Hazard ratios were calculated from Cox proportional hazards models with age as the time scale according to three study samples including 824 cases of dementia (563 cases of AD), 630 cases of dementia (435 cases of AD), and 883 cases of CIND, respectively. Adjustments were made for gender, education, lifestyle factors, comorbid diseases, and vascular risk factors.
Lower risks for AD and all-cause dementia were significantly associated with the use of any NSAIDs and the salicylates without barbiturates subgroup in the study sample including subjects with CIND at baseline. There was a weak association between any NSAIDs and the risk of CIND (hazard ratio, 0.87; 95%confidence interval, 0.76-1.00).
These results suggest that there is an association between NSAID use and a lower incidence of AD and, to a lesser extent, of CIND.