Ventilator care bundles in adult ICUs: a systematic review and meta-analysis of ventilator-associated pneumonia risk and incidence density

Background Ventilator care bundles are used to prevent ventilator-associated pneumonia (VAP), but differences in outcome denominators and study design complicate interpretation. We examined patient-level VAP risk and ventilator-time incidence density separately. Methods PubMed/MEDLINE, Embase, CENTRAL, and Web of Science were searched from inception on 11 August 2026. Eligible bundles contained at least three clinical practices for adults receiving invasive mechanical ventilation in ICUs. Nonconcurrent before-after comparisons were pooled separately for VAP risk ratios (RRs) and incidence rate ratios (IRRs). Comparisons with unclear timing, randomized results, and segmented interrupted time-series results were presented separately. Random-effects models used restricted maximum likelihood and Knapp–Hartung intervals, with modified Knapp–Hartung sensitivity analyses. ROBINS-I and GRADE assessed bias and certainty. Results Twenty-three non-randomized study families contributed to the primary meta-analyses; 22 contributed VAP data. Bundle implementation was associated with lower VAP risk (14 studies; n = 5,905; RR 0.62, 95% CI 0.51–0.76; I 2 = 9.0%; prediction interval 0.47–0.82). The pooled incidence rate ratio was 0.55 (12 studies; 95% CI 0.41–0.73; I 2 = 79.9%), but the prediction interval included the null (0.25–1.22). ICU length of stay (3 studies; MD −2.53 days, 95% CI −7.59 to 2.53) and mechanical ventilation duration (2 studies; MD −3.64 days, 95% CI −13.96 to 6.69) remained uncertain. All pooled estimates were at serious risk of bias. Certainty was low for VAP risk and very low for the other pooled outcomes. Conclusions Nonconcurrent comparisons associated bundle implementation with lower reported VAP risk, with low certainty. Incidence-density associations varied across settings, and evidence for ICU length of stay and ventilation duration was very uncertain. Evaluations with independent concurrent controls, stable surveillance, and adherence monitoring are needed. Systematic Review Registration PROSPERO (CRD420261484358).