Pantoprazole safety in a large randomised trial

Overview

Moayyedi P et al.. Gastroenterology 2019 · PMID 31152740 · DOI 10.1053/j.gastro.2019.05.056

Over three years, only gut infections were more common with pantoprazole (1.4% vs 1.0%).

Does the study support the specific claim? No · Methodological quality: high

Original publication

Plain-language summary

Scientific in-depth review

1. Full reference

Authors

Moayyedi P et al.

Journal

Gastroenterology

Year

2019

Identifier

PMID 31152740 DOI 10.1053/j.gastro.2019.05.056

Verification

verified – PubMed efetch 2026-10-25

Retracted

no

Corrections

none

Extent checked

full text

Conflicts of interest and funding

Sponsored by Bayer AG (rivaroxaban manufacturer; two Bayer employees co-authors); data analysed independently at the Population Health Research Institute. Many authors report fees/grants from Bayer and other pharmaceutical companies; first author research funding from Allergan and Takeda.

2. Research question

Are adverse events seen in observational studies confirmed in an RCT of long-term PPI?

3. Study design and population

Study design

Randomised controlled trial – 3x2 partial factorial, multicentre, double-blind, placebo-controlled RCT (COMPASS pantoprazole component; NCT01776424); safety outcomes prespecified but trial not powered for them

Population

Adults with stable coronary or peripheral artery disease (aged >=65, or younger with additional criteria) without a clinical need for long-term PPI; mean age 67.6, 78% male; excluded high bleeding risk, severe heart failure, significant renal impairment (GFR <15), dementia, severe COPD

Country

Multinational: 580 centres in 33 countries (25% Western Europe, 21.5% Eastern Europe, 25% South America, 14% North America, 14% Asia-Pacific)

Period

Recruitment March 2013 - May 2016; median follow-up 3.01 years (IQR 2.49-3.59; max ~5 years); 53,152 patient-years

Sample size

17,598 randomised (8791 pantoprazole 40 mg, 8807 placebo); 21-22% permanently discontinued study drug

Inclusion and exclusion criteria

COMPASS participants not needing PPI and willing to stop H2RA/PPI (64% of COMPASS); 0.8% were on PPI before randomisation

4. Methods and endpoints

Exposure / intervention

Pantoprazole 40 mg once daily

Comparator

Matched placebo

Endpoints

Adjudicated CV events, mortality, cancer, hospitalisation; non-adjudicated interview-reported (every 6 months) pneumonia, C. difficile, other enteric infections, fracture, gastric atrophy, CKD, diabetes, COPD, dementia

Statistical methods

ITT; stratified Cox for time-to-event outcomes; logistic regression ORs for other safety outcomes; no multiplicity adjustment; sensitivity analysis excluding discontinuers and prior PPI users; NNH by Newcombe-Wilson

5. Actual results

Results

All-cause death 630 (7.2%) vs 614 (7.0%), HR 1.03 (0.92-1.15); MI/stroke/CV death HR 1.04 (0.93-1.15); cancer HR 0.99 (0.87-1.13); GI cancers 86 vs 83. ORs (pantoprazole vs placebo): enteric infection 119 (1.4%) vs 90 (1.0%), 1.33 (1.01-1.75), NNH 301 (152-9190) over 3 years; C. difficile 9 vs 4, 2.26 (0.70-7.34); CKD 184 (2.1%) vs 158 (1.8%), 1.17 (0.94-1.45); pneumonia 1.02 (0.87-1.19); fracture 0.96 (0.79-1.17); dementia 1.20 (0.81-1.78); gastric atrophy 0.73 (0.40-1.32); COPD 1.18 (0.93-1.51); diabetes 0.96 (0.85-1.09). Results similar excluding discontinuers.

Absolute risks

Enteric infection 1.4% vs 1.0% (NNH 301 over ~3 years); CKD 2.1% vs 1.8%; deaths 7.2% vs 7.0%

Authors' conclusion

Pantoprazole was not associated with any adverse event over ~3 years except possibly enteric infections; limiting PPI prescription for fear of long-term harm is not appropriate, although PPIs should be used only when indicated.

6. Strengths

• Largest placebo-controlled PPI trial; randomised, double-blind
• Prespecified safety outcomes collected systematically every 6 months
• High retention; ITT and sensitivity analyses
• Multinational including European centres

7. Weaknesses

• Not powered for safety outcomes; rare events (C. difficile, gastric atrophy) underpowered
• Non-CV outcomes self-reported without adjudication (non-differential misclassification biases to null)
• Median 3 years; longer-term effects uncertain
• Selected population (CVD, mostly men, no PPI indication)
• CKD CI (0.94-1.45) still compatible with the 1.2-1.5-fold increases seen observationally
• No multiplicity adjustment (enteric infection finding borderline)
• No measurement of B12, iron or mineral status

8. Risk of bias

Orienting tool

RoB 2

Type of assessment

structured assessment based on the full text by a single reviewer (no formal double review)

Full formal assessment

no

Overall risk of bias

some concerns

Rationale

Single-reviewer structured RoB 2 judgement from full text; not formal (needs two independent reviewers). Overall downgraded from low to some concerns because of non-adjudicated self-reported safety outcomes; bias would tend towards the null.

Full-text review (25 October 2026): Added full safety table with ORs/CIs (CKD 1.17, 0.94-1.45; pneumonia 1.02; fracture 0.96; dementia 1.20; mortality HR 1.03), NNH 301 for enteric infection, non-adjudicated outcome ascertainment, population/exclusions, regions, Bayer sponsorship and COI; RoB 2 domains with overall changed from low to some concerns.

9. Assessment of the statistical interpretation

Mostly null with moderately wide CIs. Mortality, pneumonia and fracture effects are tightly bounded near 1; CKD (OR 1.17, 0.94-1.45) and dementia (1.20, 0.81-1.78) cannot exclude modest increases. Enteric infection excess is small in absolute terms (NNH ~300 over 3 years).

10. Relevance to the specific claim

Key test of causality for C-PPI-01, -02, -04, -05, -09.

11. Independent contrary findings

12. Transferability to Germany and Europe

High: multinational including European centres.

13. Overall methodological judgement

Strongest randomised evidence available: reassuring for mortality, pneumonia, fracture and cancer over ~3 years; supports a small excess of enteric infections; inconclusive (not exclusionary) for CKD and dementia.

14. Open questions

Outcomes beyond 3-5 years; CKD and dementia with adequate power; micronutrient status.

15. Linked fact checks

• Long-term PPI use causes chronic kidney disease and excess deaths (about 45 extra deaths per 1,000 users, mainly from cardiovascular disease, kidney disease and upper gastrointestinal cancer). – Studies contradicting or limiting the claim
• Long-term acid suppression increases bacterial overgrowth, Clostridioides difficile and other enteric infections, and changes the gut microbiome. – Studies supporting the claim
• Acid blockers stop the absorption of all minerals, iron and vitamin B12, so nothing that is eaten benefits the body, the immune system stops working, and hair falls out. – Studies contradicting or limiting the claim
• Long-term acid suppression causes hip fractures and pneumonia. – Studies contradicting or limiting the claim
• Acid blockers are 'deadly poison' that cause cancers, autoimmune diseases, infertility, erectile dysfunction, depression, fatty liver and irritable bowel; their spread since about 2004–2005 explains a doubling of cancer and erectile dysfunction. – Studies contradicting or limiting the claim