You are here

  • Home
  • Archive
  • Volume 70, Issue 7
  • Structural and functional aspects of social support as predictors of mental and physical health trajectories: Whitehall II cohort study

Article Text

Article menu

Download PDFPDF

Other topics
Structural and functional aspects of social support as predictors of mental and physical health trajectories: Whitehall II cohort study
  1. Christian Hakulinen1,2,
  2. Laura Pulkki-Råback1,3,
  3. Markus Jokela1,
  4. Jane E Ferrie4,5,
  5. Anna-Mari Aalto2,
  6. Marianna Virtanen6,
  7. Mika Kivimäki4,7,
  8. Jussi Vahtera6,8,9,
  9. Marko Elovainio1,2
  1. 1Institute of Behavioural Sciences, University of Helsinki, Helsinki, Finland
  2. 2National Institute for Health and Welfare, Helsinki, Finland
  3. 3Helsinki Collegium for Advanced Studies, University of Helsinki, Helsinki, Finland
  4. 4Department of Epidemiology and Public Health, University College London, London, UK
  5. 5School of Social and Community Medicine, University of Bristol, Bristol, UK
  6. 6Finnish Institute of Occupational Health, Helsinki, Finland
  7. 7Clinicum, Faculty of Medicine, University of Helsinki, Helsinki, Finland
  8. 8Department of Public Health, University of Turku, Turku, Finland
  9. 9Turku University Hospital, Turku, Finland
  1. Correspondence to Dr Christian Hakulinen, Institute of Behavioural Sciences, University of Helsinki, P.O. Box 9, Helsinki 00014, Finland; christian.hakulinen{at}helsinki.fi

Abstract

Background Social support is associated with better health. However, only a limited number of studies have examined the association of social support with health from the adult life course perspective and whether this association is bidirectional.

Methods Participants (n=6797; 30% women; age range from 40 to 77 years) who were followed from 1989 (phase 2) to 2006 (phase 8) were selected from the ongoing Whitehall II Study. Structural and functional social support was measured at follow-up phases 2, 5 and 7. Mental and physical health was measured at five consecutive follow-up phases (3–8).

Results Social support predicted better mental health, and certain functional aspects of social support, such as higher practical support and higher levels of negative aspects in social relationships, predicted poorer physical health. The association between negative aspects of close relationships and physical health was found to strengthen over the adult life course. In women, the association between marital status and mental health weakened until the age of approximately 60 years. Better mental and physical health was associated with higher future social support.

Conclusions The strength of the association between social support and health may vary over the adult life course. The association with health seems to be bidirectional.

  • Social and life-course epidemiology
  • SOCIAL EPIDEMIOLOGY
  • SELF-RATED HEALTH
  • MENTAL HEALTH

https://doi.org/10.1136/jech-2015-206165

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Higher levels of social support have been associated with better health.1–4 Prospective longitudinal studies have demonstrated that both structural aspects (eg, large social network size and cohesion) and functional aspects (eg, high perceived emotional support) of social support are associated with better self-reported health5 ,6 and lower levels of depression,7 coronary heart disease8 and mortality.9–11 Thus, there is clear evidence supporting an overall positive influence of social support on health.

    There have, however, been relatively few studies that have examined the association of social support with mental and physical health by considering the adult life course. To date, most studies have relied on the analysis of two measurement points and have not taken into account possible changes in the association between social support and health with age. The size of social networks and the amount of received social support tend to vary over the life course,12 although it remains unclear whether the association between social support and health strengthens, remains stable or decreases over the life course.13 This issue is of importance in the context of ageing, as changes in social network are common in later life. There is, for example, some evidence to suggest that growth of social network is associated with better self-reported health in older adults,14 but few systematic investigations on this issue are currently available.

    It is also possible that the association between social support and health is bidirectional; that is, as well as social support affecting health, health may have an impact on access to social support (eg, restrictions on mobility) or it may affect possibilities to benefit from social support (eg, cognitive decline). Bidirectional associations could lead to a vicious cycle where poor health contributes to the loss of social support over time. Lastly, few studies have examined whether structural and functional aspects of social support are associated with health equally strongly and whether structural support shapes the effect of functional support. A better understanding of directionality and major aspects of social support in terms of health would contribute to the design of health-promoting interventions.

    In the current study, we examined associations between structural and functional aspects of social support and future mental and physical health from an adult life course perspective using data from the Whitehall II Study.15 ,16 To examine possible bidirectional effects, we determined whether self-rated mental and physical health was associated with future structural and functional aspects of social support. Since previous studies have shown that social support is at least partially socioeconomically patterned,9 we controlled for the effect of socioeconomic status.

    Methods

    Study sample

    Participants were from the ongoing Whitehall II Study,15 ,16 which originally included 10 308 London-based civil servants from 20 civil service departments who were 35–55 years of age at study baseline (phase 1: 1985–1988). Data from baseline and seven follow-up phases (phases 2–8, collected in 1989–1990, 1991–1993, 1995–1996, 1997–1999, 2001, 2003–2004 and 2006, respectively) were used in this study. All participants who provided data at the baseline and at the first follow-up phase 2 (n=7578), and at any subsequent follow-up phases 3–8 (n=6797; 4788 men, 2009 women) were included. From phases 3, 4, 5, 6, 7 and 8, data were available for 6783, 6094, 5614, 5359, 5330 and 5353 participants, respectively. Ethical approval for the Whitehall II Study was obtained from the University College London Medical School Committee on the ethics of human research. Informed consent was obtained from the study participants.

    Measures

    Structural measures of social support

    Self-reported social network, which was available from phases 2, 5 and 7, and marital status were used as structural measures of social support. Social network score was obtained from questions (1) on the monthly frequency of contacts with relatives, friends and colleagues and the frequency of participation in social or religious activities and (2) on the total number of relatives or friends seen once a month or more. The scaled responses were then summed together. Marital status was dichotomised as 1=married/cohabiting; 0=never married, separated, divorced or widowed.

    Functional measures of social support

    The following three functional measures of social support were assessed at phases 2, 5 and 7 using the Close Persons Questionnaire17: confiding support, practical support and negative aspects of close relationships. Confiding support (7 items) measures included wanting to confide, confiding, sharing interests, boosting self-esteem and reciprocity. Practical support (4 items) included measures of practical help received, whereas negative aspects of close relationships (4 items) measured adverse exchanges and conflicts in relationships. Items were rated on a 4-point Likert scale, with higher scores indicating greater negative or positive support.

    Self-reported mental and physical health

    Mental health and physical health were assessed using the self-administered SF-36 questionnaire.18 ,19 Two subscales, mental health and physical functioning, which represent the main SF-36 subscales for mental and physical health, were used.18 The mental health subscale includes 5 items assessing aspects of mental well-being (eg, feeling happy, feeling nervous), and the physical health subscale includes 10 items assessing the ability to carry out daily activities (eg, difficulties in carrying groceries, difficulties in walking long distances). Cubic transformation was used to transform the negatively skewed mental and physical health scales, and both scales were then transformed into t-scores (the overall mean score across participants and follow-up phases was 50 (SD=10) and higher scores indicated better health).

    Covariates

    Age, ethnicity (0=white; 1=non-white) and socioeconomic status measured as employment grade (1=low; 2=intermediate; 3=high) were reported at the study baseline and were used as covariates in all analyses. Employment grade was based on the participant’s self-reported civil service grade title,16 ,20 which was then grouped into three grade categories using the civil service employment grade classification. Employment grade has been shown to be a broad marker of socioeconomic status as it has been associated with salary, educational level and the level of responsibility at work.16 ,21

    Statistical analysis

    All analyses were conducted separately for men and women. The association between social support with mental health and physical health trajectories was examined using longitudinal multilevel regression analyses with random intercept.22 ,23 Repeated measurements were arranged into a multilevel format in which measurements were nested within participants, that is, the same participants contributed more than one observation in the data set. Previous studies using the Whitehall II Study data have shown that mental and physical health have a non-linear relationship with age.24 ,25 Thus, a restricted cubic spline function with five knots was used to model the relationship between mental and physical health with age. To examine whether the association between social support and health strengthened or weakened with age, an interaction term between social support and the first spline age variable was introduced.

    Three separate analyses were used to examine the association between measures of social support with mental and physical health. First, measures of social support from phase 2 were used to predict mental and physical health in phases 3–8. Second, to further examine the longitudinal associations, social support as a time-dependent exposure from phases 2, 5 and 7 was used to predict mental and physical health in phases 3, 6 and 8. This enabled the examination of variation with time in the association of social support with health. Finally, to examine potential bidirectional effects, mental and physical health as time-dependent exposures from phases 4 and 6 were used to predict social support at phases 5 and 7. The longitudinal regression analyses were repeated using within-individual analysis to minimise potential confounding arising from unmeasured stable confounders. In addition to covariates, that is, age, ethnicity and employment grade, all analyses were adjusted for the effect of measurement period, that is, study phase. All statistical analyses were performed using Stata V.13.1 (StataCorp LP, College Station, Texas, USA).

    Results

    Descriptions of the study sample are shown in table 1. When compared with the original sample, participants included in the study sample were more likely to be white (p<0.001), women (p<0.001), slightly younger (mean age in years: 57.0 vs 56.8, p=0.006) and from a higher employment position (p<0.001). In addition, individuals who responded at phase 5 had a higher social network score (p<0.001), were more likely to be married or cohabiting (p<0.001), and had higher levels of emotional support (p=0.02) and practical support (p<0.001) than those who dropped out from the study. Except for emotional support, similar patterns were also observed between those who responded at phase 7 versus those who dropped out from the study.

    View this table:
    Table 1

    Descriptive statistics of the study sample, phase 2–3

    Associations between measures of social support at phase 2 with mental and physical health trajectories are shown in table 2. Higher social network score, being married or cohabiting, and higher levels of emotional support were associated with better mental health in both sexes. Higher levels of negative aspects in close relationships were in turn associated with poorer mental health. In addition, higher levels of practical support were associated with better mental health in men. Higher levels of negative aspects in close relationships were associated with poorer physical health in both sexes, whereas higher levels of emotional support were associated with better physical health among men and higher levels of practical support were associated with poorer physical health among women.

    View this table:
    Table 2

    Structural and functional aspects of social support at baseline (study phase 2) predicting mean levels of mental and physical health from study phase 3 to phase 8

    To examine whether the association between social support and health strengthens, remains stable or decreases over the life course, we ran interaction analyses (for results, see online supplementary tables S1 and S2). In men, interaction between age and practical support suggested that the association between practical support and mental health weakened with age, whereas the association between practical support and physical health strengthened with age. In women, the association between emotional support and mental health was found to weaken with age. No other interaction effects between social support with mental and physical health were found.

    Prospective longitudinal associations between repeated measures of social support and mental and physical health are shown in table 3. Higher social network score, higher levels of emotional and practical support, and lower levels of negative aspects in close relationships predicted better mental health in both sexes. Being married or cohabiting predicted better mental health only among men. Lower practical support and lower levels of negative aspects in close relationships were associated with better physical health in both sexes. However, only the associations between emotional support, practical support and negative aspects in close relationships with mental health in men and the association between practical support and physical health remained statistically significant in within-individual analyses.

    View this table:
    Table 3

    Longitudinal associations between structural and functional aspects of social support and mental health and physical health

    Interactions between age and social support predicting mental and physical health are shown in online supplementary tables S3 and S4. In women, the significant interaction between age and marital status predicting mental health indicated that the association between marital status and mental health weakened over the adult life course. There was also some indication that mental health in married women began to decline after the age of 60 years, while no such effect was found for single women. Interactions were also observed in both sexes in the association between negative aspects of close relationships and physical health; these associations strengthened over the adult life course. Interactions are illustrated in figure 1.

    Figure 1
    Figure 1

    In the left side panel (A) are predicted mental health trajectories by marital status (single vs married/cohabiting) for women and in the right side panel (B) are predicted physical health functioning trajectories by negative aspects of close relationships (low=−1 SD below the mean vs high=1 SD above the mean). Values are means. Grey shaded areas represent 95% CIs.

    Interaction analyses between structural and functional support in predicting mental and physical health are shown in online supplementary tables S5–8. Of the 24 interactions analyses, there was some evidence that high social network size buffered the effect of emotional support and negative aspects of close relationships on mental health for men. No other consistent interaction effects were found.

    Results of the possible bidirectional effects, that is, associations of mental and physical health with future structural and functional aspects of social support, are shown in table 4. Better mental health predicted a higher social network score, higher levels of emotional and practical support, and lower levels of negative aspects in close relationships in both sexes. Better physical health, in turn, predicted lower practical support and lower levels of negative aspects in social relationships in both sexes. In addition, better physical health predicted higher emotional support in men. However, only the associations between marital status and physical health remained significant in the within-individual analysis.

    View this table:
    Table 4

    Mental and physical health predicting structural and functional aspects of social support

    Discussion

    The results of this study demonstrate that there is a bidirectional association between social support and health, and that the strength of this association can vary over the adult life course. These findings, which are based on a British occupational cohort, highlight the importance of examining the role of social support over the adult life course.

    Current results are in line with previous studies showing that social support is important for mental and physical health. The role of social support in mental health has been highlighted,26 and whereas functional support was only associated with physical health, both aspects of social support clearly contributed to mental health over the adult life course. Whereas positive effects of marriage have been demonstrated in numerous studies,27 ,28 in this study, being married or cohabiting was associated only with mental, but not with physical, health. Although being married or cohabiting was a somewhat stronger predictor in men than in women, these results were, however, partly explained by the finding that the effect of marriage or cohabiting varied over the life course in women, but not in men. In addition, we found evidence that for men some effects of functional support were buffered by structural support, indicating that structural support could shape the effect of functional support on mental health.

    Our finding that higher levels of practical support were associated with poorer physical health is most likely explained by the fact that people need more practical support when they become sick (eg, assistance in transportation to healthcare). Practical support can also have a preventive effect on physical disease, as help with everyday functions may promote healthier lifestyles protective of physical disease. Negative aspects of social relationships were more strongly associated with both physical and mental health than positive social support. These results are in accordance with some previous studies showing that negative, but not positive, aspects of social support are associated with health.29–31 Negative aspects of social relationships could be especially harmful as they can be a source of stress, and thus lead to chronic strain.32

    A novel finding of this study is that the strength of social support in health can differ over the adult life course. Whereas the association between marital status and mental health in women was found to weaken with age, the role of negative aspects of close relationships in physical health strengthened gradually over the adult life course. The role of social support over the whole life course has also been emphasised previously,13 although most of these studies have focused on specific stages of the life course. For example, higher social support has been shown to predict better mental health in adolescence,33 in middle age5 and in older adulthood.34 However, social support might be of more importance among those who are more vulnerable to loss of health, such as older adults. There is also some evidence that low social support predicts faster cognitive decline,35 and as cognitive functioning is strongly linked to the ability to maintain independence, this might explain the amplifying effect of functional support at older ages.

    We obtained evidence of bidirectional association between social support and mental health; both functional and structural aspects of social support were associated with future mental health and vice versa. Similar bidirectional association was also observed between measures of social support and physical health, although the effect sizes found were considerably smaller. It has been noted previously that there is a lack of studies that have examined potential bidirectional effects between social support and health.36 However, it is most likely that the association between health and social support is considerably weaker than the association between social support and health. This is also supported by numerous studies showing the predictive strength of social support on health.4 ,26 However, our findings are of importance as they demonstrate that reverse causality, for example, health may have an impact on the availability of social support, should be taken into account in the studies of social support and health.

    Only the associations between functional support and mental health, practical support and health, and marital status with mental health were not confounded by unmeasured variables. This indicates that the discovered associations reflect mostly associations between individuals. It is possible, for example, that individuals with high social support may have better mental health across measurement times, but mental health of each person might not change as the person's social support changes from one measurement time to another.

    A number of mechanisms, not directly measured here, are likely to explain the observed associations.36 A number of psychosocial mechanisms such as social comparison (ie, individuals are likely to compare their attitudes, beliefs, and behaviours with those around them, and typically adjust their own norms to others) are likely to explain why functional aspects of social support lead to better health.37 Social support has also been shown to buffer against stressful life events,38 indicating that some beneficial effects of social support most likely resulted from coping better with difficult situations. With regard to biological mechanisms, neuroendocrine changes triggered by poor social support have been supported in studies using humans as well as animals,39 and social support has also been associated with changes in the immune systems.40

    This study has some notable strengths. Longitudinal data from over 20 years with repeated measures made it possible to examine mental and physical health trajectories. We were able to examine both structural and functional aspects of social support and potential bidirectional effects. When interpreting current findings, some limitations need to be taken into account. All measurements were based on self-reported data, which can create bias due to common method variance.41 For example, it is possible that individuals with poor mental health assess their level of social support differently from individuals with better mental health. Another potential limitation is that the measures of social support were only available from three phases, which could increase statistical error. Since the participants of this study are mainly London-based white-collar civil servants, the results may not be representative of the general British population. In addition, most women in the Whitehall II Study were from the lower occupational grades; thus, the results for women might not be generalisable to the general population of working women. Also, in a previous study using the Whitehall II Study data, higher probability of attrition has been associated with poorer mental health and poorer physical health.25 In our analyses, individuals who continued in the study had more structural and functional support than those who dropped out from the study. This can, if anything, both inflate or attenuate observed associations.

    To conclude, this study demonstrates that there is a bidirectional association between social support with mental and physical health, and that the association between social support and health may change over the adult life course. Future studies should address the mechanisms explaining the varying association of social support with health over the life course.

    What is already known on this subject

    • The positive effects of social support are well known and higher social support has been associated with better mental and physical health. However, only a few studies have examined the association of social support with health from the adult life course perspective.

    What this study adds

    • Current results show that there is a bidirectional association between social support and health. In addition, the association between social support and health was shown to vary over the life course. These findings highlight the importance of social support in public health over the adult life course.

    Acknowledgments

    The authors are grateful to all members of the Whitehall II Study team and to all participating civil servants in the Whitehall II Study; the participating civil service departments, their welfare, personnel, and establishment officers; the Occupational Health and Safety Agency; and the Council of Civil Service Unions.

    References

    1. ↵
      1. Holt-Lunstad J,
      2. Smith TB,
      3. Baker M, et al
      . Loneliness and social isolation as risk factors for mortality: a meta-analytic review. Perspect Psychol Sci 2015;10:227–37. doi:10.1177/1745691614568352
      OpenUrlAbstract/FREE Full Text
    2. ↵
      1. House JS,
      2. Landis KR,
      3. Umberson D
      . Social relationships and health. Science 1988;241:540–5. doi:10.1126/science.3399889
      OpenUrlAbstract/FREE Full Text
    3. ↵
      1. Barth J,
      2. Schneider S,
      3. von Känel R
      . Lack of social support in the etiology and the prognosis of coronary heart disease: a systematic review and meta-analysis. Psychosom Med 2010;72:229–38. doi:10.1097/PSY.0b013e3181d01611
      OpenUrlAbstract/FREE Full Text
    4. ↵
      1. Holt-Lunstad J,
      2. Smith TB,
      3. Layton JB
      . Social relationships and mortality risk: a meta-analytic review. PLoS Med 2010;7:e1000316. doi:10.1371/journal.pmed.1000316
      OpenUrlCrossRefPubMed
    5. ↵
      1. Melchior M,
      2. Berkman LF,
      3. Niedhammer I, et al
      . Social relations and self-reported health: a prospective analysis of the French Gazel cohort. Soc Sci Med 2003;56:1817–30. doi:10.1016/S0277-9536(02)00181-8
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. White AM,
      2. Philogene GS,
      3. Fine L, et al
      . Social support and self-reported health status of older adults in the United States. Am J Public Health 2009;99:1872–8. doi:10.2105/AJPH.2008.146894
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Stansfeld S,
      2. Fuhrer R,
      3. Shipley MJ
      . Types of social support as predictors of psychiatric morbidity in a cohort of British Civil Servants (Whitehall II Study). Psychol Med 1998;28:881–92. doi:10.1017/S0033291798006746
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Rosengren A,
      2. Wilhelmsen L,
      3. Orth-Gomér K
      . Coronary disease in relation to social support and social class in Swedish men. A 15 year follow-up in the study of men born in 1933. Eur Heart J 2004;25:56–63. doi:10.1016/j.ehj.2003.10.005
      OpenUrlAbstract/FREE Full Text
    9. ↵
      1. Stringhini S,
      2. Berkman L,
      3. Dugravot A, et al
      . Socioeconomic status, structural and functional measures of social support, and mortality: the British Whitehall II Cohort Study, 1985–2009. Am J Epidemiol 2012;175:1275–83. doi:10.1093/aje/kwr461
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Berkman LF,
      2. Syme SL
      . Social networks, host resistance, and mortality: a nine-year follow-up study of Alameda County residents. Am J Epidemiol 1979;109:186–204.
      OpenUrlAbstract/FREE Full Text
    11. ↵
      1. Berkman LF,
      2. Melchior M,
      3. Chastang JF, et al
      . Social integration and mortality: a prospective study of French employees of electricity of France-gas of France: the GAZEL Cohort. Am J Epidemiol 2004;159:167–74. doi:10.1093/aje/kwh020
      OpenUrlAbstract/FREE Full Text
    12. ↵
      1. Cornwell B,
      2. Schumm LP,
      3. Laumann EO, et al
      . Assessment of social network change in a national longitudinal survey. J Gerontol B Psychol Sci Soc Sci 2014;69(Suppl 2):S75–82. doi:10.1093/geronb/gbu037
      OpenUrlAbstract/FREE Full Text
    13. ↵
      1. Umberson D,
      2. Crosnoe R,
      3. Reczek C
      . Social relationships and health behavior across life course. Annu Rev Sociol 2010;36:139–57. doi:10.1146/annurev-soc-070308-120011
      OpenUrlCrossRefPubMed
    14. ↵
      1. Cornwell B,
      2. Laumann EO
      . The health benefits of network growth: new evidence from a national survey of older adults. Soc Sci Med 2015;125:94–106. doi:10.1016/j.socscimed.2013.09.011
      OpenUrlCrossRefPubMed
    15. ↵
      1. Marmot M,
      2. Brunner E
      . Cohort profile: the Whitehall II study. Int J Epidemiol 2005;34:251–6. doi:10.1093/ije/dyh372
      OpenUrlFREE Full Text
    16. ↵
      1. Marmot MG,
      2. Smith GD,
      3. Stansfeld S, et al
      . Health inequalities among British civil servants: the Whitehall II study. Lancet 1991;337:1387–93. doi:10.1016/0140-6736(91)93068-K
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. Stansfeld S,
      2. Marmot M
      . Deriving a survey measure of social support: the reliability and validity of the Close Persons Questionnaire. Soc Sci Med 1992;35:1027–35. doi:10.1016/0277-9536(92)90242-I
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Ware JE
      . SF-36 health survey update. Spine (Phila Pa 1976) 2000;25:3130–9. doi:10.1097/00007632-200012150-00008
      OpenUrlCrossRef
    19. ↵
      1. Ware JE,
      2. Snow KK,
      3. Kosinski M, et al
      . SF-36 Health Survey Manual and Interpretation Guide. Boston: New England Medical Center, 1993.
    20. ↵
      1. Ferrie JE,
      2. Shipley MJ,
      3. Davey Smith G, et al.
      Change in health inequalities among British civil servants: the Whitehall II study. J Epidemiol Community Health 2002;56:922–6. doi:10.1136/jech.56.12.922
      OpenUrlAbstract/FREE Full Text
    21. ↵
      1. Elovainio M,
      2. Ferrie JE,
      3. Singh-Manoux A, et al
      . Socioeconomic differences in cardiometabolic factors: social causation or health-related selection? Evidence from the Whitehall II Cohort Study, 1991–2004. Am J Epidemiol 2011;174:779–89. doi:10.1093/aje/kwr149
      OpenUrlAbstract/FREE Full Text
    22. ↵
      1. Singer JD,
      2. Willett JB
      . Applied longitudinal data analysis: modeling change and event occurrence. Oxford: Oxford University Press, 2003.
    23. ↵
      1. Rabe-Hesketh S,
      2. Skrondal A
      . Multilevel and longitudinal modeling using Stata. STATA Press, 2008.
    24. ↵
      1. Jokela M,
      2. Ferrie JE,
      3. Gimeno D, et al
      . From midlife to early old age: health trajectories associated with retirement. Epidemiology 2010;21:284–90. doi:10.1097/EDE.0b013e3181d61f53
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      1. Jokela M,
      2. Singh-Manoux A,
      3. Ferrie JE, et al
      . The association of cognitive performance with mental health and physical functioning strengthens with age: the Whitehall II cohort study. Psychol Med 2010;40:837–45. doi:10.1017/S0033291709991024
      OpenUrlCrossRefPubMedWeb of Science
    26. ↵
      1. Kawachi I,
      2. Berkman LF
      . Social ties and mental health. J Urban Health 2001;78:458–67. doi:10.1093/jurban/78.3.458
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Frisch M,
      2. Simonsen J
      . Marriage, cohabitation and mortality in Denmark: National cohort study of 6.5 million persons followed for up to three decades (1982–2011). Int J Epidemiol 2013;42:559–78. doi:10.1093/ije/dyt024
      OpenUrlAbstract/FREE Full Text
    28. ↵
      1. Robles TF,
      2. Kiecolt-Glaser JK
      . The physiology of marriage: pathways to health. Physiol Behav 2003;79:409–16. doi:10.1016/S0031-9384(03)00160-4
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. Newsom JT,
      2. Nishishiba M,
      3. Morgan DL, et al
      . The relative importance of three domains of positive and negative social exchanges: a longitudinal model with comparable measures. Psychol Aging 2003;18:746–54. doi:10.1037/0882-7974.18.4.746
      OpenUrlCrossRefPubMedWeb of Science
    30. ↵
      1. De Vogli R,
      2. Chandola T,
      3. Marmot MG
      . Negative aspects of close relationships and heart disease. Arch Intern Med 2007;167:1951–7. doi:10.1001/archinte.167.18.1951
      OpenUrlCrossRefPubMedWeb of Science
    31. ↵
      1. Kouvonen A,
      2. De Vogli R,
      3. Stafford M, et al
      . Social support and the likelihood of maintaining and improving levels of physical activity: the Whitehall II Study. Eur J Public Health 2012;22:514–18. doi:10.1093/eurpub/ckr091
      OpenUrlAbstract/FREE Full Text
    32. ↵
      1. Krause N,
      2. Rook KS
      . Negative interaction in late life: issues in the stability and generalizability of conflict across relationships. J Gerontol B Psychol Sci Soc Sci 2003;58:P88–99. doi:10.1093/geronb/58.2.P88
      OpenUrlAbstract/FREE Full Text
    33. ↵
      1. Rothon C,
      2. Goodwin L,
      3. Stansfeld S
      . Family social support, community ‘social capital’ and adolescents’ mental health and educational outcomes: a longitudinal study in England. Soc Psychiatry Psychiatr Epidemiol 2012;47:697–709. doi:10.1007/s00127-011-0391-7
      OpenUrlCrossRefPubMed
    34. ↵
      1. Schwarzbach M,
      2. Luppa M,
      3. Forstmeier S, et al
      . Social relations and depression in late life—a systematic review. Int J Geriatr Psychiatry 2014;29:1–21. doi:10.1002/gps.3971
      OpenUrlCrossRefPubMed
    35. ↵
      1. Liao J,
      2. Head J,
      3. Kumari M, et al
      . Negative aspects of close relationships as risk factors for cognitive aging. Am J Epidemiol 2014;180:1118–25. doi:10.1093/aje/kwu236
      OpenUrlAbstract/FREE Full Text
    36. ↵
      1. Thoits PA
      . Mechanisms linking social ties and support to physical and mental health. J Health Soc Behav 2011;52:145–61. doi:10.1177/0022146510395592
      OpenUrlAbstract/FREE Full Text
    37. ↵
      1. Berkman LF,
      2. Glass T,
      3. Brissette I, et al
      . From social integration to health: Durkheim in the new millennium. Soc Sci Med 2000;51:843–57. doi:10.1016/S0277-9536(00)00065-4
      OpenUrlCrossRefPubMedWeb of Science
    38. ↵
      1. Cohen S,
      2. Wills TA
      . Stress, social support, and the buffering hypothesis. Psychol Bull 1985;98:310–57. doi:10.1037/0033-2909.98.2.310
      OpenUrlCrossRefPubMedWeb of Science
    39. ↵
      1. Cacioppo JT,
      2. Cacioppo S,
      3. Capitanio JP, et al
      . Neuroendocrinology of social isolation. Annu Rev Psychol 2015;66:733–67. doi:10.1146/annurev-psych-010814-015240
      OpenUrlCrossRefPubMed
    40. ↵
      1. Segerstrom SC,
      2. Miller GE
      . Psychological stress and the human immune system: a meta-analytic study of 30 years of inquiry. Psychol Bull 2004;130:601–30. doi:10.1037/0033-2909.130.4.601
      OpenUrlCrossRefPubMedWeb of Science
    41. ↵
      1. Podsakoff PM,
      2. MacKenzie SB,
      3. Lee J-Y, et al
      . Common method biases in behavioral research: a critical review of the literature and recommended remedies. J Appl Psychol 2003;88:879–903. doi:10.1037/0021-9010.88.5.879
      OpenUrlCrossRefPubMedWeb of Science

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Contributors All authors participated in the design of the study. CH analysed the data, and MJ and ME supervised the analyses. CH wrote a first draft of the manuscript. All authors interpreted the findings, wrote and critically reviewed drafts of the manuscript, and approved the final version.

    • Funding MV is supported by the Academy of Finland (258598, 265174). MK is supported by the UK Medical Research Council (K013351), the Economic and Social Research Council, and NordForsk, the Nordic Council of Ministers (75021). JV is supported by the Academy of Finland (264944, 267727). ME is supported by the Academy of Finland (265977). The Whitehall II Study has been supported by grants from the Medical Research Council; the British Heart Foundation; the British Health and Safety Executive; the British Department of Health; the National Heart, Lung, and Blood Institute; and the National Institute on Aging, NIH.

    • Competing interests None declared.

    • Ethics approval University College London Medical School Committee on the ethics of human research.

    • Provenance and peer review Not commissioned; externally peer reviewed.