Integrated pharmacokinetic-pharmacodynamic modeling to evaluate empiric carbapenem therapy in bloodstream infections Tze-Peng Lim,1,2 Reyna Wang,1 Gang Quan Poh,3 Tse-Hsien Koh,4 Thean-Yen Tan,5 Winnie Lee,1 Jocelyn Qi-Min Teo,1 Yiying Cai,1 Thuan-Tong Tan,6 Pui Lai Rachel Ee,3 Andrea L Kwa1,3,7 1Department of Pharmacy, Singapore General Hospital, Singapore, Singapore; 2SingHealth Duke-NUS Medicine Academic Clinical Programme, Singapore, Singapore; 3Department of Pharmacy, Faculty of Science, National University of Singapore, Singapore, Singapore; 4Department of Microbiology, Singapore General Hospital, Singapore, Singapore; 5Department of Laboratory Medicine, Changi General Hospital, Singapore, Singapore; 6Department of Infectious Diseases, Singapore General Hospital, Singapore, Singapore; 7Emerging Infectious Diseases, Duke-NUS Medical School, Singapore, Singapore Objectives: Treatment for nosocomial bloodstream infections (BSI) caused by multidrug-resistant (MDR) Gram-negative bacteria (GNB) is challenging. Rising antimicrobial resistance, especially in extended spectrum beta-lactamase production, inadvertently increases empiric carbapenem consumption. Three antipseudomonal carbapenems (imipenem, meropenem [MER], and doripenem [DOR]) are available commercially against MDR GNB in Singapore. The study aims to determine the most optimal empiric carbapenem dosing regimens (CDR) and evaluate their cost-effectiveness for GNB-BSI in the face of increasing MDR GNB. Methods: Carbapenem minimum inhibitory concentrations (MICs) were generated for non-repeat GNB-BSI obtained in 2013–2014 from two hospitals. Monte Carlo simulations were used to assess the cumulative fraction of response (CFR) of various CDRs using the percentage of time above MIC for 40% (%T > MIC of 40%) as the pharmacokinetic (PK)–pharmacodynamic (PD) parameter for efficacy. Carbapenem costs were based on patient antibiotic costs. Antibiotic cost-effectiveness was calculated as total daily drug cost/CFR. Results: A total of 1,140 bloodstream isolates were collected. They comprised 116 Acinetobacter baumannii, 237 Pseudomonas aeruginosa, and 787 Enterobacteriaceae. All CDRs achieved ~40, ~80, and ≥90% CFRs against A. baumannii, P. aeruginosa, and Enterobacteriaceae, respectively. Against P. aeruginosa, MER 2 g every 8 h infused over 3 h and DOR 1 g every 8 h infused over 4 h achieved CFRs 84 and 81%, respectively. Against Enterobacteriaceae, the cost of MER 2 g every 8 h infused over 3 h was the lowest among the three carbapenems at $0.40/percentage of CFR. Conclusion: This study demonstrates the utility of PK–PD modeling to formulate the optimal selection of a cost-effective empiric CDR in antibiotics guidelines and formulary inclusion. The findings support the selection of high MER doses of prolonged infusions as empiric coverage for GNB-BSI in our institutions. Keywords: empiric carbapenem regimens, multidrug resistant, Gram-negative bacteria, bloodstream infections