Factors associated with good self-rated health of non-disabled elderly living alone in Japan: a cross-sectional study

In Japan, population statistics system categorizes living arrangements only according to the number of persons who live together and have the same family name. Thus, a resident will be registered as "living alone" even if actually living with his/her daughter's family or others with different family name. However, even though an individual actually lives alone, a neighbor may sometimes be his/her sibling or offspring. Therefore, we defined a person "living alone" as one who lives in the community by him/herself and has no any sibling, child, or grandchild living within a radius of 50 meters from his/her house. "Non-disabled elderly living alone" refers to elderly persons who live alone, as we defined, can perform activities of daily living (ADL, i.e. walking, eating, using the toilet, bathing and dressing) independently, have no cognitive impairment, and do not use the long-term care insurance that, since April 2000, has been provided for the elderly who meet particular physical or mental disability criteria.

The Welfare Commission is an organization through which the Japanese Ministry of Health, Labour and Welfare provides support for the elderly living alone. Every elderly person living alone is assigned a particular Welfare Commissioner whose responsibility is to offer help whenever a problem arises. According to the data of this organization, there are a total of 5943 residents aged ≥ 65 years that live alone, as we defined, in Takatsuki city, a metropolitan suburb that lies between Osaka city and Kyoto city, Japan. After obtaining the written informed consent about the conduct of the survey, questionnaires were distributed to all these individuals through the Welfare Commissioner in October 2005. We received effective responses from 4465 individuals (response rate 75.1%). Among these respondents, 3187 individuals (600 men and 2587 women) met the criteria for non-disabled elderly living alone, and these individuals became our subjects.

With referring the study conducted in middle-age Japanese [7], SRH was assessed with a single-item measure: "In general, how do you think about your health?" with 4 possible responses: (1) excellent, (2) good, (3) fair, and (4) poor; and responses were dichotomized as "good" or "poor" by combining response 1 with response 2 and response 3 with response 4.

To our knowledge, this study is the first attempt in Japan to investigate SRH among the non-disable elderly living alone. Physical condition, lifestyle factor, psychological status, and social activity were considered to explore the factors associated with SRH. Although socioeconomic status is known to correlate with SRH, the Law on Personal Information Protection implemented in Japan from April 1, 2005, does not permit us to access information on marital status, education, and income. In Japan, however, as a primary benefit of the social welfare program, all elderly persons are provided a pension regardless of whether they worked when they were young. Thus, income may not be a problem for the elderly. Nevertheless, we still use the question "Do you have to restrict meals because of economic problems?" to estimate the income status of our study population.

With respect to physical condition, the effects on SRH of visual ability, hearing ability, chewing ability, chronic disease, and inability to go out have been studied in previous surveys [8‐11]. Therefore, these factors were also assessed in the present study. Given the specific study population, i.e., those living alone, we also included risk of falling to enhance assessment of mobility. Functional status has also been reported to correlate with SRH [9]. Because our study population comprised the non-disabled elderly, we prefer to assess "high-level functional capacity" rather than "basic activities of daily living". Thus, instrumental activities of daily living (IADL) was assessed, as an indicator of "high-level functional capacity" [18]. In addition, because weight loss, especially unintentional weight loss, is common among the elderly and can reflect disease severity or undiagnosed illness, as reviewed by Alibhai SM [19], this was also assessed.

Lifestyle factors such as physical exercise [7, 13] and alcohol consumption [7] have been reported to correlate with SRH of the elderly and thereby were examined in this study. Because walking is a particularly feasible form of physical exercise for the elderly, we selected the activity of having a walk to indicate physical exercise. Also, considering that heavy drinking might be rare or not practical for an elderly person, we did not assessed alcohol consumption on the base of the volume of alcohol intake.

Depression is a major psychological disorder that affects the quality of life of the elderly [20, 21]. Thus, psychological status was assessed by examining depression in our study population.

As for social activity, social support such as having a confidant with whom to talk about the troubles, a person to take the elderly to the hospital, and a person who can help to prepare meals, has been studied among middle-age Japanese [7]. For our study subjects, the Welfare Commissioner is just the caregiver with such responsibility. Rather than social support, social integration and social networks appear to exert more influence on SRH in this population, therefore, these two items were assessed.

A questionnaire pertaining to SRH and physical condition, lifestyle factor, psychological status, and social activity was distributed to the non-disabled elderly living alone. Two-week reliability of the questionnaire was 90.3%.

Physical conditions was assessed on the basis of eight items: 1) going out alone to distant places, 2) risk of falling, 3) visual ability, 4) hearing ability, 5) chewing ability, 6) weight loss, 7) self-rated chronic disease, and 8) IADL. The item "going out alone to distant places" was assessed dichotomously as "can" if the individual reported being able to go out to distant places by him/herself or "cannot" if the individual reported being able to go only around the house, being independent indoors but unable to go out without help, or needing some help even indoors, as our previous report [22]. "Risk of falling" was examined on the basis of seven items: tripping or stumbling, inability to cross a road within the time allotted to the green signal, dizziness or faintness, inability to negotiate indoor obstacles, inability to wring out a towel, use of a cane, and knee joint pain. According to Toba K's report [23], it was dichotomized as "no fall risk" when less than three items were positive and "fall risk" when four or more items were positive. For assessment of "visual ability", participants were asked whether they can read the newspaper clearly with or without reading glasses, and for assessment of "hearing ability", participants were asked whether they need voices to be raised for conversation or the volume to be raised for watching television. "Chewing ability" was examined on the basis of the difficulty of eating different foods, as reported by Ikebe [24]. "Weight loss" was examined by asking, "Has your body weight decreased about 5 kg in the past 6 months?", for which we referred the criteria usually used in nutritional assessment in Japan [25]. "Self-rated chronic disease" was assessed dichotomously as "absence" or "presence" if any disease including hypertension, cardiovascular disease, diabetes, or stoke had been diagnosed. "IADL" was determined using the 5-item Tokyo Metropolitan Institute of Gerontology (TMIG) Index of Competence (i.e., using public transportation, shopping for daily necessities, preparing meals, paying bills, and managing deposits) [18]. A full score of five categorized the study subject as independent and a score of 0–4 categorized the subject as dependent [26].

Psychological status referred to depression. A simple Geriatric Depression Scale (GDS – 5 items) [27] was used to assess depression among the participants. Each item was coded as 0 if negative or 1 if positive. Score ≥ 2 was used to diagnose the presence of depressive symptoms, as reported by Rinaldi P [28].

Social activities comprised two items: 1) social integration, and 2) social networks. "Social integration" was assessed according to whether respondents participated in (a) a welfare center association for the aged, (b) recreational activities at a public hall or community center, or (c) a hobby association. "Social integration" was described as "attend" if the respondent reported participating in (a) or (b) at least weekly or in (c) at least monthly and as "does not attend" if not. "Social networks" was assessed according to the frequency of visual or nonvisual contact (such as by telephone, fax or mail) with relatives, friends, or confidants as ≥ 1–2 times/week [29].

Among all independent variables, items to which over 95% of individuals have the same responses were not included in the data analysis. Income estimation (no economic problem) and alcohol consumption (no heavy drinking) were thereby excluded.

The distribution of SRH in relation to age, sex, and various physical, lifestyle, psychological, and social items was examined by chi-square test. The factors associated with good SRH were identified by multivariate logistic regression with adjustment for age. The multivariate logistic regression analysis was performed in two steps: 1) fit univariate model with each variable. All variables were entered in the model separately with adjustment for age, and any variable that was significant at the 0.25 level was a potential candidate for the multivariate model. 2) Fit multivariate model. All items selected from 1) step were entered in the model. Items with p > 0.05 were eliminated one at a time in the sequence of p value. When an item was eliminated, if the change of any remaining parameter estimate was greater than 20%, this item would remain in the model as a confounder. In this study, all independent variables were found to be significant in 1) step and no confounders were found during item elimination.

Sex-specific associations with SRH were examined by stepwise multivariate logistic regression. The model included all risk factors identified as described above and the interaction terms of sex with each risk factor.

In view of the prominent difference in life expectancy between men and women (78.6 years for men; 85.5 years for women) and the sex-specific results, we performed statistical analyses not only for the total study population but also separately for men and women.

All data was shown as number (N), prevalence (%), odds ratio (OR), and 95% confidence interval (95%CI). SAS for Windows, Ver. 8.2, was used for all statistical analyses.