RESUMO
OBJECTIVE: We aimed to summarize the current evidence regarding the risk of pneumonia associated with proton pump inhibitors (PPI) treatment. METHODS: We searched PubMed, Embase, and CENTRAL from the 1970 through December 2017. We included both randomized controlled trials (RCTs) and observational studies. We used random-effect model to calculate the summary effect estimates and quantified the heterogeneity by I2 statistics. RESULTS: A total of 7,643,982 patients from 10 RCTs and 48 observational studies were included in this meta-analysis. The primary meta-analysis demonstrated PPIs use was significantly associated with increased risk of pneumonia, but the heterogeneity was high (odds ratio [OR], 1.43; 95% confidence interval [CI], 1.30-1.57; I2, 95.4%). The sensitivity analysis indicated PPIs were not statistically associated with increased risk of pneumonia among patients concomitantly taking nonsteroidal anti-inflammatory drugs (OR, 1.11; 95% CI, 0.94-1.31; I2, 5.8%). The funnel plot demonstrated significant publication bias, especially for observational studies. CONCLUSIONS: The presence of significant between-study heterogeneity and publication bias raised concerns regarding the validity of the primary meta-analytic result. Protopathic bias, or reverse causality, may cause overestimated association. Studies that adopted a design to account for protopathic bias did not show a significant association between PPI use and risk of pneumonia.