Are Proton Pump Inhibitors Associated With the Development of Community-Acquired Pneumonia?

A Meta-analysis

Christopher Giuliano; Sheila M Wilhelm; Pramodini B Kale-Pradhan


Expert Rev Clin Pharmacol. 2012;5(3):337-344. 

In This Article


The meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Guidelines.[22]

Data Sources & Searches

Our aim was to identify all relevant clinical studies that reported an association of CAP with the use of PPIs. To identify all applicable publications, a systematic literature search of PubMed and Ovid Medline was conducted by two investigators independently using the following medical subject headings and keywords: PPI, pneumonia, CAP, anti-ulcer agent, antacids, omeprazole, esomeprazole, lansoprazole, pantoprazole and rabeprazole. Databases were searched from 1988 to present (November 2011), as the first PPI was approved in 1990. Bibliographies of recent review articles and systematic reviews were hand-searched to identify any additional studies.

Study Selection

Abstracts of full-text articles were independently screened for quality and inclusion by two reviewers. This meta-analysis only included case–controlled and cohort studies, which were published in full in English and evaluated PPI use and CAP incidence in human subjects. Studies of pediatric patients, critically ill patients or Helicobacter pylori treatment were excluded. In addition, studies were excluded if the reported data were not evaluable.

Data Extraction & Quality Assessment

Two of the investigators independently extracted data from all eligible studies by using a standardized form. Accuracy of data was confirmed by a third investigator. For included studies, study design, sample size, study duration and patient demographics were collected. Data were also gathered on study setting, patient population, acid-suppressive therapy (AST) used and occurrence of CAP. The assessment of the quality of observational studies including case–control and cohort studies was done using the Newcastle–Ottawa Quality Assessment Scale (NOQAS).[101] Any discrepancy that developed regarding the inclusion, quality of a publication or interpretation of data was resolved through consensus.

Data Synthesis

The primary outcome of this meta-analysis was the association of CAP with current PPI use. We used the published adjusted odds ratio (ORs) and corresponding 95% CIs from each included study, which were log transformed and analyzed using Comprehensive Meta analysis® (Ver 2.0). We report binary outcomes as ORs. Summary effects estimates are presented with 95% CIs. For all analyses, a p-value of 0.05 or less (two-sided) was considered as significant. We assumed a class effect and pooled data for all PPIs. Data was analyzed based on our primary outcome of current use of PPIs as well as prespecified subgroup analyses including duration of PPI use (<30 or >180 days) and PPI dose (high vs low, defined as >1 defined daily dose vs ≤1 defined daily dose).[23] Tests for interactions between PPI dosage and duration subgroups were preformed. Measures were also taken to assess the individual quality of studies included in this review, as well as adjusting for heterogeneity among studies. The I2 statistic describes the percentage of the variability in effect estimates that is due to heterogeneity rather than sampling error (chance). The Cochrane handbook suggests that an I2 value greater than 50% indicates significant heterogeneity.[102] As statistical heterogeneity was expected, we used a random-effects model. In addition, a sensitivity analysis was performed based on parameters defined a priori. These parameters included a NOQAS score <6 or studies that had included non-population-based sample. A funnel plot was used to evaluate whether there was publication bias.