Clinical Outcomes of Combination Antimicrobial Therapy

Abstract

Combination antimicrobial therapy -- using two or more antibiotics simultaneously -- is standard practice for empirical sepsis management, endocarditis, tuberculosis, and increasingly for multi-drug resistant (MDR) infections, yet the evidence supporting combination over monotherapy is surprisingly inconsistent. Proponents argue that combinations broaden spectrum coverage, provide synergistic killing, and suppress resistance emergence; critics counter that combinations increase toxicity, cost, and Clostridioides difficile risk without proven mortality benefit in most clinical scenarios. We conducted a multicentre retrospective analysis of 24,862 severe bacterial infection episodes treated with combination versus monotherapy at five European academic hospitals (Rome, Zurich, Paris, Stockholm, and Berlin) between 2016 and 2022, covering bloodstream infections, hospital-acquired pneumonia, complicated intra-abdominal infections, and MDR organism infections. We compared five combination strategies: empirical broad-spectrum combination, synergy-based combination, PK/PD-optimised combination, de-escalation-guided combination, and ML-predicted optimal regimen selection. Our Combination Therapy Outcome Score (CTOS) integrates 30-day mortality, clinical cure rate, microbiological eradication, adverse event burden, and resistance emergence prevention. ML-predicted regimen selection achieved the highest CTOS (0.922) by identifying patient-pathogen-drug combinations where synergy is biologically plausible, while de-escalation-guided combination (CTOS 0.908) balanced initial broad coverage with rapid narrowing to reduce toxicity. Empirical combination without protocolised de-escalation showed no mortality benefit over monotherapy for non-MDR infections (HR 0.96, 95% CI 0.88-1.06) while doubling C. difficile rates