This study was a longitudinal analysis that collected data at the baseline and 3-year follow-up of a prospective cohort study called the Septuagenarians, Octogenarians, Nonagenarians Investigation with Centenarians (SONIC) study, a study ongoing since 2010 . The participants who were living independently were recruited from residential registries and sent letters inviting them to participate in the venue survey nearby their residential area at the baseline. The setting was two regions of western and eastern Japan, and each region was composed of both urban and rural areas: Itami City, Hyogo (western urban); Asago City, Hyogo (western rural); Itabashi ward, Tokyo (eastern urban); and Nishitama county, Tokyo (eastern rural).
The inclusion criteria of this study were as follows: 1) they were free of dementia at the baseline, 2) their completed dementia data were available, and 3) the MoCA-J score was administered both at the baseline and 3-year follow-up. The exclusion criteria were the participants who cannot participate in the consecutive 3-year at the survey venue due to severe disability/death/move to other areas including more severe due to stroke or recent stroke. All data were collected at the baseline (2010-2012) while MOCA-J was performed both at the baseline and the 3-year follow-up (2013-2015). At the baseline, a total of 2,245 participants in all age groups (69-71 years old = 1,000, 79-81 years old = 973, and 89-91 years old = 272) were included, but only 1,333 participants met the inclusion criteria and completed the 3-year follow-up (Figure 1).
The basic data were collected on the following variables including age and sex while history of stroke was determined based on interviews by the physicians or nurses at the baseline survey and the participants were classified in accordance with their yes/no responses. Participants were asked what type of stroke they had and classified to ischemic stroke, intracerebral hemorrhage, subarachnoid hemorrhage, subdural hematoma and Transient Ischemic Stroke (TIA).
Information on dementia at baseline was determined based on a combination of self-administered with their yes/no responses and information on dementia drugs at baseline in the participants’ medication record book.
Vascular risk factors
Blood pressure was measured by a physician and trained nurses. A sphygmomanometer was used to measure blood pressure twice with each arm in a sitting position. The average of the first and second measurements of each arm was used in the analysis. Hypertension was diagnosed according to the Japanese Society of Hypertension guideline 2019 , which is defined by systolic Blood Pressure (BP) ≥ 140 mmHg and diastolic BP ≥ 90 mmHg or the use of antihypertensive drugs at the first survey.
The blood samples were collected for subsequent analysis. The levels of fasting/casual blood glucose, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, and triglycerides were determined using biochemical examinations. Diabetes mellitus was defined by fasting blood glucose ≥ 126 mg/dL, casual blood glucose ≥ 200 mg/dL, hemoglobin A1c ≥ 6.5%, or use of antidiabetic drugs according to the Japan Diabetes Society . Dyslipidemia was defined by LDL-cholesterol ≥ 140 mg/dL, HDL-cholesterol < 40 mg/dL, triglycerides ≥ 150 mg/dL, or use of dyslipidemia drugs according to the Japan Atherosclerosis Society . Finally, atrial fibrillation was determined with a self-administered questionnaire with yes/no answers.
Current smoking behavior was determined based on a self-administered questionnaire and the participants were classified in accordance with their yes/no responses.
Data were collected on the following variables through self-administered questionnaires at the baseline survey: educational level (< 10 years [junior high school or less], 10-12 years [high school], or > 12 years [university or higher]), frequency of going outdoors (< 1 time/week, 1 or 2 times per week, 3 or 4 times per week, 5 or 6 times per week, and every day). The participants were asked whether they had used LTC services with a mail questionnaire. Finally, residential area was collected based on the residential registries and classified to urban or rural areas.
Assessment of cognitive functioning
The participants’ cognitive function using the MoCA-J  was performed by trained psychologists. The MoCA-J total scores (0-30 points) were used for cognitive function assessment. A higher score indicated a higher cognitive function. Generally, the MoCA-J demonstrates greater reliability and validity in the screening of MCI in community-dwelling older people than conventional cognitive tests.
The MoCA-J scores at the 3-year follow-up subtracted from the scores at the baseline were used to define changes in the MoCA-J scores. Therefore, the participants whose MoCA-J scores decreased by ≥ 2 points were defined as those with cognitive decline, while the participants whose scores decreased by < 2 points were defined as those with maintained cognition [33, 34].
After the computation of summary statistics, the Pearson’s Chi-square or Fisher’s exact test for categorical variables and the independent t-test for continuous variables were employed to compare baseline characteristics between stroke and non-stroke, maintained cognition and cognitive decline groups (based on changes in MoCA-J scores), and follow-up and dropped-out groups. Cognitive decline (MoCA-J scores deceased by ≥ 2 points) was considered the outcome variable.
Logistic regression analysis was used to determine the association, expressed as an odds ratio (OR) and 95% confidence interval (CI), between risk factors and cognitive decline. Univariate logistic regression was tested for age, sex, and MoCA-J score at the baseline. In addition, multiple logistic regression was implemented in model 1 with each variable being adjusted by age, sex, and MoCA-J score at the baseline and model 2 was adjusted by all variables (age, sex, MoCA-J score at the baseline, history of stroke, hypertension, diabetes mellitus, dyslipidemia, atrial fibrillation, current smoking, educational level, frequency of going outdoors, LTC service used, and residential areas). These statistical analyses were carried out with SPSS Statistics 24.0 (IBM Japan, Tokyo, Japan). Significance was set at .05.