This study used the STROBE-Nut as a reporting guideline (Additional file 1).
This cross-sectional study used baseline data from the Keeping Active across Generations Uniting the Youth and the Aged (KAGUYA) study. The KAGUYA study is a 5-year project to promote the integration of healthcare data and conduct “training courses for resident leaders in frailty prevention, health checks, and exercise guidance by a health support student team, and human resource development at dementia cafes” in collaboration with the town of Koryo, Nara Prefecture, and Kio University. The purpose of this project is to examine the impact of social capital on the health of residents and to use this information in constructing an effective community-based integrated care system. Although other articles have also assessed multiple health status such as functional capacity [20–22], the examination of the association with eating behavior is original to this study.
The baseline survey of the KAGUYA study was conducted in March 2016, using a self-administered questionnaire via a mail survey. The participants were all 8004 residents aged 65 years or older living in Koryo town. The town, located 25 km from Osaka, has a population of more than 30,000 residents in an area of 16 km2 comprising rural and new residential areas. To improve response rates, a combination thank you and reminder postcard was mailed to all of the participants. The questionnaire was returned by 3871 respondents (48.3% response rate).
The eligibility criteria for this study were age 65 years or older, living at home, and regularly consuming three meals daily. Accordingly, three groups of respondents were excluded from our analysis: (1) those who did not answer questions about living situations, number of people living together, whether they ate each meal, number of people eating together, and main basic characteristics (gender and age); (2) those living in nursing homes or hospitals; and (3) those who did not usually eat one or more of the three daily meals (breakfast, lunch, or dinner). We excluded those who skipped meals because the absence of eating precludes examination of effects of eating alone or with others, which is the main target of this study.
Informed consent was obtained using the documents enclosed with the questionnaire. The protocol for the KAGUYA study was approved by the Research Ethics Committee of Kio University (approval number: H27–34).
Cohabitation situation was assessed by asking how many people lived with the respondent (cohabitation). A response of zero was classified as living alone, and one or more as living with others. The survey also asked whether they usually ate three separate meals each day (breakfast, lunch, and dinner). Those who responded that they eat each meal were then asked how many people, including themselves, were present at each meal. Responses of one person were classified as eating alone, whereas responses of more than one person were classified as eating together. The KAGUYA study asked respondents about a large number of lifestyle habits that are expected to be related to health status, so we used brief binary self-reports similar to those in previous studies to assess whether respondents ate alone [18, 23].
Health status was examined by assessing subjective health, medical history, care needs, BMI, depression, and functional capacity.
Subjective health was classified as in good health for those who answered “very good,” “good,” or “not bad” to the question “How is your current health condition?” and not in good health for those who answered “not good” or “bad.” The same classification system was used in the Comprehensive Survey of Living Conditions of the Ministry of Health, Labour and Welfare, Japan .
To assess medical history of hypertension, diabetes, cardiovascular disease, stroke, osteoporosis, rheumatoid arthritis, and dyslipidemia (high cholesterol, high triglycerides, etc.), the respondents were asked, “Have you ever been diagnosed by a doctor with any of the following diseases?”
Care needs were identified according to whether the respondents had been certified as requiring long-term care or support through the long-term care insurance system of Japan .
BMI was determined by asking height (cm) and weight (kg) as integers, and dividing weight (kg) by the square of height (m).
Depression was assessed using the Japanese version  of the 5-item Geriatric Depression Scale (GDS 5) , a shortened version of the 15- and 30-item GDS [28, 29]. The GDS 5 consists of five items such as “Are you basically satisfied with your life?” The presence of depressive symptoms was defined as two or more positive answers to the depression screening questions. Diagnosis with GDS 5 has been proven to be significantly consistent with a clinical diagnosis of depression, with good interrater and test-retest reliability .
Functional capacity was assessed using the Tokyo Metropolitan Institute of Gerontology Index of Competence (TMIG-IC) . TMIG-IC is a 13-item index of competence comprising three subscales: Instrumental Self-Maintenance, such as “Can you use public transportation (bus or train) by yourself?” (5 items); Intellectual Activity, such as “Are you able to fill out forms for your pension?” (4 items); and Social Roles, such as “Do you visit the homes of friends?” (4 items). TMIG-IC measurements have high reliability according to alpha coefficients, test-retest, and correlation between subscale and total scores, as well as high construct, discriminant, and predictive validity .
High levels of functional capacity are required due to changes in the living environment and improved competence of the elderly over the past quarter century, so the Japan Science and Technology Agency Index of Competence (JST-IC) was also used . JST-IC is a 16-item index used to assess high-level competency, and consists of 4 subscales (each with four items): Social Engagement, such as “Do you participate in regional festivals or events?”; Technology Usage, such as “Can you use a mobile phone?”; Information Practice, such as “Are you interested in news and events from overseas?”; and Life Management, such as “Do you follow any measures to prevent yourself from becoming a victim of crime?” JST-IC evaluations showed moderate correlations between the size of social networks and the level of subjective health well-being, and strong correlations between TMIG-IC score, physical fitness, and health literacy . For both the TMIG-IC and JST-IC, each item is scored as 1 point for “Yes” and 0 points for “No.” Higher scores therefore indicate higher functional capabilities.
The questionnaire included socio-demographic characteristics such as gender, age (years), education history (9 years or less, 10–12 years, 13 years or more), self-assessed living conditions (difficult, not difficult), area of residence (former village areas, new residential areas), and employment status (unemployed, employed). The self-assessed living condition variable was classified as difficult for those who answered “very difficult” or “difficult” and not difficult for those who answered “normal,” “somewhat comfortable,” or “very comfortable” as in the Comprehensive Survey of Living Conditions of the Ministry of Health, Labour and Welfare, Japan . For employment status, respondents who answered “not working (including those without regular income, pensioners, and students)” were classified as unemployed, and those who indicated “farmer,” “self-employed (self-run stores, family employment, etc.),” “working (regular),” or “working (non-regular: part-time workers, side work, etc.)” were classified as employed.