In US veterans, PPIs were linked to increased all-cause and some cause-specific mortality compared with H2-blockers

The observational study cannot establish causation, and the only way to prove that proton-pump inhibitors (PPIs) cause the various adverse outcomes attributed to them is to compare them with placebo or no treatment in a randomized trial, according to an ACP Journal Club commentary.

Compared with histamine H2-receptor antagonists (H2-blockers), proton-pump inhibitors (PPIs) were associated with increased mortality risk at 10 years, according to a longitudinal observational cohort study of U.S. veterans. All-cause mortality was 39% in veterans taking PPIs and 34% in those taking H2-blockers (hazard ratio, 1.17; 95% CI, 1.10 to 1.24). There was also an excess of cause-specific mortality associated with taking PPIs for cardiovascular diseases, upper gastrointestinal cancer, and chronic kidney disease (number of attributable deaths per 1,000 PPI users, 22.91, 3.12, and 4.74, respectively).

The study was published online on May 30 by The BMJ. The following commentary by Ronald L. Koretz, MD, FACP, appeared in the ACP Journal Club section of the Sept. 17 Annals of Internal Medicine.

It is generally stated that observational studies cannot establish causation because of confounding. With the availability of large data sets (big data), sophisticated matching methodology has been proposed to overcome this problem. Using these techniques, Xie and colleagues found an association between PPI use and mortality in a population of US veterans. Although the investigators don't assert causation, it is inferred in their discussion (e.g., “the findings emphasize the need to promote awareness of potential adverse events of long term PPI use”). The analyses could not be corrected for exact reasons for ordering PPIs because these reasons were not known; also, > 50% of patients in the PPI group and about two thirds in the H2-blocker group used the drugs for reasons other than a documented gastrointestinal indication for acid suppression. Further, because the comparator was use of H2-blockers, an alternate interpretation may be that H2-blockers are associated with lower mortality.

Big data will magnify small differences regardless of whether they are true or irrelevant. The only way to prove (and not hypothesize) that PPIs cause the various adverse outcomes attributed to them is to show such effects in randomized trials comparing PPIs with placebo or no treatment. To date, such efforts have not shown that PPIs increase cause-specific or all-cause mortality. A recent large trial confirmed this, although it had only 3 years of follow-up. As for me, I have significant gastroesophageal reflux symptoms and will continue to take the PPIs that have substantially reduced those symptoms.