Combination antimicrobial treatment versus monotherapy: the contribution of meta-analyses.

Systematic reviews and meta-analyses have put into perspective the clinical implications of in vitro synergy (Box 1). Randomized, controlled trials are the cornerstone of evidence-based medicine. The trials included in the meta-analyses described in this article are the building blocks of evidence. Individual trials, however, were individually underpowered to address the broader clinical question and relevant patient-related outcomes. On the question of combination therapy, meta-analyses have shaped the complete picture. The interactions observed in vitro have not been shown to improve patient-related outcomes. Authors of systematic reviews have the privilege of considering and selecting the clinical outcomes most relevant for the individual patient. Thus, all-cause mortality, rather than treatment failure with antibiotic modifications or infection-related mortality, has been selected for the assessment of patients who had severe gram-negative infections and febrile neutropenia. Mortality and relapse were assessed for patients who had endocarditis, and clinical and lung function scores were assessed for patients who had cystic fibrosis. The authors hope that the dissemination of these reviews will lead clinicians and researchers to consider primarily these outcomes when appraising or designing clinical research. These are the outcomes that clinicians target when treating the patient. Systematic reviews have the virtue of a broad, systematic, and explicit search. In some areas, such as the use of combination therapy to treat gram-positive infections in general, and specifically to treat endocarditis and Pseudomonas aeruginosa bacteremia, the main contribution of the reviews was to show that current practice is based on very limited clinical evidence. This finding does not refute current practice but should serve to guide future trials and opens the possibility for a different choice of therapy when standard guidelines are difficult to implement. The fact that to date no evidence has been accrued for these infections is not surprising. The clinical question of combination therapy is of no major interest to pharmaceutical companies sponsoring most trials; the infections are rare; and the study design is complex. This gap in knowledge calls for a new trial paradigm: collaborative investigator-initiated, multicenter trials. When randomized, controlled trials are unfeasible, the use of novel methods for adjustments in observational studies, such as propensity analyses using large databases, might approximate the true effect of combination therapy in a wider patient population.