1. Introduction
The demographic change with an increased number of older people is an important factor for public health. Previous findings show for example that experiencing greater enjoyment in life may predict more years in good health [
1]. However, the findings are still inconsistent, and it is crucial to find predictors for healthy aging. The concept of “healthy aging” concerns several determinants such as personal and behavioral factors and the social as well as the physical environment [
2]. Specifically, the importance of having a life-course perspective is emphasized since aging is a lifelong process [
3]. Although social participation patterns remain relatively stable through people’s life course, factors in working life might also impact on opportunities for social participation in very old age [
4]. Thus, investigating social participation, which may contribute to health and well-being in old age, from a long-term perspective, is important.
Social participation is a broad concept including leisure activities or meeting with friends and consists of interactions with others in society or the community [
5]. Several studies have also found how important social participation is for life satisfaction and healthy aging [
6,
7,
8]. Moreover, social participation contributes to both cognitive and physical health in old age [
9,
10,
11,
12,
13]. Additionally, formal social participation seems to predict higher levels of quality of life and lower levels of depressive symptoms among older people [
14]. What older people value and choose to be engaged in often depends on what they have done in the past, according to the Continuity Theory of Aging [
15]. Continuity means an adaptive strategy for change in the aging process, promoted by personal preferences and social behavior based on earlier activities. Supporting the continuity theory, a longitudinal study focusing on age-related changes in leisure, including social participation, showed that participation earlier in life was a strong predictor of participation in leisure activities also in late life [
16]. Consequently, it could be important to have a repertoire of activities from the past to choose from in old age, when, for example, work tasks must be replaced with other activities [
17,
18]. Furthermore, social participation contributes to a sense of being included in a social context [
17,
19,
20], and, among very old people, social participation was shown to be preferable and of special importance for their well-being [
21]. However, to the best of our knowledge the influence of working conditions on social participation from a long-term perspective beyond working life has not yet been explored.
The Job Strain Model (JSM) explains the psychosocial aspects of work and proposes four types of psychosocial job exposure: high strain, relaxed, active, and passive [
22]. These four job types are combinations of high and low levels of psychosocial job demands and decision latitude. JSM postulates that high strain (high job demands and low decision latitude) increases the risk of ill health, and empirical support for this has been shown in epidemiological studies of, e.g., coronary heart disease, stroke, diabetes, depression, and neck and shoulder disorders (for recent reviews see [
23,
24,
25]. Job strain has also been related to poor survival in a long-term follow-up after retirement [
26].
Among people aged 45–64 years, associations between psychosocial work conditions and social participation using the Job Strain Model have been shown [
22,
27]. More specifically, Lindström [
27] found that passive and high strain jobs were negatively associated with social participation, and that active and relaxed jobs were associated with higher levels of social participation. Work stressors among working adults aged 57–65 was also found to be predictors of limited physical functioning 20 years later [
28]. Thus, there may be factors in working life that also have effects on social participation in later life.
Summing up, several studies found that social participation is important for health [
6,
8,
11,
13,
29], and that high strain jobs impact negatively on mental and physical health [
24,
25]. However, knowledge about the impact of job strain on social participation from a long-term perspective is lacking. The Scania Public Health Cohort, with 10-year follow-up data, provides a unique opportunity to assess social participation and earlier work-related determinants. In line with previous findings in a cross-sectional study design [
27], we hypothesize that there are associations between psychosocial work conditions and social participation from a longitudinal perspective. More precisely, our hypothesis is that low decision latitude, passive job, and job strain are negatively associated with high social participation, and that high decision latitude, active job, and relaxed job are associated with higher levels of social participation.
The purpose of the current study was to investigate the associations between psychosocial working conditions and social participation from a long-term perspective. More specifically the study aimed
to investigate whether psychosocial demands and their combinations predict social participation among 55-year or older working people in a 10-year follow-up when they were not working.
to investigate if high decision latitude was associated with social participation at baseline and predicted high social participation at follow-up.
2. Material and Methods
2.1. Sample and Settings
Comprehensive public health questionnaires, “
The Scania Public Health Survey”, were sent out, by post, in 2000, 2005, and 2010 to a non-proportional geographically stratified sample of inhabitants in 33 municipalities of the county of Scania in the south of Sweden [
30]. These individuals were randomly selected from the population register, such that equal representation was achieved from all 33 municipalities in the region of Scania, Sweden. Details according to design have been described elsewhere [
30]. In total, 24,922 subjects born 1919–1981 (age 18–80) were asked to participate in 2000 and of these 13,604 responded (58% response rate). In 2010, an identical questionnaire was sent out to the 12,117 respondents from the first wave who were still alive and living in Scania, which was responded to by 9103 subjects (75% response rate).
In the present study we included respondents who were 55+ and still working at least 10 h/week at baseline in 2000 and who did not work at follow-up in 2010. The final cohort ended up being 1098 respondents of whom 51% were men and 49% were women (
Figure 1).
2.2. Outcome Variables
Social participation (during the past year) describes how actively a person has taken part in activities in society. The social participation variable consisted of 13 items: participation in study circle/course at work, study circle/course at leisure time, union meeting, meeting of other organization, theatre/cinema, arts exhibition, church, sports event, had written letter to editor of a newspaper/magazine, demonstration of any kind, visited public event (night club, dance or similar), larger family gathering, or been at a private party. Items were dichotomized (yes/no) and summed up, and if three or less were indicated, the social participation of that person was classified as low, and if four or more were indicated, the social participation of that person was classified as high [
31]. This question has been used in Sweden since 1960s and has been validated in an earlier study [
32].
2.3. Exposure Variable
Psychosocial working conditions were measured with a Swedish translation of the original Job Content Questionnaire (JCQ) [
33]. JCQ is based on the JSM [
34] and was further developed [
22] with a focus on psychosocial demands and control. High demands refer to intensive or rapid work where the employee may experience conflicting demands. Job control refers to the degree of decision-making authority and skill discretion of the employee, i.e., decision latitude. The JCQ items consist of 14 statements where respondents were asked to either agree or disagree on a four level Likert scale (1–4). Thus, the answer is based on the individual’s own experience of demands and control in the working environment. Consequently, there could be variations in the same profession. Two continuous variables reflecting psychosocial job demands and decision latitude were thus created, and both were dichotomized at the median level. Following the demand-control model, four different job types were defined by combining psychosocial demands and decision latitude. That is, high strain job is a combination of high demands and low decision latitude,
relaxed job is a combination of low demands and high decision latitude,
active job is a combination of high demands and high decision latitude, and passive job is a combination of low demands and low decision latitude.
2.4. Other Baseline Characteristics
Demographic characteristics considered sex, married/cohabitating versus single, and length of education (dichotomized into 12 years and less, corresponding to primary and secondary school, vs. 13 years or more corresponding to university).
Financial stress was captured by the question “How often during the past 12 months have you had difficulties paying your bills (rent, electricity, telephone, mortgage, insurance, etc.)?” with response alternatives “Every month”, “About half of the months”, “A few times” and “Never”. The answer was considered as financial stress if the respondent had answered “Every month” or “About half of the months”, and as “No financial stress” if the answer was “A few times” or “Never”.
To capture the family situation the question was posed, “Do you have any old or sick relative that you need to help, refer to or care for?” with the response alternatives yes or no.
Physical activity was measured by a single question asking about leisure time activity (household work excluded) with the response alternatives: mostly sedentary leisure time activities, moderate leisure time physical activities, regular exercise, hard or competitive sports/training regularly or several times a week. Answers were dichotomized, as physically active (last three alternatives) vs. Not physically active (first alternative).
Self-rated health was measured with the question, “In general, how do you rate your current health status” with five response alternatives “Excellent”, “Good”, “Fair”, “Bad”, and “Very bad” [
35,
36,
37]. This single question is considered to be the most reliable and valid item estimate of the self-rated health status [
38]. Answers were dichotomized as “Good self-rated health” if the respondent had answered “excellent” and “good”, and Poor self-rated health if the answer was “fair”, “bad”, and “very bad” in any of the two waves 1999 and 2010.
2.5. Statistical Analysis
Kruskal Wallis test was used to detect differences between the four job types (high strain, relaxed, active, and passive) in social participation rates at baseline and McNemar’s test to detect within each job type group changes in social participation rates between baseline and follow-up 10 years later. Bivariate logistic regression was used to test whether the potential confounders, sex, self-rated health, marital status, not caring for a sick relative, education level, financial stress, and physical activity at baseline, were associated with social participation at follow up. Thereafter, a stepwise multivariate logistic regression analysis was performed to test if high decision latitude at baseline was associated with high social participation at baseline and follow-up. The model was adjusted for the confounders, whose p-values in the bivariate logistic regression analyses were <0.10, i.e., good self-rated health, not caring for a sick relative, high educational level, and physically active. Low decision latitude with the lowest social participation rates was selected as the reference category.
To test for a possible effect modification, i.e., the effect of having two factors worse than additive, a synergy index (SI) was calculated as proposed by Rothman [
39]. The following algorithm was used where SI = 1 meant no additive effect, SI > 1 meant a signified synergistic effect, and SI < 1 meant an antagonistic effect.
RR = risk ratio; Ab = exposed to one of the factors; aB = exposed to the other factor; AB = exposed to both factors.
The two factors included in this calculation were self-rated health and educational level at baseline. The level of statistical significance was set at p < 0.05. The statistical analyses were conducted with SPSS, version 24 (IBM Corp., Armonk, NY, USA).
2.6. Ethics
The study was conducted in accordance with the Helsinki Declaration and The Regional Ethical Review Board in Lund approved the study (2016/720).