Dependent variable

Our outcome variable was dementia incidence based on each municipality’s publicly available long-term care insurance records. Japan’s Ministry of Health, Labor and Welfare (MHLW) established a scale called the Degree of Independency in Daily Lives of Demented Individuals (hereafter “dementia scale”) [13], and each local government sent personal investigators to participants’ homes to assess their eligibility for nursing care, such as need for home helpers. The investigators evaluated physical function, daily life ability, cognitive function, mental and behavioral disorders, adaptation to social life, and special medical treatment within 14 days [14, 15]. Following the assessment, the investigators classified the participants into one of eight ranks on the dementia scale according to their cognitive disability (Supplementary Table S1) [15, 16].

The dementia scale correlated highly with the Mini-mental State Examination (Spearman rank correlation ρ = − 0.74) [17]. Moreover, another study, where neuropsychiatrists conducted clinical interviews as defined by the International Psychogeriatric Association, found that the dementia scale was an effective indicator for the clinical diagnosis. Sensitivity and specificity were 73 and 96% for rank II dementia [18]. According to MHLW, we defined rank II dementia or above as showing some symptoms, behaviors, or communication problems during daily lives [15, 19]. We discontinued data tracking participants who did not develop dementia before they died or were lost to follow-up during the 6-year study period.