variable. Area under the curve (AUC) and 95% confidence interval were calculated. The identified cut-off risk values were used for stratifying patients and assessing the impact on clinical/microbiological outcome in each sub-cohort of patients receiving targeted monotherapy. Results: Overall, 209 and 203 patients receiving CI piperacillin-tazobactam and meropenem were included, respectively. Early aggressive PK/PD target non-attainment was reported in 33 cases (15.8%) receiving piperacillin-tazobactam and in 8 (3.9%) of those treated with meropenem. A score threshold ≥ 2 points for piperacillin-tazobactam (AUC 0.81; 95% CI 0.75-0.86; p < 0.0001) and ≥ 3 points for patients treated with meropenem (AUC 0.96; 95% CI 0.93-0.99; p < 0.0001) were significantly associated with early aggressive PK/PD target non-attainment. Patients achieving the cut-off predictive score showed a significant higher microbiological failure rate in both piperacillin-tazobactam (56.3% vs. 28.6%, p = 0.044) and meropenem sub-cohorts (50.0% vs. 10.4%, p = 0.028). Conclusions: Our findings suggest that the proposed predictive cut-off risk score may represent a valuable tool for identifying promptly critically ill patients at high risk of early aggressive PK/PD target non-attainment with CI piperacillin-tazobactam or meropenem for whom a more intensified CI dosing regimens should be promptly applied bedside.