Two January studies took apart the link between acid drugs and stomach cancer

Proton pump inhibitors revolutionised the treatment of acid-related disorders, and widespread overuse has since raised concerns about their long-term safety [s2]. Among those concerns is a run of observational studies reporting associations between long-term use and stomach cancer. Those reports have been hard to interpret, because the reason people take these drugs — persistent upper gastrointestinal symptoms — is also an early symptom of the cancer in question.

Three papers published in January approach that problem from different directions and reach the same conclusion. The design detail is the story.

The Israeli study: watching an association disappear

A matched case-control study published in PLOS Medicine on 6 January used electronic health records from a national health organisation in Israel [s1]. Cases were 875 adults with incident upper gastrointestinal cancer — oesophageal, gastric or duodenal — diagnosed between 2003 and 2024, mean age 63.0 (SD 11.9) years and 62.5% male [s1]. Each was matched to 10 cancer-free controls, for 8,750 controls in total, on age, sex, ethnic sector, socioeconomic status and year of enrolment [s1].

The design feature that matters is that exposure was modelled in discrete pre-diagnosis windows: 0 to 6 months, 6 to 12 months, 1 to 3 years and 3 to 10 years [s1].

In models that did not adjust for symptom-related diagnoses, the associations were large. Esomeprazole carried an adjusted odds ratio of 4.01 (95% CI 3.20 to 5.03, p < 0.001) and omeprazole 2.38 (1.99 to 2.85, p < 0.001) [s1]. Those are the kind of numbers that generate headlines.

Then the windows were separated. Associations diminished for exposures more than a year before diagnosis [s1]. After excluding the final year before diagnosis and adjusting for symptom-related diagnoses — reflux, gastritis, peptic ulcer disease — the researchers did not detect any harmful association [s1].

Remote use, more than three years before diagnosis, was instead associated with lower odds: omeprazole 0.62 (95% CI 0.51 to 0.75, p < 0.001), with similar patterns in a gastric-only subgroup of 701 cases and 7,010 controls [s1].

The authors' conclusion is that apparent harm was concentrated in the months immediately before diagnosis and disappeared once diagnostic context was accounted for, and that the practical implication is to investigate new upper gastrointestinal symptoms rather than attribute cancer risk to acid-suppressive therapy [s1].

What the pattern means

This is textbook reverse causation, caught in the act. A person develops an early stomach cancer. It causes indigestion. A doctor prescribes a proton pump inhibitor. Months later the cancer is diagnosed. In a study that counts any exposure before diagnosis, the drug looks like a cause of the disease that in fact prompted the prescription.

A commentary in the same journal, published a day later, situates this in the wider literature [s2]. It notes that proton pump inhibitors revolutionised treatment of acid-related disorders, that widespread overuse has raised concerns about long-term safety, and that the Israeli study presents new evidence assessing whether the links between prolonged use and upper gastrointestinal cancer are causal [s2].

The Nordic study: the same answer at scale

Two weeks later The BMJ published a study built explicitly to correct the methodological weaknesses of the existing literature [s3].

It used prospectively collected data from complete nationwide registries in five Nordic countries — Denmark, Finland, Iceland, Norway and Sweden — covering all healthcare in those systems [s3]. Cases were people with gastric adenocarcinoma, each matched to 10 controls drawn at random from the entire population of their country, on age, sex, calendar year and country [s3]. The study included 17,232 cases of gastric non-cardia adenocarcinoma and 172,297 controls [s3].

Four design decisions distinguish it. Exposure was defined as long-term use, more than a year, with the 12 months before the diagnosis or inclusion date excluded [s3]. Gastric cardia adenocarcinoma was excluded entirely to avoid confounding by indication from reflux [s3]. Histamine-2-receptor antagonists were analysed in parallel as a specificity check [s3]. And adjustment covered Helicobacter pylori treatment, peptic ulcer disease, smoking-related diseases, alcohol-related diseases, obesity or type 2 diabetes, and treatment with metformin, non-steroidal anti-inflammatory drugs and statins [s3].

Long-term proton pump inhibitor use occurred in 1,766 cases (10.2%) and 16,312 controls (9.5%) [s3]. The adjusted odds ratio was 1.01 (95% CI 0.96 to 1.07) — no association [s3]. The result for histamine-2-receptor antagonists was similar at 1.03 (0.86 to 1.23) [s3].

The authors go further than reporting a null. They identify the specific sources of error that produced false positive associations in earlier work: including proton pump inhibitor use shortly before diagnosis, counting short-term use, including cardia adenocarcinoma, and failing to adjust for Helicobacter pylori-related variables [s3]. Their conclusion is stated as a possibility rather than a certainty — long-term use may not be associated with increased risk [s3].

What this does not resolve

None of these papers says proton pump inhibitors are free of adverse effects. They address one outcome. The Israeli study's finding of reduced odds with remote use should not be read as protection; residual confounding is the more likely explanation, and the authors list residual confounding, absence of dietary and family history data, and incomplete capture of over-the-counter use among their limitations [s1].

The Nordic study is a case-control analysis of registry data, which cannot capture over-the-counter purchases or actual adherence either. And a null result with a confidence interval of 0.96 to 1.07 rules out a large effect, not a small one [s3].

Neither is a reason for anyone to start or stop a medication, and neither addresses the separate and well-founded concern that these drugs are frequently continued for years without anyone revisiting whether they are still needed.

What to watch

Whether guidelines and patient information materials that currently mention gastric cancer risk are revised, and whether the reverse-causation correction is applied retrospectively to the other outcomes — dementia, kidney disease, fracture — where similar observational associations have been reported.

Sources

1. [s1] Association between proton pump inhibitor use and upper gastrointestinal cancer: A matched case-control study accounting for reverse causation and confounding by indication. PLOS Medicine, 6 January 2026. https://doi.org/10.1371/journal.pmed.1004842
2. [s2] Context matters: Causality and global epidemiology of proton pump inhibitor safety. PLOS Medicine, 7 January 2026. https://doi.org/10.1371/journal.pmed.1004862
3. [s3] Long term use of proton pump inhibitors and risk of stomach cancer: population based case-control study in five Nordic countries. BMJ, 21 January 2026. https://doi.org/10.1136/bmj-2025-086384