Effect of Antibiotic Discontinuation Strategies on Mortality and Infectious Complications in Critically Ill Septic Patients: A Meta-Analysis and Trial Sequential Analysis*

Objective: To investigate methods of antibiotic duration minimization and their effect on mortality and infectious complications in critically ill patients. Data Sources: A systematic search of PubMed, Embase (via Ovid), clinicaltrials.gov, and the Cochrane Central Register of Controlled Trials (via Wiley) (CENTRAL, Issue 2, 2015). Study Selection: Randomized clinical trials comparing strategies to minimize antibiotic duration (days) for patients with infections or sepsis in intensive care. Data Extraction: A systematic review with meta-analyses and trial sequential analyses of randomized clinical trials. Dichotomous data are presented as relative risk (95% CIs) and p value, and continuous data are presented as mean difference (CI) and p value. Data Synthesis: We included 22 randomized clinical trials (6,046 patients). Strategies to minimize antibiotic use included procalcitonin (14 randomized clinical trials), clinical algorithms (two randomized clinical trials), and fixed-antibiotic duration (six randomized clinical trials). Procalcitonin (–1.23 [–1.61 to –0.85]; p < 0.001), but not clinical algorithm–guided antibiotic therapy (–7.41 [–18.18 to 3.37]; p = 0.18), was associated with shorter duration of antibiotic therapy. The intended reduction in antibiotic duration ranged from 3 to 7 days in fixed-duration antibiotic therapy randomized clinical trials. Neither procalcitonin-guided antibiotic treatment (0.91 [0.82–1.01]; p = 0.09), clinical algorithm–guided antibiotic treatment (0.67 [0.30–1.54]; p = 0.35), nor fixed-duration antibiotics (1.21 [0.90–1.63]; p = 0.20) were associated with reduction in mortality. Z -curve for trial sequential analyses of mortality associated with procalcitonin-guided therapy did not reach the trial sequential monitoring boundaries for benefit, harm, or futility (adjusted CI, 0.72–1.10). Trial sequential analyses for mortality associated with clinical algorithm and fixed-duration treatment accumulated less than 5% of the required information size. Despite shorter antibiotic duration, neither procalcitonin-guided therapy (0.93 [0.84–1.03]; p = 0.15) nor fixed-duration antibiotic therapy (1.06 [0.74–1.53]; p = 0.75) was associated with treatment failure. Conclusions: Although the duration of antibiotic therapy is reduced with procalcitonin-guided therapy or prespecified limited duration, meta-analysis and trial sequential analyses are inconclusive for mortality benefit. Data on clinical algorithms to guide antibiotic cessation are limited.