|| Checking for direct PDF access through Ovid
Background: Aspirin may help to prevent dementia through its antiplatelet and anti-inflammatory effects, but findings from previous studies are mixed. Considering the length of the preclinical phase of dementia, early initiation of regular aspirin use may be more protective against dementia than later initiation. In this study, we explored the association of regular aspirin use initiated during and after midlife with the risk of dementia.Methods: Using data from 1996 to 2013 on 10,398 ARIC participants, we assessed the association of self-reported prior duration of regular aspirin use which was collected during 1996-1998 (as the baseline of this study) with incident dementia after the baseline. Dementia cases were identified based on: in-person dementia ascertainment from 2011 to 2013; the Telephone Interview for Cognitive Status (TICS) or informant interviews on dementia; and hospital discharge codes/death certificate code by surveillance through 2013. The relative hazards of dementia across categories of the duration of regular aspirin use prior to baseline (<2 years, 2 to <5 years, and ≥ 5 years) were compared to no regular use by Cox proportional hazard models. We adjusted for potential confounders, including demographic variables and major dementia risk factors at baseline. In sensitivity analyses, propensity score weighting was applied. Subgroup analyses were conducted by age, gender, race, APOE, and diabetes status.Results: Participants had a median age of 62 (IQR: 58, 67) at baseline. 55.8% (5,805 of 10,398) were female and 20.6% (2,147 of 10,398) were African American. Lower but non-significant relative hazards (HR) were observed among participants with regular aspirin use for < 2 years (HR: 0.85; 95% CI: 0.69 to 1.04) and ≥ 5 years (HR: 0.87; 95% CI: 0.68 to 1.11), but not for the 2-5 year interval. With propensity score weighting, regular aspirin use < 2 years achieved marginal significance (HR: 0.79; 95% CI: 0.63 to 0.99), while the effect of regular aspirin use ≥ 5 years changed towards null (HR: 1.04; 95%CI: 0.75 to 1.46). Without a prior hypothesis, an interaction was found for diabetes status (p = 0.016) with a significant protective association of ever regular aspirin use compared to no regular use among non-diabetic participants (HR: 0.82; 95%CI: 0.69 to 0.98).Conclusions: We found a non-significant protective association of regular aspirin use at late-middle life with onward incident dementia in a community-dwelling population. Data suggest persons with fewer comorbidities, specifically without diabetes, might be more susceptible to the preventive effect of aspirin use on dementia. Our results are subject to confounding by indication, suggesting the need for additional studies in populations which have different covariate structures than ARIC.