Background: This cohort study estimated the population attributable fraction (PAF) of seven combined major risk factors for incident dementia.
Methods: We conducted a cohort study of 8563 community-dwelling individuals aged ≥ 65 years. In a baseline survey (2006), we collected data on major seven risk factors: diabetes mellitus, hypertension, obesity, physical inactivity, severe psychological distress, smoking, and low educational attainment. The total number of risk factors was applied as an exposure variable. Subjects were categorized into four groups according to the total number of risk factors they possessed (0, 1, 2, ≥ 3 risk factors). Data on incident dementia were retrieved from the public Long-term Care Insurance database. Hazard ratios (HRs) and 95% confidence interval (95% CI) were estimated using the Cox proportional regression model. We also calculated the PAF using HRs and the prevalences in our cohort data.
Results: The number of cases of incident dementia was 577 (6.7%). A dose-response relationship between the total number of risk factors and incident dementia was observed; in comparison with no risk factors (reference), the age- and sex-adjusted HRs (95% CIs) were 1.25 (0.92-1.70) for one risk factor, 1.59 (1.18-2.15) for two, and 2.21 (1.62-3.01) for three or more (P trend < 0.001). If subjects had adhered to none of the risk factors, the PAF would have been 32.2%. If subjects had improved toward one better category, the PAF would have been 23.0%.
Conclusion: Our findings suggest that reducing the combined number of risk factors would contribute significantly to reducing the incidence of dementia.
Keywords: Attributable fraction; Cohort; Dementia; Joint effect; Risk factors.