17,598 people were randomised to a PPI or placebo for three years. One harm signal emerged

Proton pump inhibitors have been linked, in observational studies, to fractures, dementia, chronic kidney disease, pneumonia and a long list of other harms. When the question was put to an adequately powered randomised trial — 17,598 participants, a median of 3.01 years of follow-up — the only outcome that differed significantly between the drug and placebo groups was enteric infections [s1].

That is one finding out of the long list of adverse-event categories the investigators collected, and it is a materially different picture from the one that reached the public.

The trial design

The analysis was built on a 3 × 2 partial factorial double-blind trial in participants with stable cardiovascular disease and peripheral artery disease [s1]. Participants were randomly assigned to pantoprazole 40 mg daily (8,791 people) or placebo (8,807 people), and independently randomised to rivaroxaban 2.5 mg twice daily with aspirin 100 mg once daily, rivaroxaban 5 mg twice daily, or aspirin 100 mg alone [s1]. It is registered as NCT01776424 [s1].

Data were collected every six months on the development of pneumonia, Clostridium difficile infection, other enteric infections, fractures, gastric atrophy, chronic kidney disease, diabetes, chronic obstructive lung disease, dementia, cardiovascular disease, cancer, hospitalisations and all-cause mortality [s1]. Follow-up ran to a median of 3.01 years, accumulating 53,152 patient-years [s1].

The design is what makes this informative. Randomisation removes the problem that has always dogged PPI safety research: the people prescribed these drugs long-term differ systematically from those who are not, in ways that also predict fractures, kidney disease and dementia.

The result

There was no statistically significant difference between the pantoprazole and placebo groups on any safety outcome except enteric infections, which occurred in 1.4% of the pantoprazole group against 1.0% of the placebo group — an odds ratio of 1.33 (95% CI 1.01 to 1.75) [s1].

That result is biologically coherent. Stomach acid is a barrier to swallowed organisms, and suppressing it should plausibly let more of them through. It is also the one finding here that the authors treat as real.

C. difficile infection was approximately twice as common in the pantoprazole group as in the placebo group — but there were only 13 events in total, so the difference was not statistically significant [s1]. That is an important number to state precisely: a doubling built on 13 events is not evidence of a doubling, and it is not evidence of no effect either. The trial was simply too small on that specific outcome to say.

The authors' conclusion is that pantoprazole is not associated with any adverse event when used for three years, with the possible exception of an increased risk of enteric infections [s1].

What the trial does not rule out

Three years is not a lifetime. Many people take PPIs for far longer, and this trial cannot speak to what happens at ten or twenty years. Outcomes with long latency — cancer being the obvious one — are precisely the ones a three-year trial is least able to detect, even at this sample size.

The population is also specific: adults with stable cardiovascular disease and peripheral artery disease, all of whom were simultaneously taking rivaroxaban, aspirin or both [s1]. That is an older, sicker cohort than the average person taking an over-the-counter PPI, which cuts both ways — more events to detect, but less generalisability.

And the C. difficile result stands as an explicit gap rather than a reassurance [s1].

Where the guideline landed

The American College of Gastroenterology's guideline on gastro-oesophageal reflux disease, published in November 2021, describes the state of the argument directly: scrutiny of PPIs has increased considerably, and although they remain the medical treatment of choice for GERD, multiple publications have raised questions about adverse events, raised doubts about the safety of long-term use, and increased concern about overprescribing [s2].

The guideline holds both positions at once — the drugs are the treatment of choice, and overprescribing is a real concern — and grades its recommendations using the GRADE system, while flagging separately those key concepts and suggestions that did not have sufficient evidence to be graded at all [s2].

What this changes for a reader

The gap between the observational literature and the randomised result is the story here. A long list of associations reported in cohort studies did not reproduce when the same drug was assigned at random and the same outcomes were counted over three years [s1]. That does not prove those associations were entirely artefacts — no single trial does — but it substantially weakens each of them.

What survives is a modest increase in enteric infections, on a mechanism that makes sense, at an rate of 1.4% against 1.0% over a median of 3.01 years [s1]. And an open question about C. difficile that this trial did not have the events to answer [s1].

The separate question of whether a given person needs a PPI at all is not one the trial addresses. Overprescribing, which the ACG names as a concern, is a question about indication rather than safety [s2].

This article is informational and is not medical advice. Do not start or stop an acid-suppressing medication on the basis of an article; that decision belongs with a reader and their clinician.

Sources

• Safety of Proton Pump Inhibitors Based on a Large, Multi-Year, Randomized Trial of Patients Receiving Rivaroxaban or Aspirin — Gastroenterology, 2019-05-29
• ACG Clinical Guideline for the Diagnosis and Management of Gastroesophageal Reflux Disease — American Journal of Gastroenterology, 2021-11-22