Objective: To compare specific antibiotic regimens, and monotherapy vs. combination therapy, for the empirical treatment of ventilator-associated pneumonia (VAP).
Design: Meta-analysis.
Data source: Medline, Embase, Cochrane register of controlled trials, study authors, and review articles.
Study selection: We included randomized controlled trials that evaluated empirical parenteral antibiotic regimens for adult patients with clinically suspected VAP.
Data selection: Two independent review groups searched the literature, extracted data, and evaluated trial quality. The primary outcome was all-cause mortality; secondary outcomes included treatment failure. Relative risks were pooled using a random effects model.
Results: We identified 41 trials randomizing 7,015 patients and comparing 29 unique regimens. Methodological quality was low, reflecting low rates of complete follow-up (43.9%), use of a double-blinded interventional strategy (14.6%), and randomization concealment (48.6%). Overall mortality was 20.3%; treatment failure occurred in 37.4% of patients who could be evaluated microbiologically. No mortality differences were observed between any of the regimens compared. Only one of three pooled comparisons yielded a significant difference for treatment failure: The combination of ceftazidime/aminoglycoside was inferior to meropenem (two trials, relative risk 0.70, 95% confidence interval 0.53-0.93). Rates of mortality and treatment failure for monotherapy compared with combination therapy were similar (11 trials, relative risk for mortality of monotherapy 0.94, confidence interval 0.76-1.16; and relative risk of treatment failure for monotherapy 0.88, confidence interval 0.72-1.07).
Conclusions: Monotherapy is not inferior to combination therapy in the empirical treatment of VAP. Available data neither identify a superior empirical regimen nor conclusively conclude that available regimens result in equivalent outcomes. Larger and more rigorous trials evaluating the choice of, and even need for, empirical therapy for VAP are needed.