LONDON – The use of acid-suppression therapy was associated with a significantly increased risk for Clostridium difficile infection in a systematic review and meta-analysis of data from 30 studies.
Previous observational studies have suggested a link between proton pump inhibitor (PPI) use and an increased risk of developing Clostridium difficile infection (CDI). In February, the Food and Drug Administration issued an alert about the use of PPIs because of the increased CDI risk.
This new meta-analysis supports this association, with a significantly increased risk of CDI with acid suppression (odds ratio, 1.58; P less than .05), Dr. Iman Tleyjeh said at the European Congress of Clinical Microbiology and Infectious Diseases.
"We found a robust association between acid-suppression therapy and the appearance of CDI," said Dr. Tleyjeh, director of the research and scientific publication center and a consultant in infectious disease at King Fahad Medical City, Riyadh, Saudi Arabia.
"Our findings have major global implications, particularly that most of the use of acid-suppression therapy [is] unjustified based on available evidence," he added.
The analysis included studies published through January 2012. Inclusion criteria were that each study had to be an analytical study reporting the effect of acid suppression on CDI incidence, and had to be adjusted for confounders such as antibiotic use.
A total of 30 studies from the United States, United Kingdom, Canada, and Israel met the criteria. There were no randomized clinical trials. Twenty-one were cohort studies, and nine were case-control studies, with a total of 52 effect estimates. All of the included studies were of very good quality, said Dr. Tleyjeh, who is also with the Mayo Clinic, Rochester, Minn.
Separate analyses were done for the different types of studies, and the investigators addressed statistical issues including heterogeneity, publication bias, and residual confounding. Twenty-four studies were from single centers, five were multicenter, two were from general practice research databases, and one was from a general practice clinic. Five studies exclusively addressed community-acquired CDI, 24 addressed hospital-acquired CDI, 1 covered both community- and hospital-acquired CDI, and 1 examined CDI rates in a long-term care facility.
A random-effect meta-analysis revealed that use of acid-suppression therapy (PPIs or H2 blockers) was associated with a significantly increased risk for CDI, with an odds ratio of 1.58 (95% confidence interval [CI], 1.38-1.80). The pooled proportion of CDI patients who were exposed to antibiotics was 0.70 (95% CI, 0.64-0.75). Thus, on average, 30% of CDI patients were not exposed to antibiotics, based on pharmacy records, discharge summaries, or other data sources, Dr. Tleyjeh noted.
Each of the separate analyses showed a significant association (P less than 0.05) between acid suppression and CDI, with pooled effect estimates of 1.67 for studies investigating PPIs, 1.44 for those looking at H2 blockers, 1.46 for the cohort studies, and 1.63 for the case-control studies.
Analysis of the heterogeneity between studies revealed that the main driver was an overall lower pooled effect size for the Canadian studies compared with those from the other three countries. Among the Canadian studies, the odds ratio for the effect of acid suppression on CDI was just 1.08, compared with 1.79 for the U.S. studies, 1.91 for the U.K. studies, and 2.52 for the Israeli studies.
The "meta-regression analysis" technique was used to further explore the heterogeneity between studies with regard to study design, study location, journal impact factor, method of ascertainment of antibiotic use, and proportion with antibiotic exposure.
That analysis revealed that the country where the study was performed was a significant independent variable, whereas patient’s sex had a small independent impact. A comparison of different country-specific study characteristics revealed that CDI cases in Canadian studies had a higher exposure to antibiotics than did those in the three other countries, 90% vs. 72%.
Further analysis also revealed that potential publication bias, the tendency for negative studies to remain unpublished, and adjustment for that factor using a novel regression-based method resulted in an adjusted average odds ratio of 1.38 (95% CI, 1.21-1.58). "So, there is resistant association even if we assume there is [publication] bias and we adjust for it," Dr. Tleyjeh said.
Because the included studies were all observational, a final sensitivity analysis was done to assess the possible effect of unknown and unmeasured confounders. This revealed that an unmeasured confounder would have to be severely imbalanced between the acid-suppression users and nonusers, by an odds ratio of 10, or would have to increase the risk of CDI by at least twofold to account for this association. Such a confounder would have to be even stronger than antibiotic use, he commented.