Participants were elderly individuals aged 65 or older who lived in Chiba Prefecture adjacent to Tokyo.

The survey was conducted from October to December 2015.

The questionnaire was distributed to participants via presidents of district senior citizen clubs. Participants were asked to fill out the questionnaire and return it using a return envelope.

The questionnaire assessed participants’ basic attributes, state of health, HRQOL, social support network, and spirituality.

Basic attributes

For basic attributes of participants, we asked about age, gender, household structure, job, years of residence, economic affluence, education level, and religion. Regarding job, we asked whether the respondent had a job with income. Years of residence concerned the number of years the respondent had lived at the current address, and religion asked whether the respondent had a religion and, if so, what type it was. As for economic affluence, we asked about it with a 4-point scale, ranging from “Not really affluent” to “Quite affluent.”

State of health

For state of health, we asked whether the respondent was eligible for long- term care, had any disease under treatment, and had any history of present illness or subjective symptoms. For history of present illness, we asked about common diseases among elderly people such as heart disease, stroke, cancer, gastrointestinal diseases, and osteoarthritis. For symptoms, we asked about lower back pain, headache, knee pain, difficulty seeing, difficulty hearing, and so on. These diseases have been chosen in previous studies as well [11] [19] .

Health-related quality of life

To minimize the burden on participants, we used the abbreviated version of the standard MOS 36-Item Short Form Health Survey, the MOS 8-Item Short Form Health Survey Japanese version (SF-8) [20] to evaluate HRQOL. The SF-8 has eight items, which include physical functioning, physical role, bodily pain, general health perception, vitality, social functioning, role emotional, and mental health. These items are separated into two components, physical component summary (PCS) and mental component summary (MCS) scores (SF-36.org, 2014).

Social support network

To evaluate social support networks, we used the Japanese version of the Abbreviated Lubben Social Network Scale (LSNS-6) [21] . The LSNS-6 measures the sizes of one’s family network and friend network with the number of individuals included in each network. The summed score ranges from 0 to 30 points, and less than 12 points (less than 6 points for subscales) indicates social isolation.

Spirituality

For measuring spirituality, we used the Spirituality Rating Scale Related to Health in the Elderly (SP Health Scale) by Takeda et al. [15] . They defined spirituality as “something inherent in all human beings and, while it tends to surface when one faces the crisis of life, it also exists in ordinary daily life and is involved in enhancing the sense of happiness with the quality and spiritual aspects of life,” and developed a scale to measure spirituality among the elderly by considering the socio-cultural background of Japan. Its reliability and validity have been demonstrated. This 5-point scale has 18 items and consists of six subscales: meaning and purpose of living, attitudes toward death and dying, self- transcendence, accord with others, spiritual support, and harmony with nature (Table 1). The item scores are simply summed as subscale scores to create an overall score. Higher scores indicate higher spirituality.

The difference in SF-8 score (PCS and MCS) was compared by category for each survey item using a t-test or one-way ANOVA. Continuous variables were divided into “high” and “low” values based on the median, such that age was divided into two at 75, and LSNS-6 was divided into two at 12 points (6 points for subscales). This was done to divide elderly participants into old-old vs. young-old populations as well as not socially isolated vs. socially isolated. In addition, a multiple comparison test was performed using the Tukey method when the ANOVA was significant.

Next, aiming to understand the extent of each variable’s effect, regression analyses were performed with the forced entry option by setting the basic attributes, state of health, LSNS-6, and SP Health Scale as independent variables and PCS and MCS as dependent variables. Lastly, aiming to further investigate the relation between spirituality and health-related quality of life, partial correlation analysis was conducted with controlling possible confounding factors (age, gender, household structure, economic affluence, job with income, years of residence, education level, and religion).

Statistical analysis was performed using IBM SPSS Statistics Ver. 20 with the significance level set at less than 0.05.

The survey was carried out after receiving approval from the Ethics Review Committees of the University of Human Arts and Sciences (No. 475) and Kameda College of Health Sciences (No. 2015 A 008).

The mean age of participants was 77.5 ± 5.8 years, and there were 291 men (45.7%) and 346 women (54.3%). Examining the participants by age group, there were 206 (32.3%) former-stage elderly respondents aged between 65 and 74 and 425 (66.7%) latter-stage elderly respondents aged 75 or older. There were 189 (29.7%) respondents who had a job with income. As for subjective sense of economic affluence, 43 (6.9%) of the respondents said they were “quite affluent” while 300 (48.0%) said they were “slightly affluent.” Regarding religion, 255 (40.0%) reported having a religion, and among them, 226 were Buddhists (35.5%). As for state of health, 593 (93.1%) said they had medical conditions, and the mean number of diseases was 2.3 ± 1.4 per person. See Table 2 and Table 3 for details.

The mean PCS and MCS were 45.2 ± 7.7 and 51.4 ± 5.9, respectively. The mean score for the 18 items on the SP Health Scale was 67.2 ± 8.0. The mean score of the LSNS-6 was 17.4 ± 5.5, and 13.7% had a score less than 12, which indicates social isolation.

The mean HRQOL scores for all respondents were 45.2 ± 7.7 (PCS) and 51.4 ±

a: As most Japanese have a pension, they can live without a job.

5.9 (MCS). In addition, when t-tests and one-way ANOVAs were used to compare the HRQOL scores by each independent variable, significant differences were found in the following basic attributes: age, economic affluence, job, years of residence, and education level. First, in terms of age, PCS was significantly higher among those aged between 65 and 74 (p < 0.05). In terms of economic affluence, both PCS and MCS were higher among those who were affluent (PCS: p < 0.01; MCS: p < 0.001). In addition, the respondents with jobs had a higher MCS (p < 0.05), those who had lived for less time at the current address had a higher PCS (p < 0.05), and high school and college graduates had a higher PCS than did elementary school and middle school graduates (p < 0.001). See Table 4 for details.

Next, in terms of state of health, both PCS and MCS were significantly lower among those who had some kind of disease (PCS: p < 0.001; MCS: p < 0.001), symptoms of lower back pain (PCS: p < 0.001; MCS: p < 0.01), headache (PCS: p < 0.001; MCS: p < 0.05), and difficulty seeing (PCS: p < 0.05; MCS: p < 0.05). In addition, those with a heart disease (p < 0.05), cancer (p < 0.05), gastrointestinal disease (p < 0.001), osteoarthritis (p < 0.001), knee pain (p < 0.001), difficulty hearing (p < 0.01), and those who were eligible for long-term care (p < 0.01) had a significantly lower PCS but not MCS. See Table 5 for details.

In terms of the SP Health Scale, when the respondents were divided into two groups of “high” and “low” (high and low spirituality groups) based on the me-

PCS: physical component summary; MCS: mental component summary. An F-test was used for economic affluence and education level (multiple comparison tests were done using the Bonferroni method), and a t-test was used for the rest. a. Not affluent < Slightly affluent**; b. Not affluent < Slightly affluent*, Not affluent < Quite affluent**, Not really affluent < Slightly affluent**, Not really affluent < Quite affluent**; c. Elementary and middle school < High school**, Elementary and middle school

PCS: physical component summary; MCS: mental component summary. *: p < 0.05, **: p < 0.01, ***: p < 0.001.

dian of each subscale. The “high” group had a significantly higher PCS and MCS for meaning and purpose of living (PCS: p < 0.01; MCS: p < 0.001) and spiritual support (PCS: p < 0.05; MCS: p < 0.001). In addition, MCS was higher for the “high” group for accord with others (p < 0.05). See Table 6 for details.

Finally, with the LSNS-6, when the respondents were divided into “high” and “low” groups (high = not socially isolated, low = socially isolated) at a cut-off point of 12 for the overall score and 6 for the subscale scores, PCS was significantly higher for the socially isolated group in terms of family (p < 0.05), and MCS was significantly higher for the not socially isolated group in terms of friends (p < 0.01). See Table 6 for details.

Using basic attributes, state of health, SP Health Scale, and LSNS-6 as independent variables and SF-8 scores (PCS and MCS) as the dependent variables, mul-

PCS: physical component summary; MCS: mental component summary; LSNS-6: Abbreviated Lubben Social Network Scale; SP: Spirituality Rating Scale Related to Health in the Elderly. *: p < 0.05, **: p < 0.01, ***: p < 0.001.

tiple regression analyses were performed with the forced entry option.

The factors that increased PCS were economic affluence (β = 0.094, p < 0.05) and meaning and purpose of living on the SP Health Scale (β = 0.165, p < 0.01). The factors that reduced PCS were age (β = −0.206, p < 0.001), heart disease (β = −0.139, p < 0.01), gastrointestinal disease (β = −0.096, p < 0.05), osteoarthritis (β = −0.180, p < 0.001), knee pain (β = −0.209, p < 0.001), lower back pain (β = −0.188, p < 0.001), and eligibility for long-term care (β = −0.145, p < 0.01). The only factor that increased MCS was economic affluence (β = 0.152, p < 0.01), and the factors that decreased MCS were cancer (β = −0.104, p < 0.05), headache (β = −0.139, p < 0.01), and self-transcendence (β = −0.148, p < 0.05). As the variance inflation factor ranged between 1.055 and 2.239, it was deemed that there was no multicollinearity among the independent variables. See Table 7 for details.

In order to investigate the relation between spirituality and health-related quality

PCS: physical component summary; MCS: mental component summary; LSNS-6: Abbreviated Lubben Social Network Scale; SP: Spirituality Rating Scale Related to Health in the Elderly. a. Male = 0, Female = 1; b. Living alone = 0, Not living alone = 1; c. Not affluent = 1, Not really affluent = 2, Slightly affluent = 3, Quite affluent = 4; d. No = 0, Yes = 1; e. Elementary and middle school = 1, High school = 2, College = 3.

PCS: physical component summary; MCS: mental component summary; Controlled variables: age, gender, household structure, economic affluence, job with income, years of residence, education level, religion.

of life, partial correlation analysis was conducted with controlling basic attributes (age, gender, household structure, economic affluence, job with income, years of residence, education level, and religion).

The result showed that PCS was positively associated with most of SP subscales (r = 0.137 - 0.224) while MCS was positively associated with only SP (Meaning and purpose of living) and SP (Spiritual support) (r = 0.141 - 0.153) (Table 8).

The standard values of PCS and MCS are 46.25 ± 7.10 and 51.50 ± 5.75, respectively, for Japanese individuals aged 70 to 79 [22] . As the mean age of the participants in this study was 77.5, and PCS and MCS were 45.16 ± 7.73 and 51.40 ± 5.91, respectively, there was no substantial difference. Participants had a normal level of physical and mental wellness, even though 93.1% were being treated for multiple diseases.

The mean score for the 18 items on the SP Health Scale was 67.2 ± 8.0. This was not much different from the results of a survey conducted on elderly people living in the East Asian region including Japan, Korea, and China by Takeda [23] (68.4 ± 6.6).

The mean score of the LSNS-6 for the participants in this study was 17.4 ± 5.5, and 13.7% had a score below 12, which indicates social isolation. As the average score was 16.2 ± 5.1, with 19.4% scoring below 12, in a previous study by Kurimoto et al. [21] on community-dwelling elderly individuals in Japan, the participants in this study had slightly larger social support networks.

The above results indicate almost the same scores for HRQOL, SP Health Scale, and LSNS-6 as in previous studies. Therefore, the participants in this study can be regarded as representative of the community-dwelling elderly population of Japan.

Those who were younger than 65, more economically affluent, lived shorter in the current place, had higher education level, higher SP Health Scale scores (meaning and purpose of living and spiritual support), and less family support had higher PCS scores. In contrast, having a disease under treatment or symptoms such as heart disease, cancer, gastrointestinal disease, osteoarthritis, lower back pain, knee pain, headache, difficulty seeing, difficulty hearing, and eligibility for long-term care was associated with reduced PCS scores.

On the other hand, those who were more economically affluent, had a job, higher SP Health Scale scores (meaning and purpose of living, accord with others, and spiritual support), and more friend support had higher MCS scores. In contrast, having a disease under treatment or symptoms such as lower back pain, headache, and difficulty seeing were associated with lower MCS scores.

The results are consistent with a previous study by Hatayama et al. [5] , who showed that education level and income had positive effects, while chronic disease, pain, difficulty seeing, and difficulty hearing had negative effects on HRQOL. Tsutsumi’s [22] comparison between elderly people with a certain degree of independence and elderly people requiring long-term care also showed that the latter had significantly lower PCS scores. In addition, some previous studies reported that aging was not related to MCS, although it lowered PCS [6] [24] , which was also observed here.

The results of multiple regression analyses showed that the factors predicting PCS scores were basic attributes (age and economic affluence), state of health (heart disease, gastrointestinal disease, osteoarthritis, lower back pain, knee pain, and eligible for long-term care), and SP Health Scale (meaning and purpose of living). Of those, economic affluence and meaning and purpose of living had positive effects, and the rest had negative effects. These results were consistent with the results of the t-tests.

The finding that economic affluence has a positive effect on PCS is consistent with de Belvis’s report [6] indicating that subjective income has a positive effect on PCS. The result was also consistent with reports that those with higher incomes have lower rates of mortality and eligibility for long-term care [25] as well as higher basic and instrumental activities of daily living, social activity, sense of wellness, and less disease [4] , and that those who feel their life circumstances are poor tend to have a lower sense of wellness [26] .

Meaning and purpose of living―a subscale of the SP Health Scale―also had a positive effect on PCS. Considering that these components are fundamental to human existence and that humans with fuller meaning and purpose should be able to lead a life full of vitality, the results appear reasonable.

Meanwhile, the factors that predicted lower PCS were age, heart disease, gastrointestinal disease, osteoarthritis, lower back pain, knee pain, and eligibility for long-term care. These factors likely reduce PCS because they reduce daily life activities, motor functions, and vitality.

On the other hand, factors predicting MCS were basic attributes (economic affluence), state of health (cancer and headache), and SP Health Scale (self- transcendence), and all except economic affluence had a negative effect. Although economic affluence and headache were consistent with the results of the t-tests, cancer and self-transcendence were not.

The finding that economic affluence had a positive effect on MCS is consistent with de Belvis’s finding [6] that subjective income had a positive effect on MCS. This is also consistent with Fujiwara [27] , who found that income and life circumstances affect mental health (Geriatric Depression Scale scores and anxiety about the future).

Headache is closely related to mental health [28] , and cancer is linked directly to life and death; therefore, it is possible that their effects on mental health are large. However, the t-test showed no significant relationship with MCS.

Meanwhile, self-transcendence, a subscale in the SP Health Scale, also had a negative effect on MCS. We can interpret that this result was obtained because mentally exhausted elderly individuals are more likely to think about their connection with ancestors and descendants or feel support from power beyond them. However, this result is contrary to the theory and previous studies that elderly people solve spiritual problems in relation to something that transcends themselves [29] . In addition, as the t-test did not indicate a significant relationship, we cannot conclude that self-transcendence has a negative effect on the mental health of elderly individuals.

Although meaning and purpose of living was not significant, its p-value (0.058) is quite close to 0.05, and the absolute value of its coefficient (0.136) is almost same as that of self-transcendence (−0.148). These imply that if sample size had been large enough, this factor could have a significant effect on MCS.

The result of partial correlation analysis showed that both PCS and MCS were positively related with spirituality, but the former were stronger. The result that SP (Meaning and purpose of living) and SP (Spiritual support) were positively related with both PCS and MCS was consistent with that of t test (Table 6). Moreover, the finding that SP (Meaning and purpose of living) was positively related with PCS was consistent with that of multiple regression analysis (Table 7).

In sum, the results concerning the effect of spirituality suggest that individuals with a higher score on meaning and purpose of living on the SP Health Scale had higher HRQOL. Therefore, to improve HRQOL in elderly individuals, it might be effective to provide support, such as creating opportunities and forums to identify one’s purpose in life and sources of hope, as well as offering methods for recollection and life review to identify the meaning and purpose of one’s life [30] .

In the past, spirituality has been discussed only in settings of intractable diseases and end of life. Therefore, it is a unique contribution of this study to have shown the potential to enhance physical and mental health of community-dwelling elderly individuals by organically integrating aspects of body, mind, and society from the perspective of spirituality [31] .