Last year, a widely reported epidemiology study came to the troubling conclusion that elderly people who regularly took proton pump inhibitors (PPIs) were at increased risk of dementia. Now, a study published in the June 7 Journal of the American Geriatrics Society challenges this association. Led by Felicia Goldstein at Emory University in Atlanta, researchers found that in patients monitored at U.S. Alzheimer’s Disease Centers (ADCs), PPI intake correlated with, if anything, a slightly lower risk of developing mild cognitive impairment or dementia.

Combing through the National Alzheimer’s Coordinating Center (NACC) data from 33 federally funded ADCs, Goldstein and colleagues examined the records of 10,486 people aged 50 or older who had had two to six annual medical visits over 10 years. The NACC unites the collections of the neuropathology cores at NIA-funded ADCs into a single uniform data set that includes standardized assessments collected longitudinally. ADCs use standardized neuropsychological tests, including the clinical dementia rating scale for dementia diagnosis. About 2,000 of the patients tracked took PPIs intermittently, and 884 reported taking PPIs regularly. These include over-the-counter drugs such as omeprazole (Prilosec in the United States), lansoprazole (Prevacid in the United States), and others.

The researchers first applied regression analysis to look for associations between PPI use and cognitive decline in people who were cognitively normal or had MCI at the time of their first visit. They adjusted their analysis to control for age, race, sex, education, vascular diseases, depression, or anticholinergic or histamine-2 receptor antagonist use. The latter drug also reduces acid reflux, and some studies have implicated it in increasing risk of cognitive impairment. The present analysis found no association between regular or intermittent PPI use and cognitive decline. In fact, the data indicated a slight dip in the probability of PPI users progressing to MCI or dementia, with a hazard ratio of 0.78 for continuous and of 0.84 for intermittent use.

A similar analysis associated both intermittent and regular PPI use with a lower hazard ratio, 0.82, of being diagnosed with AD. Goldstein and colleagues also analyzed data from people with normal cognition during their first visit separately from those who already had MCI, to see if either group might be more vulnerable to PPIs. They found no increased risk for cognitive decline in either group. Lastly, the authors focused on people 75 and older to more closely mirror the populations tracked in previous studies. Once again, they detected no PPI-associated increase in dementia risk.

The findings contrast with those reported last year by Britta Haenisch and colleagues at the German Center for Neurodegenerative Diseases, Bonn (see Feb 2016 news). 

Which study is correct? Both Lewis Kuller, University of Pittsburgh, and John Breitner, McGill University, Montreal, said it’s impossible to tell at this point. “This [new] study doesn’t really close the gate,” said Kuller. Besides Goldstein and Haenisch’s work, previous studies have suggested positive, negative, or neutral associations between PPIs and cognitive decline (Nevado-Hurtado et al., 2016; Tai et al., 2017; Booker et al., 2016; Boustani et al., 2007). Last April, a consensus panel of nine international experts in gastroesophageal reflux deemed the results too inconclusive to change their safety recommendations for use of over-the-counter PPIs (Johnson et al., 2017). “The whole literature is very confused, and this new study doesn’t add clarity,” said Breitner.

A major strength of Goldstein’s new study, noted Kuller and Breitner, is the quality of the cognitive diagnoses, including comprehensive neuropsychological tests and patient interviews. The German study relied on health insurance medical records for dementia diagnosis, which can often be unreliable, researchers agreed. However, the NACC data set showed significant variability in follow-up visits, raising the question of whether patients in the U.S. study were sufficiently tracked to capture associations. “It’s unclear how many patients dropped out of the study and at what times,” said Kuller.

Another limitation, acknowledged by Goldstein and colleagues, is that PPI usage data came from lengthy questionnaires filled out by the patients themselves, rather than pharmacy dispensing databases or physician records. This is particularly problematic in patients with early or advanced signs of cognitive impairment, wrote Haenisch. Under these circumstances, PPI use might have been under- or overreported. On that note, Goldstein et al. also failed to find an association between anticholinergic medications and MCI or AD, a link that has been widely reported in the past.

Moreover, although the authors adjusted for histamine-2 receptor antagonist (H2RA) use, Breitner suspects that PPIs, which came on the market more recently, could be a surrogate for prior H2RA use. While the hazard ratios reported by Goldstein and colleagues are statistically significant, they are weak and within the range that could be due to confounders, such as previous use of H2RAs, suggested Breitner.

Limitations aside, the study offers a new perspective on the potential risks of PPIs, which are used by as many as 5 percent of adults over 65. “It is somewhat reassuring that there probably isn’t a powerful effect,” said Kuller. A prospective, randomized, longitudinal trial, ideally with AD biomarker data, might clarify the risk PPIs pose, though Kuller thinks the likelihood is slim that such a study would be funded.—Marina Chicurel


  1. Actually the following statement is not quite correct per UDS protocol: 

    "Another limitation, acknowledged by Goldstein and colleagues, is that PPI usage data came from lengthy questionnaires filled out by the patients themselves, rather than pharmacy dispensing databases or physician records." 

    These data are not supposed to be the result of "lengthy questionnaires filled out by the patients themselves." The pharmacy data, as well as other data captured by the UDS, are supposed to be collected as part of the clinical interview, usually by clinic staff, nurse, or examining physician, and available records. If the patient appears cognitively impaired, the information may be obtained from the co-participant (study partner) interview and may be supplemented by available medical records.  It would be decidedly "off protocol" to hand a study subject a UDS packet, or medications data form, and ask them to complete it on their own.

Make a Comment

To make a comment you must login or register.


Brain Bank Citations

  1. National Alzheimer’s Coordinating Center

News Citations

  1. Can Common Heartburn Drugs Raise Risk for Dementia?

Paper Citations

  1. . Commonly prescribed drugs associate with cognitive function: a cross-sectional study in UK Biobank. BMJ Open. 2016 Nov 30;6(11):e012177. PubMed.
  2. . Risk of dementia from proton pump inhibitor use in Asian population: A nationwide cohort study in Taiwan. PLoS One. 2017;12(2):e0171006. Epub 2017 Feb 15 PubMed.
  3. . Risk factors for dementia diagnosis in German primary care practices. Int Psychogeriatr. 2016 Jul;28(7):1059-65. Epub 2016 Jan 8 PubMed.
  4. . The association between cognition and histamine-2 receptor antagonists in African Americans. J Am Geriatr Soc. 2007 Aug;55(8):1248-53. PubMed.
  5. . The Safety of Appropriate Use of Over-the-Counter Proton Pump Inhibitors: An Evidence-Based Review and Delphi Consensus. Drugs. 2017 Apr;77(5):547-561. PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. . Proton Pump Inhibitors and Risk of Mild Cognitive Impairment and Dementia. J Am Geriatr Soc. 2017 Sep;65(9):1969-1974. Epub 2017 Jun 7 PubMed.