Adverse life events are among the best-known risk factors for common mental health problems across the life-course.Reference Kendler, Karkowski and Prescott1, Reference Miloyan, Bienvenu, Brilot and Eaton2 Positive social relationships may mitigate such adverse effects by providing networks of social ties and sources of emotional, informational and practical support.Reference Cohen and Wills3 However, some key issues remain to be clarified. First, although the majority of adults have some positive social relationships in their lives, it is unclear whether higher levels of personal support, including both qualitative and structural aspects of social relationships, confer additional benefits especially perhaps in the face of higher levels of stress. Such effects have been found for other health-promoting factors, for example, engaging in higher levels of physical activity provides additional benefits to health;4 we hypothesised that similar processes operate in relation to a combination of quality and the extent of individuals' social relationships, which we label here social resources. Second, current evidence rests heavily on cross-sectional studies, where the direction of association is uncertain:Reference Gariépy, Honkaniemi and Quesnel-Vallée5 mental health problems may have an impact on the social resources available to individuals, as well as the reverse. Third, although child and adolescent precursors of social resources in earlier adulthood have been identified,Reference Pulkkinen, Nygren and Kokko6, Reference Englund, Kuo, Puig and Collins7 little is known about whether they remain important at mid-life. Long-term longitudinal studies are needed to shed light on these issues and derive pointers for interventions.
In this study, we used data from a prospective national birth cohort to investigate whether higher than average levels of adult social resources confer additional benefits to mental health among individuals experiencing stressful life events. Specifically, we examined the concurrent association between variation in social resources and affective symptoms at age 45. We further investigated prospective associations between resources at age 45 and psychological distress at age 50. In both cases, we took account of variations in prior psychological distress to clarify the direction of associations. Finally, we used follow-back analyses to explore child, adolescent and early adult precursors of higher than average social resources at mid-life.
Method
Participants
Participants were from the National Child Development Study, which surveyed 18 558 babies born in 1 week in 1958 (98% of live births) in England, Scotland and Wales.Reference Power and Elliott8 Subsequent sweeps took place when participants were ages 7, 11 and 16 during childhood, and throughout adulthood at ages 23, 33, 42, 45, 50 and 55. At age 45, a subsample of over 9000 participants took part in a survey designed to provide more objective measures of biomedical risk, from which several of this study's measures were derived. We drew on data from the childhood, adolescent, age 42 and age 50 sweeps (accessed adhering to the terms of the Economic and Social Data Service End User License agreement), and the age 45 biomedical sweep, which had more stringent access requirements covered by a ‘special licence’.
Measures
Social resources
There is considerable variation in how social resources are conceptualised and measured across studies.Reference Valtorta, Kanaan, Gilbody and Hanratty9 By drawing on the Close Persons QuestionnaireReference Stansfeld and Marmot10 and the Berkman-Syme Social Network Index,Reference Berkman and Syme11 we chose to incorporate both structural and qualitative aspects of a person's social resources. At age 45, the Close Persons Questionnaire was used to assess the quality of personal support received from the individual reported as being closest to the participant during the previous 12-month period. In most cases, this person was a spouse (79.7%), followed by a parent (3.8%), an offspring (3.6%) or a sibling (3.3%). The questionnaire captures three main constructs of social support: confiding/emotional (for example ‘In the last 12 months, could you rely on this person when needed?’), practical (for example ‘In the last 12 months, did this person give you help with small things when needed?’) and negative (for example ‘In the last 12 months, did talking to this person make things worse?’). Items assessing negative aspects of social support were reverse-coded to ensure higher scores represented more positive support. A selection of items from the Berkman-Syme Social Network Index were also included to assess the structural aspects of a person's social resources. The measure assessed the number of social ties, and the frequency of contact with family, friends and acquaintances outside the household, as well as the degree of participation in social, recreational or political groups.
A summed score of items from both measures included confiding/emotional, practical and reverse-coded negative items assessing personal support received by the individual, along with items assessing their wider social network, to create an overall indicator of social resources (mean 49.27, s.d. = 8.51, range 13–70). Higher scores represent better-quality support, as well as a greater number and more frequent contact with social ties. The reliability of the scale was α = 0.80. To facilitate comparisons among individuals with differing levels of social resources, we trichotomised the indexReference Stouthamer-Loeber, Loeber, Farrington, Zhang, van Kammen and Maguin12, Reference Brumley and Jaffee13 to identify those with poor (bottom 25%, range 13–44), typical (middle 50%, range 45–55) and rich (top 25%, range 55.5–70) resources. People with typical social resources had some contact with a wider network of social ties and received good personal support. We refer to those in the upper quarter of the distribution as having ‘rich’ social resources as they represent individuals who had more frequent contact with a larger network and received better-quality support than those with typical resources. We refer to those in the bottom quarter of the distribution as having ‘poor’ social resources, as they had relatively less contact with a smaller network and received poorer-quality support than those with typical resources.
Men and women were equally likely to have rich resources (odds ratios (OR) = 1.05, 95% CI 0.90–1.22, P>0.05). We did not find significant interactions between gender and social resources in association with affective symptoms at age 45 (χ2 = 2.29, P>0.05) and psychological distress at age 50 (χ2 = 0.79, P>0.05). Therefore, we presented findings for men and women combined.
Stressful life events
Stressful life events experienced in the past 6 months were self-reported at age 45, including events concerning health (for example ‘Have you yourself suffered serious illness, injury or assault?’), employment (for example ‘Were you sacked from your job?’), criminality (for example ‘Have you had problems with police and a court appearance?’), partnerships (for example ‘Have you broken off a steady relationship?’) and other relationships (for example ‘Has a close family friend or another relative died?’). Nearly half of the total sample reported no exposure to adverse events in the past 6 months, 41.3% reported experiencing one or two, and 13.4% had experienced three or more.
Mental health symptoms
Affective symptoms were assessed at age 45 using the Clinical Interview Schedule-Revised (CIS-R), a validated measure designed to identify common mental disorders using structured interviews.Reference Lewis, Pelosi, Araya and Dunn14 A shortened version of the CIS-R was administered in this sample, focusing on symptoms of anxiety and depression experienced in the past week.
Psychological distress was assessed at age 42 and at the age-50 follow-ups using a nine-item version of the Malaise Inventory;Reference Rutter, Tizard and Whitmore15 this commonly used, self-reported screening tool has robust psychometric properties in this sample.Reference Rodgers, Pickles, Power, Collishaw and Maughan16
Precursors
A comprehensive range of child, family and adult factors were examined as potential precursors of social resources at mid-life and are listed and described in supplementary Table 1 available at https://doi.org/10.1192/bjp.2019.25.
Statistical analyses
We tested the association between different levels of social resources and mental health symptoms among 4997 participants who reported exposure to either one or two stressful life events, or three or more. We used negative binomial regressions to account for the overdispersed distribution of the outcome measures.
First, we examined concurrent associations by comparing levels of affective symptoms at age 45 between individuals with typical and rich levels of social resources with people with poor levels. We made a further comparison between individuals with typical and rich levels of resources. Second, we tested longitudinal associations between different levels of social resources and psychological distress at age 50. We further adjusted for variations in prior mental health symptoms to shed light on the direction of association between social resources and mental health symptoms at ages 45 and 50. Third, we conducted multinomial logistic regression analyses to examine a range of potential child, family and adult precursors among the full range of participants with available data on social resources at mid-life (n = 8507). We aimed to identify domains of precursors that were associated with an increased likelihood of having rich resources, compared with typical levels, while also reducing the risk of having poor resources. We examined each domain separately by conducting a series of bivariate analyses with each variable.
In a last step, we repeated the initial analyses looking at social resources and mental health outcomes adding controls for the significant precursors of social resources. We adjusted all analyses for gender and social class at age 42 (i.e. professional, managerial and technical, skilled non-manual, skilled manual, partly skilled and unskilled occupations) to account for differences in mental health symptoms experienced by men and women, and across varying levels of social class. Stata V.15.0 was used for all analyses.
Attrition
Sample retention rates were high in childhood,Reference Englund, Kuo, Puig and Collins7 but fell during adulthood. Data were available on 53% of participants at age 50 (9788/18 558). We used logistic regression analyses to predict availability of complete data at mid-life; data availability was unrelated to exposure to stress and social resources at mid-life, but was predicted by male gender, low parental social class and lower child reading scores. We created outcome-specific inverse probability weightsReference Seaman and White17 using the variables listed above, in order to take some account of bias associated with missingness. We used these weights in all analyses.
Results
As expected, levels of affective symptoms increased with greater exposure to stressful life events (Table 1). Higher levels of social resources were associated with lower levels of prior and subsequent psychological distress, and with lower levels of contemporaneous affective symptoms. Social resources were not associated with stressful life events.
Table 1 Correlations between social resources, stressful life events and mental health symptoms
Significant findings are in bold, P<0.001.
Benefits of rich social resources on concurrent affective symptoms
Individuals with higher levels of social resources had lower levels of affective symptoms than those with poorer resources (Fig. 1(a)). Among individuals exposed to one or two stressful life events, having either rich or typical levels of social resources was associated with a reduction in affective symptoms at age 45, compared with having poor resources (see supplementary Table 2). Rich resources did not confer any additional benefits by contrast with typical levels. However, among individuals exposed to three or more stressors, only rich social resources conferred benefits in terms of symptom levels: individuals with typical and poor resources had similarly high levels of symptoms, whereas it was only those with rich resources that had lower levels in this group.
Fig. 1 Mean scores for levels of social resources by stressful life events in (a) affective symptoms and (b) psychological distress.
*P<0.05, **P<0.01, ***P<0.001.
Benefits of rich social resources on subsequent psychological distress
To assess whether the beneficial effects of rich social resources persisted over time, we examined levels of psychological distress 5 years after individuals were exposed to stress. The findings followed a similar pattern to those of the concurrent analyses. Among individuals who were previously exposed to one or two stressful life events, both rich and typical levels of social resources were associated with lower levels of psychological distress; rich resources did not confer any additional benefits (Fig. 1(b)). However, among individuals previously exposed to three or more stressful life events, once again it was only those people with rich social resources who showed evidence of benefits: individuals with typical and poor resources had similarly high levels of symptoms, whereas lower levels of psychological distress were only observed among those with rich resources.
Effects of accounting for prior psychological distress
We further tested whether the benefits conferred by rich social resources remained after adjusting for variations in individuals' prior psychological distress assessed at age 42 (see supplementary Table 2). The benefits associated with having rich resources were reduced but remained. Overall, findings show that social resources reduced the risk of both concurrent and later symptoms, even for those who experienced prior psychological distress.
Life-course precursors of rich social resources
Our analyses identified precursors of rich mid-life social resources across development; they included greater sociability in childhood (for example seeing friends outside school) and in adolescence (for example going to discos or parties), as well as better-quality social resources earlier in adult life (such as confiding relationships) (Table 2). We tested the extent to which the effect of social resources on mid-life outcomes was attributable to these earlier precursors, focusing in particular on developmental indicators of sociability in childhood and adolescence (Table 3). Associations between social resources and measures of affective symptoms at mid-life became non-significant after adjusting for earlier sociability (i.e. seeing friends outside school in childhood and going to discos or parties during adolescence); although social resources are helpful at mid-life, determinants appear to be rooted in early life.
Table 2 Life-course precursors of social resources at age 45a
a. Weighted relative risk ratios are reported. Significant findings are in bold.
* P<0.05, **P<0.01, ***P<0.001.
Table 3 Group comparisons across levels of social resources for participants exposed to (a) one or two, or (b) three or more stressful life events, adjusted for earlier sociabilitya
a. Earlier sociability included the following variables: seeing friends outside school in childhood and going to discos or parties during adolescence. Analyses are adjusted for gender, social class and prior mental health symptoms. Weighted incidence rate ratios are reported. No results were significant at the P<0.05 level.
Discussion
Our findings in a large-scale, representative cohort underscore the importance of more and good-quality social resources in mitigating mental health difficulties under particularly stressful circumstances. We also contribute longitudinal evidence to show that individuals can benefit from positive social resources in the longer term, as effects on mental health persisted 5 years following exposure to stress. In addition, we provide evidence that social resources mitigate risk for mental health symptoms even taking account of prior mental health. By taking a life-course approach, we show that individuals who display sociability during childhood and adolescence have the most advantageous social resources by mid-life. This finding highlights that there may be advantages to intervening prior to adulthood, to ensure individuals develop social resources that help them cope with stress later in life.
Benefits conferred by rich social resources
Previous studies commonly used either dichotomous or linear approaches to characterise variations in the size of networks, frequency of contacts and the degree of personal support that individuals received.Reference Bélanger, Ahmed, Vafaei, Curcio, Phillips and Zunzunegui18, Reference Pettit, Roberts, Lewinsohn, Seeley and Yaroslavsky19 We tested a more fine-grained approach, differentiating people with rich social resources from those with poor and more typical levels. This distinction mapped on to the distribution of social resources represented in the population we studied. It also made it possible to test whether the social resources available to the majority of adults are sufficient in mitigating risks, or whether richer resources are needed under especially stressful circumstances. Taking this approach, we corroborated previous findings that social resources have their limits,Reference Burton, Stice and Seeley20 as typical levels did not mitigate risks under particularly stressful circumstances. As most prior studies did not explicitly differentiate between rich and typical levels of social resources, the association between stress and social resources proved to be more nuanced in our case. Namely, we show that rich social resources are beneficial – and indeed necessary – under particularly stressful circumstances.
Our study contributes new evidence that people who are best able to weather exposure to multiple stressors are those who have access to better-quality emotional and practical support from at least one person – typically their spouse or partner – as well as a broader network of social ties including friends, relatives and social groups. This suggests that a combination of better-quality personal support and regular contact with a larger network of social ties is important. In practice, of course, these two aspects of social resources are often linked. Individuals who have regular contact with a larger network of social ties may have greater potential to access different types of support from several sources, and may consequently be better able to cope with stress.Reference Ostberg and Lennartsson21 This may be especially important in the case of relationship breakdowns or losses, which are not only stressors in themselves but may also entail a loss of key emotional and practical support. An individual with a larger social network may be best able to compensate for losses of this kind by drawing on the resources available to them from their wider social network.
Persisting effects of rich social resources
Our findings also show that individuals who have regular contact with a larger network of social ties and more supportive relationships reap the benefits, not only at the time of exposure to stress, but also over the longer term. This was the case – although associations were statistically marginal – even for individuals who might be expected to be more vulnerable because of their prior mental health symptoms, and consequently may be more likely to be exposed to stressReference Hammen22 and receive less support.Reference Hakulinen, Pulkki-Råback, Jokela, Ferrie, Aalto and Virtanen23 There may be a variety of mechanisms involved in explaining how social resources yield health advantages in the longer term. Supportive relationships may positively influence individuals' response to stress by promoting use of healthy behaviours and discouraging risky behaviours; they may also bolster individuals' beliefs that they are able to cope effectively with stress, and instil a sense of responsibility towards their social ties (for example needing to stay healthy to provide for a family).Reference Thoits24, Reference Umberson, Crosnoe and Reczek25 It has also been proposed that social resources may reduce the burden of allostatic load – the cumulative ‘wear and tear’ on physiological systems as a result of exposure to stress – on health.Reference Wiley, Bei, Bower and Stanton26 One such mechanism could be the reduction of harmful dysregulation of the hypothalamic–pituitary–adrenal system through altering the individual's appraisal of stress.
Life-course precursors of rich social resources
Whereas the majority of existing studies tend to follow-up into earlier stages of adulthood only, we show that early precursors may continue to be important for developing good social resources up to and including mid-life. We show that more sociable children and adolescents have more and better-quality social resources by mid-life, adding further evidence to the growing body of studies showing that precursors of adult social resources are rooted in early life. Our findings also corroborate and further research showing that children and adolescents who spend more time with their peers procure health benefits in later life.Reference Cundiff and Matthews27
Equipping less sociable children with the skills to initiate and maintain relationships with their peers may have far-reaching benefits across the life-course. It may also be possible to strengthen social resources at older ages as our findings showed that adults with several sources of personal support prior to mid-life (typically their partner or spouse, parents or in-laws, other relatives, and friends or neighbours) were also more likely to have good social resources later in life. Interventions designed to strengthen adult social resources may be useful in ensuring individuals have the tools to elicit more or better-quality support from their existing networks when needed.Reference Brand, Lakey and Berman28
Limitations
Our study has some limitations. First, we were only able to examine longitudinal effects at a 5-year follow-up at mid-life. Although we showed that social resources are beneficial several years later, the extent to which these effects persist over longer time periods remains uncertain. This limitation is particularly salient given that the social resources available to individuals may become more limited with advancing age, because of health problems and the death of family. Second, our study did not include repeated measures of social resources at mid-life and so it remains unclear how the observed long-term effects operated. Cohorts with measures of social support assessed at different stages in the life-course would provide opportunities for the exploration of these issues.
Furthermore, the measure of support received focused on only one nominated person (typically the spouse or partner), thereby precluding reports of multiple sources of support. However, spouses and partners are the most probable source of support for adults and are most consistently shown to be important in protecting against depression.Reference Gariépy, Honkaniemi and Quesnel-Vallée5 We conducted sensitivity analyses to take account of whether individuals were in a partnership at age 45, and the findings remained unchanged (see supplementary Table 3).
Third, our study focused on indicators of mental health outcomes at mid-life, based on the measures available in this 50-year longitudinal study. Future studies should explore the effect of social resources – in the context of stress – on other mental health outcomes. Fourth, common method variance may confound the results as we used self-reported information to identify stress exposure, levels of social resources and mental health symptoms. It may be useful for future studies to gather data from multiple informants to increase the robustness of findings. Fifth, attrition was unavoidable in our 50-year-long study. We used inverse probability weights to take some account of any associated selection bias. Sixth, our study did not assess the role of online communities and social media platforms. This is particularly important as young people today increasingly use these platforms to create and maintain social ties.
Implications
Our study has some clinical implications. We show that both the quality and quantity of social resources available to individuals may be important resilience-related factors. It may therefore be beneficial for clinicians to assess an individual's close relationships, as well as the extent to which they engage with a wider network of social ties. This is especially relevant if the individual has been exposed to several stressful life events in quick succession, as the social resources available to the majority of people may no longer be effective in mitigating risk. Under these circumstances, prompt intervention is required to prevent the onset, or worsening, of mental health problems. Effective interventions for mental health include those designed to increase opportunities for individuals to elicit support from their community,Reference Dawson, Bryant, Harper, Kuowei, Rahman and Schafer29, Reference Hogan, Linden and Najarian30 underscoring the importance of encouraging activities that enhance social inclusion as outlined by the Care Programme Approach framework. Examples of such interventions include activities around social skills, training in assertiveness and conflict resolution and group interventions that provide support through peers. Promoting social inclusion is increasingly important given recent societal changes, including technological advances, greater geographic mobility and rising economic pressures. Some individuals may consequently be at greater risk of being socially isolated (for example those who are unemployed) and experiencing mental health symptoms upon exposure to stress.
Our findings also emphasise the need for further longitudinal research to better understand the complex interplay between social resources, stress and mental health. Longitudinal studies offer unique opportunities to assess changes in the nature of a person's social resources across age, which may inform a more targeted approach to addressing deficits in support at different life stages. It may also be possible to examine the extent to which interventions designed to improve access to support (such as peer support programmes) engender resilience when people are exposed to stress.
Although bolstering adult social resources at the time of exposure to stress is worthwhile, our findings suggest that early interventions may provide the best opportunity to benefit future mental health. Schools and educational professionals should be aware that children with poorer social skills may be less well-equipped to cope with stress later in life. Engaging children in structured activities such as volunteering and active citizenship is one such intervention that provides opportunities for children to lead or participate in social action projects in the community (for example National Citizen Service and Supporting Inclusion Programme). Another example is peer-mentoring by older children matched with younger mentees based on similar characteristics such as gender, hobbies, personalities or academic subjects. Children may benefit from school-based (for example aggression or bullying prevention) or out-of-school (for example mentoring or arts/sports-based activities) social and emotional learning interventions.Reference Clarke, Morreale, Field, Hussein and Barry31 Such interventions may provide children with the skills to forge more and better-quality social ties with others well into adulthood.
Funding
The work was supported by the Economic and Social Research Council (grant number ES/P010113/1). The biomedical examination in the National Child Development Study was funded by the Medical Research Council (MRC) grant G0000934. The funders did not have any role in the analysis, interpretation of the results or preparation of the manuscript. L.A. is the Economic and Social Research Council Mental Health Leadership Fellow.
Acknowledgements
We thank all the participants in the National Child Development Study; the Centre for Longitudinal Studies (CLS), UCL Institute of Education for use of the data; and the Economic and Social Data Service (ESDS) for facilitating access to the data.
Supplementary material
Supplementary material is available online at https://doi.org/10.1192/bjp.2019.25.
eLetters
No eLetters have been published for this article.