OBJECTIVES: The association between proton pump inhibitors (PPIs) and dementia in the elderly have been called into question, yet available studies have methodological limitations and their findings are conflicting. This study aims to identify distinct trajectories of longitudinal PPIs use and to examine whether these use patterns associated with risk of dementia.

METHODS: Adults aged 65 years or older who were newly prescribed oral PPIs in 2005 were identified from Taiwan’s National Health Insurance Research Database. The date of the first PPIs prescription was defined as the index date for each subject. Monthly use of PPIs during the trajectory period, defined as 3 years of follow-up since the index date, was determined. A group-based trajectory modeling with a zero-inflated Poisson distribution and third order polynomials was used to identify distinct groups with regard to longitudinal PPIs use over 3 years. All dementia occurring during the 5-year follow-up after trajectory period were identified. Time-varying Cox proportional hazards model with adjustment of medications related to cognitive decline was used to examine the association between distinct trajectories of PPIs use and dementia.

RESULTS: Among 10,533 eligible elderlies, we identified 3 trajectories of longitudinal PPIs use over a 3-year trajectory period: short-term users (n=7406, 70.3 %), fluctuated users (n=1528, 14.5 %) and long-term users (n=1599, 15.2 %). Long-term users (HR=0.99 [95 % CI 0.93-1.17]) and fluctuated users (HR=0.91 [95 % CI 0.76-1.09]) were not associated with risk of dementia as compared with short-term users. Similar results were observed at sub-distribution proportional hazards model as sensitivity analysis.

CONCLUSIONS: With novel analysis method and full adjustment for medications related to cognitive decline, we found PPIs were not associated with risk of dementia, in contrast to some of previous findings.