This study was performed and prepared according to the guidelines proposed by Cochrane Collaboration in the Cochrane Handbook for Systematic Reviews of Interventions (http://www.cochrane handbook.org) and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [9,10] (e-Figure 1).
Search strategy
We searched articles of all languages published from inception to March 27, 2022 in PubMed, Embase, Ovid, Medline and Cochrane Central Register of Controlled Trials by a combination of MESH terms and keywords (e-Appendix 1). Additional potentially relevant references were identified through manually forward and backward citation tracking of included papers.
Study selection criteria
Inclusion and exclusion criteria
Observational and interventional trials included a control group were eligible for review inclusion, if they compared the effect of prophylactic antibiotics administration with delayed/ clinically-driven/ even no administration of antibiotics in adult patients following OHCA. Trials reported only as abstracts were excluded. Case reports and case series were not eligible for inclusion.
Data extraction
Data were independently extracted by the first and the second authors. Extracted data consisted of the name of first author, year of publication, study design, participants, interventions, clinical parameters and adverse events. We resolved disagreements through discussions until a consensus was reached. We did not contact study authors for further data.
Outcome measurements and definitions
The primary outcome was the incidence of pneumonia (including early-onset pneumonia). Secondary outcomes included survival and survival with good neurological outcome, length of ICU and hospital stay, incidence of sepsis and bacteremia. Early-onset pneumonia was considered pneumonia occurs within the first 5 days when recruited.
Assessment of risk of bias
We used the Cochrane Collaboration tool to assess the risk of bias in randomized controlled trials (RCTs) [10, 11]. Domains containing random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of the outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), and other bias were assessed. The remaining observational trials were assessed using the ROBINS-I tool [12]. Domains include bias due to confounding, bias in selection of participants into the study, bias in classification of interventions, bias due to deviation from intend intervention, bias due to missing data, bias in measurement of outcomes, and bias in selection of the reported results. We rated each domain of the trials as low risk, unclear, or high risk. Trials were considered low risk when each independent domain was rated as low risk. Any domain rated as unclear or high risk increased the overall risk score.
Statistical analysis
Data were analyzed using Statistics/Data Analysis 15.1. The results of dichotomous data were presented as forest plots through the odds ratios (ORs) with 95% confidence intervals (CIs). Forest plots using Weighted Mean Difference (WMD) with 95% CI were performed for the assessment of continuous data. We used random (M-H heterogeneity) model to assess the effects since clinical heterogeneity including study setting, study design, intervention was high. A Eegger test was conducted to assess the publication bias. A p value ≤ 0.05 was considered statistically significant, except when otherwise specified.
Assessment of the certainty of the evidence
Grading of Recommendations Assessment, Development and Evaluate system (GRADE) was used to evaluating the quality of evidence, based on the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness and publication bias) [13]. The first and the second authors independently rated the quality of evidence for the outcomes. The level of evidence was classified into four categories as high, moderate, low or very low. We resolved disagreements through discussions until a consensus was reached.