Does omeprazole cause dementia? What the conflicting studies actually show

Short answer: Probably not, and the evidence has been moving away from the idea rather than towards it. The studies genuinely conflict: a 2026 Taiwanese cohort of 2.4 million adults found dementia risk climbing with cumulative use, up to 66% higher in the heaviest-exposure group — while the largest pooled analysis to date, 18 studies and more than 6.3 million people, found no statistically significant association at all (RR 1.14, 95% CI 0.98–1.33). An umbrella review of eleven separate reviews concluded that a true causal relationship "appears unlikely," and genetic studies find no causal signal. Most usefully of all: when researchers compared people taking a proton pump inhibitor against people taking a different acid-suppressing drug — people with the same underlying problem — the association essentially vanished.

This is educational content, not medical advice, and it is not a reason to stop anything. Proton pump inhibitors treat conditions — ulcers, bleeding risk, severe reflux, Barrett's oesophagus — where stopping carries real consequences. Stopping abruptly also causes a rebound surge of acid. Any change belongs with the clinician who prescribed it.

Why the same question keeps getting opposite answers

Proton pump inhibitors — omeprazole (Prilosec), esomeprazole (Nexium), lansoprazole, pantoprazole, rabeprazole — are among the most-prescribed drugs on earth, and they are taken for years by people who are, on average, older and in worse health than people who take nothing.

That is the whole problem in one sentence. Heartburn is not randomly distributed. People on long-term acid suppression are more likely to be overweight, to have diabetes, to be on several other medicines, to smoke, to drink, and to have the kind of health record that produces more doctor visits — and more doctor visits produce more diagnoses of everything, dementia included.

So when a study finds that PPI users get more dementia, there are three possible explanations, and researchers spend most of their effort trying to tell them apart:

1. The drug does something to the brain.
2. The people who take the drug were already at higher risk for other reasons.
3. The early, unnamed phase of dementia itself leads to more prescribing and more diagnosing.

The honest state of the field is that explanations 2 and 3 account for most of what has been observed, and nobody has convincingly isolated explanation 1.

The studies behind the frightening headlines

The 2016 German study is where the "44%" came from. Analysing claims data from Germany's largest statutory insurer, researchers followed 73,679 people aged 75 and over and reported that those on regular PPI prescriptions had a 44% higher rate of dementia (HR 1.44, 95% CI 1.36–1.52). It is worth knowing how that study was built: only 2,950 of those people counted as regular users, exposure came from prescription records with no information on over-the-counter use, and the analysis could not adjust for education, weight, smoking or diet. The authors' own concluding sentence went considerably further than a claims-database analysis can support — a point other researchers made at the time, in print.

The 2026 Taiwanese study is the newest and by far the largest of the positive findings. Using national insurance and preventive-health databases, it followed 2,444,999 adults aged 40 and over; 137,126 developed dementia. Cumulative PPI exposure tracked with risk in a dose-dependent way, reaching HR 1.66 (95% CI 1.64–1.69) in the highest-exposure group.

That is a large, careful, dose-responsive finding and it deserves to be taken seriously. But it contains a detail that cuts against the simple interpretation, and the authors report it plainly: the association was strongest in the youngest people. Adults under 65 in the high-exposure group had HR 1.97 (1.89–2.06), and the association weakened with advancing age.

If a drug were steadily damaging ageing brains, you would expect the opposite — the effect should be at least as visible in the people whose brains are most vulnerable. An association that is strongest in the under-65s, in a database study where younger people with heavy long-term medication use are unusual and unusually unwell, looks more like a marker of who gets prescribed a lot of medicine than like pharmacology. (The same inverted age pattern shows up in the sleeping-pill literature, for the same reason.)

The studies that looked hardest and found nothing

Three pieces of evidence pull firmly the other way, and they are stronger designs than the claims-database studies.

The Kaiser Permanente study — and why it matters more than its size suggests. Researchers in Seattle followed 3,484 adults aged 65 and over for an average of 7.5 years, during which 827 developed dementia. Crucially, dementia was not read off an insurance code: participants were screened every two years and anyone who screened positive was properly evaluated. Exposure was measured as total standardised daily doses from pharmacy records. The result was flat at every level of use — about one year of daily use, HR 0.87 (0.65–1.18); three years, 0.99 (0.75–1.30); five years, 1.13 (0.82–1.56). No trend, no signal, nothing for Alzheimer's either.

Why that study carries extra weight here: it is the same research group, the same cohort and the same dose-measuring method that produced the well-known finding that heavy long-term anticholinergic use is associated with dementia. The same tool, pointed at anticholinergics, found a dose-response. Pointed at PPIs, it found a flat line. That is about as close as observational research gets to a built-in control.

The ASPREE analysis. A post-hoc analysis of a large randomised trial of older adults — 18,934 community-based people aged 65 and over in the US and Australia, with medications reviewed in person at annual visits and dementia diagnosed against formal criteria — found PPI use associated with HR 0.88 (0.72–1.08) for incident dementia. No association with cognitive impairment short of dementia, and no difference in how cognitive test scores changed over time. The authors described their findings as providing "reassurance about the safety of long-term use of PPIs among older adults."

The genetic evidence. Mendelian randomisation uses inherited genetic variants — fixed at conception, and therefore not influenced by your lifestyle, weight or health record — as a natural experiment. A 2024 analysis applied it to five PPI drugs and all-cause dementia plus five subtypes. A couple of weak signals appeared initially, but none survived correction for multiple testing. The authors' conclusion was that it is "inappropriate to restrict clinically justified PPI prescriptions merely due to potential cognitive risks."

The single most useful comparison, and almost nobody quotes it

Here is the test that does the most work, and it appears nowhere on the first page of search results.

If the problem with studying PPI users is that they differ from non-users in a hundred ways, then the fix is obvious: don't compare them with non-users. Compare them with people taking a different drug for the same complaint.

That comparison exists. A 2023 systematic review and meta-analysis in the British Journal of Clinical Pharmacology pooled nine prospective studies covering 204,108 dementia cases. Against non-users, PPI use gave RR 1.16 (95% CI 1.00–1.35) — an interval whose lower bound sits exactly on no-effect.

Then the authors ran the comparison that matters: PPI users against users of H2-receptor antagonists (famotidine and its relatives), a different class of acid-suppressing drug taken by people with, broadly, the same reason to be taking something. The result was RR 1.03 (95% CI 0.66–1.62) — no association whatsoever.

When you compare people with reflux to people with reflux, the signal disappears. When you compare people with reflux to the general population, a small signal appears. That pattern is what confounding by indication looks like, and it is the most informative single fact in this literature.

The same review's own summary is appropriately careful: no clear evidence of an association, but because the sensitivity analyses disagreed with each other, some risk could not be entirely ruled out.

What the "33%" on page one actually is

Most of the consumer articles you will find on this question — including several hospital-system blogs and the big health portals — rest on one 2023 study, and report it as "a third higher risk."

That study is a well-conducted analysis of the long-running ARIC cohort: 5,712 dementia-free participants, median follow-up 5.5 years, 585 cases. Two findings, and the popular coverage usually carries only one of them:

• People currently taking PPIs were not at significantly higher risk — HR 1.1 (95% CI 0.9–1.3).
• People who had used them for more than 4.4 cumulative years beforehand were at higher risk — HR 1.3 (95% CI 1.0–1.8).

That second confidence interval reaches down to 1.0. It is a result on the edge of statistical significance, from an analysis the journal itself classified as Class III evidence — a rating meaning, roughly, "suggestive, not dependable." Shorter durations of use showed nothing. The authors' own discussion asks for future research to understand possible pathways; they do not claim to have found one.

"Long-term use linked to 33% higher dementia risk" is a fair headline for that study. It is not a fair summary of the evidence base, which now includes a pooled analysis six times larger finding nothing.

What proton pump inhibitors are known to do

There are real reasons to review long-term acid suppression, and none of them depend on the dementia question being settled. Long-term use is associated with lower vitamin B12 absorption, lower magnesium, and — with varying degrees of certainty — more fractures, more Clostridioides difficile infection and kidney problems. Guidelines have for years advised using the lowest effective dose for the shortest necessary time.

The B12 point is the one most worth acting on, because it connects directly to memory and because it is checkable and correctable. PPIs work by suppressing the stomach acid needed to release B12 from food, and a genuine B12 deficiency can cause memory problems, confusion and fatigue that improve when it is corrected. If you have been on acid suppression for years and your memory feels different, that is a specific blood test to ask about — not a reason to assume the worst. We cover it in low vitamin B12 and memory.

What to do if you have been taking one for years

Do not stop on your own. PPIs cause rebound acid hypersecretion: stop suddenly after months of use and acid production overshoots for a few weeks, symptoms come back worse than before, and people conclude they need the drug permanently. That rebound is a predictable pharmacological effect, not proof the drug is essential.

What is reasonable to do:

1. Ask what the indication is, and whether it still applies. Some people take a PPI for a defined, ongoing reason — Barrett's oesophagus, a previous bleeding ulcer, long-term anti-inflammatory use. For them, stopping would be the risky choice. Others were started on one for a symptom years ago and nobody ever revisited it. Those are completely different situations and only your prescriber can tell you which one you are in.

2. Ask about stepping down rather than stopping. The usual approach is a gradual dose reduction, sometimes moving to on-demand use or to an H2 blocker such as famotidine, rather than an abrupt halt. Done with a clinician, this is routine.

3. Raise B12 if you have been on it for years. A simple blood test, and a fixable problem if it turns up.

4. Put the risk in proportion. Acid suppression is not on the Lancet Commission's list of major modifiable dementia risk factors — the fourteen that are are things like blood pressure, hearing, blood sugar, physical activity, smoking, social connection and LDL cholesterol. If you spend your energy on the medicine and not on those, you are optimising the wrong variable.

So — does omeprazole cause dementia?

On the evidence as it stands: there is no good reason to believe that it does.

• The pooled estimate across 18 studies and 6.3 million people is not statistically significant, with enormous disagreement between studies.
• An umbrella review of eleven reviews concluded that most find no link and that a causal relationship "appears unlikely."
• The best-measured cohorts, where dementia was diagnosed properly rather than read from billing codes, found nothing.
• The active-comparator analysis — PPI users against people taking a different acid drug — found nothing.
• Genetic analyses find no causal effect.
• The largest positive study is real and large, but its association is strongest in the under-65s, which is not how drug harm usually behaves.

None of that makes long-term acid suppression something to stay on without review. It makes dementia the wrong reason to review it. The good reasons — B12, magnesium, bone health, and the simple question of whether you still need it — are better established and more actionable.

How we research and review our content →

An independent physician audited the Solenna app. We published all nine findings →