Contributions of Modifiable Risk Factors to Dementia Incidence: A Bayesian Network Analysis

Abstract

Objective: To determine and compare the contributions of modifiable risk factors (RFs) with the prevention of dementia in older adults.

Design: A systematic review and Bayesian network meta-analysis (NMA). The observational group was set as a reference to collect all existing RFs and compare them with each other.

Setting and participants: An exhaustive and comprehensive literature search strategy was used to identify relevant prospective cohort studies from several online databases from their inception to May 1, 2019. Participants without dementia were adults aged greater than 50 years.

Measures: The required data were extracted from the eligible studies to facilitate the Bayesian NMA.

Results: Forty-three cohort studies with 277,294 participants were included in this NMA. Using the observation group as the reference, all defined RFs, except for antioxidants, were associated with lower risks of all-cause dementia [no sleep disturbances (odds ratio, OR 0.43, 95% credible interval, CrI 0.24-0.62), a high level of education (OR 0.50, 95% CrI 0.34-0.66), no history of diabetes (OR 0.57, 95% CrI 0.36-0.78), nonobese patients (OR 0.61, 95% CrI 0.39-0.83), no smoking history (OR 0.62, 95% CrI 0.45-0.79), living with family members (OR 0.67, 95% CrI 0.45-0.89), participation in physical exercise (OR 0.73, 95% CrI 0.46-0.94), abstinence from drinking (OR 0.78, 95% CrI 0.56-0.99), and no history of hypertension (OR 0.80, 95% CrI 0.65-0.96)].

Conclusions/relevance: The findings provide reliable support for the hypothesis that modifiable somatic and lifestyle factors are strong predictors of all-cause dementia.