The effects of social relationships and social support are well documented in a large number of studies. Social networks or social support exchanges in supportive social relationships are related to better health. For example, those with more social networks have a lower risk of mortality [1
], morbidity [2
], functional disabilities [3
], and cognitive impairment [6
]. When people interact, something is exchanged, namely, social support. In other words, social support is embedded in social relationships [1
]. Studies have also reported the association between social support and cognition. Seeman et al. [7
] reported that emotional support was a significant predictor of cognitive functioning at the 7.5-year follow-up, using MacArthur Studies of Successful Aging data. Another 12-year follow-up study [8
] demonstrated that baseline exposure to emotional support was independently associated with a better Mini-Mental State Examination (MMSE) score at follow-up. According to a systematic review of 19 longitudinal studies that investigated the association between social relationship factors and incident dementia, not network size per se but lack of social interaction seems to be associated with incident dementia [6
However, given the nature of social support, cultural or gender factors cannot be ignored when assessing its effect on health. A study in the United States [9
] demonstrated that gender moderates the relationship between social support and cognition; higher levels of emotional support are associated with better cognition only among females. Support sources are also important factors to consider. In general, studies in Western nations have reported that support from a wide variety of sources, including friends or neighbors, can be beneficial for the health of elderly populations [1
]. On the contrary, studies in the East have reported the relative importance of family support. A cross-sectional study in China [10
] demonstrated that family support is the strongest predictor of cognitive function among older persons, whereas support from friends was not. Cheng et al. [11
] suggested, based on their conclusions in another cross-sectional study in China, that interactions with friends are less well-being enhancing in Asian societies compared with their Western counterparts. The authors speculated that, in societies where social harmony and reciprocity are valued, like those across Asia, the cost of seeking help from others, especially from non-family members, might be higher compared with more individualistic societies. Fiori et al. suggested that the relative importance of filial duty in Asian cultures might also explain the result [12
]. Thus, the association between social support and cognition might be different depending on support sources in Asian societies such as China, Korea, and Japan compared with their non-Asian counterparts.
However, as far as we know, few studies have examined the longitudinal effect of social support on dementia in Asia. Only one study in Taiwan reported that, along with other risk factors such as physical activity or depression, social support is related to late life cognitive decline [13
]. Saito et al.’s study demonstrated the relative importance of diverse social networks in preventing dementia. However, the study did not consider sources of support. In addition, gender was treated as a covariate and gender differences were unclear [14
The purpose of this study is to assess the effect of social support on incident dementia, taking into consideration support sources by gender. Our hypothesis was that the protective effect of social support on dementia differs by source and gender, and that the effect of family support would be stronger compared with non-family support.
2. Materials and Methods
The present study is part of the Aichi Gerontological Evaluation Study (AGES) project [15
]. The AGES project is a community-based prospective cohort study in Japan in which investigators evaluate factors associated with incident functional disability or dementia among noninstitutionalized older people aged 65 years old or above. In the baseline year of 2003, questionnaires were sent to a random sampling of community-living older adults aged 65 years or older in six large municipalities and a complete census of four small cities. The response rate was 52.1%. Detailed descriptions of the project and questions on the survey have been published [16
]. After excluding those with incomplete data on sex and age, 15,313 people (7381 men and 7932 women) were introduced in the cohort and followed for about 10 years from 1 November 2003, to 28 March 2013. We obtained information regarding incident functional disability, dementia, death, and relocation of participants (e.g., moving out of the study area) from the Long-Term Care Insurance (LTCI) system database maintained by municipalities. For the analyses, we eliminated those with limitations in basic activities of daily living, such as using the toilet, bathing, or transferring at the baseline year of 2003. This procedure left 14,088 people, or 92% of the total sample of this cohort. Study participants were comparable to entire older Japanese populations in terms of age and sex [15
2.1. Brief Description of Japan’s LTCI System
Japan’s LTCI system is a government-operated national insurance system for long-term care and was introduced in April 2000 to entitle every Japanese person aged 65 years and older with functional limitations or dementia to care in basic activities of daily living [17
]. In this system, certification of long-term care needs is based on an evaluation of each applicant’s degree of physical and mental disability, determined by a home-visit interview and a diagnosis from a primary care physician. A municipality certification committee determines the eligibility for receiving services [17
]. As receipt of benefits under this LTCI system is on an application basis, some people do not receive benefits under the system for various reasons, such as the availability of family members to provide care or financial burden (a 10% coinsurance is required to use services under the LTCI). To minimize such bias, we asked about basic activities of daily living, such as using the toilet, bathing, or transferring, in the 2003 survey and then excluded those already functionally impaired at baseline.
2.2. Incident Dementia
Incident dementia was ascertained when study participants became eligible for Japan’s public LTCI system, Level II or higher, on the index for the evaluation of care needs for people with dementia. The index was developed by the Ministry of Health and Welfare, based on observations of symptoms and behaviors that cause daily life impediment and degradation of cognitive functions along with communication difficulty. This index was validated using the MMSE and Revised Hasegawa Dementia Scale (HDS-R). The correlation coefficients with each scale were −0.744 and −0.735, respectively, indicating strong correlations with clinically used instruments [18
]. Insurance data were provided by insurers (municipalities) per the study agreement with the AGES project [16
2.3. Social Support
To elucidate social support, we asked respondents about the five types of perceived social support with respect to persons in three social support sources: co-residing family, family/relatives living apart, and friends/neighbors. “Co-residing family” refers to spouse/partner, children, or others living in the same household. “Family/relatives living apart” refers to adult children, siblings, or others not living in the same household. The types of support were emotional (providing/receiving), instrumental (providing/receiving), and appraisal (receiving). The act of “listening to concerns and complaints” was regarded as emotional support and that of “looking after when sick in bed for a few days” as instrumental support. Appraisal support was elicited by asking, “Do you have someone who acknowledges your existence and value?” These types of supports were often used in previous research studies [1
]. Answering categories for these social support variables were coded dichotomously as “support available” (coded as 1) and “no support available” (coded as 0).
Age and health status are important confounders when assessing the relationship between social support and health [1
]. Number of illnesses was ascertained by the question, “Are you currently receiving any medical treatment, and if so, for which illnesses?” For analyses, we calculated the number of illnesses, treated as a continuous variable. We adjusted for depression as well, as depression coexists with or predicts dementia [20
]. Depression was assessed by the 15-item geriatric depression scale (GDS-15), an instrument to screen depression among community-living older persons [21
]. The score was transformed into a dichotomous variable (<5: no depression; 5+: depression).
Subjective cognitive complaints predict dementia among the elderly [22
]. Thus, we asked respondents the following three questions: “Do you often get into trouble when you leave your belongings behind somewhere?”; “Do you often get times or places confused?”; and “Do you often forget things that happened recently (e.g., what you had for breakfast)?” The possible highest score was three, suggesting more cognitive impairments. As living arrangement is strongly associated with health [19
], we adjusted for its effect as well by stratifying those living alone and those living with someone else. We adjusted for marital status as well, since a small number (n
= 291) of respondents did not live with their spouse even while they were married, probably owing to hospitalization or institutionalization of their spouse. Engagement in community activities was ascertained by asking respondents about the number of community groups they participated in, such as sports, hobby, or local associations, as such participation is protective against dementia [20
]. Health behaviors such as smoking status, daily physical activity (average walking time per day), alcohol consumption, and education were also considered in the model.
2.5. Statistical Analyses
As studies have indicated that the association between social support and health differ by gender [1
], Cox proportional hazard models were employed stratified by gender to assess the association between baseline social support and incident dementia. Those who died or moved away from the study site during the follow-up period were considered as censored cases. To test if the effects of social support from each source were independent of the influence of others, we entered social support in the model along with other covariates stratified by support source. As our main aim was to study the relative effect of social support, social support variables were aggregated to create social support scores by support source (co-residing family members, family/relatives living apart, and friends/neighbors).
Moreover, we tested which support type was the most beneficial when coming from the same support source. The correlation coefficients of social support variables were as high as 0.705, for example, between “providing instrumental support” and “receiving instrumental support” for family members living together. As such, we constructed three models stratified by support source to avoid multi-collinearity. Then, we investigated the association between each social support and dementia. We used SPSS 21.0J (SPSS, Chicago, IL, USA) for statistical analyses. A p-value of less than 0.10 was considered marginally significant, and a p-value less than 0.05 was considered statistically significant.
2.6. Ethical Issues
The study protocol and informed consent procedure were approved by the Nihon Fukushi University Ethics Committee (#10-05). The study was conducted in compliance with the fifth revision of the Declaration of Helsinki.
shows baseline characteristics by gender. During the 10-year follow-up, 14.6% of men and 18.7% of women developed dementia. Men were slightly younger and participated in more community groups but smoked and drank more alcoholic beverages. More women lived alone and had more cognitive complaints than did men. In addition, women were significantly more depressed and had more illnesses.
describes social support by type and source. As for social support, men exchanged more support with co-residing family members, whereas women had more ties with family/relatives living apart or friends/neighbors. When subdivided by sources, men exchanged more emotional and instrumental support with their co-residing family and received more appraisal support from all sources than women did. Women exchanged more emotional and instrumental support with someone outside of their own households, namely a family member/relative who lives apart or friends/neighbors. When considering the effect of support sources, men benefit more from support exchanges with their co-residing family members, as shown in Table 3
. As for types of support, providing support to co-residing family was a significant protector against dementia among men, whereas among women, providing emotional support to family/relatives who live apart and receiving emotional support from friends or neighbors were protective against dementia (Table 4
). Contrary to our expectations, instrumental support exchanges with friends or neighbors were risks for dementia among men. Among women, receiving emotional support from co-residing family members raised the risk of dementia, whereas providing support to a family/relative who lives apart and receiving emotional support from friends or neighbors were protective of dementia (Table 4
Gender differences were observed in the association between social support and incident dementia. Support from co-residing family members was protective among men, whereas among women, no significant effect of social support on dementia was observed. Considering the gender differences in the association between social support and dementia and the possibility of misclassification of cases, the use of objective diagnosis data might be necessary. Also, given the fact that the number of older people living alone is increasing [2
], there is a greater need to investigate the effect of promoting social ties with outside family members. As men benefit more from support from co-residing family members, different strategies might be needed in designing intervention programs for men. Deteriorating cognitive or physical functions might undermine their self-image, leading to poor health; this occurs especially among men who want to be strong and do not want to show their weakness to others [24
]. In our study, appraisal support from someone outside of their own homes was related to a lower hazard for dementia, although it was not significant. If support is provided with respect, or if recipients are given more of a chance to give back support, this might preserve autonomy and mitigate the negative effect of receiving support, especially among men. Such support could enhance their self-esteem and compensate for the threat of loss of autonomy from receiving support.