A network meta-analysis of antibiotics for treatment of hospitalised patients with suspected or proven meticillin-resistant Staphylococcus aureus infection.

Infections due to meticillin-resistant Staphylococcus aureus (MRSA) pose a serious health risk. Novel methods for assessing comparative effectiveness and safety may provide valuable insights into therapeutic choices. We did a systematic review searching electronic databases including the archives of FDA/CDER and performed a Bayesian network meta-analysis to compare parenteral antibiotics used for treating hospitalised adults with complicated skin and soft-tissue infections (cSSTIs) or hospital-acquired or ventilator-associated pneumonia (HAP/VAP). Models were adjusted for clinical heterogeneity due to between-arm differences in the proportion of patients with diabetes (for cSSTI) and in those requiring mechanical ventilation (for pneumonia). Treatments were ranked on efficacy, defined as clinical success in the modified intention-to-treat population (MITT) and in the MITT population with MRSA at baseline (MRSA m-MITT), on all-cause mortality (in pneumonia only), and on serious adverse events and withdrawals due to adverse events. We identified 24 randomised controlled trials (17 for cSSTI and 10 for HAP/VAP) comparing one of six antibiotics with vancomycin. The network meta-analysis indicated that vancomycin ranked third (of six antibiotics) in cSSTI and second (of four) in pneumonia on both efficacy and safety. However, direct pairwise meta-analyses remained inconclusive as evidenced by the adjusted median odds ratios (ORs) and their 95% credible intervals. In cSSTI, linezolid and ceftaroline were non-significantly more effective than vancomycin. Linezolid ORs were 1.15 (0.74-1.71) and 1.01 (0.42-2.14) and ceftaroline ORs were 1.12 (0.78-1.64) and 1.59 (0.68-3.74) in the MITT and MRSA m-MITT populations, respectively. For HAP/VAP, linezolid was non-significantly better than vancomycin, with ORs of 1.05 (0.72-1.57) and 1.32 (0.71-2.48) in the MITT and MRSA m-MITT populations, respectively. We suspect performance and detection bias in cSSTI trials involving linezolid, but regression methods could not adjust for this potential bias. In these clinical trials, the preferred agents for treating serious MRSA infections were ceftaroline (for cSSTI, not studied in HAP/VAP) and linezolid. However, translation of these findings into practice should consider the small size of the evidence networks and the consequent uncertainty associated with the parameter estimates, the lack of evidence for ceftaroline in patients with severe renal impairment, and the lower internal validity of some of the linezolid trials.