This study employed a cross-sectional design and a self-administered questionnaire without identifiable markings.

The participants were older adults living in a town with a population of 27,000, located in a rural area with a 40% aging rate. They were members of a community-based organization supported by national and local governments to promote enjoyable activities for older adults (Japan Federation of Senior Citizens Clubs, n.d.). The reason for selecting a single municipality was to avoid bias in public support for older adults. Participants were divided into two groups for exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). EFA required at least 10 participants per item (Carneiro, 2003) and CFA 200 participants (Anderson and Gerbing, 1984). Therefore, over 370 participants were required for EFA and CFA in the current study. The response rate we expected for the questionnaires was ~45% based on previous studies (Ozone et al., 2022). Furthermore, as some older adults who use long-term care insurance have cognitive decline (Ministry of Health, 2002), we excluded older adults with this type of insurance from the analysis, as their responses may not have been valid. In Japan, the certification rate for long-term care insurance is 18.4% (Cabinet Office, 2022). Based on the above, the questionnaire needed to be distributed to over 995 older adults; it was ultimately distributed to 1,614 older adults.

Following distribution, the questionnaires were collected by the club's senior officers. Participants were asked to hand in their sealed questionnaires to a senior club officer. Data were collected from November 2022 to January 2023.

The survey items included questions on basic characteristics, original draft of the DASFWB, and three different scales used to assess criterion-related validity.

IBM SPSS Statistics standard Grand Pack 28.0 and IBM SPSS Amos version 29.0 were used for the data analysis.

This study involving human participants was approved by the ethics committee of epidemiology research at Hiroshima University (approval number E2022-0020). Participants were informed of the purpose and methods of the study and told that they were free to participate or withdraw from the study at any point without penalty or consequences. Furthermore, individuals would be unable to be identified based on their data. The participants provided their written informed consent to participate in this study. This study was conducted in accordance with the ethical standards established by the Declaration of Helsinki.

Of the 1,614 questionnaires distributed, responses from 815 individuals were analyzed (valid response rate: 50.5%). Participant characteristics are presented in Table 1. This study included 382 male (46.9%) and 433 female (53.1%) participants. The age group with the highest number of participants was 75–85 years, comprising 453 individuals (55.6%). Of the participants, 150 (18.5%) were living alone, 663 (81.5%) were living together, 668 (82.2%) reported having an illness under treatment, and 160 (19.6%) had subjective symptoms of dementia. No differences were found between the 408 EFA and 407 CFA participants.

In this study, three out of the 17 total items exhibited ceiling effects (item 3: 4.11 ± 1.016, item 16: 4.450 ± 0.819, item 17: 4.020 ± 0.984). No floor effect items were present. The ceiling effect item was retained after careful consideration of its content by the experts. No item exhibited r > 0.75 in the inter-item correlations, and the I–T correlations with r < 0.30 (item 4: r = 0.249, item 11: r = 0.043) were deleted (Table 2).

EFA was performed using maximum likelihood and Promax rotation, while the KMO and Bartlett's sphericity tests were conducted prior to EFA to determine the factor analysis' goodness of fit; the KMO value was 0.885, and Bartlett's sphericity test was p < 0.001, while EFA implementation was reasonable. The number of factors was extracted by adopting a three-factor structure that exhibited more than 1 in the eigenvalue analysis of the correlation matrix. Items with commonality < 0.20 or less, and factor loadings < 0.30, were deleted, resulting in a three-factor, 12-item scale (Table 3). The items deleted were item 1, “I won't feel depressed if I have dementia,” item 2, “I believe I'm still valuable even if I have dementia,” and item 6, “I think my family will assist me even if I forget things due to dementia.”

Factor 1 comprised five items denoting “willingness to engage with society,” as it revealed a distinct way of living and interacting with the community after dementia onset. Factor 2 comprised four items and was labeled “dispelling anxiety about dementia,” owing to its emphasis on accepting the disease and avoiding concerns about failure. Factor 3 comprised three items; it was named “actions to take to cope with dementia,” since it indicated efforts to prevent dementia and a willingness to promote health, such as seeking treatment and consultation, when symptoms of dementia were perceived. The Cronbach's alpha coefficients for the scale and individual factors were 0.843 for Factor 1, 0.723 for Factor 2, 0.644 for Factor 3, and 0.857 for the overall scale. The coefficient correlations ranged from ρ = 0.310–0.526, indicating a moderate to weak correlation.

CFA was conducted to test the goodness of fit of the three-factor structures obtained in the EFA. The results indicate that the 12 items of the three factors demonstrated a favorable goodness of fit, with χ² = 146.574, df = 51, p < 0.001, Tucker-Lewis index = 0.909, comparative of fit index = 0.930, goodness of fit index = 0.945, and root mean square error of approximation = 0.068 (Figure 1).

Convergent validity was assessed by analyzing the AVE and CR of the three factors. The AVE values for Factors 1, 2, and 3 were 0.49, 0.46, and 0.53, respectively, while the CR values were 0.82, 0.76, and 0.77, respectively. For Factors 1 and 2, the AVE ranged from 0.36 to 0.5, which was slightly below the criterion of 0.5. However, CR was over 0.7 for all factors (Table 4).

Regarding discriminant validity, the square root of the AVE for all factors was greater than their corresponding correlation coefficients (Table 4).

To evaluate the criterion-related validity, we calculated the correlation coefficients between the newly created scale and the PGC-MS, LSNS-6, and UCLA-Loneliness (Table 4). The correlation with PGC-MS was weak for the entire scale, ρ = 0.341 (p < 0.001), and for Factors 1–3, ranging from ρ = 0.232 to 0.290 (p < 0.001). For the LSNS-6, the overall score was ρ = 0.191 (p < 0.001), with Factors 1 and 3 having correlation coefficients of ρ = 0.183 (p < 0.001) and ρ = 0.244 (p < 0.001), respectively; Factor 2, however, was uncorrelated, with a coefficient of ρ = 0.055. UCLA-Loneliness had correlation coefficients of ρ = −0.245 (p < 0.001) for the entire scale, ρ = −0.250 (p < 0.001) for Factor 1, ρ = −0.127 (p < 0.05) for Factor 2, and ρ = −0.168 (p < 0.001) for Factor 3.

The feelings of older adults with dementia used in the draft scale comprised four categories, whereas the DASFWB consisted of three factors and 12 items identified through the EFA. The DASFWB consists of three factors: “willingness to engage in society,” “dispelling anxiety about dementia,” and “actions to take to cope with dementia.” All items assume that a person without dementia has developed it. In addition, the focus on well-being is expected to foster positive attitudes toward dementia and help develop an inclusive society.

Factor 1, “willingness to engage in society,” which comprised one category of the conceptual construct, namely the “desire to live authentically in a familiar community,” and one item, the “emotion of hope and despair toward their family.” This factor represents an aspiration to reside in the community and interact with other community members. The “willingness to engage in society” factor in this scale reflects the view that social interactions play an important role in well-being. A sense of connection and belonging in a society where people's dignity is respected and are encouraged to engage in social networks is essential for their well-being and quality of life (Wiersma and Denton, 2016). However, the stigma associated with dementia can lead to social exclusion, abuse, and discrimination (WHO, 2021b), negatively affecting the well-being of people with dementia. Therefore, maintaining connections with the society is considered an important factor for well-being.

Factor 2, “dispelling anxiety about dementia,” includes an item in the construct category “anxiety about socializing.” This item is related to psychological well-being, such as being able to live one's life without worrying about what others think of one's dementia and being able to accept the illness. Despite the ongoing global efforts aimed at eliminating stigma toward people with dementia, inadequate awareness of dementia continues to subject those with dementia and their caregivers and families to prejudice and discrimination in their communities (WHO, 2021b). The stigma associated with dementia is negatively correlated with quality of life (Lion et al., 2020) and can lead to psychological distress, such as depression and anxiety (Sibley et al., 2021). Such psychological burdens can lead to a decline in well-being; “Dispelling anxiety about dementia” becomes an important factor for the well-being of individuals who hypothetically develop the condition.

Factor 3, “actions to take to cope with dementia,” assesses whether respondents can take actions on their own to prevent dementia and adopt countermeasures when they observe a decline in cognitive function. As everyone is at risk of developing dementia, it is vital to take measures to prevent and manage cognitive decline (Lisko et al., 2021). However, according to the report, ~20% of people with dementia wish to keep their diagnosis a secret (Alzheimer's Disease International, 2019). Shame may serve as an underlying mechanism through which stigma is enacted and perpetuated, resulting in delays in accessing diagnosis and support services (Lopez et al., 2020). The willingness to seek medical advice or consultation without hesitation when suspecting the onset of dementia is an important factor for well-being.

The reliability and validity of the DASFWB were tested using items, exploratory and confirmatory factors, constructs, and criterion-related analyses. After analyzing the items and conducting EFA, we identified a structure comprising 12 items and three factors from the draft of 17 items. Cronbach's alpha coefficients exceeded 0.7 overall for Factor 1, “willingness to engage in society,” and Factor 2, “dispelling anxiety about dementia.” However, Factor 3, “actions to take to cope with dementia,” yielded a lower alpha coefficient of 0.644, falling short of the established criterion of 0.7. Although an alpha coefficient of 0.60–0.74 is still deemed clinically significant (Cicchetti, 1994), the scale's reliability remained within an acceptable range.

CFA revealed that the model fit met statistical standards. The convergent validity results indicated that all factors had a CR > 0.7. Regarding Factor 1, “willingness to engage in society,” and Factor 2, “dispelling anxiety about dementia,” the AVE values were slightly below 0.5 but above 0.38. The square roots of the above AVE values were greater than the correlation coefficients, demonstrating good discriminant validity (Shrestha, 2021). Therefore, convergent and discriminant validity were considered to be acceptable.

Criterion-related validity demonstrated correlations between the newly developed scale and existing measures from the PGC-MG, LSNS-6, and UCLA-Loneliness. The LSNS-6 correlations indicated that an increased social network was associated with greater dementia-related well-being. Conversely, the UCLA-Loneliness correlations showed that increased loneliness was associated with decreased dementia-related well-being. However, only the subscale “dispelling anxiety about dementia” did not exhibit a correlation with LSNS-6. This may be because network size alone cannot explain psychological factors (Kino et al., 2023), as isolation does not necessarily affect physical or mental health. Based on these results, we consider that criterion-related validity is confirmed.