(1227) A Network Meta-Analysis to Determine Optimal Antibiotic Choices for Nosocomial Pneumonia

Introduction: Several novel antibiotics have gained approval for the treatment of nosocomial pneumonia (both hospital-acquired and ventilator-associated) in recent years, including cefiderocol, imipenem/cilastatin/relebactam, ceftazidime/avibactam, and ceftolozane/tazobactam. These newer antibiotics have not been directly compared to one another or to many other conventional agents; thus, uncertainty persists as to the optimal therapy. We performed a network meta-analysis of previously published randomized-controlled trials (RCTs) to elucidate which antibiotic is superior for gram negative (GN) and gram positive (GP) nosocomial pneumonia (NP).

Methods: Ovid Medline, Scopus, Embase, and Cochrane databases were searched to identify 31 RCTs in NP. Separate networks were made for GN and GP bacteria. A network meta-analysis was performed using frequentist random effect models to evaluate clinical efficacy, mortality, and adverse events.

Results: The GN network included 5,596 patients and 11 antibiotic treatments in 19 RCTs published between 1996 and 2021. A Q statistic of 11.07 (p = 0.27) and I2 of 18.7% indicated low heterogeneity. The GP network contained 3,817 patients and 7 antibiotics in 12 RCTs, and had a Q statistic of 5.71 (p = 0.456) with an I2 of 0%. In the GN network, meropenem was considered the reference treatment. Among all 11 GN treatments, only the combination of ceftazidime + aminoglycoside had an increased risk of clinical failure in a random effects model (OR 1.86, 95% confidence interval (CI) 1.14-3.02), and none had worse mortality. In the subgroup of 700 patients with P. aeruginosa, no single agent demonstrated superior efficacy. When compared to vancomycin, linezolid demonstrated a decreased risk of clinical failure (OR 0.73, 95% CI 0.59-0.91) and acute kidney injury (OR 0.44, 95% CI 0.29-0.65).

Conclusions: All antibiotics targeting GN bacteria (except the combination of ceftazidime + aminoglycoside) had equivalent clinical efficacy and mortality in NP, including for the treatment of P. aeruginosa. Linezolid showed superior efficacy and a lower risk of acute kidney injury in presumed GP NP compared to vancomycin. These results suggest that when GN NP is suspected, empiric administration of novel antibiotics is not superior to treating with conventional anti-pseudomonal agents.