Proton Pump Inhibitors and Risk of Mild Cognitive Impairment and Dementia

Felicia C. Goldstein, PhD; Kyle Steenland, PhD; Liping Zhao, MSPH; Whitney Wharton, PhD; Allan I. Levey, MD, PhD; Ihab Hajjar, MD, MS


J Am Geriatr Soc. 2017;65(9):1969-1974. 

In This Article


The results of this study do not confirm recent reports[1,2] that the use of PPIs is linked to greater risk of dementia and AD. In this dataset, individuals without dementia at baseline who were taking PPIs at every follow-up occasion or were intermittently taking PPIs had lower risk of cognitive decline than those were not taking PPIs. In addition, consistent and intermittent use of PPIs was associated with lower risk of cognitive decline with a suspected AD etiology. When the sample was further divided into those with normal cognition or MCI, the same finding of lower risk for cognitive decline was observed for both groups taking PPIs.

A strength of this study was the well-phenotyped sample of individuals that relied on diagnoses of cognitive status by a team of experienced clinicians in academic medical centers. ADC clinicians rely on a comprehensive neuropsychological battery and an independent interview with a study partner to obtain information about a person's functional status before making a diagnosis. Two of the previous investigations[1,2] in individuals in primary care relied on administrative databases using the diagnoses of a diverse group of practitioners presumably with varying degrees of experience in diagnosing dementia. Studies performed in the United States indicate that cognitive impairment and dementia are underdiagnosed and underreported in individuals in primary care.[13,14] Another strength was the availability of a broad age range of individuals; prior studies were limited to persons in their mid 70's.[1,2,5] H2RA use was also controlled for, which was not done in prior studies.[1,2] Other investigations included polypharmacy, defined as taking five or more drugs, as a covariate instead. This may or may not have included H2Ras, which are an older class of drugs used to treat gastrointestinal disorders and are still prescribed. One of these studies[7] found that use of these medications more than doubled the risk of incident cognitive impairment in community-residing African Americans, whereas there was no greater risk with PPIs. Failure of prior studies to control for H2RAs could have produced varying results depending on the mix of such medications in their polypharmacy category. The current study, unlike the previous study,[7] found that H2RAs decreased the risk of MCI and dementia, including that with AD as the suspected etiology. The current study sample was more heterogeneous in age and race, which may have contributed to the differences in findings. Another study[15] more recently found that H2RA use was not a risk factor for dementia or AD in Adult Changes in Thought (ACT) study participants. Strengths included the ability to evaluate exposure over a decade before enrollment in ACT and detailed information using pharmacy records that allowed them to examine dose-response relationships as risk factors.

The current study findings replicate the results of a recent study,[5] but they conflict with those of two others[1,2] demonstrating a detrimental effect of PPIs. The conflicting findings from these two studies cannot be attributed to differences in the classes of PPI medications used for analyses, because they were the same in the current study. Information concerning dosage and schedule of PPI use was not available in the NACC database. This lack of dispensing data is also a limitation of these other recent published reports.[1,2,5] Another limitation concerns a reliance on PPI use based on self-report. This may have resulted in misclassification bias due to persons forgetting their medications, especially in those with MCI. This bias could have occurred as well in the finding of one of the previous studies of detrimental associations between PPIs and cognition because they relied on interviews.[1] All studies, including the current one, lacked information on adherence to PPIs. Finally, all available information was used in the analyses; the missed visits were not considered. For example, a participant who was cognitively normal at baseline may have skipped a Year 2 follow-up. If there were records of PPI use at all the available visits, that participant was considered a PPI user at Year 2. If the diagnosis was MCI at Year 3, conversion would have been classified at Year 3 despite the lack of Year 2 follow-up information. This could have led to misclassification of PPI use or time to diagnosis. One of the previous studies[1] used a last observation carried forward approach, which could have led to similar misclassification.

Other risk factors for dementia and AD were replicated across all studies, including older age, depression, and diagnoses of diabetes mellitus and stroke. As in all observational pharmacoepidemiological studies, indication bias may be skewing these results to suggest that PPIs are not detrimental to cognition. The current analysis found that PPI users were at higher risk of dementia due to their higher frequency of cardiovascular disease, hypertension, diabetes mellitus, and depression. Hence, it is unlikely that the observation was related to a bias to prescribing PPIs to healthier individuals. It is possible that the PPI users, because of the greater frequency of cardiovascular risk factors, received better health care than non-PPI users. This, in turn, could have reduced their risk of dementia.

The current findings do not support that PPIs are associated with greater risk of dementia despite mechanisms proposed as to why they should be. It has been hypothesized that PPIs are associated with higher beta amyloid levels, which are involved in the pathogenesis of AD. Higher beta amyloid levels were observed in an amyloid cell model and in mice after PPI treatment.[16] A human clinical trial demonstrating changes in cognitive functioning and cerebrospinal fluid and neuroimaging biomarkers of AD linked to PPI use would support these findings. Another potential mechanism involves the role of vitamin B12, given findings of an association between use of PPIs and vitamin B12 deficiency.[17] Certain PPIs, including lansoprazole and omeprazole, have been found to cross the blood–brain barrier, demonstrating that they directly affect the brain.[18,19] Caution needs to be exercised when speculating about the effect of PPIs on brain functioning until a randomized, prospective clinical trial elucidates the effect of PPIs on cognition.