1 SUBMITTED 1 OCT 21 1 REVISIONS REQ. 8 DEC 21 & 20 FEB 22; REVISIONS RECD. 11 JAN & 3 MAR 22 2 ACCEPTED 27 APR 22 3 ONLINE-FIRST: MAY 2022 4 DOI: https://doi.org/10.18295/squmj.5.2022.035 5 6 Social Relationships and Onset of Functional Limitation Among Older 7 Adults with Chronic Conditions 8 Does Gender Matter? 9 Dandan Jiao,1,9 Kumi Watanabe,3 Yuko Sawada,4 Munenori Matsumoto,1 10 Ammara Ajmal,1 Emiko Tanaka,5 Taeko Watanabe,6 Yuka Sugisawa,7 11 Sumio Ito,8 Rika Okumura,8 Yuriko Kawasaki,8 *Tokie Anme2* 12 1Graduate School of Comprehensive Human Sciences and 2Faculty of Medicine, University of 13 Tsukuba, Tsukuba, Japan; 3RIKEN Center for Advanced Intelligence Project, Tokyo, Japan; 14 4Department of Physical Therapy, Morinomiya University of Medical Sciences, Osaka, 15 Japan; 5Department of Community Nursing, Musashino University, Tokyo, Japan; 6College of 16 Nursing and Nutrition, Shukutoku University, Chiba, Japan; 7Department of Nursing, 17 Tsukuba International University, Tsukuba, Japan; 8Department of Public Welfare, 18 Tobishima, Japan; 9Department of Nursing, The First Affiliated Hospital, and College of 19 Clinical Medicine of Henan University of Science and Technology, Luoyang, China. 20 *Corresponding Author’s e-mail: tokieanme@gmail.com 21 22 Abstract 23 Objective: This study aimed to examine the longitudinal association between social 24 relationships and physical functioning among community-dwelling older adults with chronic 25 conditions. Methods: Self-reported questionnaires were distributed and collected between 26 2014 and 2017 from participants aged 65 years and older. The Index of Social Interaction was 27 used to evaluate social relationships, and the instrumental activities of daily living (IADL) 28 mailto:tokieanme@gmail.com 2 subscale of the Tokyo Metropolitan Institute of Gerontology Index of Competence was used 29 to examine functional status. Data from 422 participants (190 men and 232 women) were 30 included in the final analysis. Results: High social relationships demonstrated significant 31 adverse effects (OR = 0.77, 95% CI = 0.64 – 0.93) on the decline of IADL in the overall 32 sample, particularly for women (OR = 0.71, 95% CI = 0.55 – 0.93) but not for men ( p = 33 0.131). Conclusion: The finding suggests that functional limitation was influenced by social 34 relationships among the disabled older adults, and the influence of social relationships on 35 functional limitation differed based on gender. 36 Keywords: Interpersonal Relations; Functional Status; Gender; Aged; Chronic Disease; 37 Longitudinal Studies; Health Behaviour. 38 39 Advances in Knowledge 40 1. This study found that social relationships may prevent physical deterioration among 41 people with chronic conditions. 42 2. This positive association was observed among a female group, whereas no effect was 43 found among a male group. 44 3. Taking measures to promote social relationships and being aware of gender differences 45 may improve the physical function of older adults with chronic conditions. 46 47 Application to Patient Care 48 1. Physicians, nurses, and other health professionals should encourage older people to 49 engage in social relations that will benefit disabled older people’s health outcomes. 50 2. Social relationships should be a focus of chronic disease management. 51 52 Introduction 53 Aging is a global public issue, with at least one person out of eleven people being above 65 54 years of age in 2019. This number is expected to rise, with the older person being out of only 55 six people, by 2050.1 In Japan, the aging rate is 28.8% (about one out of four people ) of the 56 total population as of 2020.2 Most countries are experiencing not only a rapidly ageing 57 population but also the impact of related functional limitations. The latter refers to physical 58 3 restrictions in performing fundamental activities required in daily life (e.g., stooping, 59 climbing stairs and ambulating).3 Functional decline begins earlier and manifests more 60 severely in older adults with chronic diseases.4 It has been argued that individuals with 61 functional limitations have higher risks of falling 5 and mortality.6 However, older adults with 62 chronic conditions are also willing to maintain their level of functioning,7 and therefore, 63 identifying factors that can be linked the maintenance of their physical function are needed. 64 65 Social relationships are interactions between individuals and their social environment. 66 Numerous studies have documented the effects of poor social relationships on health 67 outcomes (e.g., depressive symptoms8 and mortality9). Additionally, social relationships have 68 an impact on functional limitations. For example, a prospective study indicated that baseline 69 social isolation predicted physical function decline after four years.10 Another study suggested 70 that lack of social relationships is linked to poor activities of daily living (ADL) and 71 instrumental activities of daily living (IADL).11 These studies have mainly focused on the 72 general older population. However, little is known about the beneficial effects of social 73 relationships and functional status on older adults with chronic health conditions. As 74 individuals with chronic conditions generally have restricted social participation,12 it is 75 necessary to investigate how social relationships affect them. A recent study suggested that 76 cancer survivors have higher chances of functional impairment when frequent contact with 77 others declines,13 and adults with type 2 diabetes with low levels of social support reported 78 higher functional disabilities.14 However, these studies only investigated the social relations of 79 individuals with specific chronic diseases (e.g., cancer, diabetes). As multiple chronic 80 diseases are increasing, studies investigating the comprehensive experiences of chronic 81 diseases are required. 82 83 Regarding the gender differences in social relationships and health, it has been demonstrated 84 that women spend more time and resources building social relationships than men. 85 Consequently, it is plausible that the negative impact of poor social relationships is greater for 86 women.15 For example, Lee et al.16 found that social relations were significantly associated 87 with cognitive function in women only. However, Hajek et al.17 found that decreased social 88 4 support impacts functional impairment in both genders. Hence, consistent results regarding 89 the influence of gender on social relationships and health have not been established. Further, 90 gender differences have not been thoroughly investigated in studies on individuals with 91 chronic diseases. 92 93 To address these gaps in the research, the current study examined the association between 94 social relationships and functional status among older adults with chronic diseases and 95 assessed the effects of gender on these associations. The results can be used to further the 96 understanding of social relationships and provide evidence for disease management to delay 97 deterioration and improve health-related quality of life for older adults with chronic diseases. 98 99 Methods 100 Design and participants 101 Data for this three-year longitudinal study were extracted from a single-centre cohort project, 102 the Community Empowerment and Care for well-being and Health Longevity (CEC), which 103 was established in Japan in 1991. The CEC aims to explore the factors related to the well-104 being of residents in the context of the low birth rate, ageing population, and high medical 105 expenditure. This project was conducted in Tobishima in Aichi Ken, central Japan; the entire 106 population of around 4,800 participated in the survey. In 2020, the ageing rate was 107 approximately 28.0%. Our research project was conducted in collaboration with the local 108 government municipality’s health policy evaluation, and our survey period was consistent 109 with that of the municipal survey. Before 2011, the survey was conducted every one or two 110 years. Since 2011, the survey has been conducted every three years. In each wave, the survey 111 is performed from April to May. To avoid selection bias, all the residents were invited to and 112 agreed to participate in the survey. Questionnaires were mailed to all residents. The 113 questionnaire encompassed demographic characteristics, nutrition, lifestyle habits, long-term 114 care needs evaluation, evaluation of local services, social relationships, and medical 115 conditions. 116 117 Unlike previous research related to this project, in the present study, we focused on older 118 5 adults living with chronic disease. We used data collected from individuals aged 65 years and 119 older with at least one chronic medical condition in 2014. The inclusion criteria were: (1) 120 people with at least one chronic disease (hypertension, stroke, heart disease, diabetes, 121 hyperlipidaemia, lung disease, arthritis, cancer, immune disease, depression, eye disease, and 122 ear disease), (2) physically independent at baseline, and (3) no missing information on IADL. 123 In 2014, 523 individuals who fit the inclusion criteria were enrolled. The chronic conditions 124 were determined using the question, ‘Do you have an illness that is being treated, or do you 125 suffer from an illness sequela?’ In 2017, a follow-up study was conducted to assess the 126 participants for symptoms of physical functional decline. The data were collected from 1 127 April to 15 May in 2014 and 2017. Between 2014 and 2017, 55 participants could not be 128 reached for the follow-up, and 46 participants were excluded owing to missing IADL 129 information (Figure 1). We calculated the required sample size using the G*power. A 130 minimal sample size of 191 was required based on the expected 0.8 power and a 0.05 level of 131 significance. 132 133 Measurements 134 Functional competence was assessed by the IADL subscale of the Tokyo Metropolitan 135 Institute of Gerontology Index of Competence. The IADL subscale comprises five items: 136 using public transportation, shopping, preparing meals, paying bills, and individual banking 137 management. For each item, a positive response was coded as 1, and a negative response was 138 coded as 0. For example, regarding the use of public transportation, the response options to 139 the question ‘Do you use public transportation (bus or train) to go out on your own?’ were ‘I 140 can and do’, ‘I can but do not’, and ‘I cannot’. A response of ‘I cannot’ received 0 point, 141 whereas the other two responses received 1 point. The total score ranged from 0 to 5, and a 142 score of 5 was considered a normal IADL, while a score of 0–4 indicated a low IADL.18 143 144 The Index of Social Interaction (ISI) was used to evaluate social relationships.19 The ISI 145 includes five subscales and 18 items. The Independence subscale has four items to assess 146 motivation to live, motivation to maintain a healthy life, taking an active approach towards 147 life, and having a regular lifestyle. The Social Curiosity subscale measures habits of reading 148 6 newspapers and books, using new equipment (e.g., a video system), hobbies, and a feeling of 149 importance in society. The Interaction subscale measures communication with family 150 members and non-family members and interacting with non-family members. The 151 Participation subscale measures participation in social groups, in neighbourhood groups, 152 watching television, and taking an active social role. The Feelings of Safety subscale tests if 153 participants have someone who can offer counselling and provide support during 154 emergencies. For all items, a positive response was coded as 1, and a negative response was 155 coded as 0. Taking ‘Do you have someone to counsel in a difficult situation?’ as an example, 156 the response options were ‘always’, ‘frequently’, ‘sometimes’, and ‘never’. A response of 157 ‘never’ received 0 point, while the other three received 1 point. The total score was 18, with a 158 higher score indicating good social interaction. The ISI was taken as a continuous variable in 159 the analysis. 160 161 In accordance with previous studies,12,20 we considered age, sex, living status, exercise, 162 smoking, and drinking as covariates. Age was measured as a continuous variable. Exercise 163 was evaluated by the question ‘Do you usually exercise?’ and was categorised as ‘yes’ if their 164 response were ‘always’ or ‘sometimes’ and ‘no’ if otherwise. Living status was evaluated by 165 asking if the participants lived alone or with others. Living with others included spouses, 166 sons, daughters, daughters-in-law, grandsons, brothers/sisters, and others. Smoking was 167 evaluated by the question ‘Do you smoke?’; those who responded ‘every day’ or ‘sometimes’ 168 were regarded as current smokers, ‘previously did but have stopped now’ as ex-smokers, and 169 ‘do not smoke’ as non-smokers. Drinking was assessed by an item ‘Do you drink?’ and was 170 dichotomised as yes if participants answered ‘every day’ or ‘sometimes’ and no if they 171 answered ‘do not drink’. 172 173 Statistical analysis 174 Baseline demographic information between gender groups was compared by chi-square (𝜒2) 175 tests or non-parametric tests. These tests were also implemented to examine demographic 176 information and the IADL for categorical and continuous variables, respectively. A multiple 177 logistic regression analysis was performed to examine the association between social 178 7 relationships and declining IADL after controlling for covariates that were statistically 179 significantly associated with IADL in the chi-square or non-parametric tests. We then fitted 180 additional models to examine gender differences in the association between the ISI and IADL 181 using a gender stratified analysis. A sensitive analysis was also completed to compare the 182 demographic differences between included and excluded cases. All the analyses were 183 performed using IBM SPSS 26.0. 184 185 Ethical consideration 186 This study was approved by the University of Tsukuba Ethics Committee (No. 1331-1). The 187 survey data were anonymously provided by the local government. 188 189 Results 190 Data from 422 individuals were included in the analysis. Of the 422 participants, most were 191 women, not living alone, doing exercise, non-drinkers, non-smokers, and diagnosed with one 192 chronic disease. While social relationships, age, living status, exercise, and disease status did 193 not differ between women and men, drinking and smoking differed significantly. Specifically, 194 men reported higher percentages of smoking and drinking [Table 1]. 195 196 The bivariate analysis demonstrated that age, exercise, and social relationships were 197 associated with the IADL after three years [Tables 2 and 3]. After controlling for age and 198 exercise in the logistic regression models, the results demonstrated that higher social 199 relationship index was (OR = 0.77, 95% CI =0.64–0.93) inversely associated with a low 200 IADL [Table 4]. 201 202 The sex-stratified data revealed different results. In Model 1, higher social relationship index 203 demonstrated a significant inverse association with a low IADL in men (OR= 0.77, 95% CI = 204 0.61–0.98) and women (OR= 0.69, 95% CI = 0.56–0.85). After adjusting for age (Model 2) 205 and exercise (Model 3), higher social relationship index was inversely related to low IADL in 206 women (OR = 0.66, 95% CI = 0.51–0.86 and OR = 0.71, 95% CI = 0.55–0.93, respectively). 207 However, social relationships index adjusted for age and exercise demonstrated no statistical 208 8 association with a low IADL in men, with p = 0.070 and p = 0.131 (Table 4). 209 210 The sensitive analysis demonstrated the difference in age between the excluded case (72.9 ± 211 6.0) and included case (75.9 ± 6.6). The excluded group is older than the included cases (p < 212 0.05), and a gender difference was not observed. 213 214 Discussion 215 This study examined the effects of social relationships on the functional status of older adults 216 with chronic conditions. The results demonstrated that higher social relationship index could 217 reduce older adults’ functional decline. However, the beneficial effects of social relationship 218 index are significantly evident only among women. 219 220 The stress-buffering model and main effect model are the two main theoretical models linking 221 social relationships to health outcomes.21 The stress-buffering model posits that social 222 resources, such as social companionship, may reduce stress by increasing positive moods. The 223 main effect model proposes that social support can reduce problematic behaviours (e.g., 224 smoking, drinking, and not seeking medical help) associated with poor health outcomes. 225 Based on these theoretical models, chronic diseases and their consequences are stressors, 226 which may be reduced by their social resources. Meanwhile, social relationships can directly 227 benefit functional status. 228 229 Social relationships seem to have adverse effects on the functional decline among older adults 230 with chronic diseases. This result aligns with previous research22 conducted in six countries, 231 which demonstrated that social capital and well-being were positively associated, regardless 232 of chronic diseases. A systematic review indicated that social relationships play an important 233 role in improving well-being and mental health among people with disabilities.23 Moreover, 234 another previous study also demonstrated that strong social relationships could decrease 235 functional decline even among individuals with multiple chronic diseases.20 Social relations 236 affect health outcomes through a reciprocity exchange24— social relations might enhance 237 resources, including transportation support and caring, which can affect health-related 238 9 behaviours. For instance, social contacts may allow individuals with chronic diseases to 239 undergo medical check-ups, access important health-related information, and gain confidence 240 in health-promoting behaviours, which can delay the onset of physical decline. 241 242 Chronic diseases are usually lifelong and cause earlier functional limitations; subsequently, 243 measures should be taken to prevent the deterioration of these conditions. This study 244 contributes to the existing literature through the identification of a cost-effective method to 245 gather evidence for chronic disease management. Being aware of the benefits of social 246 relationships for health outcomes may provide evidence and directions for chronic disease 247 management and suggests that health promotion programmes should be expanded to include 248 social relationships and traditionally modified factors (e.g., physical activity and healthy 249 eating behaviour). 250 251 Our findings further indicated that higher social relationship index could reduce the 252 probability of functional decline among women but not men. This finding aligns with a 253 previous 10-year longitudinal study25 that found that women with positive social support had 254 lower mortality risks than those with poor social support; however, this association was not 255 observed among men. Additionally, another longitudinal study examining the effects of 256 participation in social activities and cognitive decline among older adults found that social 257 activities impacted cognitive decline only among women.26 The benefits women experience 258 due to strong social relationships are particularly relevant in potentially improving their 259 conditions as women with chronic diseases are more likely to experience functional 260 disabilities.27 One plausible reason is that women generally maintained stability in their social 261 activities, whereas men’s social activities declined over time.28 Moreover, older women tend 262 to have larger social networks than older men.29 These findings support our results on the 263 gender-moderated effects of social relationships on functional status. 264 265 It is already known that behaviour change and improvements in the treatment of chronic 266 diseases may improve quality of life.30 Our study demonstrated that social relationship index, 267 which encompass independence, social curiosity, interaction with others, participating in 268 10 social activities, and feeling safe in daily life, are effective in maintaining physical function. 269 Regarding the practical implications of these findings, physicians, nurses, and social workers 270 need to consider the impact of social relationships on health and develop chronic disease 271 management interventions such as promoting interactions with family and non-family 272 members, using new equipment (e.g., video camera, internet), and reading newspapers. 273 Moreover, team-based strategies should be developed; these must include various parties such 274 as medical staff, social workers, health policymakers, and other health stakeholders. Further, 275 considering the gender differences in the effects of social relationships on health, more 276 integration measures should be included for men. Additionally, longitudinal research could 277 facilitate the identification of other relevant aspects of men’s social relations. 278 279 This study has several limitations. First, disease severity and duration were not examined, 280 which may have affected the results. Individuals with more severe and longer-lasting chronic 281 diseases may experience worse functional deterioration. Subsequently, it is unclear if the 282 effects of social relationships differ among people with chronic diseases of different severity 283 and duration. Further studies taking disease severity and duration into consideration are 284 therefore necessary. Second, although some diseases can affect individuals’ social 285 interactions, we did not explore the effect of social relationships on functional limitation by 286 individual diseases. Identifying the effects of specific diseases may facilitate the development 287 of targeted measures. Third, the specific types and frequency of social relationships index 288 were not examined. Knowing the specific source of social relationships would yield more 289 specific evidence for intervening to promote social engagement. Fourth, only one indicator 290 was used to examine functional status; combining subjective and objective measurements for 291 functional indicators might strengthen the reliability of the results. Fifth, even though we 292 highlighted the importance of social relationship index with regards to decreasing the odds of 293 functional decline, we did not examine interaction effects. Thus, studies examining the 294 interaction effects (multiplicative or additive) of social relationship index and chronic 295 conditions, which might offer deeper insight into the target population, will be significant for 296 public health and psychosocial research. Finally, this study was conducted in one area, which 297 may limit the generalisability of the results. Considering that culture is crucial to social 298 11 relationships, studies using data from various countries to provide insight into the 299 comparative impacts of social relationships on health outcomes across cultures are necessary. 300 301 Despite these gaps, this study addresses the association between social relationships and 302 functional status among community-dwelling older adults with chronic conditions and gender 303 differences. Our results further imply that it may be beneficial to consider gender differences 304 when encouraging social relationships for disease management. A better understanding of 305 gender-based differences can help develop further interventional programmes and studies to 306 promote health outcomes linked to gender differences. 307 308 Conclusion 309 Social relationships influence physical functioning among older adults with chronic diseases. 310 Based on the results of this study, among people with chronic disease, social relationships 311 were related to functional decline after three years; further, this association differed by 312 gender. Professionals should assist older adults with developing, maintaining, and 313 strengthening their social relations in daily life through interventions such as using new 314 equipment, interacting with family and non-family members, participating in various 315 activities, receiving support from others, and taking an active approach to life, to increase the 316 sustainability of their functional independence. Given their significance, social relationships 317 should form an important part of health policy decisions. Further, being aware of the 318 moderating effects of gender differences may facilitate the development of more appropriate 319 measures to reduce functional decline among older adults, and encouraging older women to 320 maintain social relationships may be effective in reducing their vulnerability to functional 321 limitations. 322 323 Conflict of Interest 324 The authors declare no conflicts of interest. 325 326 Funding 327 This work was supported by JSPS KAKENHI (Grant Number JP21K18449) and in part by 328 12 the JST SPRING (Grant Number JPMJSP2124). 329 330 Authors’ Contribution 331 DJ and TA conceptualised and designed the study. ET, TW and YS collected the data. DJ and 332 KW analysed and interpreted the data. All authors were involved in the study investigation. 333 SI, RO, YK and TA handled the project administration. DJ drafted the manuscript. YS, MM 334 and AA reviewed and edited the manuscript. TA supervised the study and acquired the 335 funding. All authors approved the final version of the manuscript. 336 337 Acknowledgement 338 We are grateful to all the participants and research members for their participation in this 339 study. The first author, Dandan Jiao would like to express her deep appreciation to receive the 340 fellowship from the Sasagawa scholarship from the Japan-China Medical Association. 341 342 References 343 1. World population ageing 2019: highlights New York City United Nations, department of 344 economic and social affairs, population division; 2019. From: 345 https://www.un.org/en/development/desa/population/publications/pdf/ageing/WorldPopul346 ationAgeing2019-Highlights.pdf Accessed: Jul 2020. 347 2. Statistic Bureau of Japan. Statistic handbook of Japan 2021. From: 348 https://www.stat.go.jp/english/data/handbook/c0117.html Accessed: Dec 2021. 349 3. Verbrugge LM, Jette AM. The disablement process. Soc Sci Med 1994; 38(1):1–14. 350 https://doi.org/10.1016/0277-9536(94)90294-1. 351 4. Fong JH. Disability incidence and functional decline among older adults with major 352 chronic diseases. BMC Geriatr 2019; 19(1):323. https://doi.org/10.1186/s12877-019-353 1348-z. 354 5. Deandrea S, Bravi F, Turati F, Lucenteforte E, La Vecchia C, Negri E. Risk factors for 355 falls in older people in nursing homes and hospitals. A systematic review and meta-356 analysis. Arch Gerontol Geriatr 2013; 56(3):407–415. 357 https://doi.org/10.1016/j.archger.2012.12.006. 358 https://www.un.org/en/development/desa/population/publications/pdf/ageing/WorldPopulationAgeing2019-Highlights.pdf https://www.un.org/en/development/desa/population/publications/pdf/ageing/WorldPopulationAgeing2019-Highlights.pdf https://www.stat.go.jp/english/data/handbook/c0117.html 13 6. Mutikainen S, Rantanen T, Alén M, Kauppinen M, Karjalainen J, Kaprio J, et al. Walking 359 ability and all-cause mortality in older women. Int J Sports Med 2011; 32(3):216–222. 360 https://doi.org/10.1055/s-0030-1268506. 361 7. McGilton KS, Vellani S, Yeung L, Chishtie J, Commisso E, Ploeg J, et al. Identifying 362 and understanding the health and social care needs of older adults with multiple chronic 363 conditions and their caregivers: a scoping review. BMC Geriatr 2018; 18(1):231. 364 https://doi.org/10.1186/s12877-018-0925-x. 365 8. Conde-Sala JL, Garre-Olmo J, Calvó-Perxas L, Turró-Garriga O, Vilalta-Franch J. 366 Course of depressive symptoms and associated factors in people aged 65+ in Europe: A 367 two-year follow-up. J Affect Disord 2019; 245:440–450. 368 https://doi.org/10.1016/j.jad.2018.10.358. 369 9. Takahashi S, Ojima T, Kondo K, Shimizu S, Fukuhara S, Yamamoto Y. Social 370 participation and the combination of future needs for long-term care and mortality among 371 older Japanese people: A prospective cohort study from the Aichi Gerontological 372 Evaluation Study (AGES). BMJ Open 2019; 9:e030500. 373 https://doi.org/10.1136/bmjopen-2019-030500. 374 10. Yu B, Steptoe A, Niu K, Jia X. Social Isolation and Loneliness as Risk Factors for Grip 375 Strength Decline Among Older Women and Men in China. J Am Med Dir Assoc 2020; 376 21(12):1926-1930. https://doi.org/10.1016/j.jamda.2020.06.029. 377 11. Cwirlej-Sozanska A, Wisniowska-Szurlej A, Wilmowska-Pietruszynska A, Sozanski B. 378 Determinants of ADL and IADL disability in older adults in southeastern Poland. BMC 379 Geriatr 2019; 19(1):297. https://doi.org/10.1186/s12877-019-1319-4. 380 12. Griffith LE, Raina P, Levasseur M, Sohel N, Payette H, Tuokko H, et al. Functional 381 disability and social participation restriction associated with chronic conditions in 382 middle-aged and older adults. J Epidemiol Community Health 2017; 71(4):381–389. 383 https://doi.org/10.1136/jech-2016-207982. 384 13. Guida JL, Holt CL, Dallal CM, He X, Gold R, Liu H. Social relationships and functional 385 impairment in aging cancer survivors: A longitudinal social network study. Gerontologist 386 2020; 60(4):607–616. https://doi.org/10.1093/geront/gnz051. 387 14. Levy AM, Deschênes SS, Burns RJ, Elgendy R, Schmitz N. Trajectories of social support 388 14 in adults with type 2 diabetes: Associations with depressive symptoms and functional 389 disability. Int J Geriatr Psychiatry 2019; 34(3):480–487. 390 https://doi.org/10.1002/gps.5042. 391 15. Chemaitelly H, Kanaan C, Beydoun H, Chaaya M, Kanaan M, Sibai AM, et al. The role 392 of gender in the association of social capital, social support, and economic security with 393 self-rated health among older adults in deprived communities in Beirut. Qual Life Res 394 2013; 1371–1379. https://doi.org/10.1007/s11136-012-0273-9. 395 16. Lee S, Lee S, Lee E, Youm Y, Cho HS, Kim WJ. Gender differences in social network of 396 cognitive function among community-dwelling older adults. Geriatr Gerontol Intl 2020; 397 20(5):467–473. https://doi.org/10.1111/ggi.13906. 398 17. Hajek A, Brettschneider C, Mallon T, van der Leeden C, Mamone S, Wiese B, et al. How 399 does social support affect functional impairment in late life? Findings of a multicenter 400 prospective cohort study in Germany. Age Ageing 2017; 46(5):813–820. 401 https://doi.org/10.1093/ageing/afx012. 402 18. Ishizaki T, Kai I, Kobayashi Y, Imanaka Y. Functional transitions and active life 403 expectancy for older Japanese living in a community. Arch Gerontol Geriatr 2002; 404 35(2):107–120. https://doi.org/10.1016/s0167-4943(02)00002-x. 405 19. Anme T, Shimada C. Social interaction and mortality in a five-year longitudinal study of 406 the elderly. Jpn J Public Health 2000; 47:127–133. [In Japanese] 407 20. Jiao D, Watanabe K, Sawada Y, Tanaka E, Watanabe T, Tomisaki E, et al. 408 Multimorbidity and functional limitation: The role of social relationships. Arch Gerontol 409 Geriatr 2020; 92:104249. https://doi.org/10.1016/j.archger.2020.104249. Epub ahead of 410 print. 411 21. Cohen S, Wills TA. Stress, social support, and the buffering hypothesis. Psychol Bull 412 1985; 98(2):310–357. 413 22. Christian AK, Sanuade OA, Okyere MA, Adjaye-Gbewonyo K. Social capital is 414 associated with improved subjective well-being of older adults with chronic non-415 communicable disease in six low- and middle-income countries. Global Health 2020; 416 16(1):2. https://doi.org/10.1186/s12992-019-0538-y. 417 23. Tough H, Siegrist J, Fekete C. Social relationships, mental health and well-being in 418 https://doi.org/10.1007/s11136-012-0273-9 15 physical disability: A systematic review. BMC Pub Health 2017; 17(1): 414. 419 https://doi.org/10.1186/s12889-017-4308-6. 420 24. Cockerham WC, Hamby BW, Oates GR. The social determinants of chronic disease. Am 421 J Prev Med 2017; 52(1s1):S5–S12. https://doi.org/10.1016/j.amepre.2016.09.010. 422 25. Lyyra TM, Heikkinen RL. Perceived social support and mortality in older people. J 423 Gerontol B Psychol Sci Soc Sci 2006; 61(3):S147–152. 424 https://doi.org/10.1093/geronb/61.3.s147. 425 26. Hwang J, Park S, Kim S. Effects of participation in social activities on cognitive function 426 among middle-aged and older adults in Korea. Int J Environ Res Public Health 2018; 427 15(10):2315. https://doi.org/10.3390/ijerph15102315. 428 27. Calderón-Larrañaga A, Santoni G, Wang HX, Welmer AK, Rizzuto D, Vetrano DL, et al. 429 Rapidly developing multimorbidity and disability in older adults: Does social background 430 matter?. J Intern Med 2018; 283(5):489–499. https://doi.org/10.1111/joim.12739. 431 28. Finkel D, Andel R, Pedersen NL. Gender differences in longitudinal trajectories of 432 change in physical, social, and cognitive/sedentary leisure activities. J Gerontol B 433 Psychol Sci Soc Sci 2018; 73(8):1491–1500. https://doi.org/10.1093/geronb/gbw116. 434 29. Pillemer S, Ayers E, Holtzer R. Gender-stratified analyses reveal longitudinal 435 associations between social support and cognitive decline in older men. Aging Ment 436 Health. 2019;23(10):1326-1332. https://doi: 10.1080/13607863.2018.1495178. 437 30. Preventing non-communicable diseases. [Internet]. Geneva: World Health Organization. 438 Available from: https://www.who.int/activities/preventing-noncommunicable-diseases 439 Accessed: Jan 2020. 440 16 Figure 1: Flow chart of participants Older adults aged ≥ 65 years n = 1004 Provided valid questionnaire n = 976 Excluding: People with no chronic disease (n = 267) Poor IADL (n = 174) Missing information on IADL (n = 12) Target population at baseline (2014) n = 523 Excluding: Lost to follow-up (n = 34) Dead (n = 21) Participants included in the analysis n = 422 Completed follow-up (2017) n = 468 Excluding: Missing information on IADL (n = 46) 17 Table 1: Baseline characteristics of the participants Variables Category Total (n = 422) Men (n = 190) Women (n = 232) P n % n % n % Age (mean ± SD) 72.8 ± 6.1 72.7 ± 6.1 72.9 ± 6.1 0.680 Living status Not alone 388 94.4 176 95.1 212 93.8 0.560 Alone 23 5.6 9 4.9 14 6.2 Missing 11 5 6 Exercise Yes 254 62.1 115 63.2 139 61.2 0.686 No 155 37.9 67 36.8 88 38.8 Missing 13 8 5 Drinking Yes 142 35.1 106 41.1 36 16.1 < 0.001 No 262 64.9 74 58.9 188 83.9 Missing 18 10 8 Smoking Current 32 8.2 30 17.0 2 0.9 < 0.001 Ex-smoker 99 25.4 93 52.5 6 2.8 Non- smoker 259 66.4 54 30.5 205 96.3 Missing 32 13 19 ISI (mean ± SD) 16.5 ± 1.6 16.6 ± 1.5 16.4 ± 1.6 0.763 Missing 45 15 30 Number of diseases 1 235 55.7 103 54.2 132 56.9 0.581 2 187 44.3 87 45.8 100 43.1 SD: standard deviation; ISI: Index of Social Interaction. 18 Table 2: Baseline characteristics of participants per follow-up level of IADL Variables Category Normal (n =359) Low (n = 63) 2/Z P n % n % Age (mean ± SD) 71.9  5.4 78.5  6.8 -6.995 < 0.001 Living status Not alone 329 94.0 59 96.7 0.728 0.552 Alone 21 6.0 2 3.3 Missing 9 2 Exercise Activity 227 65.0 27 45.0 8.739 0.003 Inactivity 122 35.0 33 55.0 Missing 10 3 Drinking Yes 118 65.8 24 59.3 0.927 0.336 No 227 34.2 35 40.7 Missing 14 4 Smoking Current 27 8.1 5 8.8 0.322 0.851 Ex-smoker 83 24.9 16 28.1 Non-smoker 223 67.0 36 63.1 Missing 26 6 ISI (mean ± SD) 16.6  1.4 15.7  1.9 -3.317 0.001 Missing 41 4 Number of diseases 1 198 55.2 37 58.7 0.278 0.598 2 161 44.8 26 41.3 IADL: instrumental activities of daily living; SD: standard deviation; ISI: index of social interaction. 19 Table 3: Baseline characteristics and follow-up IADL by gender Variables Category Men Women Normal (n = 161) Low (n = 29) 2/Z P Normal (n = 198) Low (n = 34) 2/Z P n % n % n % n % Age (mean  SD) 71.9  5.5 77.1  7.2 -3.686 < 0.001 71.85.3 79.66.3 -6.153 <0.001 Living status Not alone 147 94.2 29 100.0 1.759 0.359 182 93.8 30 93.8 0.000 1.000 Alone 9 5.8 0 0.0 12 6.2 2 6.2 Missing 5 0 4 2 Exercise Activity 102 65.4 13 50.0 2.268 0.132 125 64.8 14 41.2 6.777 0.009 Inactivity 54 34.6 13 50.0 68 35.2 20 58.8 Missing 5 3 5 0 Drinking Yes 87 56.5 19 73.1 2.527 0.112 31 16.2 5 15.2 0.024 0.876 No 67 43.5 7 26.9 160 83.8 28 84.8 Missing 7 3 7 1 Smoking Current 25 16.5 5 19.2 1.834 0.400 2 1.1 0 0.0 1.416 0.493 Ex-smoker 77 51.0 16 61.6 6 3.3 0 0.0 Non-smoker 49 32.5 5 19.2 174 95.6 31 100.0 Missing 10 3 16 3 ISI (mean ± SD) 14.92.3 13.8  2.6 -2.080 0.038 14.72.6 13.12.4 -2.656 0.008 Missing 15 0 26 4 Number of diseases 1 88 54.6 15 51.7 0.085 0.770 110 55.6 22 64.7 0.991 0.320  73 45.4 14 48.3 88 44.4 12 35.3 IADL: instrumental activities of daily living; SD: standard deviation; ISI: index of social interaction. Table 4: Logistic regression models of the association between baseline ISI and follow-up low IADL Variables Model 1 Model 2 Model 3 OR 95% CI P OR 95% CI P OR 95% CI P Total ISI 0.73 0.62 - 0.85 < 0.001 0.73 0.61 - 0.87 < 0.001 0.77 0.64 - 0.93 0.006 Age 1.18 1.12 - 1.24 < 0.001 1.17 1.11 - 1.23 <0.001 Exercise 1.67 0.86 - 3.26 0.130 Men ISI 0.77 0.61 - 0.98 0.035 0.79 0.61 - 1.02 0.070 0.80 0.60 - 1.07 0.131 Age 1.13 1.06 - 1.21 < 0.001 1.11 1.04 - 1.19 0.003 Exercise 1.21 0.44 - 3.28 0.714 Women ISI 0.69 0.56 - 0.85 0.001 0.66 0.51 - 0.86 0.002 0.71 0.55 - 0.93 0.011 Age 1.24 1.14 - 1.35 < 0.001 1.25 1.45 - 1.36 < 0.001 Exercise 2.56 0.98 - 6.65 0.054 IADL: instrumental activities of daily living; OR: odds ratio; CI: confidence interval; ISI: index of social interaction.