548749
research-article2014
SJP0010.1177/1403494814548749F. E. Jonsson et al.F. E. Jonsson et al.
Scandinavian Journal of Public Health, 2014; 42: 581–588
Original Article
Social capital across the life course and functional somatic symptoms in mid-adulthood
FRIDA JONSSON, Anne Hammarström & Per E. Gustafsson Department of Public Health and Clinical Medicine, Family Medicine, Umeå University, Umeå, Sweden
Abstract Aim: To examine social capital across life and functional somatic symptoms in middle-age, according to life-course models of cumulative risk and sensitive periods. Methods: Data from the 26-year prospective study the Northern Swedish Cohort enabled complete case analyses on 940 individuals (451 women and 489 men) participating in questionnaire surveys at ages 16, 21, 30 and 42. Social capital was operationalized at the individual level, comprising items on social participation, social influence and social support. Functional somatic symptoms were a summary measure of self-reported physical symptoms, palpitation and sleeping difficulties occuring during the 12 months prior to the data collection. Linear regression was used as the main statistical method, examining the relationship between functional somatic symptoms at age 42 and social capital across life. Results: Lower levels of social capital accumulated over the life course were associated with higher levels of functional somatic symptoms at age 42, for both women and men. Social capital was, especially among adolescent men, related to functional somatic symptoms at age 42, independently of social capital later in life and baseline material circumstances. Conclusions: The health impact of poor social capital may be due to accumulation across the life course and to adolescence being a particularly sensitive period. It is relevant for preventive work to acknowledge effects of social capital throughout life. Key Words: Social epidemiology, human development, life course, mental health, midlife events, models, the Northern Swedish Cohort, prospective study, social capital, somatic symptoms
Introduction A large body of mainly cross-sectional research links individual social capital to physical and mental health in adolescence and adulthood [1–4]. Whether social capital from earlier in life impacts health in adulthood is a central, but as of yet unresolved question within the field [5]. Following two life-course models, this prospective study examines if social capital, in adolescence and accumulated over the life course, relates to functional somatic symptoms in adulthood. The concept of social capital originates from ideas by Bourdieu and Coleman about social networks [6] and social structures [7], respectively, combined with research by Putnam [8] on the importance of trust and reciprocity within communities. It can be
described as “resources embedded in a social structure which are accessed and/or mobilized in purposive actions” [9], but due to its complexity, no consensus definition exists. Within research, social capital is generally differentiated as either social cohesion or networks, studied as a property of people or of communities and neighborhoods. Considering these differentiations, we studied social capital through its network aspects, approaching it as the integrated effects of social influence, social participation and social support at an individual level [10]. Social capital is assumed to benefit health by protecting against stress, creating a sense of belonging and providing information and material resources. Thus, the general assumption that social relations and networks
Correspondence: Frida Jonsson, Department of Public Health and Clinical Medicine, Family Medicine, Umeå University, SE-901 85 Umeå, Sweden. Email: [email protected] (Accepted 13 July 2014) © 2014 the Nordic Societies of Public Health DOI: 10.1177/1403494814548749
Downloaded from sjp.sagepub.com at University of New England on June 8, 2015
582 F. Jonsson et al. generate symbolic, psychological and material recourses; and that these, via strengthening and buffering mechanisms, may protect people from the pathologic consequences of long-term stress, connects social capital to health [10,11]. But while a recent meta-analysis suggests cumulative associations and a strong relationship between social capital and self-rated health, as well as mortality [2], the effects on mental health remain inconclusive [12]. Still, social capital has been found to improve mental well-being [5], while associations between poor social capital and depression also have been revealed [13]. Longitudinal research is scarce within the field, making us unaware as to whether the impact of social capital on health early in life persists until adulthood. As suggested by a review from Nyqvist et al. [5], lifecourse studies on social capital are required, which would be especially beneficial if applied on mental health outcomes. Generally defined as a clustering of physical symptoms with no or unknown pathology, functional somatic symptoms relate to mental health by being considered a measure of distress and by being closely linked to depression and anxiety [14,15]. But while this health problem is known to depend upon various social determinants [16], associations to social capital have been examined only in a few studies on adolescents [17]. In addition, although functional somatic symptoms are often a temporary state present early as well as later in life, it may be the subject of continuity along the life course, with enduring distress as a potential consequence [16,18]. Life-course epidemiology acknowledges not only contemporary exposures, but circumstances prior in life as having implications for health [19]. Through general theoretical models highlighting the importance of duration, magnitude and timing of exposure, this framework offers hypotheses on how health may develop over life. Thus, by following two conceptual life-course models, we analysed in what ways social capital across life may relate to functional somatic symptoms in adulthood. Applying the cumulative risk model, addressing whether factors act on health cumulatively [20], we proposed that social capital accumulates in number and/or severity (magnitude), but also across time (duration), with implications for mental health in mid-adulthood. This model focuses on amount of exposure, while timing of impact is ignored. In addition, the sensitive period model highlights exposure during certain life-course periods, especially childhood and adolescence as particularly critical, with enduring health effects independent from later conditions as a consequence [19]. Through this model, we hypothesized that functional somatic symptoms in adulthood were related to social capital
during adolescence, independently of social capital further along the life course. Aim Contributing to the so far nonexistent knowledge about the life-course association between social capital and health, we especially focused on its relationship to mental health and functional somatic symptoms. The aim was to examine if an association between social capital across life and functional somatic symptoms in middle-age can be explained by the cumulative risk and/or sensitive period life-course models. Methods Sample and procedures The Northern Swedish Cohort provides prospective data, collected five times (in 1981, 1983, 1986, 1995 and 2007) during 26 years, on all students whom attended or should have attended the 9th grade of the Swedish compulsory school in the municipality of Luleå, in 1981 (at that time, aged 16) [21]. Out of 1071 students, 94.3% of those alive chose to participate across the entire period (n = 1010). The present study used the self-administrated questionnaire data from the 1981, 1986, 1995 and 2007 surveys. Ethical approval was granted by the Regional Ethical Review Board in Umeå. Outcome Functional somatic symptoms during the last 12 months at age 42 formed an index (including 10 items scored 0–2; answered ‘no’, ‘yes light’ or ‘yes severe’) on different self-reported physical symptoms (headache/migraine, stomach ache, nausea, backache/hip pain/sciatica, fatigue, breathlessness, dizziness and overstrain), occurrence of palpitations and how often the respondent had suffered from difficulties sleeping. When summarized, Cronbach’s alpha revealed an internal consistency of 0.782 and the index ranged from 0–18 for women and 0–15 for men, where higher values equaled more somatic problems. Exposures Social capital was operationalized at the individual level, comprising items on social participation, social influence and social support; and aimed at broadly covering this a priori concept [10]. Items from questionnaires at ages 16, 21, 30 and 42 were combined
Downloaded from sjp.sagepub.com at University of New England on June 8, 2015
Social capital across the life course and functional somatic symptoms in mid-adulthood 583 (9, 6, 12 and 12 items selected for each age, respectively) into age-specific, as well as one cumulative measure of social capital. Included items at age 16 were: 1. Social participation, items on involvement in associational and leisure activities; 2. Social influence, items on perceived opportunity to decide in school and at home; and 3. Social support, items on perceived contact with mother, father and friends.
self-employed parents, while the manual work group represented low socioeconomic status. Material adversity at age 16 was a measurement of unfavorable material circumstances, operationalized (for details, see Gustafsson, et al. [22]) as the count (0–3) of items on: •• Parental unemployment; •• Poor material standard of living; and •• Residential crowding. Data analysis
Items selected at age 21 were: 1. Social participation involvement in associations; 2. Social influence, whether the respondent felt that he/she could decide sufficiently often, speak their mind, was appreciated by others and whether it was hard to get others to listen; and 3. Social support, (‘are you mostly alone or with friends during spare time’). The items at age 30 and 42 were identical: 1. Social participation, involvement in associations; 2. Social influence, whether the respondent felt that he/she could decide sufficiently often, speak their mind, was appreciated by others and whether it was hard to get others to listen; and 3. Social support including seven questions on the number of and perceived support from close contacts. All items had 4–6 Likert scale response options, coded from 1–6, where a higher value equaled less social capital. Each variable was then standardized and summarized into an index representing cumulative social capital. Although internal consistency was low (α < 0.05) the operationalization was guided by a theoretical framework; and an impact of exposure may still act on health through common pathways, despite a lack of co-occurrence. Covariates Functional somatic symptoms at age 16 were measured identically as at age 42, but based on questionnaires completed at age 16. Socioeconomic disadvantage was operationalized based on parent(s)-reported occupation, separating between manual-working parents (blue-collar), at least one parent being a non-manual employee (whitecollar) and self-employed. A high socioeconomic group contained those with non-manual and
Descriptive statistics for all variables are displayed in Table I. Complete case analysis was possible for 940 participants (451 women and 489 men). We examined selection bias regarding missing data by using separate simple logistic regression with inclusion/exclusion as the outcome, and main variables as the predictors. The analysis revealed no evidence of systematic drop-out (p > 0.218) with regard to social capital measures (cumulative and at age 16, 21, 30 and 42), material adversity and socioeconomic disadvantage at age 16, or sex; although the functional somatic symptoms at age 16 was borderline significant (n = 52 excluded; p = 0.080). All analyses were performed in SPSS 20 through linear regression on the full sample and stratified by sex, as both mental health problems and its determinants may differ between women and men [23]. We assessed the hypothesis of cumulative risk by regressing functional somatic symptoms at age 42 on cumulative social capital (Model 0); with additions of covariates sex, socioeconomic disadvantage and material adversity at age 16 (Model 1) and functional somatic symptoms at age 16 (Model 2). The sensitive period hypothesis tested the association between social capital at age 16 and functional somatic symptoms at age 42, with a successive introduction of social capital variables at later ages (age 21 (Model 1); 21 and 30 (Model 2); 21, 30 and 42 (Model 3)); as well as sex, socioeconomic disadvantage and material adversity at age 16 (Model 4), and baseline functional somatic symptoms (Model 5). Multi-collinearity was at moderate levels; variance inflation factor (VIF) was < 1.51 in the full sample, < 1.61 for women and < 1.45 for men. Functional somatic symptoms at age 42 were slightly positively skewed, but considered to fulfill assumptions of approximate normality, after reviewing Q-Q plot and histogram, and analyzing the residuals.
Downloaded from sjp.sagepub.com at University of New England on June 8, 2015
584 F. Jonsson et al. Table I. Descriptive statistics for levels of social capital at age 16, 21, 30 and 42, functional somatic symptoms at age 16 and 42, socioeconomic disadvantage and material adversities at age 16; in the full sample (n = 940), and between women (n = 451) and men (n = 489) Estimates are mean (SD). Full sample Social capital Age 16 Age 21 Age 30 Age 42 Cumulativec Functional somatic symptoms Age 16 Age 42 Material adversities Age 16 Socioeconomic disadvantage Age 16
Women
Men
Range
Estimate
Range
Estimate
Range
1–4.22 1–3.83 1–3.92 1–4.33 −7.2–12.12
2.89 (0.411) 2.20 (0.483) 2.11 (0.479) 2.09 (0.549) −0.004 (2.81)
1–4.22 1–3.83 1–3.92 1–4.18 −7.28–10.72
2.96 (0.380) 2.23 (0.478) 2.12 (0.463) 2.09 (0.528) 0.25 (2.73)
1–4.11 1–3.83 1–3.92 1–4.33 −7.04–12.12
0.00–16.00 0.00–18.00
3.35 (2.540) 4.24 (3.306)
0.00–16.00 3.70 (2.510) 0.00–12.00 0.00–18.00 4.77 (3.503) 0.00–15.00
0.00–3.00
0.48 (0.677)
0.00–3.00
37.7% low
0.59 (0.726) 0.00–3.00
36.6% low
39.4% low
Difference Estimate
P-value
2.83 (0.429) < 0.001a 2.16 (0.486) 0.022a 2.10 (0.494) 0.460a 2.10 (0.569) 0.852a −0.24 (2.87) 0.006a 3.02 (2.526) < 0.001a 3.75 (3.032) < 0.001a 0.39 (0.614) < 0.001a 0.354b
on t-test. on chi-square test. cIndex of standardized social capital variables. aBased
bBased
Results The descriptive analysis (Table I) showed stable, but slightly increasing levels of social capital across the life course, differing between the sexes at age 16 (p < 0.001) and 21 (p = 0.022), and with regard to cumulative social capital (p = 0.006); with men reporting significantly lower levels. Women presented higher levels (p < 0.001) of functional somatic symptoms at ages 16 and 42, but also a larger increase across life than men. Zero-order correlations (by sex) presented in Table II reveal weak-to-moderate correlations between social capital at different periods and in relation to the covariates, indicating a measure of continuity of social capital from adolescence to mid-adulthood. The cumulative risk model was examined by regressing functional somatic symptoms at age 42 on cumulative social capital (Table III). In the full sample, cumulative social capital significantly predicted higher levels of functional somatic symptoms at age 42, before (0.331, in Model 0) and after (0.287, in Model 2) controlling for sex, socioeconomic disadvantage, material adversities and baseline functional somatic symptoms. Stratifying the analysis by sex revealed similar results for women and men. To examine the sensitive period hypothesis, functional somatic symptoms at age 42 was first regressed separately on social capital at ages 16, 21, 30 and 42, using simple regression. This analysis showed significant associations between lower
levels of social capital at each age (Table IV, Model 0), with estimates being 0.180, 0.192, 0.199 and 0.346, respectively, in the full sample. Thereafter, by using multiple regression, we tested whether the relationship between social capital during adolescence and functional somatic symptoms in midadulthood existed independently of social capital across the life course. Our analysis revealed a significant association between social capital at age 16 and functional somatic symptoms at age 42, after controlling for social capital at ages 21, 30 and 42 (Models 1–3). The further addition of covariates (Model 4) did not alter the results substantially, whereas the final inclusion of baseline functional somatic symptoms (Model 5) caused a moderate attenuation with social capital at age 16 becoming borderline significant (confidence interval (CI) = -0.003-0.123). In addition, throughout the series of analyses, social capital at age 42 was found to be the main independent predictor with an estimate of 0.310 in the final model (5, top panel). The stratified analysis revealed similar patterns for men (Table IV, lower panel). Social capital at age 16 was initially significantly associated with functional somatic symptoms at age 42, independently of later social capital, and after adjusting for socioeconomic and material conditions; whereas the addition of baseline functional somatic symptoms in the final model (Model 5) made the association drop to borderline significance (CI = − 0.010–0.143). The same progression was not evident for women, where only contemporary levels of social capital remained consistently significant, throughout the series of models.
Downloaded from sjp.sagepub.com at University of New England on June 8, 2015
Social capital across the life course and functional somatic symptoms in mid-adulthood 585 Table II. Zero-order correlations (Pearson’s r) between all variables (SC, SD, MA and FSS) in women (above diagonal) and men (below diagonal). Variable
SC age 16
SC age 21
SC age 30
SC age 42
SC Cum
SD age 16
MA age 16
FSS age 16
FSS age 42
SC age 16 SC age 21 SC age 30 SC age 42 SC Cum SD age 16 MA age 16 FSS age 16 FSS age 42
–
0.302b
0.236b
0.225b
0.610b
0.100a
0.174a
0.170b
0.244b
–
0.392b
0.347b
0.724b
0.154b
0.139b
0.103b
0.231b
0.389b
–
0.556b
0.768b
0.155b
0.193b
0.134b
0.208b
0.316b
0.490b
–
0.746b
0.111b
0.087
0.085
0.607b
0.696b
0.761b
0.721b
–
0.191b
0.203b
0.172b
0.077
0.125b
0.085
0.070
0.134b
–
0.190b
0.021
0.100a
0.082
0.124b
0.154b
0.167b
0.238b
–
0.068
0.136b
0.086
0.074
0.092a
0.142b
0.069
0.077
–
0.173b
0.192b
0.190b
0.315b
0.318b
0.123b
0.092a
0.253b
0.148b 0.174b 0.208b 0.393b 0.323b 0.027 0.081 0.181b –
ap
< 0.05 (2-tailed). < 0.01 (2-tailed). Cum: Cumulative; FSS: functional somatic symptoms; MA: material adversity; SC: social capital; SD: socioeconomic disadvantage. bp
Table III. Linear regressions of functional somatic symptoms at age 42 on cumulative social capital (Model 0), adjusted for sex, socioeconomic disadvantage and material adversity at age 16 (Model 1) and functional somatic symptoms at age 16 (Model 2) in the full sample (n = 940). Analysis stratified by sex: women (n = 451) and men (n = 489), excluding sex as a covariate (Model 1). Predictor estimates are standardized regression coefficients with 95% confidence intervals. Estimate Full sample Cumulative social capital, age 16–42 Model R2 Model p Women Cumulative social capital, age 16–42 Model R2 Model p Men Cumulative social capital, age 16–42 Model R2 Model p
Model 0
Model 1
0.331 (0.271–0.392)
0.312 (0.250–0.374)
0.109 –
0.125 (0.001)
0.355 (0.260–0.451)
0.358 (0.259–0.457)
0.107 –
0.108 (0.745)
0.291 (0.215–0.368)
0.277 (0.199–0.355)
0.102 –
0.111 (0.103)
Discussion As far as we know, the present report is the first to examine the relationship between social capital over the life course and health, here exemplified by selfreported functional somatic symptoms in midadulthood. Our study suggested that social capital acts cumulatively on functional somatic symptoms across life, in both women and men. In addition, although
Model 2 0.287 (0.226–0.349) 0.152 (
research-article2014
SJP0010.1177/1403494814548749F. E. Jonsson et al.F. E. Jonsson et al.
Scandinavian Journal of Public Health, 2014; 42: 581–588
Original Article
Social capital across the life course and functional somatic symptoms in mid-adulthood
FRIDA JONSSON, Anne Hammarström & Per E. Gustafsson Department of Public Health and Clinical Medicine, Family Medicine, Umeå University, Umeå, Sweden
Abstract Aim: To examine social capital across life and functional somatic symptoms in middle-age, according to life-course models of cumulative risk and sensitive periods. Methods: Data from the 26-year prospective study the Northern Swedish Cohort enabled complete case analyses on 940 individuals (451 women and 489 men) participating in questionnaire surveys at ages 16, 21, 30 and 42. Social capital was operationalized at the individual level, comprising items on social participation, social influence and social support. Functional somatic symptoms were a summary measure of self-reported physical symptoms, palpitation and sleeping difficulties occuring during the 12 months prior to the data collection. Linear regression was used as the main statistical method, examining the relationship between functional somatic symptoms at age 42 and social capital across life. Results: Lower levels of social capital accumulated over the life course were associated with higher levels of functional somatic symptoms at age 42, for both women and men. Social capital was, especially among adolescent men, related to functional somatic symptoms at age 42, independently of social capital later in life and baseline material circumstances. Conclusions: The health impact of poor social capital may be due to accumulation across the life course and to adolescence being a particularly sensitive period. It is relevant for preventive work to acknowledge effects of social capital throughout life. Key Words: Social epidemiology, human development, life course, mental health, midlife events, models, the Northern Swedish Cohort, prospective study, social capital, somatic symptoms
Introduction A large body of mainly cross-sectional research links individual social capital to physical and mental health in adolescence and adulthood [1–4]. Whether social capital from earlier in life impacts health in adulthood is a central, but as of yet unresolved question within the field [5]. Following two life-course models, this prospective study examines if social capital, in adolescence and accumulated over the life course, relates to functional somatic symptoms in adulthood. The concept of social capital originates from ideas by Bourdieu and Coleman about social networks [6] and social structures [7], respectively, combined with research by Putnam [8] on the importance of trust and reciprocity within communities. It can be
described as “resources embedded in a social structure which are accessed and/or mobilized in purposive actions” [9], but due to its complexity, no consensus definition exists. Within research, social capital is generally differentiated as either social cohesion or networks, studied as a property of people or of communities and neighborhoods. Considering these differentiations, we studied social capital through its network aspects, approaching it as the integrated effects of social influence, social participation and social support at an individual level [10]. Social capital is assumed to benefit health by protecting against stress, creating a sense of belonging and providing information and material resources. Thus, the general assumption that social relations and networks
Correspondence: Frida Jonsson, Department of Public Health and Clinical Medicine, Family Medicine, Umeå University, SE-901 85 Umeå, Sweden. Email: [email protected] (Accepted 13 July 2014) © 2014 the Nordic Societies of Public Health DOI: 10.1177/1403494814548749
Downloaded from sjp.sagepub.com at University of New England on June 8, 2015
582 F. Jonsson et al. generate symbolic, psychological and material recourses; and that these, via strengthening and buffering mechanisms, may protect people from the pathologic consequences of long-term stress, connects social capital to health [10,11]. But while a recent meta-analysis suggests cumulative associations and a strong relationship between social capital and self-rated health, as well as mortality [2], the effects on mental health remain inconclusive [12]. Still, social capital has been found to improve mental well-being [5], while associations between poor social capital and depression also have been revealed [13]. Longitudinal research is scarce within the field, making us unaware as to whether the impact of social capital on health early in life persists until adulthood. As suggested by a review from Nyqvist et al. [5], lifecourse studies on social capital are required, which would be especially beneficial if applied on mental health outcomes. Generally defined as a clustering of physical symptoms with no or unknown pathology, functional somatic symptoms relate to mental health by being considered a measure of distress and by being closely linked to depression and anxiety [14,15]. But while this health problem is known to depend upon various social determinants [16], associations to social capital have been examined only in a few studies on adolescents [17]. In addition, although functional somatic symptoms are often a temporary state present early as well as later in life, it may be the subject of continuity along the life course, with enduring distress as a potential consequence [16,18]. Life-course epidemiology acknowledges not only contemporary exposures, but circumstances prior in life as having implications for health [19]. Through general theoretical models highlighting the importance of duration, magnitude and timing of exposure, this framework offers hypotheses on how health may develop over life. Thus, by following two conceptual life-course models, we analysed in what ways social capital across life may relate to functional somatic symptoms in adulthood. Applying the cumulative risk model, addressing whether factors act on health cumulatively [20], we proposed that social capital accumulates in number and/or severity (magnitude), but also across time (duration), with implications for mental health in mid-adulthood. This model focuses on amount of exposure, while timing of impact is ignored. In addition, the sensitive period model highlights exposure during certain life-course periods, especially childhood and adolescence as particularly critical, with enduring health effects independent from later conditions as a consequence [19]. Through this model, we hypothesized that functional somatic symptoms in adulthood were related to social capital
during adolescence, independently of social capital further along the life course. Aim Contributing to the so far nonexistent knowledge about the life-course association between social capital and health, we especially focused on its relationship to mental health and functional somatic symptoms. The aim was to examine if an association between social capital across life and functional somatic symptoms in middle-age can be explained by the cumulative risk and/or sensitive period life-course models. Methods Sample and procedures The Northern Swedish Cohort provides prospective data, collected five times (in 1981, 1983, 1986, 1995 and 2007) during 26 years, on all students whom attended or should have attended the 9th grade of the Swedish compulsory school in the municipality of Luleå, in 1981 (at that time, aged 16) [21]. Out of 1071 students, 94.3% of those alive chose to participate across the entire period (n = 1010). The present study used the self-administrated questionnaire data from the 1981, 1986, 1995 and 2007 surveys. Ethical approval was granted by the Regional Ethical Review Board in Umeå. Outcome Functional somatic symptoms during the last 12 months at age 42 formed an index (including 10 items scored 0–2; answered ‘no’, ‘yes light’ or ‘yes severe’) on different self-reported physical symptoms (headache/migraine, stomach ache, nausea, backache/hip pain/sciatica, fatigue, breathlessness, dizziness and overstrain), occurrence of palpitations and how often the respondent had suffered from difficulties sleeping. When summarized, Cronbach’s alpha revealed an internal consistency of 0.782 and the index ranged from 0–18 for women and 0–15 for men, where higher values equaled more somatic problems. Exposures Social capital was operationalized at the individual level, comprising items on social participation, social influence and social support; and aimed at broadly covering this a priori concept [10]. Items from questionnaires at ages 16, 21, 30 and 42 were combined
Downloaded from sjp.sagepub.com at University of New England on June 8, 2015
Social capital across the life course and functional somatic symptoms in mid-adulthood 583 (9, 6, 12 and 12 items selected for each age, respectively) into age-specific, as well as one cumulative measure of social capital. Included items at age 16 were: 1. Social participation, items on involvement in associational and leisure activities; 2. Social influence, items on perceived opportunity to decide in school and at home; and 3. Social support, items on perceived contact with mother, father and friends.
self-employed parents, while the manual work group represented low socioeconomic status. Material adversity at age 16 was a measurement of unfavorable material circumstances, operationalized (for details, see Gustafsson, et al. [22]) as the count (0–3) of items on: •• Parental unemployment; •• Poor material standard of living; and •• Residential crowding. Data analysis
Items selected at age 21 were: 1. Social participation involvement in associations; 2. Social influence, whether the respondent felt that he/she could decide sufficiently often, speak their mind, was appreciated by others and whether it was hard to get others to listen; and 3. Social support, (‘are you mostly alone or with friends during spare time’). The items at age 30 and 42 were identical: 1. Social participation, involvement in associations; 2. Social influence, whether the respondent felt that he/she could decide sufficiently often, speak their mind, was appreciated by others and whether it was hard to get others to listen; and 3. Social support including seven questions on the number of and perceived support from close contacts. All items had 4–6 Likert scale response options, coded from 1–6, where a higher value equaled less social capital. Each variable was then standardized and summarized into an index representing cumulative social capital. Although internal consistency was low (α < 0.05) the operationalization was guided by a theoretical framework; and an impact of exposure may still act on health through common pathways, despite a lack of co-occurrence. Covariates Functional somatic symptoms at age 16 were measured identically as at age 42, but based on questionnaires completed at age 16. Socioeconomic disadvantage was operationalized based on parent(s)-reported occupation, separating between manual-working parents (blue-collar), at least one parent being a non-manual employee (whitecollar) and self-employed. A high socioeconomic group contained those with non-manual and
Descriptive statistics for all variables are displayed in Table I. Complete case analysis was possible for 940 participants (451 women and 489 men). We examined selection bias regarding missing data by using separate simple logistic regression with inclusion/exclusion as the outcome, and main variables as the predictors. The analysis revealed no evidence of systematic drop-out (p > 0.218) with regard to social capital measures (cumulative and at age 16, 21, 30 and 42), material adversity and socioeconomic disadvantage at age 16, or sex; although the functional somatic symptoms at age 16 was borderline significant (n = 52 excluded; p = 0.080). All analyses were performed in SPSS 20 through linear regression on the full sample and stratified by sex, as both mental health problems and its determinants may differ between women and men [23]. We assessed the hypothesis of cumulative risk by regressing functional somatic symptoms at age 42 on cumulative social capital (Model 0); with additions of covariates sex, socioeconomic disadvantage and material adversity at age 16 (Model 1) and functional somatic symptoms at age 16 (Model 2). The sensitive period hypothesis tested the association between social capital at age 16 and functional somatic symptoms at age 42, with a successive introduction of social capital variables at later ages (age 21 (Model 1); 21 and 30 (Model 2); 21, 30 and 42 (Model 3)); as well as sex, socioeconomic disadvantage and material adversity at age 16 (Model 4), and baseline functional somatic symptoms (Model 5). Multi-collinearity was at moderate levels; variance inflation factor (VIF) was < 1.51 in the full sample, < 1.61 for women and < 1.45 for men. Functional somatic symptoms at age 42 were slightly positively skewed, but considered to fulfill assumptions of approximate normality, after reviewing Q-Q plot and histogram, and analyzing the residuals.
Downloaded from sjp.sagepub.com at University of New England on June 8, 2015
584 F. Jonsson et al. Table I. Descriptive statistics for levels of social capital at age 16, 21, 30 and 42, functional somatic symptoms at age 16 and 42, socioeconomic disadvantage and material adversities at age 16; in the full sample (n = 940), and between women (n = 451) and men (n = 489) Estimates are mean (SD). Full sample Social capital Age 16 Age 21 Age 30 Age 42 Cumulativec Functional somatic symptoms Age 16 Age 42 Material adversities Age 16 Socioeconomic disadvantage Age 16
Women
Men
Range
Estimate
Range
Estimate
Range
1–4.22 1–3.83 1–3.92 1–4.33 −7.2–12.12
2.89 (0.411) 2.20 (0.483) 2.11 (0.479) 2.09 (0.549) −0.004 (2.81)
1–4.22 1–3.83 1–3.92 1–4.18 −7.28–10.72
2.96 (0.380) 2.23 (0.478) 2.12 (0.463) 2.09 (0.528) 0.25 (2.73)
1–4.11 1–3.83 1–3.92 1–4.33 −7.04–12.12
0.00–16.00 0.00–18.00
3.35 (2.540) 4.24 (3.306)
0.00–16.00 3.70 (2.510) 0.00–12.00 0.00–18.00 4.77 (3.503) 0.00–15.00
0.00–3.00
0.48 (0.677)
0.00–3.00
37.7% low
0.59 (0.726) 0.00–3.00
36.6% low
39.4% low
Difference Estimate
P-value
2.83 (0.429) < 0.001a 2.16 (0.486) 0.022a 2.10 (0.494) 0.460a 2.10 (0.569) 0.852a −0.24 (2.87) 0.006a 3.02 (2.526) < 0.001a 3.75 (3.032) < 0.001a 0.39 (0.614) < 0.001a 0.354b
on t-test. on chi-square test. cIndex of standardized social capital variables. aBased
bBased
Results The descriptive analysis (Table I) showed stable, but slightly increasing levels of social capital across the life course, differing between the sexes at age 16 (p < 0.001) and 21 (p = 0.022), and with regard to cumulative social capital (p = 0.006); with men reporting significantly lower levels. Women presented higher levels (p < 0.001) of functional somatic symptoms at ages 16 and 42, but also a larger increase across life than men. Zero-order correlations (by sex) presented in Table II reveal weak-to-moderate correlations between social capital at different periods and in relation to the covariates, indicating a measure of continuity of social capital from adolescence to mid-adulthood. The cumulative risk model was examined by regressing functional somatic symptoms at age 42 on cumulative social capital (Table III). In the full sample, cumulative social capital significantly predicted higher levels of functional somatic symptoms at age 42, before (0.331, in Model 0) and after (0.287, in Model 2) controlling for sex, socioeconomic disadvantage, material adversities and baseline functional somatic symptoms. Stratifying the analysis by sex revealed similar results for women and men. To examine the sensitive period hypothesis, functional somatic symptoms at age 42 was first regressed separately on social capital at ages 16, 21, 30 and 42, using simple regression. This analysis showed significant associations between lower
levels of social capital at each age (Table IV, Model 0), with estimates being 0.180, 0.192, 0.199 and 0.346, respectively, in the full sample. Thereafter, by using multiple regression, we tested whether the relationship between social capital during adolescence and functional somatic symptoms in midadulthood existed independently of social capital across the life course. Our analysis revealed a significant association between social capital at age 16 and functional somatic symptoms at age 42, after controlling for social capital at ages 21, 30 and 42 (Models 1–3). The further addition of covariates (Model 4) did not alter the results substantially, whereas the final inclusion of baseline functional somatic symptoms (Model 5) caused a moderate attenuation with social capital at age 16 becoming borderline significant (confidence interval (CI) = -0.003-0.123). In addition, throughout the series of analyses, social capital at age 42 was found to be the main independent predictor with an estimate of 0.310 in the final model (5, top panel). The stratified analysis revealed similar patterns for men (Table IV, lower panel). Social capital at age 16 was initially significantly associated with functional somatic symptoms at age 42, independently of later social capital, and after adjusting for socioeconomic and material conditions; whereas the addition of baseline functional somatic symptoms in the final model (Model 5) made the association drop to borderline significance (CI = − 0.010–0.143). The same progression was not evident for women, where only contemporary levels of social capital remained consistently significant, throughout the series of models.
Downloaded from sjp.sagepub.com at University of New England on June 8, 2015
Social capital across the life course and functional somatic symptoms in mid-adulthood 585 Table II. Zero-order correlations (Pearson’s r) between all variables (SC, SD, MA and FSS) in women (above diagonal) and men (below diagonal). Variable
SC age 16
SC age 21
SC age 30
SC age 42
SC Cum
SD age 16
MA age 16
FSS age 16
FSS age 42
SC age 16 SC age 21 SC age 30 SC age 42 SC Cum SD age 16 MA age 16 FSS age 16 FSS age 42
–
0.302b
0.236b
0.225b
0.610b
0.100a
0.174a
0.170b
0.244b
–
0.392b
0.347b
0.724b
0.154b
0.139b
0.103b
0.231b
0.389b
–
0.556b
0.768b
0.155b
0.193b
0.134b
0.208b
0.316b
0.490b
–
0.746b
0.111b
0.087
0.085
0.607b
0.696b
0.761b
0.721b
–
0.191b
0.203b
0.172b
0.077
0.125b
0.085
0.070
0.134b
–
0.190b
0.021
0.100a
0.082
0.124b
0.154b
0.167b
0.238b
–
0.068
0.136b
0.086
0.074
0.092a
0.142b
0.069
0.077
–
0.173b
0.192b
0.190b
0.315b
0.318b
0.123b
0.092a
0.253b
0.148b 0.174b 0.208b 0.393b 0.323b 0.027 0.081 0.181b –
ap
< 0.05 (2-tailed). < 0.01 (2-tailed). Cum: Cumulative; FSS: functional somatic symptoms; MA: material adversity; SC: social capital; SD: socioeconomic disadvantage. bp
Table III. Linear regressions of functional somatic symptoms at age 42 on cumulative social capital (Model 0), adjusted for sex, socioeconomic disadvantage and material adversity at age 16 (Model 1) and functional somatic symptoms at age 16 (Model 2) in the full sample (n = 940). Analysis stratified by sex: women (n = 451) and men (n = 489), excluding sex as a covariate (Model 1). Predictor estimates are standardized regression coefficients with 95% confidence intervals. Estimate Full sample Cumulative social capital, age 16–42 Model R2 Model p Women Cumulative social capital, age 16–42 Model R2 Model p Men Cumulative social capital, age 16–42 Model R2 Model p
Model 0
Model 1
0.331 (0.271–0.392)
0.312 (0.250–0.374)
0.109 –
0.125 (0.001)
0.355 (0.260–0.451)
0.358 (0.259–0.457)
0.107 –
0.108 (0.745)
0.291 (0.215–0.368)
0.277 (0.199–0.355)
0.102 –
0.111 (0.103)
Discussion As far as we know, the present report is the first to examine the relationship between social capital over the life course and health, here exemplified by selfreported functional somatic symptoms in midadulthood. Our study suggested that social capital acts cumulatively on functional somatic symptoms across life, in both women and men. In addition, although
Model 2 0.287 (0.226–0.349) 0.152 (
