Methodology
Various methods have been used in Japan to clarify the experiences of community-dwelling older people in the real world. Repeated access to community residents is often obtained using epidemiological surveys. In such surveys, researchers re-access baseline survey respondents. However, it is often difficult to access people who are no longer living in a particular community (i.e., those who have moved to geriatric institutions or died). Attempts to access hospitals or community support centers listed in a baseline survey are often unproductive; the staff of such institutions cannot provide information because of privacy protection. As a result, conventional social surveys are of limited use in determining real-world outcomes for community-dwelling older people at risk for discontinuation of community living.
To overcome these limitations, we used a community-based participatory research (CBPR) framework.9 In addition to using conventional mailing and telephone contact to re-access residents, we created a base camp for our research that provided a comfortable place for community residents to spend time, and in which researchers and community workers could collaborate with community residents. Although this center welcomed all residents regardless of age or address, participants from the original study were repeatedly informed about the center. In summary, we acted as both community-service providers and researchers to build trust with stakeholders. We held monthly case conferences with the comprehensive support centers from the catchment area, which enabled us to gather information about our study participants.
Because this was a CBPR study, the sample size was limited as we had to maintain close relationships with the participants. Because our main focus was continuation (or not) of living in the community, in the analysis, we regarded the outcome of discontinuation of living in the community as one variable (i.e., as all-cause discontinuation, which comprised death, institutionalization, hospitalization, and moving away).
Japanese context and setting
In Japan, modernization has led to smaller family sizes, which has resulted in more older people living by themselves. In addition, weakening social ties mean that older people tend to be isolated in the community.
Our study was conducted in Takashimadaira, which is located in the northwest area of metropolitan Tokyo. Takashimadaira contains the largest housing complex district in Japan, which was built during the 1970s. The aging rate (i.e., the percentage of the total population aged ≥ 65 years, a widely used indicator of aging in Japan) of this area is approximately 40%. This rate is the same as the predicted aging rate for 2055, which is when societal aging is expected to plateau. We chose this area because it is considered to resemble the Japanese society of the future. An administrative corporation currently manages the housing complex. Because new residents are not required to have a guarantor (guarantors are customary in Japanese business) many new residents are older people who do not have relatives on whom they can rely for financial support.
Participants
Before the study, we conducted a three-step survey of all community residents who were aged 70 years or over in 2016. Figure 1 shows the participant selection flow. Briefly, in the first step, questionnaires were sent to 7,614 residents, and 5,430 responses were received. In the second step, 2,020 residents completed face-to-face surveys in the community center. This assessment included the MMSE. Those with an MMSE score < 24, which is a commonly used cutoff criterion, were potential participants in the subsequent survey. In the third step, a research team that comprised a certified psychiatrist and a gerontologist or a public health nurse made home visits to 198 participants with cognitive impairment. In 2019, 3 years after the baseline survey, we re-accessed these 198 people with cognitive impairment (80 men and 118 women).
Measurement
The main outcome was participants’ current status (i.e., community living, moved to another community, hospitalized, institutionalized, deceased, or unknown).
Covariates
The following factors were examined in the baseline survey:
(1) Sociodemographic variables
We used questionnaires to obtain information about participants’ sociodemographic characteristics. Information about long-term care insurance for each participant was obtained from the government.
(2) Cognitive assessment
MMSE10,11 assessments were conducted by a psychologist or a public health nurse under the supervision of a psychologist.
(3) Psychological assessment
Depressive symptoms were assessed using the 15-item Geriatric Depression Scale (GDS-15).12 The total score ranges from 0 to 15, and scores above 5 are considered to indicate the presence of depressive symptoms. We also assessed participants’ mental well-being using the simplified Japanese version of the World Health Organization (WHO)-Five Well-being Index (S-WHO-5-J). 13
(4) Physical health-related assessment
Self-perceived health was measured using a four-point Likert-type scale. Answers of “very good” or “good” were recoded as “good” and other answers as “not good.”
Frailty was assessed using the Kihon Checklist (KCL),14 which was developed by the Japanese Ministry of Health, Labour and Welfare. Satake et al.15 found that total KCL scores were closely correlated with frailty, as defined in the Cardiovascular Health Study criteria. In this study, we used total KCL score cutoffs of 7/8 and 3/4 to identify frailty and potential frailty, respectively.
(5) Sociological variables
(i) Relationship with the community
We assessed community participation using eight items on attendance of neighborhood associations, social clubs, sport clubs, volunteer clubs, senior clubs, alumni associations, occupational associations, and other social groups. Individuals who answered “no” to all items were considered as not showing any community participation. We also assessed trust among neighbors using the item “Do you trust your neighbors?” which was developed for this study. Responses were on a five-point Likert-type scale. Those who answered “disagree” or “strongly disagree” were regarded as “lack of trust in neighbors”.
(ii) Socioeconomic status
We asked participants about their perceived current socioeconomic status using a five-point Likert-type scale. Individuals who answered “somewhat poor” and “poor” were regarded as having financial disadvantage. We also asked participants to report their annual income range; those who reported < 1,000,000 yen (equivalent to 9,200 USD at a currency rate of 108.12 yen/dollar) were regarded as having a low income. We used this threshold because the average disposable income for older people in Japan is 2,100,000 yen.16
(6) Dementia diagnosis judged by geriatric psychiatric specialists at participants’ homes
The visiting geriatric psychiatric specialist diagnosed participants using the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-5).17 The final diagnosis was made in an interdisciplinary research meeting in the community center that included more than two certified psychiatrists. We also used the Clinical Dementia Rating (CDR) scale, which is a widely used measure.18 Using information obtained from participants, family members (if accessible), and prescriptions (if possible), we also clarified whether participants had received a diagnosis of dementia in a clinical setting. If so, we attempted to obtain details about the neurological basis of the diagnosis.
(7) Need for social support
Participants’ need for social support at home was evaluated on nine domains:8 1) dementia subtype diagnosis, 2) medical check-up for physical conditions, 3) continuous medical care, 4) daily living support, 5) support for family members, 6) housing support, 7) long-term care insurance, 8) financial support, and 9) rights protection (see supplementary file 1). This evaluation was made by two or more visiting experts.
Statistical analysis
There are various reasons for discontinuation of community living because every person has a unique background. As the ultimate aim of our CBPR approach was to develop an inclusive community in which older people can continue to live, we compared descriptive characteristics of participants who continued community living with those who discontinued community living for any reason (moved to another community, hospitalized, institutionalized, and deceased were combined as all-cause discontinuation in the main analysis). Participants with unknown outcomes were excluded from the analysis. T-tests were used for continuous variables and chi-square tests were used for nominal variables. Next, we performed a multiple logistic regression analysis that included factors significant in the bivariate analyses (threshold set at a p-value < 0.05). To avoid multicollinearity, CDR score was not included in the multiple logistic regression analysis because it was clinically obvious that it correlated with dementia diagnosis. We then confirmed that the variance inflation factor was less than 2 for all the independent variables included in the multiple logistic regression analysis. Continuous variables that did not have a cutoff criterion (i.e., MMSE, S-WHO-5-J) were converted into two-item variables and divided into two groups with cutoff points based on the average score. That is, because the average MMSE score was 20.1, the cutoff point was set at 20/21. Similarly, the average S-WHO-5-J score was 8.8, and the cutoff point was set at 8/9. A p-value < 0.05 was regarded as statistically significant.