Unemployment and work disability due to common mental disorders among young adults: selection or causation?

Unemployment and work disability due to common mental disorders among young adults: selection or... Abstract Background Unemployment in early adulthood is associated with higher rate of disability due to common mental disorders (CMDs). We investigated to what extent the association between unemployment and sub-sequent long-term sickness absence due to CMDs is direct or whether it is dependent on accumulation of mental health problems and socioeconomic disadvantage. Methods In this longitudinal study, a population-based 60% sample of Finnish young adults born between 1983 and 1985 (N = 116 878) was followed up for the incidence of CMDs from 2006 to 2010. Sociodemographic and health-related covariates were identified using several nationwide registers. Hazard ratios (HRs) with 95% confidence intervals (CIs), and survival and cumulative hazard functions for CMD were calculated. A matching procedure was applied to account for the systematic differences in the distribution of the baseline characteristics. Results A total of 1416 (2.4%) of men and 2539 (4.4%) of women were granted a long-term sickness allowance for CMD during the follow-up. After matching, HR (95% CI) of CMD for men decreased from 2.38 (2.12–2.68) to 1.31 (1.03–1.67) and for women from 1.97 (1.79–2.18) to 1.39 (1.18–1.65). Approximately half of the effect of the unemployment on CMDs was explained by the background variables. Conclusion Using a causal approach, our study suggests that unemployment is consistently associated with an increased risk of work disability due to CMDs. Considering the young unemployed as a risk group may help in targeting interventions promoting mental health and improving educational and employment opportunities. Introduction The majority of the burden of mental illness in the community arises from the less severe but more frequent ‘neurotic’ disorders, dominated by anxiety, depression or a combination of both. These conditions are referred to as the ‘common mental disorders’ (CMDs).1 Mental disorders are characterized by an early onset, and CMDs are a leading health problem experienced by young adults in the OECD countries.2 One indication of the topicality of this issue is the considerable increase in the prevalence of sickness and disability benefits, such as long-term sickness allowances, in several developed countries.3 In Finland, the annual proportion of young adults receiving long-term (14 days or over) sickness allowance nearly doubled from 0.27 to 0.52% between 1995 and 2012.4 This trend, typically accountable for mental and behavioural disorders,4 has been associated with actual changes in young adult’s health, but also with factors playing a role in the social determination of health such as the changes in the labour market, the overall economic development, as well as with legislative changes.3,5 It is well-established that the distribution of CMDs follows the socioeconomic gradient, and that employment status is one of key ‘independent markers’ in the gradient.1,6 Meta-analyses have shown the bi-directionality of the association between unemployment and mental ill-health in young adults, and that the effect of unemployment on health may be stronger in younger than older age groups.7,8 Successful completion of education and transition from education to the labour market positively affects mental health and well-being of young adults.7 In contrast, job loss and unemployment exposes young adults to impaired mental health which in turn reduces the likelihood of getting (back) into employment.5,7 All in all, young adults with mental health problems often have a weaker employment history which, in turn, poses them into higher risk of sustained mental ill-health. For several decades, different views of the direction of the causal relationship between unemployment and ill health have been debated. Various studies, mostly focusing on middle-aged populations and using questionnaires on self-rated health, have produced evidence that the association is at least partly an effect of health selection.7,9,10 On the other hand, some studies have given evidence for full health selection11,12 or indirect health selection via socioeconomic processes.13 Studies focusing on sickness benefits indicate that unemployment at young age increases the risk of later sickness allowances due to mental disorders, which in turn increase the risk for job termination and unemployment.14,15 Results using administrative data are, however, inconclusive. For example, a recent twin-study indicated that familial circumstances operate as a common cause for both low socio-economic status (SES) in young adulthood and sub-sequent sick leaves granted for mental disorders.16 Mental disorders constitute a central risk factor for unstable position in the labour market measured either as long-term unemployment, sickness absence, or disability pension.17–19 However, to our best knowledge there are no studies examining the relationship between unemployment and sickness allowances due to CMDs among young adults with an advanced causal statistical approach. A recent review on the effects of unemployment and pre-carious employment on health of young adults recommended studies that can enhance the causal model by including a gender perspective, longitudinal data, more indicators on pre-cariousness and third factor explanations.20 Understanding the role and the direction of causation at early life stages is important as young adulthood is a critical period both for the development of organ systems and for psychological and social development.21 In this study, we focus on the transition to adulthood phase of individuals aged 17–26 years. The hypotheses in this study were that the young adults experiencing unemployment, compared with their counterparts, would have an increased risk of CMDs; and that the selection based on socioeconomic factors such as low educational attainment, poor labour market integration as well as exposure to prior health problems would partly explain this relationship. The main interest was to find out to which extent the selection into unemployment causally explains the relationship, and through which mechanisms possible selection into the unemployment occurs. Methods Study population This was a longitudinal 60% sample of Finnish residents, born between 1983 and 1985. The size of the total population was 119 061 in 2005. After the exclusion of those with a record of rehabilitation for severe disability or a disability pension prior to the follow-up (n = 2232) the size of the analytical sample became 116 829. Several nationwide population and health registers were used. All the registers include the subject’s unique personal ID number, which allowed deterministic linkages between the registers.22 The register holders have approved the use of the data for research. The study was approved by the Ethics Committee of the Rehabilitation Foundation of Finland. Predictor The main predictor was an unemployment benefit claim that was in effect within the time period between October and December 2005. We used unemployment as a dichotomous and continuous variable as long-term unemployment is expected to have more negative effects on CMDs.7,8 The continuous variable was the total length of the unemployment period meeting the inclusion criteria. Lowering the continuous predictor to the power of ¼ provided most accurate predictions, indicating a curvilinear association. The non-exposure group consisted of those who were employed and those outside the labour market for other reasons, such as students. The measure of unemployment status refers to the exposure period only which in this case means that we do not make a distinction between those unemployment periods that ended before the onset of a CMD during the follow-up period and those that were continuous until the onset. In Finland, the unemployment benefit is either a means-tested or an insurance based financial assistance, which is paid as compensation for loss of income due to unemployment. The data were obtained from the Unemployment Register maintained by the Ministry of Economic Affairs and Employment. Outcome The outcome was CMD, which was defined as a sickness allowance period of at least 14 days with a classification of F30-F48 in International Classification of Diseases (ICD-10), the diagnostic criteria used in Finnish medical and social insurance system. The outcome date was the first date of the sickness allowance period during the follow-up. All residents of Finland are insured under the Health Insurance Act and the sickness allowance is paid as compensation for loss of income due to work disability. Also full-time students, unemployed, and those doing household work are eligible for the benefit. The data were obtained from Social Insurance Institution of Finland (SII). Follow-up period The beginning of follow-up time for CMD was the 1st of January 2006 (the index date) when the age of the cohort averaged 22 years. The incidence of a CMD related sickness allowance period was treated as the outcome and the survival time was the distance between the index date and the date of CMD occurrence. The follow-up ended at the end of 2010. Covariates Data on sociodemographic and socioeconomic factors included educational level, residential area, residential status, and the number of years registered as not in employment or education, measured as labour market status in the last week of each calendar year and obtained from Statistic Finland. Health related covariates included annual data on purchases of psychotropic medication, participation in psychiatric rehabilitation provided by SII and previous sickness allowances due to mental disorders (ICD-10 codes F00-F99). The data were obtained from SII and were used as an indicator of prior mental disorders. Statistical analysis Hazard ratios (HRs) with 95% confidence intervals (CIs) were calculated for unemployment (dichotomy). Kaplan–Meier product-limit estimates and Nelson–Aalens cumulative incidence rates were calculated as point estimates for the non-exposed and for individuals unemployed for three months and 12 months (continuous). All analyses were performed separately for men and women. Proportional hazards assumption was tested with Schoenfeld residuals test, which showed no indication of violation of proportionality. All analyses were performed using Stata 12.1 software. Matching procedure The robustness of the results gained from Cox regression models was estimated with matched samples to control selection bias resulting from the differentiated sociodemographic and health profiles of the exposure groups. Matching produced matching weights that balanced the groups nearly identical on all baseline covariates (tables 1 and 2; data shown for the full sample and the final matched sample). Matching was implemented with the Coarsened Exact Matching (CEM) statistical package that has been reported to produce higher quality estimates compared with propensity score matching.23 The matching procedure defined statistically an area of common support containing both treated and control units. The region of common support included 88% of non-exposed men, 72% of exposed men, 85% of non-exposed women and 77% of exposed women. The proportion of selection bias was assessed comparing the difference in cumulative hazard estimates between the exposed and non-exposed in the full and matched samples, first, matching with sociodemographic and socioeconomic covariates and, second, matching with both sociodemographic, socioeconomic and health-related covariates. Table 1 Comparison of baseline characteristics between exposed and non-exposed subjects in the full and matched sample, men Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 59 660 51 634 8026 45 215 5775 Months at risk 3 523 955 3 056 685 467 270 2 664 692 338 240 CMDs, n 1416 1038 378 1374 229 CMDs, % 2.37 2.01 4.71 3.04 3.97 Variable Unemployment at baseline (months)     0 86.5 100.0 0.0 100.0 0.0     1––3 5.6 0.0 41.3 0.0 44.7     3– 7.9 0.0 58.7 0.0 55.3 Education     Secondary education 78.0 79.6 67.4 73.5 73.5     Less than secondary 22.0 20.4 32.6 26.5 26.5 Residential area     South Finland 46.4 47.6 38.7 40.7 40.7     West Finland 26.7 26.8 26.2 26.9 26.9     East Finland 12.7 12.0 17.6 16.8 16.8     North Finland 13.7 13.2 17.2 15.4 15.4     Åland 0.4 0.5 0.2 0.2 0.2 Language, %     Finnish 92.4 92.2 94.1 96.2 96.2     Swedish 5.1 5.6 2.3 1.7 1.7     Other 2.5 2.3 3.6 2.0 2.0 Residential status, %     Parental home 46.3 46.9 42.3 48.2 48.2     Single 32.3 31.7 36.6 35.2 35.2     Own family 20.3 20.6 18.3 15.8 15.8     Without permanent residence 1.1 0.9 2.8 0.8 0.8 Unemployment 2003-baseline (months)     0 51.0 55.8 20.4 27.5 27.5     1–6 27.5 27.3 28.8 35.3 34.7     6– 21.5 16.9 50.8 37.1 37.7 Years ‘not in employment or education’     0 78.8 86.2 31.2 30.9 30.9     1 15.8 12.0 40.8 50.6 50.6     2+ 5.4 1.8 27.9 18.5 18.5 Years of psychotropic drug purchases     0 95.8 96.6 90.7 95.0 95.0     1 2.5 2.0 5.4 3.4 3.4     2+ 1.7 1.4 3.9 1.6 1.6 F-diag sickness allowance 2004–2005     No 99.4 99.5 98.9 99.8 99.8     Yes 0.6 0.5 1.1 0.2 0.2 Vocational rehabilitation     No 98.6 98.7 97.5 99.1 99.1     Yes 1.4 1.3 2.5 0.9 0.9 Psychotherapy     No 99.7 99.7 99.6 99.9 99.9     Yes 0.3 0.3 0.4 0.1 0.1 Any health related indicator     No 93.6 94.6 86.8 93.1 93.1     Yes 6.4 5.4 13.2 6.9 6.9 Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 59 660 51 634 8026 45 215 5775 Months at risk 3 523 955 3 056 685 467 270 2 664 692 338 240 CMDs, n 1416 1038 378 1374 229 CMDs, % 2.37 2.01 4.71 3.04 3.97 Variable Unemployment at baseline (months)     0 86.5 100.0 0.0 100.0 0.0     1––3 5.6 0.0 41.3 0.0 44.7     3– 7.9 0.0 58.7 0.0 55.3 Education     Secondary education 78.0 79.6 67.4 73.5 73.5     Less than secondary 22.0 20.4 32.6 26.5 26.5 Residential area     South Finland 46.4 47.6 38.7 40.7 40.7     West Finland 26.7 26.8 26.2 26.9 26.9     East Finland 12.7 12.0 17.6 16.8 16.8     North Finland 13.7 13.2 17.2 15.4 15.4     Åland 0.4 0.5 0.2 0.2 0.2 Language, %     Finnish 92.4 92.2 94.1 96.2 96.2     Swedish 5.1 5.6 2.3 1.7 1.7     Other 2.5 2.3 3.6 2.0 2.0 Residential status, %     Parental home 46.3 46.9 42.3 48.2 48.2     Single 32.3 31.7 36.6 35.2 35.2     Own family 20.3 20.6 18.3 15.8 15.8     Without permanent residence 1.1 0.9 2.8 0.8 0.8 Unemployment 2003-baseline (months)     0 51.0 55.8 20.4 27.5 27.5     1–6 27.5 27.3 28.8 35.3 34.7     6– 21.5 16.9 50.8 37.1 37.7 Years ‘not in employment or education’     0 78.8 86.2 31.2 30.9 30.9     1 15.8 12.0 40.8 50.6 50.6     2+ 5.4 1.8 27.9 18.5 18.5 Years of psychotropic drug purchases     0 95.8 96.6 90.7 95.0 95.0     1 2.5 2.0 5.4 3.4 3.4     2+ 1.7 1.4 3.9 1.6 1.6 F-diag sickness allowance 2004–2005     No 99.4 99.5 98.9 99.8 99.8     Yes 0.6 0.5 1.1 0.2 0.2 Vocational rehabilitation     No 98.6 98.7 97.5 99.1 99.1     Yes 1.4 1.3 2.5 0.9 0.9 Psychotherapy     No 99.7 99.7 99.6 99.9 99.9     Yes 0.3 0.3 0.4 0.1 0.1 Any health related indicator     No 93.6 94.6 86.8 93.1 93.1     Yes 6.4 5.4 13.2 6.9 6.9 Table 1 Comparison of baseline characteristics between exposed and non-exposed subjects in the full and matched sample, men Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 59 660 51 634 8026 45 215 5775 Months at risk 3 523 955 3 056 685 467 270 2 664 692 338 240 CMDs, n 1416 1038 378 1374 229 CMDs, % 2.37 2.01 4.71 3.04 3.97 Variable Unemployment at baseline (months)     0 86.5 100.0 0.0 100.0 0.0     1––3 5.6 0.0 41.3 0.0 44.7     3– 7.9 0.0 58.7 0.0 55.3 Education     Secondary education 78.0 79.6 67.4 73.5 73.5     Less than secondary 22.0 20.4 32.6 26.5 26.5 Residential area     South Finland 46.4 47.6 38.7 40.7 40.7     West Finland 26.7 26.8 26.2 26.9 26.9     East Finland 12.7 12.0 17.6 16.8 16.8     North Finland 13.7 13.2 17.2 15.4 15.4     Åland 0.4 0.5 0.2 0.2 0.2 Language, %     Finnish 92.4 92.2 94.1 96.2 96.2     Swedish 5.1 5.6 2.3 1.7 1.7     Other 2.5 2.3 3.6 2.0 2.0 Residential status, %     Parental home 46.3 46.9 42.3 48.2 48.2     Single 32.3 31.7 36.6 35.2 35.2     Own family 20.3 20.6 18.3 15.8 15.8     Without permanent residence 1.1 0.9 2.8 0.8 0.8 Unemployment 2003-baseline (months)     0 51.0 55.8 20.4 27.5 27.5     1–6 27.5 27.3 28.8 35.3 34.7     6– 21.5 16.9 50.8 37.1 37.7 Years ‘not in employment or education’     0 78.8 86.2 31.2 30.9 30.9     1 15.8 12.0 40.8 50.6 50.6     2+ 5.4 1.8 27.9 18.5 18.5 Years of psychotropic drug purchases     0 95.8 96.6 90.7 95.0 95.0     1 2.5 2.0 5.4 3.4 3.4     2+ 1.7 1.4 3.9 1.6 1.6 F-diag sickness allowance 2004–2005     No 99.4 99.5 98.9 99.8 99.8     Yes 0.6 0.5 1.1 0.2 0.2 Vocational rehabilitation     No 98.6 98.7 97.5 99.1 99.1     Yes 1.4 1.3 2.5 0.9 0.9 Psychotherapy     No 99.7 99.7 99.6 99.9 99.9     Yes 0.3 0.3 0.4 0.1 0.1 Any health related indicator     No 93.6 94.6 86.8 93.1 93.1     Yes 6.4 5.4 13.2 6.9 6.9 Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 59 660 51 634 8026 45 215 5775 Months at risk 3 523 955 3 056 685 467 270 2 664 692 338 240 CMDs, n 1416 1038 378 1374 229 CMDs, % 2.37 2.01 4.71 3.04 3.97 Variable Unemployment at baseline (months)     0 86.5 100.0 0.0 100.0 0.0     1––3 5.6 0.0 41.3 0.0 44.7     3– 7.9 0.0 58.7 0.0 55.3 Education     Secondary education 78.0 79.6 67.4 73.5 73.5     Less than secondary 22.0 20.4 32.6 26.5 26.5 Residential area     South Finland 46.4 47.6 38.7 40.7 40.7     West Finland 26.7 26.8 26.2 26.9 26.9     East Finland 12.7 12.0 17.6 16.8 16.8     North Finland 13.7 13.2 17.2 15.4 15.4     Åland 0.4 0.5 0.2 0.2 0.2 Language, %     Finnish 92.4 92.2 94.1 96.2 96.2     Swedish 5.1 5.6 2.3 1.7 1.7     Other 2.5 2.3 3.6 2.0 2.0 Residential status, %     Parental home 46.3 46.9 42.3 48.2 48.2     Single 32.3 31.7 36.6 35.2 35.2     Own family 20.3 20.6 18.3 15.8 15.8     Without permanent residence 1.1 0.9 2.8 0.8 0.8 Unemployment 2003-baseline (months)     0 51.0 55.8 20.4 27.5 27.5     1–6 27.5 27.3 28.8 35.3 34.7     6– 21.5 16.9 50.8 37.1 37.7 Years ‘not in employment or education’     0 78.8 86.2 31.2 30.9 30.9     1 15.8 12.0 40.8 50.6 50.6     2+ 5.4 1.8 27.9 18.5 18.5 Years of psychotropic drug purchases     0 95.8 96.6 90.7 95.0 95.0     1 2.5 2.0 5.4 3.4 3.4     2+ 1.7 1.4 3.9 1.6 1.6 F-diag sickness allowance 2004–2005     No 99.4 99.5 98.9 99.8 99.8     Yes 0.6 0.5 1.1 0.2 0.2 Vocational rehabilitation     No 98.6 98.7 97.5 99.1 99.1     Yes 1.4 1.3 2.5 0.9 0.9 Psychotherapy     No 99.7 99.7 99.6 99.9 99.9     Yes 0.3 0.3 0.4 0.1 0.1 Any health related indicator     No 93.6 94.6 86.8 93.1 93.1     Yes 6.4 5.4 13.2 6.9 6.9 Table 2 Comparison of baseline characteristics between exposed and non-exposed subjects in the full and matched sample, women Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 57 169 50 838 6331 43 352 4869 Months at risk 3 335 736 2 972 953 362 783 2 527 751 281 951 CMDs, n 2539 2046 493 1973 307 CMDs, % 4.44 4.02 7.79 4.55 6.31 Variable Unemployment at baseline (months)     0 88.9 100.0 0.0 100.0 0.0     1–3 4.9 0.0 43.9 0.0 46.1     3– 6.2 0.0 56.1 0.0 53.9 Education     Secondary education 84.4 85.3 77.0 83.2 83.2     Less than secondary 15.6 14.7 23.0 16.8 16.8 Residential area     South Finland 48.7 50.0 37.8 38.2 38.2     West Finland 25.9 25.5 29.8 31.1 31.1     East Finland 12.3 11.9 15.6 14.7 14.7     North Finland 12.7 12.2 16.5 15.8 15.8     Åland 0.4 0.4 0.3 0.1 0.1 Language, %     Finnish 92.9 92.6 95.2 97.1 97.1     Swedish 4.9 5.2 1.8 1.3 1.3     Other 2.3 2.2 3.0 1.6 1.6 Residential status, %     Parental home 25.3 25.9 20.8 20.2 20.2     Single 36.2 36.0 37.9 34.8 34.8     Own family 37.6 37.3 39.9 44.6 44.6     Without permanent residence 0.9 0.8 1.5 0.3 0.3 Unemployment 2003-baseline (months)     0 54.4 57.9 25.7 32.2 32.2     1–6 26.6 26.0 31.1 36.1 35.0     6– 19.0 16.0 43.2 31.7 32.8 Years ‘not in employment or education’     0 80.6 86.0 37.1 35.3 35.3     1 13.7 10.4 40.8 47.3 47.3     2+ 5.7 3.6 22.2 17.3 17.3 Years of psychotropic drug purchases     0 92.3 93.0 87.0 91.3 91.3     1 4.6 4.2 7.5 6.0 6.0     2+ 3.1 2.8 5.5 2.8 2.8 F-diag sickness allowance 2004-2005     No 98.7 98.9 97.1 99.0 99.0     Yes 1.3 1.1 2.9 1.0 1.0 Vocational Rehabilitation     No 98.2 98.4 96.8 99.0 99.0     Yes 1.8 1.6 3.2 1.0 1.0 Psychotherapy     No 98.6 98.7 98.6 99.5 99.5     Yes 1.4 1.3 1.4 0.5 0.5 Any health related indicator     No 88.2 89.0 81.4 87.7 87.7     Yes 11.8 11.0 18.6 12.3 12.3 Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 57 169 50 838 6331 43 352 4869 Months at risk 3 335 736 2 972 953 362 783 2 527 751 281 951 CMDs, n 2539 2046 493 1973 307 CMDs, % 4.44 4.02 7.79 4.55 6.31 Variable Unemployment at baseline (months)     0 88.9 100.0 0.0 100.0 0.0     1–3 4.9 0.0 43.9 0.0 46.1     3– 6.2 0.0 56.1 0.0 53.9 Education     Secondary education 84.4 85.3 77.0 83.2 83.2     Less than secondary 15.6 14.7 23.0 16.8 16.8 Residential area     South Finland 48.7 50.0 37.8 38.2 38.2     West Finland 25.9 25.5 29.8 31.1 31.1     East Finland 12.3 11.9 15.6 14.7 14.7     North Finland 12.7 12.2 16.5 15.8 15.8     Åland 0.4 0.4 0.3 0.1 0.1 Language, %     Finnish 92.9 92.6 95.2 97.1 97.1     Swedish 4.9 5.2 1.8 1.3 1.3     Other 2.3 2.2 3.0 1.6 1.6 Residential status, %     Parental home 25.3 25.9 20.8 20.2 20.2     Single 36.2 36.0 37.9 34.8 34.8     Own family 37.6 37.3 39.9 44.6 44.6     Without permanent residence 0.9 0.8 1.5 0.3 0.3 Unemployment 2003-baseline (months)     0 54.4 57.9 25.7 32.2 32.2     1–6 26.6 26.0 31.1 36.1 35.0     6– 19.0 16.0 43.2 31.7 32.8 Years ‘not in employment or education’     0 80.6 86.0 37.1 35.3 35.3     1 13.7 10.4 40.8 47.3 47.3     2+ 5.7 3.6 22.2 17.3 17.3 Years of psychotropic drug purchases     0 92.3 93.0 87.0 91.3 91.3     1 4.6 4.2 7.5 6.0 6.0     2+ 3.1 2.8 5.5 2.8 2.8 F-diag sickness allowance 2004-2005     No 98.7 98.9 97.1 99.0 99.0     Yes 1.3 1.1 2.9 1.0 1.0 Vocational Rehabilitation     No 98.2 98.4 96.8 99.0 99.0     Yes 1.8 1.6 3.2 1.0 1.0 Psychotherapy     No 98.6 98.7 98.6 99.5 99.5     Yes 1.4 1.3 1.4 0.5 0.5 Any health related indicator     No 88.2 89.0 81.4 87.7 87.7     Yes 11.8 11.0 18.6 12.3 12.3 Table 2 Comparison of baseline characteristics between exposed and non-exposed subjects in the full and matched sample, women Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 57 169 50 838 6331 43 352 4869 Months at risk 3 335 736 2 972 953 362 783 2 527 751 281 951 CMDs, n 2539 2046 493 1973 307 CMDs, % 4.44 4.02 7.79 4.55 6.31 Variable Unemployment at baseline (months)     0 88.9 100.0 0.0 100.0 0.0     1–3 4.9 0.0 43.9 0.0 46.1     3– 6.2 0.0 56.1 0.0 53.9 Education     Secondary education 84.4 85.3 77.0 83.2 83.2     Less than secondary 15.6 14.7 23.0 16.8 16.8 Residential area     South Finland 48.7 50.0 37.8 38.2 38.2     West Finland 25.9 25.5 29.8 31.1 31.1     East Finland 12.3 11.9 15.6 14.7 14.7     North Finland 12.7 12.2 16.5 15.8 15.8     Åland 0.4 0.4 0.3 0.1 0.1 Language, %     Finnish 92.9 92.6 95.2 97.1 97.1     Swedish 4.9 5.2 1.8 1.3 1.3     Other 2.3 2.2 3.0 1.6 1.6 Residential status, %     Parental home 25.3 25.9 20.8 20.2 20.2     Single 36.2 36.0 37.9 34.8 34.8     Own family 37.6 37.3 39.9 44.6 44.6     Without permanent residence 0.9 0.8 1.5 0.3 0.3 Unemployment 2003-baseline (months)     0 54.4 57.9 25.7 32.2 32.2     1–6 26.6 26.0 31.1 36.1 35.0     6– 19.0 16.0 43.2 31.7 32.8 Years ‘not in employment or education’     0 80.6 86.0 37.1 35.3 35.3     1 13.7 10.4 40.8 47.3 47.3     2+ 5.7 3.6 22.2 17.3 17.3 Years of psychotropic drug purchases     0 92.3 93.0 87.0 91.3 91.3     1 4.6 4.2 7.5 6.0 6.0     2+ 3.1 2.8 5.5 2.8 2.8 F-diag sickness allowance 2004-2005     No 98.7 98.9 97.1 99.0 99.0     Yes 1.3 1.1 2.9 1.0 1.0 Vocational Rehabilitation     No 98.2 98.4 96.8 99.0 99.0     Yes 1.8 1.6 3.2 1.0 1.0 Psychotherapy     No 98.6 98.7 98.6 99.5 99.5     Yes 1.4 1.3 1.4 0.5 0.5 Any health related indicator     No 88.2 89.0 81.4 87.7 87.7     Yes 11.8 11.0 18.6 12.3 12.3 Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 57 169 50 838 6331 43 352 4869 Months at risk 3 335 736 2 972 953 362 783 2 527 751 281 951 CMDs, n 2539 2046 493 1973 307 CMDs, % 4.44 4.02 7.79 4.55 6.31 Variable Unemployment at baseline (months)     0 88.9 100.0 0.0 100.0 0.0     1–3 4.9 0.0 43.9 0.0 46.1     3– 6.2 0.0 56.1 0.0 53.9 Education     Secondary education 84.4 85.3 77.0 83.2 83.2     Less than secondary 15.6 14.7 23.0 16.8 16.8 Residential area     South Finland 48.7 50.0 37.8 38.2 38.2     West Finland 25.9 25.5 29.8 31.1 31.1     East Finland 12.3 11.9 15.6 14.7 14.7     North Finland 12.7 12.2 16.5 15.8 15.8     Åland 0.4 0.4 0.3 0.1 0.1 Language, %     Finnish 92.9 92.6 95.2 97.1 97.1     Swedish 4.9 5.2 1.8 1.3 1.3     Other 2.3 2.2 3.0 1.6 1.6 Residential status, %     Parental home 25.3 25.9 20.8 20.2 20.2     Single 36.2 36.0 37.9 34.8 34.8     Own family 37.6 37.3 39.9 44.6 44.6     Without permanent residence 0.9 0.8 1.5 0.3 0.3 Unemployment 2003-baseline (months)     0 54.4 57.9 25.7 32.2 32.2     1–6 26.6 26.0 31.1 36.1 35.0     6– 19.0 16.0 43.2 31.7 32.8 Years ‘not in employment or education’     0 80.6 86.0 37.1 35.3 35.3     1 13.7 10.4 40.8 47.3 47.3     2+ 5.7 3.6 22.2 17.3 17.3 Years of psychotropic drug purchases     0 92.3 93.0 87.0 91.3 91.3     1 4.6 4.2 7.5 6.0 6.0     2+ 3.1 2.8 5.5 2.8 2.8 F-diag sickness allowance 2004-2005     No 98.7 98.9 97.1 99.0 99.0     Yes 1.3 1.1 2.9 1.0 1.0 Vocational Rehabilitation     No 98.2 98.4 96.8 99.0 99.0     Yes 1.8 1.6 3.2 1.0 1.0 Psychotherapy     No 98.6 98.7 98.6 99.5 99.5     Yes 1.4 1.3 1.4 0.5 0.5 Any health related indicator     No 88.2 89.0 81.4 87.7 87.7     Yes 11.8 11.0 18.6 12.3 12.3 Results Among the 59 660 men and 57 169 women who remained in the study population for 59 months on average, 13.5% of men and 11.1% of women had experienced unemployment (tables 1 and 2). Men had more adverse sociodemographic profiles with more adverse educational attainment and prior unemployment profiles. Baseline ill-health-related indicators were more common in women. 6.4% of men and 11.8% of women had at least one health related problem. In both genders, the unemployed were more disadvantaged in terms of all baseline characteristics. 2.0% of non-exposed men and 4.0% of non-exposed women were granted sickness allowance for CMD during the follow-up. Corresponding figures were 4.7% for exposed men and 7.8% for exposed women. Survival estimates before and after matching Before the matching procedure, the subjects with unemployment had a significantly higher hazard for CMD with HRs of 2.38 (95% CI 2.12–2.68) for men and 1.97 (95% CI 1.79–2.18) for women, respectively (table 3). After the full matching procedure HRs were reduced to 1.31 (1.03–1.67) for men and 1.39 (1.18–1.65) for women, indicating partial causal effects. Those with longer unemployment periods experienced a higher risk of CMD. In terms of cumulative incidence rates, the surplus effect of unemployment on the rate of CMDs at 60 months of survival time was 2.5 for men with three months of unemployment and 4.2 for those with 12 months of unemployment compared with their non-exposed counterparts in the full sample. The corresponding figures were 3.6 and 5.8 for women. Unemployment created excess risk of 1.0 for men with three months of unemployment and 1.6 for those with 12 months of unemployment. The corresponding figures were 1.8 and 2.8 for women. The latter figures represent the genuine effect of unemployment on CMDs, also indicating that the risk of CMD rises with the length of unemployment period even after taking the selection into account. The prolongation of unemployment appeared to have a more negative effect for women. Table 3 HRs for CMDs with 95% CIs; Nelson–Aalens cumulative hazards at 60 months; and the difference of cumulative hazards between the exposure groups and samples Unemployment Nelson–Aalen Cum. Haz. at 60 months Difference in cumulative hazard vs. non-exposeda Proportion of the effect assignable to background covariatesb HR 95% CI Non-exposed Exposed (3 mc) Exposed (12 m) Exposed (3 m) Exposed (12 m) Exposed (3 m) Exposed (12 m) Gender Males     Full sample 2.38 (2.12–2.68) 2.05 4.53 6.27 2.48 4.23 – –     Matched sample 1d 1.65 (1.33–2.04) 2.05 3.90 4.96 1.85 2.92 0.63 1.31     Matched sample 2e 1.31 (1.03–1.67) 2.05 3.09 3.61 1.04 1.56 0.81 1.35 Females     Full sample 1.97 (1.79–2.18) 4.15 7.73 9.99 3.58 5.84 – –     Matched sample 1 1.80 (1.54–2.10) 4.15 7.45 9.43 3.30 5.29 0.28 0.55     Matched sample 2 1.39 (1.18–1.65) 4.15 5.98 6.93 1.83 2.78 1.47 2.51 Unemployment Nelson–Aalen Cum. Haz. at 60 months Difference in cumulative hazard vs. non-exposeda Proportion of the effect assignable to background covariatesb HR 95% CI Non-exposed Exposed (3 mc) Exposed (12 m) Exposed (3 m) Exposed (12 m) Exposed (3 m) Exposed (12 m) Gender Males     Full sample 2.38 (2.12–2.68) 2.05 4.53 6.27 2.48 4.23 – –     Matched sample 1d 1.65 (1.33–2.04) 2.05 3.90 4.96 1.85 2.92 0.63 1.31     Matched sample 2e 1.31 (1.03–1.67) 2.05 3.09 3.61 1.04 1.56 0.81 1.35 Females     Full sample 1.97 (1.79–2.18) 4.15 7.73 9.99 3.58 5.84 – –     Matched sample 1 1.80 (1.54–2.10) 4.15 7.45 9.43 3.30 5.29 0.28 0.55     Matched sample 2 1.39 (1.18–1.65) 4.15 5.98 6.93 1.83 2.78 1.47 2.51 a Difference in Nelson–Aalen cumulative hazard estimate between the exposed and non-exposed (‘average treatment effect’ of the unemployment). b Proportion of the difference assignable to background covariates (‘difference-in-differences’). c Estimate for those with three months of unemployment at the baseline. d Matching based on sociodemographic and socioeconomic covariates. e Matching based on sociodemographic, socioeconomic and health covariates. Table 3 HRs for CMDs with 95% CIs; Nelson–Aalens cumulative hazards at 60 months; and the difference of cumulative hazards between the exposure groups and samples Unemployment Nelson–Aalen Cum. Haz. at 60 months Difference in cumulative hazard vs. non-exposeda Proportion of the effect assignable to background covariatesb HR 95% CI Non-exposed Exposed (3 mc) Exposed (12 m) Exposed (3 m) Exposed (12 m) Exposed (3 m) Exposed (12 m) Gender Males     Full sample 2.38 (2.12–2.68) 2.05 4.53 6.27 2.48 4.23 – –     Matched sample 1d 1.65 (1.33–2.04) 2.05 3.90 4.96 1.85 2.92 0.63 1.31     Matched sample 2e 1.31 (1.03–1.67) 2.05 3.09 3.61 1.04 1.56 0.81 1.35 Females     Full sample 1.97 (1.79–2.18) 4.15 7.73 9.99 3.58 5.84 – –     Matched sample 1 1.80 (1.54–2.10) 4.15 7.45 9.43 3.30 5.29 0.28 0.55     Matched sample 2 1.39 (1.18–1.65) 4.15 5.98 6.93 1.83 2.78 1.47 2.51 Unemployment Nelson–Aalen Cum. Haz. at 60 months Difference in cumulative hazard vs. non-exposeda Proportion of the effect assignable to background covariatesb HR 95% CI Non-exposed Exposed (3 mc) Exposed (12 m) Exposed (3 m) Exposed (12 m) Exposed (3 m) Exposed (12 m) Gender Males     Full sample 2.38 (2.12–2.68) 2.05 4.53 6.27 2.48 4.23 – –     Matched sample 1d 1.65 (1.33–2.04) 2.05 3.90 4.96 1.85 2.92 0.63 1.31     Matched sample 2e 1.31 (1.03–1.67) 2.05 3.09 3.61 1.04 1.56 0.81 1.35 Females     Full sample 1.97 (1.79–2.18) 4.15 7.73 9.99 3.58 5.84 – –     Matched sample 1 1.80 (1.54–2.10) 4.15 7.45 9.43 3.30 5.29 0.28 0.55     Matched sample 2 1.39 (1.18–1.65) 4.15 5.98 6.93 1.83 2.78 1.47 2.51 a Difference in Nelson–Aalen cumulative hazard estimate between the exposed and non-exposed (‘average treatment effect’ of the unemployment). b Proportion of the difference assignable to background covariates (‘difference-in-differences’). c Estimate for those with three months of unemployment at the baseline. d Matching based on sociodemographic and socioeconomic covariates. e Matching based on sociodemographic, socioeconomic and health covariates. For both genders, roughly half of the observed effect of unemployment on CMDs may be interpreted as a contribution of background factors. Health factors prior or at the time of unemployment were the main contributors to this bias; with men having higher proportion of the effect assignable to sociodemographic and socioeconomic factors and women to health status at baseline. We found higher selection bias in the estimates for 12 months of unemployment compared with the estimates for three months of unemployment. The background characteristics may explain not only the exposure to unemployment but the length of unemployment period which in turn is associated with a higher incidence of CMD. The plotted Kaplan–Meier estimates in figure 1 give a visual presentation how the amount of selection bias was dependent on the length of unemployment. The figure confirms our observation of the gendered pattern of the association. Figure 1 View largeDownload slide Kaplan–Meier curves showing CMDs based on full/matched sample comparison and length of exposure to unemployment at the baseline (months), by gender Figure 1 View largeDownload slide Kaplan–Meier curves showing CMDs based on full/matched sample comparison and length of exposure to unemployment at the baseline (months), by gender Discussion This large population-based study adds to the existing evidence on the association between unemployment and CMDs. We found that unemployment is independently associated with CMD related work disability in young adults. Although several studies have suggested that selection into unemployment may explain why the unemployed experience more mental health problems compared with other population groups,7,11–18 no previous study has investigated the selection bias using a large register based longitudinal data, using a causal statistical approach and following a cohort of young adults over a long period of time. Neither has any study, to our best knowledge, studied the association between unemployment and work CMD related work disability in the critical period of early adulthood. We used survival analysis for assessing the cumulative risk of CMD in combination with matching and reweighting estimators, which offered an advanced approach to examine causal inference with observational data. We found that, even though a considerable part of the unemployment—mental ill health association was explained by selection mechanism, there was also a consistent causal association between unemployment and CMDs. We found a dose–response effect that increased with the duration of unemployment. Both causal effect and selection mechanism followed distinct gendered patterns. Strengths and limitations In this study, we used nationwide registers containing information of the use of social security benefits (unemployment benefit and sickness allowance). Sickness allowance does indicate a medically relevant serious CMD. However, as the measure of health problems it does not reflect CMDs that do not prevent people from working and imposes certain degree of under-coverage as the vast majority of those with a health problem do not claim sickness allowance.15,24 We cannot ascertain that factors related to benefit seeking such as personal coping capabilities, relationships, or legislative factors related to receiving such benefits15 are equal between the individuals in different exposure groups. Furthermore, those in pre-carious employment positions, as employed young adults often are, may have disincentives for claiming sickness allowance25 or sickness allowance may mask unemployment.3 All these factors may affect the reliability of the estimates. The using of register data has its strengths. The cohort included over 110 000 young adults, and provided a sufficient number of individuals exposed to different lengths of unemployment which allowed us to use unemployment both as a dichotomous and a continuous variable, as well as to use an advanced matching strategy to account for systematic differences in background characteristics. With a direct link to the implementation of social policy measures, our approach using register data may be seen as a pragmatic way for investigating unemployment and CMDs within the public health context. A further benefit from the policy perspective is that sickness allowance utilization has been found to be a predictor for later disability pension and death15,26 and our results may thus be used in planning preventive programmes. The fact that our baseline health indicators were health interventions highlights the importance to enhance the cooperation between health, employment and social policies to prevent vicious circles in which unemployment may trigger processes leading to mental ill-health. Comparison with previous studies and theoretical implications The health selection hypothesis identifies health as a causative factor to unemployment and thus unemployment may merely ‘mask’ the underlying causes behind the association between unemployment and mental ill-health. We found that a considerable proportion of such an effect behind CMD related work disability indicating that poor mental health may influence the risk and duration of unemployment27,28 and that a larger proportion of individuals with impaired health are being selected in the stock of unemployed.28 Health may be a significant independent factor in the process of becoming unemployed in early adulthood but may also be associated with a trajectory of under-employment, including unemployment, insufficient working hours, low pay and intermittent unemployment spells, which, too, constitute a risk factor for mental ill-health.20,29 Understanding socio-economic mechanisms occurring in early adulthood is an important point of targeted prevention. However, socioeconomic disparities in health grow gradually across the life course, being small in early adulthood and increasing through middle adulthood and early old age.30 Unfortunately, the nature of our data focusing only on birth cohorts 1983–85 did not allow comparisons between different birth cohorts, that is, to make age-group comparisons. Neither did it allow us to follow the cohort through middle adulthood nor to early old age. Our data did not allow to control for many important background factors such as poor health in childhood31 or health behaviours such as risky drinking or sedentary lifestyle,32,33 or the strong influence of familial factors,18 such as growing up in families experiencing economic hardship or in which the safety or balanced development has been compromised with their adverse effects on both educational attainment and mental health.20,34,35 However, we used matching and reweighting estimators which, if unbiased, would balance unobserved variables to the extent they are correlated with those that are observed.36 Consistent with previous studies, we found gender differences in vulnerability to the effects of unemployment.37,38 We observed gender-related differences in the distribution of family formation, educational attainment, employment status, and psychotropic medication at baseline. Given these gendered differences, all analyses were conducted separately for men and women. Unemployed women had a considerably higher risk of CMD even after taking into account of selection bias. Furthermore, we found that, for men, the effect of unemployment on CMDs was more pronouncedly mediated through adverse sociodemographic and socioeconomic indicators and for women through health selection. Prior research has found more pronounced association for unemployment and health behaviour among men and for unemployment and health among women.9,39 Our finding that unemployed men and women have different background profiles may be partly explained by similar gender difference between health and health behaviours prior to unemployment. However, we do not know the extent our results are related to such differences or gendered differences in social insurance practices, disease patterns, or work-related factors.40 The main finding in our study is that both the selection mechanism for unemployment and the effects of the unemployment in young adulthood appear to be gender specific. Conclusions We found a causal association between unemployment and CMDs related to work disability among young adults. The deteriorating effect of unemployment at the transition-to-adulthood period on sub-sequent mental health warrant well-targeted preventive policies. These policies include a set of policies as unemployment may be at the same time an exogenous ‘shock’ leading to financial, material, social, psychological or behavioural changes that will harm individual’s mental health and be a marker of a longer process of accumulation of disadvantage. The latter indicates a need for policies targeting at socio-economic and health conditions prior to unemployment, possibly during the secondary education or already in childhood. A wide array of policy measures that affect SES and the health of young adults have been suggested in public health literature. These include active labour market programmes to prevent long periods of unemployment, training programmes, programmes enhancing young people’s psychosocial resources, social security measures to offset the economic consequences of unemployment, work re-design to reduce social stressors at work, social support and autonomy, specific health programmes and services and support for youth career initiatives.20 Policies that affect labour market trajectories of young people should be targeted especially to those unemployed young adults with no educational qualifications, pre-carious employment histories and early mental disorders. Funding This work was supported by Social Insurance Institution of Finland (SII) (grant KKRL/6/26/2001 to J.H.), Medical Research Council (MRC) (grant MR/K023241/1 to A.K.), Economic and Social Research Council (ESRC) (grant ES/L007509/1 to A.K.), Academy of Finland (grants 258598, 265174 to M.V.). Conflicts of interest: None declared. Key points This large register study examined the effect of unemployment on work disability due to common mental disorders (CMDs) in early adulthood with population level register data. Although part of the associations were explained by prior health status, unemployment increased the incidence of CMDs even after the selection to unemployment was taken into the account with a matching procedure. Unemployment was less prevalent in women but showed to be more detrimental for them in terms of their effect on CMD-related work disability. Public health policy could acknowledge that unemployment in young adulthood may be at the same time an exogenous shock leading to work disability and a ‘marker’ in a process of accumulation of disadvantage leading to work disability. References 1 Fryers T , Melzer D , Jenkins R , Brugha T . The distribution of the common mental disorders: social inequalities in Europe . Clin Pract Epidemiol Ment Health 2005 ; 1 : 14 . Google Scholar CrossRef Search ADS PubMed 2 OECD . Sick on the Job? Myths and Realities about Mental Health and Work . Paris : OECD Publishing , 2012 . 3 Kaltenbrunner Bernitz B , Grees N , Jakobsson Randers M , et al. Young adults on disability benefits in 7 countries . Scand J Public Health 2013 ; 41 : 3 – 26 . Google Scholar CrossRef Search ADS PubMed 4 Koskenvuo K , Hytti H , Autti-Rämö I . Alle 25-vuotiaiden nuorten työkyvyttömyys-ja kuntoutusetuuksien käytön kehitys. Kela.Työpapereita 2013 ;50. [The trends in disability and rehabilitation benefits for young people under 25 years. SII Working Papers.] 5 Reneflot A , Evensen M . Unemployment and psychological distress among young adults in the Nordic countries: a review of the literature . Int J Soc Wel 2014 ; 23 : 3 – 15 . Google Scholar CrossRef Search ADS 6 Lorant V , Deliège D , Eaton W , et al. Socioeconomic inequalities in depression: a meta-analysis . Am J Epidemiol 2003 ; 157 : 98 – 112 . Google Scholar CrossRef Search ADS PubMed 7 Paul KI , Moser K . Unemployment impairs mental health: meta-analyses . J Vocat Behav 2009 ; 74 : 264 – 82 . Google Scholar CrossRef Search ADS 8 McKee-Ryan F , Song Z , Wanberg CR , Kinicki AJ . Psychological and physical well-being during unemployment: a meta-analytic study . J Appl Psychol 2005 ; 90 : 53 – 76 . Google Scholar CrossRef Search ADS PubMed 9 Janlert U , Winefield AH , Hammarström A . Length of unemployment and health-related outcomes: a life-course analysis . Eur J Public Health 2015 ; 25 : 662 – 7 . Google Scholar CrossRef Search ADS PubMed 10 Virtanen P , Janlert U , Hammarström A . Health status and health behaviour as predictors of the occurrence of unemployment and prolonged unemployment . Public Health 2013 ; 127 : 46 – 52 . Google Scholar CrossRef Search ADS PubMed 11 Böckerman P , Ilmakunnas P . Unemployment and self-assessed health: evidence from panel data . Health Econ 2009 ; 18 : 161 – 79 . Google Scholar CrossRef Search ADS PubMed 12 Schmitz H . Why are the unemployed in worse health? The causal effect of unemployment on health . Lab Econ 2011 ; 18 : 71 – 8 . Google Scholar CrossRef Search ADS 13 Steele F , French R , Bartley M . Adjusting for selection bias in longitudinal analyses using simultaneous equations modeling: the relationship between employment transitions and mental health . Epidemiology 2013 ; 24 : 703 – 11 . Google Scholar CrossRef Search ADS PubMed 14 Helgesson M , Johansson B , Nordqvist T , et al. Unemployment at a young age and later sickness absence, disability pension and death in native Swedes and immigrants . Eur J Public Health 2013 ; 23 : 606 – 10 . Google Scholar CrossRef Search ADS PubMed 15 Helgesson M , Johansson B , Nordqvist T , et al. Sickness absence at a young age and later sickness absence, disability pension, death, unemployment and income in native Swedes and immigrants . Eur J Public Health 2015 ; 25 : 688 – 92 . Google Scholar CrossRef Search ADS PubMed 16 Torvik FA , Ystrom E , Czajkowski N , Tambs K , Røysamb E , Ørstavik R , Knudsen GP , Kendler KS , Reichborn-Kjennerud T . Socioeconomic status and sick leave granted for mental and somatic disorders: a prospective study of young adult twins . BMC Public Health 2015 ; 15 : 13 . Google Scholar CrossRef Search ADS PubMed 17 Helgesson M , Tinghög P , Niederkrotenthaler T , et al. Labour-market marginalisation after mental disorders among young natives and immigrants living in Sweden . BMC Public Health 2017 ; 17 : 593 . Google Scholar CrossRef Search ADS PubMed 18 Hultin H , Lindholm C , Möller J . Is there an association between long-term sick leave and disability pension and unemployment beyond the effect of health status? A cohort study . PLoS ONE 2012 ; 7 : e35614 . Google Scholar CrossRef Search ADS PubMed 19 Harkko J , Kouvonen A , Virtanen M . Educational attainment, labour market position and mental ill health as pathways from adversities in adolescence to disability pension in early adulthood: a Finnish cohort study using register data . Scand J Public Health 2016 ; 44 ; 678 – 87 . Google Scholar CrossRef Search ADS 20 Vancea M , Utzet M . How unemployment and precarious employment affect the health of young people: a scoping study on social determinants . Scand J Public Health 2017 ; 45 : 73 – 84 . Google Scholar CrossRef Search ADS PubMed 21 Bartley M , Blane D , Montgomery S . Health and the life course: why safety nets matter . BMJ 1997 ; 314 : 1194 – 6 . Google Scholar CrossRef Search ADS PubMed 22 Gissler M , Haukka J . Finnish health and social welfare registers in epidemiological research . Norsk Epidemiologi 2004 ; 14 : 113 – 20 . 23 Iacus SM , King G , Porro G . Causal inference without balance checking: coarsened exact matching . Polit Anal 2012 ; 20 : 1 – 24 . Google Scholar CrossRef Search ADS 24 Aittomäki A , Martikainen P , Laaksonen M , et al. Household economic resources, labour-market advantage and health problems – a study on causal relationships using prospective register data . Soc Sci Med 2012 ; 75 : 1303 – 10 . Google Scholar CrossRef Search ADS PubMed 25 Virtanen P , Vahtera J , Nakari R , et al. Economy and job contract as contexts of sickness absence practices: revisiting locality and habitus . Soc Sci Med 2004 ; 58 : 1219 – 29 . Google Scholar CrossRef Search ADS PubMed 26 Laaksonen M , Blomgren J , Tuulio-Henriksson A . Sickness allowance histories among disability retirees due to mental disorders: a retrospective case–control study . Scand J Public Health 2016 ; 44 : 291 – 9 . Google Scholar CrossRef Search ADS PubMed 27 Stewart JM . The impact of health status on the duration of unemployment spells and the implications for studies of the impact of unemployment on health status . J Health Econ 2001 ; 20 : 781 – 96 . Google Scholar CrossRef Search ADS PubMed 28 Butterworth P , Leach LS , Pirkis J , Kelaher M . Poor mental health influences risk and duration of unemployment: a prospective study . Soc Psychiatry Psychiatr Epidemiol 2012 ; 47 : 1013 – 21 . Google Scholar CrossRef Search ADS PubMed 29 Dooley D , Prause J . The Social Costs of Underemployment: Inadequate Employment as Disguised Unemployment . Cambridge : Cambridge University Press , 2004 . 30 House JS , Lantz PM , Herd P . Continuity and change in the social stratification of aging and health over the life course: evidence from a nationally representative longitudinal study from 1986 to 2001/2002 (Americans' Changing Lives Study) . J Gerontol B Psychol Sci Soc Sci 2005 ; 60 : S15 – 26 . Google Scholar CrossRef Search ADS 31 Haas SA . Health selection and the process of social stratification: the effect of childhood health on socioeconomic attainment . J Health Soc Behav 2006 ; 47 : 339 – 54 . Google Scholar CrossRef Search ADS PubMed 32 Böckerman P , Hyytinen A , Maczulskij T . Devil in disguise: does drinking lead to a disability pension? Prev Med 2016 ; 86 : 130 – 5 . Google Scholar CrossRef Search ADS PubMed 33 Böckerman P , Maczulskij T . The Education-health Nexus: fact and fiction . Soc Sci Med 2016 ; 150 : 112 – 6 . Google Scholar CrossRef Search ADS PubMed 34 Harkonmäki K , Korkeila K , Vahtera J , et al. Childhood adversities as a predictor of disability retirement . J Epidemiol Community Health 2007 ; 61 : 479 – 84 . Google Scholar CrossRef Search ADS PubMed 35 Kestilä L . Pathways to health: Determinants of health, health behaviour and health inequalities in early adulthood. Doctoral Thesis, Univ. Helsinki. 2008 . 36 Stuart EA . Matching methods for causal inference: a review and a look forward . Stat Sci 2010 ; 25 : 1 – 21 . Google Scholar CrossRef Search ADS PubMed 37 Backhans MC , Hemmingsson T . Unemployment and mental health-who is (not) affected? Eur J Pub Health 2012 ; 22 : 429 . Google Scholar CrossRef Search ADS 38 Tøge AG , Blekesaune M . Unemployment transitions and self-rated health in Europe: a longitudinal analysis of EU-SILC from 2008 to 2011 . Soc Sci Med 2015 ; 143 : 171 – 8 . Google Scholar CrossRef Search ADS PubMed 39 Reine I , Novo M , Hammarström A . Unemployment and ill health – a gender analysis: results from a 14-year follow-up of the Northern Swedish Cohort . Public Health 2013 ; 127 : 214 – 22 . Google Scholar CrossRef Search ADS PubMed 40 Karlsson N . Prospective cohort studies of disability pension and mortality in a Swedish county. Doctoral Thesis, Karolinska University Press, Stockholm. 2008 . © The Author(s) 2018. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The European Journal of Public Health Oxford University Press

Unemployment and work disability due to common mental disorders among young adults: selection or causation?

Loading next page...
 
/lp/ou_press/unemployment-and-work-disability-due-to-common-mental-disorders-among-066tGBr1F1
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved.
ISSN
1101-1262
eISSN
1464-360X
D.O.I.
10.1093/eurpub/cky024
Publisher site
See Article on Publisher Site

Abstract

Abstract Background Unemployment in early adulthood is associated with higher rate of disability due to common mental disorders (CMDs). We investigated to what extent the association between unemployment and sub-sequent long-term sickness absence due to CMDs is direct or whether it is dependent on accumulation of mental health problems and socioeconomic disadvantage. Methods In this longitudinal study, a population-based 60% sample of Finnish young adults born between 1983 and 1985 (N = 116 878) was followed up for the incidence of CMDs from 2006 to 2010. Sociodemographic and health-related covariates were identified using several nationwide registers. Hazard ratios (HRs) with 95% confidence intervals (CIs), and survival and cumulative hazard functions for CMD were calculated. A matching procedure was applied to account for the systematic differences in the distribution of the baseline characteristics. Results A total of 1416 (2.4%) of men and 2539 (4.4%) of women were granted a long-term sickness allowance for CMD during the follow-up. After matching, HR (95% CI) of CMD for men decreased from 2.38 (2.12–2.68) to 1.31 (1.03–1.67) and for women from 1.97 (1.79–2.18) to 1.39 (1.18–1.65). Approximately half of the effect of the unemployment on CMDs was explained by the background variables. Conclusion Using a causal approach, our study suggests that unemployment is consistently associated with an increased risk of work disability due to CMDs. Considering the young unemployed as a risk group may help in targeting interventions promoting mental health and improving educational and employment opportunities. Introduction The majority of the burden of mental illness in the community arises from the less severe but more frequent ‘neurotic’ disorders, dominated by anxiety, depression or a combination of both. These conditions are referred to as the ‘common mental disorders’ (CMDs).1 Mental disorders are characterized by an early onset, and CMDs are a leading health problem experienced by young adults in the OECD countries.2 One indication of the topicality of this issue is the considerable increase in the prevalence of sickness and disability benefits, such as long-term sickness allowances, in several developed countries.3 In Finland, the annual proportion of young adults receiving long-term (14 days or over) sickness allowance nearly doubled from 0.27 to 0.52% between 1995 and 2012.4 This trend, typically accountable for mental and behavioural disorders,4 has been associated with actual changes in young adult’s health, but also with factors playing a role in the social determination of health such as the changes in the labour market, the overall economic development, as well as with legislative changes.3,5 It is well-established that the distribution of CMDs follows the socioeconomic gradient, and that employment status is one of key ‘independent markers’ in the gradient.1,6 Meta-analyses have shown the bi-directionality of the association between unemployment and mental ill-health in young adults, and that the effect of unemployment on health may be stronger in younger than older age groups.7,8 Successful completion of education and transition from education to the labour market positively affects mental health and well-being of young adults.7 In contrast, job loss and unemployment exposes young adults to impaired mental health which in turn reduces the likelihood of getting (back) into employment.5,7 All in all, young adults with mental health problems often have a weaker employment history which, in turn, poses them into higher risk of sustained mental ill-health. For several decades, different views of the direction of the causal relationship between unemployment and ill health have been debated. Various studies, mostly focusing on middle-aged populations and using questionnaires on self-rated health, have produced evidence that the association is at least partly an effect of health selection.7,9,10 On the other hand, some studies have given evidence for full health selection11,12 or indirect health selection via socioeconomic processes.13 Studies focusing on sickness benefits indicate that unemployment at young age increases the risk of later sickness allowances due to mental disorders, which in turn increase the risk for job termination and unemployment.14,15 Results using administrative data are, however, inconclusive. For example, a recent twin-study indicated that familial circumstances operate as a common cause for both low socio-economic status (SES) in young adulthood and sub-sequent sick leaves granted for mental disorders.16 Mental disorders constitute a central risk factor for unstable position in the labour market measured either as long-term unemployment, sickness absence, or disability pension.17–19 However, to our best knowledge there are no studies examining the relationship between unemployment and sickness allowances due to CMDs among young adults with an advanced causal statistical approach. A recent review on the effects of unemployment and pre-carious employment on health of young adults recommended studies that can enhance the causal model by including a gender perspective, longitudinal data, more indicators on pre-cariousness and third factor explanations.20 Understanding the role and the direction of causation at early life stages is important as young adulthood is a critical period both for the development of organ systems and for psychological and social development.21 In this study, we focus on the transition to adulthood phase of individuals aged 17–26 years. The hypotheses in this study were that the young adults experiencing unemployment, compared with their counterparts, would have an increased risk of CMDs; and that the selection based on socioeconomic factors such as low educational attainment, poor labour market integration as well as exposure to prior health problems would partly explain this relationship. The main interest was to find out to which extent the selection into unemployment causally explains the relationship, and through which mechanisms possible selection into the unemployment occurs. Methods Study population This was a longitudinal 60% sample of Finnish residents, born between 1983 and 1985. The size of the total population was 119 061 in 2005. After the exclusion of those with a record of rehabilitation for severe disability or a disability pension prior to the follow-up (n = 2232) the size of the analytical sample became 116 829. Several nationwide population and health registers were used. All the registers include the subject’s unique personal ID number, which allowed deterministic linkages between the registers.22 The register holders have approved the use of the data for research. The study was approved by the Ethics Committee of the Rehabilitation Foundation of Finland. Predictor The main predictor was an unemployment benefit claim that was in effect within the time period between October and December 2005. We used unemployment as a dichotomous and continuous variable as long-term unemployment is expected to have more negative effects on CMDs.7,8 The continuous variable was the total length of the unemployment period meeting the inclusion criteria. Lowering the continuous predictor to the power of ¼ provided most accurate predictions, indicating a curvilinear association. The non-exposure group consisted of those who were employed and those outside the labour market for other reasons, such as students. The measure of unemployment status refers to the exposure period only which in this case means that we do not make a distinction between those unemployment periods that ended before the onset of a CMD during the follow-up period and those that were continuous until the onset. In Finland, the unemployment benefit is either a means-tested or an insurance based financial assistance, which is paid as compensation for loss of income due to unemployment. The data were obtained from the Unemployment Register maintained by the Ministry of Economic Affairs and Employment. Outcome The outcome was CMD, which was defined as a sickness allowance period of at least 14 days with a classification of F30-F48 in International Classification of Diseases (ICD-10), the diagnostic criteria used in Finnish medical and social insurance system. The outcome date was the first date of the sickness allowance period during the follow-up. All residents of Finland are insured under the Health Insurance Act and the sickness allowance is paid as compensation for loss of income due to work disability. Also full-time students, unemployed, and those doing household work are eligible for the benefit. The data were obtained from Social Insurance Institution of Finland (SII). Follow-up period The beginning of follow-up time for CMD was the 1st of January 2006 (the index date) when the age of the cohort averaged 22 years. The incidence of a CMD related sickness allowance period was treated as the outcome and the survival time was the distance between the index date and the date of CMD occurrence. The follow-up ended at the end of 2010. Covariates Data on sociodemographic and socioeconomic factors included educational level, residential area, residential status, and the number of years registered as not in employment or education, measured as labour market status in the last week of each calendar year and obtained from Statistic Finland. Health related covariates included annual data on purchases of psychotropic medication, participation in psychiatric rehabilitation provided by SII and previous sickness allowances due to mental disorders (ICD-10 codes F00-F99). The data were obtained from SII and were used as an indicator of prior mental disorders. Statistical analysis Hazard ratios (HRs) with 95% confidence intervals (CIs) were calculated for unemployment (dichotomy). Kaplan–Meier product-limit estimates and Nelson–Aalens cumulative incidence rates were calculated as point estimates for the non-exposed and for individuals unemployed for three months and 12 months (continuous). All analyses were performed separately for men and women. Proportional hazards assumption was tested with Schoenfeld residuals test, which showed no indication of violation of proportionality. All analyses were performed using Stata 12.1 software. Matching procedure The robustness of the results gained from Cox regression models was estimated with matched samples to control selection bias resulting from the differentiated sociodemographic and health profiles of the exposure groups. Matching produced matching weights that balanced the groups nearly identical on all baseline covariates (tables 1 and 2; data shown for the full sample and the final matched sample). Matching was implemented with the Coarsened Exact Matching (CEM) statistical package that has been reported to produce higher quality estimates compared with propensity score matching.23 The matching procedure defined statistically an area of common support containing both treated and control units. The region of common support included 88% of non-exposed men, 72% of exposed men, 85% of non-exposed women and 77% of exposed women. The proportion of selection bias was assessed comparing the difference in cumulative hazard estimates between the exposed and non-exposed in the full and matched samples, first, matching with sociodemographic and socioeconomic covariates and, second, matching with both sociodemographic, socioeconomic and health-related covariates. Table 1 Comparison of baseline characteristics between exposed and non-exposed subjects in the full and matched sample, men Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 59 660 51 634 8026 45 215 5775 Months at risk 3 523 955 3 056 685 467 270 2 664 692 338 240 CMDs, n 1416 1038 378 1374 229 CMDs, % 2.37 2.01 4.71 3.04 3.97 Variable Unemployment at baseline (months)     0 86.5 100.0 0.0 100.0 0.0     1––3 5.6 0.0 41.3 0.0 44.7     3– 7.9 0.0 58.7 0.0 55.3 Education     Secondary education 78.0 79.6 67.4 73.5 73.5     Less than secondary 22.0 20.4 32.6 26.5 26.5 Residential area     South Finland 46.4 47.6 38.7 40.7 40.7     West Finland 26.7 26.8 26.2 26.9 26.9     East Finland 12.7 12.0 17.6 16.8 16.8     North Finland 13.7 13.2 17.2 15.4 15.4     Åland 0.4 0.5 0.2 0.2 0.2 Language, %     Finnish 92.4 92.2 94.1 96.2 96.2     Swedish 5.1 5.6 2.3 1.7 1.7     Other 2.5 2.3 3.6 2.0 2.0 Residential status, %     Parental home 46.3 46.9 42.3 48.2 48.2     Single 32.3 31.7 36.6 35.2 35.2     Own family 20.3 20.6 18.3 15.8 15.8     Without permanent residence 1.1 0.9 2.8 0.8 0.8 Unemployment 2003-baseline (months)     0 51.0 55.8 20.4 27.5 27.5     1–6 27.5 27.3 28.8 35.3 34.7     6– 21.5 16.9 50.8 37.1 37.7 Years ‘not in employment or education’     0 78.8 86.2 31.2 30.9 30.9     1 15.8 12.0 40.8 50.6 50.6     2+ 5.4 1.8 27.9 18.5 18.5 Years of psychotropic drug purchases     0 95.8 96.6 90.7 95.0 95.0     1 2.5 2.0 5.4 3.4 3.4     2+ 1.7 1.4 3.9 1.6 1.6 F-diag sickness allowance 2004–2005     No 99.4 99.5 98.9 99.8 99.8     Yes 0.6 0.5 1.1 0.2 0.2 Vocational rehabilitation     No 98.6 98.7 97.5 99.1 99.1     Yes 1.4 1.3 2.5 0.9 0.9 Psychotherapy     No 99.7 99.7 99.6 99.9 99.9     Yes 0.3 0.3 0.4 0.1 0.1 Any health related indicator     No 93.6 94.6 86.8 93.1 93.1     Yes 6.4 5.4 13.2 6.9 6.9 Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 59 660 51 634 8026 45 215 5775 Months at risk 3 523 955 3 056 685 467 270 2 664 692 338 240 CMDs, n 1416 1038 378 1374 229 CMDs, % 2.37 2.01 4.71 3.04 3.97 Variable Unemployment at baseline (months)     0 86.5 100.0 0.0 100.0 0.0     1––3 5.6 0.0 41.3 0.0 44.7     3– 7.9 0.0 58.7 0.0 55.3 Education     Secondary education 78.0 79.6 67.4 73.5 73.5     Less than secondary 22.0 20.4 32.6 26.5 26.5 Residential area     South Finland 46.4 47.6 38.7 40.7 40.7     West Finland 26.7 26.8 26.2 26.9 26.9     East Finland 12.7 12.0 17.6 16.8 16.8     North Finland 13.7 13.2 17.2 15.4 15.4     Åland 0.4 0.5 0.2 0.2 0.2 Language, %     Finnish 92.4 92.2 94.1 96.2 96.2     Swedish 5.1 5.6 2.3 1.7 1.7     Other 2.5 2.3 3.6 2.0 2.0 Residential status, %     Parental home 46.3 46.9 42.3 48.2 48.2     Single 32.3 31.7 36.6 35.2 35.2     Own family 20.3 20.6 18.3 15.8 15.8     Without permanent residence 1.1 0.9 2.8 0.8 0.8 Unemployment 2003-baseline (months)     0 51.0 55.8 20.4 27.5 27.5     1–6 27.5 27.3 28.8 35.3 34.7     6– 21.5 16.9 50.8 37.1 37.7 Years ‘not in employment or education’     0 78.8 86.2 31.2 30.9 30.9     1 15.8 12.0 40.8 50.6 50.6     2+ 5.4 1.8 27.9 18.5 18.5 Years of psychotropic drug purchases     0 95.8 96.6 90.7 95.0 95.0     1 2.5 2.0 5.4 3.4 3.4     2+ 1.7 1.4 3.9 1.6 1.6 F-diag sickness allowance 2004–2005     No 99.4 99.5 98.9 99.8 99.8     Yes 0.6 0.5 1.1 0.2 0.2 Vocational rehabilitation     No 98.6 98.7 97.5 99.1 99.1     Yes 1.4 1.3 2.5 0.9 0.9 Psychotherapy     No 99.7 99.7 99.6 99.9 99.9     Yes 0.3 0.3 0.4 0.1 0.1 Any health related indicator     No 93.6 94.6 86.8 93.1 93.1     Yes 6.4 5.4 13.2 6.9 6.9 Table 1 Comparison of baseline characteristics between exposed and non-exposed subjects in the full and matched sample, men Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 59 660 51 634 8026 45 215 5775 Months at risk 3 523 955 3 056 685 467 270 2 664 692 338 240 CMDs, n 1416 1038 378 1374 229 CMDs, % 2.37 2.01 4.71 3.04 3.97 Variable Unemployment at baseline (months)     0 86.5 100.0 0.0 100.0 0.0     1––3 5.6 0.0 41.3 0.0 44.7     3– 7.9 0.0 58.7 0.0 55.3 Education     Secondary education 78.0 79.6 67.4 73.5 73.5     Less than secondary 22.0 20.4 32.6 26.5 26.5 Residential area     South Finland 46.4 47.6 38.7 40.7 40.7     West Finland 26.7 26.8 26.2 26.9 26.9     East Finland 12.7 12.0 17.6 16.8 16.8     North Finland 13.7 13.2 17.2 15.4 15.4     Åland 0.4 0.5 0.2 0.2 0.2 Language, %     Finnish 92.4 92.2 94.1 96.2 96.2     Swedish 5.1 5.6 2.3 1.7 1.7     Other 2.5 2.3 3.6 2.0 2.0 Residential status, %     Parental home 46.3 46.9 42.3 48.2 48.2     Single 32.3 31.7 36.6 35.2 35.2     Own family 20.3 20.6 18.3 15.8 15.8     Without permanent residence 1.1 0.9 2.8 0.8 0.8 Unemployment 2003-baseline (months)     0 51.0 55.8 20.4 27.5 27.5     1–6 27.5 27.3 28.8 35.3 34.7     6– 21.5 16.9 50.8 37.1 37.7 Years ‘not in employment or education’     0 78.8 86.2 31.2 30.9 30.9     1 15.8 12.0 40.8 50.6 50.6     2+ 5.4 1.8 27.9 18.5 18.5 Years of psychotropic drug purchases     0 95.8 96.6 90.7 95.0 95.0     1 2.5 2.0 5.4 3.4 3.4     2+ 1.7 1.4 3.9 1.6 1.6 F-diag sickness allowance 2004–2005     No 99.4 99.5 98.9 99.8 99.8     Yes 0.6 0.5 1.1 0.2 0.2 Vocational rehabilitation     No 98.6 98.7 97.5 99.1 99.1     Yes 1.4 1.3 2.5 0.9 0.9 Psychotherapy     No 99.7 99.7 99.6 99.9 99.9     Yes 0.3 0.3 0.4 0.1 0.1 Any health related indicator     No 93.6 94.6 86.8 93.1 93.1     Yes 6.4 5.4 13.2 6.9 6.9 Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 59 660 51 634 8026 45 215 5775 Months at risk 3 523 955 3 056 685 467 270 2 664 692 338 240 CMDs, n 1416 1038 378 1374 229 CMDs, % 2.37 2.01 4.71 3.04 3.97 Variable Unemployment at baseline (months)     0 86.5 100.0 0.0 100.0 0.0     1––3 5.6 0.0 41.3 0.0 44.7     3– 7.9 0.0 58.7 0.0 55.3 Education     Secondary education 78.0 79.6 67.4 73.5 73.5     Less than secondary 22.0 20.4 32.6 26.5 26.5 Residential area     South Finland 46.4 47.6 38.7 40.7 40.7     West Finland 26.7 26.8 26.2 26.9 26.9     East Finland 12.7 12.0 17.6 16.8 16.8     North Finland 13.7 13.2 17.2 15.4 15.4     Åland 0.4 0.5 0.2 0.2 0.2 Language, %     Finnish 92.4 92.2 94.1 96.2 96.2     Swedish 5.1 5.6 2.3 1.7 1.7     Other 2.5 2.3 3.6 2.0 2.0 Residential status, %     Parental home 46.3 46.9 42.3 48.2 48.2     Single 32.3 31.7 36.6 35.2 35.2     Own family 20.3 20.6 18.3 15.8 15.8     Without permanent residence 1.1 0.9 2.8 0.8 0.8 Unemployment 2003-baseline (months)     0 51.0 55.8 20.4 27.5 27.5     1–6 27.5 27.3 28.8 35.3 34.7     6– 21.5 16.9 50.8 37.1 37.7 Years ‘not in employment or education’     0 78.8 86.2 31.2 30.9 30.9     1 15.8 12.0 40.8 50.6 50.6     2+ 5.4 1.8 27.9 18.5 18.5 Years of psychotropic drug purchases     0 95.8 96.6 90.7 95.0 95.0     1 2.5 2.0 5.4 3.4 3.4     2+ 1.7 1.4 3.9 1.6 1.6 F-diag sickness allowance 2004–2005     No 99.4 99.5 98.9 99.8 99.8     Yes 0.6 0.5 1.1 0.2 0.2 Vocational rehabilitation     No 98.6 98.7 97.5 99.1 99.1     Yes 1.4 1.3 2.5 0.9 0.9 Psychotherapy     No 99.7 99.7 99.6 99.9 99.9     Yes 0.3 0.3 0.4 0.1 0.1 Any health related indicator     No 93.6 94.6 86.8 93.1 93.1     Yes 6.4 5.4 13.2 6.9 6.9 Table 2 Comparison of baseline characteristics between exposed and non-exposed subjects in the full and matched sample, women Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 57 169 50 838 6331 43 352 4869 Months at risk 3 335 736 2 972 953 362 783 2 527 751 281 951 CMDs, n 2539 2046 493 1973 307 CMDs, % 4.44 4.02 7.79 4.55 6.31 Variable Unemployment at baseline (months)     0 88.9 100.0 0.0 100.0 0.0     1–3 4.9 0.0 43.9 0.0 46.1     3– 6.2 0.0 56.1 0.0 53.9 Education     Secondary education 84.4 85.3 77.0 83.2 83.2     Less than secondary 15.6 14.7 23.0 16.8 16.8 Residential area     South Finland 48.7 50.0 37.8 38.2 38.2     West Finland 25.9 25.5 29.8 31.1 31.1     East Finland 12.3 11.9 15.6 14.7 14.7     North Finland 12.7 12.2 16.5 15.8 15.8     Åland 0.4 0.4 0.3 0.1 0.1 Language, %     Finnish 92.9 92.6 95.2 97.1 97.1     Swedish 4.9 5.2 1.8 1.3 1.3     Other 2.3 2.2 3.0 1.6 1.6 Residential status, %     Parental home 25.3 25.9 20.8 20.2 20.2     Single 36.2 36.0 37.9 34.8 34.8     Own family 37.6 37.3 39.9 44.6 44.6     Without permanent residence 0.9 0.8 1.5 0.3 0.3 Unemployment 2003-baseline (months)     0 54.4 57.9 25.7 32.2 32.2     1–6 26.6 26.0 31.1 36.1 35.0     6– 19.0 16.0 43.2 31.7 32.8 Years ‘not in employment or education’     0 80.6 86.0 37.1 35.3 35.3     1 13.7 10.4 40.8 47.3 47.3     2+ 5.7 3.6 22.2 17.3 17.3 Years of psychotropic drug purchases     0 92.3 93.0 87.0 91.3 91.3     1 4.6 4.2 7.5 6.0 6.0     2+ 3.1 2.8 5.5 2.8 2.8 F-diag sickness allowance 2004-2005     No 98.7 98.9 97.1 99.0 99.0     Yes 1.3 1.1 2.9 1.0 1.0 Vocational Rehabilitation     No 98.2 98.4 96.8 99.0 99.0     Yes 1.8 1.6 3.2 1.0 1.0 Psychotherapy     No 98.6 98.7 98.6 99.5 99.5     Yes 1.4 1.3 1.4 0.5 0.5 Any health related indicator     No 88.2 89.0 81.4 87.7 87.7     Yes 11.8 11.0 18.6 12.3 12.3 Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 57 169 50 838 6331 43 352 4869 Months at risk 3 335 736 2 972 953 362 783 2 527 751 281 951 CMDs, n 2539 2046 493 1973 307 CMDs, % 4.44 4.02 7.79 4.55 6.31 Variable Unemployment at baseline (months)     0 88.9 100.0 0.0 100.0 0.0     1–3 4.9 0.0 43.9 0.0 46.1     3– 6.2 0.0 56.1 0.0 53.9 Education     Secondary education 84.4 85.3 77.0 83.2 83.2     Less than secondary 15.6 14.7 23.0 16.8 16.8 Residential area     South Finland 48.7 50.0 37.8 38.2 38.2     West Finland 25.9 25.5 29.8 31.1 31.1     East Finland 12.3 11.9 15.6 14.7 14.7     North Finland 12.7 12.2 16.5 15.8 15.8     Åland 0.4 0.4 0.3 0.1 0.1 Language, %     Finnish 92.9 92.6 95.2 97.1 97.1     Swedish 4.9 5.2 1.8 1.3 1.3     Other 2.3 2.2 3.0 1.6 1.6 Residential status, %     Parental home 25.3 25.9 20.8 20.2 20.2     Single 36.2 36.0 37.9 34.8 34.8     Own family 37.6 37.3 39.9 44.6 44.6     Without permanent residence 0.9 0.8 1.5 0.3 0.3 Unemployment 2003-baseline (months)     0 54.4 57.9 25.7 32.2 32.2     1–6 26.6 26.0 31.1 36.1 35.0     6– 19.0 16.0 43.2 31.7 32.8 Years ‘not in employment or education’     0 80.6 86.0 37.1 35.3 35.3     1 13.7 10.4 40.8 47.3 47.3     2+ 5.7 3.6 22.2 17.3 17.3 Years of psychotropic drug purchases     0 92.3 93.0 87.0 91.3 91.3     1 4.6 4.2 7.5 6.0 6.0     2+ 3.1 2.8 5.5 2.8 2.8 F-diag sickness allowance 2004-2005     No 98.7 98.9 97.1 99.0 99.0     Yes 1.3 1.1 2.9 1.0 1.0 Vocational Rehabilitation     No 98.2 98.4 96.8 99.0 99.0     Yes 1.8 1.6 3.2 1.0 1.0 Psychotherapy     No 98.6 98.7 98.6 99.5 99.5     Yes 1.4 1.3 1.4 0.5 0.5 Any health related indicator     No 88.2 89.0 81.4 87.7 87.7     Yes 11.8 11.0 18.6 12.3 12.3 Table 2 Comparison of baseline characteristics between exposed and non-exposed subjects in the full and matched sample, women Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 57 169 50 838 6331 43 352 4869 Months at risk 3 335 736 2 972 953 362 783 2 527 751 281 951 CMDs, n 2539 2046 493 1973 307 CMDs, % 4.44 4.02 7.79 4.55 6.31 Variable Unemployment at baseline (months)     0 88.9 100.0 0.0 100.0 0.0     1–3 4.9 0.0 43.9 0.0 46.1     3– 6.2 0.0 56.1 0.0 53.9 Education     Secondary education 84.4 85.3 77.0 83.2 83.2     Less than secondary 15.6 14.7 23.0 16.8 16.8 Residential area     South Finland 48.7 50.0 37.8 38.2 38.2     West Finland 25.9 25.5 29.8 31.1 31.1     East Finland 12.3 11.9 15.6 14.7 14.7     North Finland 12.7 12.2 16.5 15.8 15.8     Åland 0.4 0.4 0.3 0.1 0.1 Language, %     Finnish 92.9 92.6 95.2 97.1 97.1     Swedish 4.9 5.2 1.8 1.3 1.3     Other 2.3 2.2 3.0 1.6 1.6 Residential status, %     Parental home 25.3 25.9 20.8 20.2 20.2     Single 36.2 36.0 37.9 34.8 34.8     Own family 37.6 37.3 39.9 44.6 44.6     Without permanent residence 0.9 0.8 1.5 0.3 0.3 Unemployment 2003-baseline (months)     0 54.4 57.9 25.7 32.2 32.2     1–6 26.6 26.0 31.1 36.1 35.0     6– 19.0 16.0 43.2 31.7 32.8 Years ‘not in employment or education’     0 80.6 86.0 37.1 35.3 35.3     1 13.7 10.4 40.8 47.3 47.3     2+ 5.7 3.6 22.2 17.3 17.3 Years of psychotropic drug purchases     0 92.3 93.0 87.0 91.3 91.3     1 4.6 4.2 7.5 6.0 6.0     2+ 3.1 2.8 5.5 2.8 2.8 F-diag sickness allowance 2004-2005     No 98.7 98.9 97.1 99.0 99.0     Yes 1.3 1.1 2.9 1.0 1.0 Vocational Rehabilitation     No 98.2 98.4 96.8 99.0 99.0     Yes 1.8 1.6 3.2 1.0 1.0 Psychotherapy     No 98.6 98.7 98.6 99.5 99.5     Yes 1.4 1.3 1.4 0.5 0.5 Any health related indicator     No 88.2 89.0 81.4 87.7 87.7     Yes 11.8 11.0 18.6 12.3 12.3 Full sample Matched sample (CEM) Total Non-exposed Exposed Non-exposed Exposed n 57 169 50 838 6331 43 352 4869 Months at risk 3 335 736 2 972 953 362 783 2 527 751 281 951 CMDs, n 2539 2046 493 1973 307 CMDs, % 4.44 4.02 7.79 4.55 6.31 Variable Unemployment at baseline (months)     0 88.9 100.0 0.0 100.0 0.0     1–3 4.9 0.0 43.9 0.0 46.1     3– 6.2 0.0 56.1 0.0 53.9 Education     Secondary education 84.4 85.3 77.0 83.2 83.2     Less than secondary 15.6 14.7 23.0 16.8 16.8 Residential area     South Finland 48.7 50.0 37.8 38.2 38.2     West Finland 25.9 25.5 29.8 31.1 31.1     East Finland 12.3 11.9 15.6 14.7 14.7     North Finland 12.7 12.2 16.5 15.8 15.8     Åland 0.4 0.4 0.3 0.1 0.1 Language, %     Finnish 92.9 92.6 95.2 97.1 97.1     Swedish 4.9 5.2 1.8 1.3 1.3     Other 2.3 2.2 3.0 1.6 1.6 Residential status, %     Parental home 25.3 25.9 20.8 20.2 20.2     Single 36.2 36.0 37.9 34.8 34.8     Own family 37.6 37.3 39.9 44.6 44.6     Without permanent residence 0.9 0.8 1.5 0.3 0.3 Unemployment 2003-baseline (months)     0 54.4 57.9 25.7 32.2 32.2     1–6 26.6 26.0 31.1 36.1 35.0     6– 19.0 16.0 43.2 31.7 32.8 Years ‘not in employment or education’     0 80.6 86.0 37.1 35.3 35.3     1 13.7 10.4 40.8 47.3 47.3     2+ 5.7 3.6 22.2 17.3 17.3 Years of psychotropic drug purchases     0 92.3 93.0 87.0 91.3 91.3     1 4.6 4.2 7.5 6.0 6.0     2+ 3.1 2.8 5.5 2.8 2.8 F-diag sickness allowance 2004-2005     No 98.7 98.9 97.1 99.0 99.0     Yes 1.3 1.1 2.9 1.0 1.0 Vocational Rehabilitation     No 98.2 98.4 96.8 99.0 99.0     Yes 1.8 1.6 3.2 1.0 1.0 Psychotherapy     No 98.6 98.7 98.6 99.5 99.5     Yes 1.4 1.3 1.4 0.5 0.5 Any health related indicator     No 88.2 89.0 81.4 87.7 87.7     Yes 11.8 11.0 18.6 12.3 12.3 Results Among the 59 660 men and 57 169 women who remained in the study population for 59 months on average, 13.5% of men and 11.1% of women had experienced unemployment (tables 1 and 2). Men had more adverse sociodemographic profiles with more adverse educational attainment and prior unemployment profiles. Baseline ill-health-related indicators were more common in women. 6.4% of men and 11.8% of women had at least one health related problem. In both genders, the unemployed were more disadvantaged in terms of all baseline characteristics. 2.0% of non-exposed men and 4.0% of non-exposed women were granted sickness allowance for CMD during the follow-up. Corresponding figures were 4.7% for exposed men and 7.8% for exposed women. Survival estimates before and after matching Before the matching procedure, the subjects with unemployment had a significantly higher hazard for CMD with HRs of 2.38 (95% CI 2.12–2.68) for men and 1.97 (95% CI 1.79–2.18) for women, respectively (table 3). After the full matching procedure HRs were reduced to 1.31 (1.03–1.67) for men and 1.39 (1.18–1.65) for women, indicating partial causal effects. Those with longer unemployment periods experienced a higher risk of CMD. In terms of cumulative incidence rates, the surplus effect of unemployment on the rate of CMDs at 60 months of survival time was 2.5 for men with three months of unemployment and 4.2 for those with 12 months of unemployment compared with their non-exposed counterparts in the full sample. The corresponding figures were 3.6 and 5.8 for women. Unemployment created excess risk of 1.0 for men with three months of unemployment and 1.6 for those with 12 months of unemployment. The corresponding figures were 1.8 and 2.8 for women. The latter figures represent the genuine effect of unemployment on CMDs, also indicating that the risk of CMD rises with the length of unemployment period even after taking the selection into account. The prolongation of unemployment appeared to have a more negative effect for women. Table 3 HRs for CMDs with 95% CIs; Nelson–Aalens cumulative hazards at 60 months; and the difference of cumulative hazards between the exposure groups and samples Unemployment Nelson–Aalen Cum. Haz. at 60 months Difference in cumulative hazard vs. non-exposeda Proportion of the effect assignable to background covariatesb HR 95% CI Non-exposed Exposed (3 mc) Exposed (12 m) Exposed (3 m) Exposed (12 m) Exposed (3 m) Exposed (12 m) Gender Males     Full sample 2.38 (2.12–2.68) 2.05 4.53 6.27 2.48 4.23 – –     Matched sample 1d 1.65 (1.33–2.04) 2.05 3.90 4.96 1.85 2.92 0.63 1.31     Matched sample 2e 1.31 (1.03–1.67) 2.05 3.09 3.61 1.04 1.56 0.81 1.35 Females     Full sample 1.97 (1.79–2.18) 4.15 7.73 9.99 3.58 5.84 – –     Matched sample 1 1.80 (1.54–2.10) 4.15 7.45 9.43 3.30 5.29 0.28 0.55     Matched sample 2 1.39 (1.18–1.65) 4.15 5.98 6.93 1.83 2.78 1.47 2.51 Unemployment Nelson–Aalen Cum. Haz. at 60 months Difference in cumulative hazard vs. non-exposeda Proportion of the effect assignable to background covariatesb HR 95% CI Non-exposed Exposed (3 mc) Exposed (12 m) Exposed (3 m) Exposed (12 m) Exposed (3 m) Exposed (12 m) Gender Males     Full sample 2.38 (2.12–2.68) 2.05 4.53 6.27 2.48 4.23 – –     Matched sample 1d 1.65 (1.33–2.04) 2.05 3.90 4.96 1.85 2.92 0.63 1.31     Matched sample 2e 1.31 (1.03–1.67) 2.05 3.09 3.61 1.04 1.56 0.81 1.35 Females     Full sample 1.97 (1.79–2.18) 4.15 7.73 9.99 3.58 5.84 – –     Matched sample 1 1.80 (1.54–2.10) 4.15 7.45 9.43 3.30 5.29 0.28 0.55     Matched sample 2 1.39 (1.18–1.65) 4.15 5.98 6.93 1.83 2.78 1.47 2.51 a Difference in Nelson–Aalen cumulative hazard estimate between the exposed and non-exposed (‘average treatment effect’ of the unemployment). b Proportion of the difference assignable to background covariates (‘difference-in-differences’). c Estimate for those with three months of unemployment at the baseline. d Matching based on sociodemographic and socioeconomic covariates. e Matching based on sociodemographic, socioeconomic and health covariates. Table 3 HRs for CMDs with 95% CIs; Nelson–Aalens cumulative hazards at 60 months; and the difference of cumulative hazards between the exposure groups and samples Unemployment Nelson–Aalen Cum. Haz. at 60 months Difference in cumulative hazard vs. non-exposeda Proportion of the effect assignable to background covariatesb HR 95% CI Non-exposed Exposed (3 mc) Exposed (12 m) Exposed (3 m) Exposed (12 m) Exposed (3 m) Exposed (12 m) Gender Males     Full sample 2.38 (2.12–2.68) 2.05 4.53 6.27 2.48 4.23 – –     Matched sample 1d 1.65 (1.33–2.04) 2.05 3.90 4.96 1.85 2.92 0.63 1.31     Matched sample 2e 1.31 (1.03–1.67) 2.05 3.09 3.61 1.04 1.56 0.81 1.35 Females     Full sample 1.97 (1.79–2.18) 4.15 7.73 9.99 3.58 5.84 – –     Matched sample 1 1.80 (1.54–2.10) 4.15 7.45 9.43 3.30 5.29 0.28 0.55     Matched sample 2 1.39 (1.18–1.65) 4.15 5.98 6.93 1.83 2.78 1.47 2.51 Unemployment Nelson–Aalen Cum. Haz. at 60 months Difference in cumulative hazard vs. non-exposeda Proportion of the effect assignable to background covariatesb HR 95% CI Non-exposed Exposed (3 mc) Exposed (12 m) Exposed (3 m) Exposed (12 m) Exposed (3 m) Exposed (12 m) Gender Males     Full sample 2.38 (2.12–2.68) 2.05 4.53 6.27 2.48 4.23 – –     Matched sample 1d 1.65 (1.33–2.04) 2.05 3.90 4.96 1.85 2.92 0.63 1.31     Matched sample 2e 1.31 (1.03–1.67) 2.05 3.09 3.61 1.04 1.56 0.81 1.35 Females     Full sample 1.97 (1.79–2.18) 4.15 7.73 9.99 3.58 5.84 – –     Matched sample 1 1.80 (1.54–2.10) 4.15 7.45 9.43 3.30 5.29 0.28 0.55     Matched sample 2 1.39 (1.18–1.65) 4.15 5.98 6.93 1.83 2.78 1.47 2.51 a Difference in Nelson–Aalen cumulative hazard estimate between the exposed and non-exposed (‘average treatment effect’ of the unemployment). b Proportion of the difference assignable to background covariates (‘difference-in-differences’). c Estimate for those with three months of unemployment at the baseline. d Matching based on sociodemographic and socioeconomic covariates. e Matching based on sociodemographic, socioeconomic and health covariates. For both genders, roughly half of the observed effect of unemployment on CMDs may be interpreted as a contribution of background factors. Health factors prior or at the time of unemployment were the main contributors to this bias; with men having higher proportion of the effect assignable to sociodemographic and socioeconomic factors and women to health status at baseline. We found higher selection bias in the estimates for 12 months of unemployment compared with the estimates for three months of unemployment. The background characteristics may explain not only the exposure to unemployment but the length of unemployment period which in turn is associated with a higher incidence of CMD. The plotted Kaplan–Meier estimates in figure 1 give a visual presentation how the amount of selection bias was dependent on the length of unemployment. The figure confirms our observation of the gendered pattern of the association. Figure 1 View largeDownload slide Kaplan–Meier curves showing CMDs based on full/matched sample comparison and length of exposure to unemployment at the baseline (months), by gender Figure 1 View largeDownload slide Kaplan–Meier curves showing CMDs based on full/matched sample comparison and length of exposure to unemployment at the baseline (months), by gender Discussion This large population-based study adds to the existing evidence on the association between unemployment and CMDs. We found that unemployment is independently associated with CMD related work disability in young adults. Although several studies have suggested that selection into unemployment may explain why the unemployed experience more mental health problems compared with other population groups,7,11–18 no previous study has investigated the selection bias using a large register based longitudinal data, using a causal statistical approach and following a cohort of young adults over a long period of time. Neither has any study, to our best knowledge, studied the association between unemployment and work CMD related work disability in the critical period of early adulthood. We used survival analysis for assessing the cumulative risk of CMD in combination with matching and reweighting estimators, which offered an advanced approach to examine causal inference with observational data. We found that, even though a considerable part of the unemployment—mental ill health association was explained by selection mechanism, there was also a consistent causal association between unemployment and CMDs. We found a dose–response effect that increased with the duration of unemployment. Both causal effect and selection mechanism followed distinct gendered patterns. Strengths and limitations In this study, we used nationwide registers containing information of the use of social security benefits (unemployment benefit and sickness allowance). Sickness allowance does indicate a medically relevant serious CMD. However, as the measure of health problems it does not reflect CMDs that do not prevent people from working and imposes certain degree of under-coverage as the vast majority of those with a health problem do not claim sickness allowance.15,24 We cannot ascertain that factors related to benefit seeking such as personal coping capabilities, relationships, or legislative factors related to receiving such benefits15 are equal between the individuals in different exposure groups. Furthermore, those in pre-carious employment positions, as employed young adults often are, may have disincentives for claiming sickness allowance25 or sickness allowance may mask unemployment.3 All these factors may affect the reliability of the estimates. The using of register data has its strengths. The cohort included over 110 000 young adults, and provided a sufficient number of individuals exposed to different lengths of unemployment which allowed us to use unemployment both as a dichotomous and a continuous variable, as well as to use an advanced matching strategy to account for systematic differences in background characteristics. With a direct link to the implementation of social policy measures, our approach using register data may be seen as a pragmatic way for investigating unemployment and CMDs within the public health context. A further benefit from the policy perspective is that sickness allowance utilization has been found to be a predictor for later disability pension and death15,26 and our results may thus be used in planning preventive programmes. The fact that our baseline health indicators were health interventions highlights the importance to enhance the cooperation between health, employment and social policies to prevent vicious circles in which unemployment may trigger processes leading to mental ill-health. Comparison with previous studies and theoretical implications The health selection hypothesis identifies health as a causative factor to unemployment and thus unemployment may merely ‘mask’ the underlying causes behind the association between unemployment and mental ill-health. We found that a considerable proportion of such an effect behind CMD related work disability indicating that poor mental health may influence the risk and duration of unemployment27,28 and that a larger proportion of individuals with impaired health are being selected in the stock of unemployed.28 Health may be a significant independent factor in the process of becoming unemployed in early adulthood but may also be associated with a trajectory of under-employment, including unemployment, insufficient working hours, low pay and intermittent unemployment spells, which, too, constitute a risk factor for mental ill-health.20,29 Understanding socio-economic mechanisms occurring in early adulthood is an important point of targeted prevention. However, socioeconomic disparities in health grow gradually across the life course, being small in early adulthood and increasing through middle adulthood and early old age.30 Unfortunately, the nature of our data focusing only on birth cohorts 1983–85 did not allow comparisons between different birth cohorts, that is, to make age-group comparisons. Neither did it allow us to follow the cohort through middle adulthood nor to early old age. Our data did not allow to control for many important background factors such as poor health in childhood31 or health behaviours such as risky drinking or sedentary lifestyle,32,33 or the strong influence of familial factors,18 such as growing up in families experiencing economic hardship or in which the safety or balanced development has been compromised with their adverse effects on both educational attainment and mental health.20,34,35 However, we used matching and reweighting estimators which, if unbiased, would balance unobserved variables to the extent they are correlated with those that are observed.36 Consistent with previous studies, we found gender differences in vulnerability to the effects of unemployment.37,38 We observed gender-related differences in the distribution of family formation, educational attainment, employment status, and psychotropic medication at baseline. Given these gendered differences, all analyses were conducted separately for men and women. Unemployed women had a considerably higher risk of CMD even after taking into account of selection bias. Furthermore, we found that, for men, the effect of unemployment on CMDs was more pronouncedly mediated through adverse sociodemographic and socioeconomic indicators and for women through health selection. Prior research has found more pronounced association for unemployment and health behaviour among men and for unemployment and health among women.9,39 Our finding that unemployed men and women have different background profiles may be partly explained by similar gender difference between health and health behaviours prior to unemployment. However, we do not know the extent our results are related to such differences or gendered differences in social insurance practices, disease patterns, or work-related factors.40 The main finding in our study is that both the selection mechanism for unemployment and the effects of the unemployment in young adulthood appear to be gender specific. Conclusions We found a causal association between unemployment and CMDs related to work disability among young adults. The deteriorating effect of unemployment at the transition-to-adulthood period on sub-sequent mental health warrant well-targeted preventive policies. These policies include a set of policies as unemployment may be at the same time an exogenous ‘shock’ leading to financial, material, social, psychological or behavioural changes that will harm individual’s mental health and be a marker of a longer process of accumulation of disadvantage. The latter indicates a need for policies targeting at socio-economic and health conditions prior to unemployment, possibly during the secondary education or already in childhood. A wide array of policy measures that affect SES and the health of young adults have been suggested in public health literature. These include active labour market programmes to prevent long periods of unemployment, training programmes, programmes enhancing young people’s psychosocial resources, social security measures to offset the economic consequences of unemployment, work re-design to reduce social stressors at work, social support and autonomy, specific health programmes and services and support for youth career initiatives.20 Policies that affect labour market trajectories of young people should be targeted especially to those unemployed young adults with no educational qualifications, pre-carious employment histories and early mental disorders. Funding This work was supported by Social Insurance Institution of Finland (SII) (grant KKRL/6/26/2001 to J.H.), Medical Research Council (MRC) (grant MR/K023241/1 to A.K.), Economic and Social Research Council (ESRC) (grant ES/L007509/1 to A.K.), Academy of Finland (grants 258598, 265174 to M.V.). Conflicts of interest: None declared. Key points This large register study examined the effect of unemployment on work disability due to common mental disorders (CMDs) in early adulthood with population level register data. Although part of the associations were explained by prior health status, unemployment increased the incidence of CMDs even after the selection to unemployment was taken into the account with a matching procedure. Unemployment was less prevalent in women but showed to be more detrimental for them in terms of their effect on CMD-related work disability. Public health policy could acknowledge that unemployment in young adulthood may be at the same time an exogenous shock leading to work disability and a ‘marker’ in a process of accumulation of disadvantage leading to work disability. References 1 Fryers T , Melzer D , Jenkins R , Brugha T . The distribution of the common mental disorders: social inequalities in Europe . Clin Pract Epidemiol Ment Health 2005 ; 1 : 14 . Google Scholar CrossRef Search ADS PubMed 2 OECD . Sick on the Job? Myths and Realities about Mental Health and Work . Paris : OECD Publishing , 2012 . 3 Kaltenbrunner Bernitz B , Grees N , Jakobsson Randers M , et al. Young adults on disability benefits in 7 countries . Scand J Public Health 2013 ; 41 : 3 – 26 . Google Scholar CrossRef Search ADS PubMed 4 Koskenvuo K , Hytti H , Autti-Rämö I . Alle 25-vuotiaiden nuorten työkyvyttömyys-ja kuntoutusetuuksien käytön kehitys. Kela.Työpapereita 2013 ;50. [The trends in disability and rehabilitation benefits for young people under 25 years. SII Working Papers.] 5 Reneflot A , Evensen M . Unemployment and psychological distress among young adults in the Nordic countries: a review of the literature . Int J Soc Wel 2014 ; 23 : 3 – 15 . Google Scholar CrossRef Search ADS 6 Lorant V , Deliège D , Eaton W , et al. Socioeconomic inequalities in depression: a meta-analysis . Am J Epidemiol 2003 ; 157 : 98 – 112 . Google Scholar CrossRef Search ADS PubMed 7 Paul KI , Moser K . Unemployment impairs mental health: meta-analyses . J Vocat Behav 2009 ; 74 : 264 – 82 . Google Scholar CrossRef Search ADS 8 McKee-Ryan F , Song Z , Wanberg CR , Kinicki AJ . Psychological and physical well-being during unemployment: a meta-analytic study . J Appl Psychol 2005 ; 90 : 53 – 76 . Google Scholar CrossRef Search ADS PubMed 9 Janlert U , Winefield AH , Hammarström A . Length of unemployment and health-related outcomes: a life-course analysis . Eur J Public Health 2015 ; 25 : 662 – 7 . Google Scholar CrossRef Search ADS PubMed 10 Virtanen P , Janlert U , Hammarström A . Health status and health behaviour as predictors of the occurrence of unemployment and prolonged unemployment . Public Health 2013 ; 127 : 46 – 52 . Google Scholar CrossRef Search ADS PubMed 11 Böckerman P , Ilmakunnas P . Unemployment and self-assessed health: evidence from panel data . Health Econ 2009 ; 18 : 161 – 79 . Google Scholar CrossRef Search ADS PubMed 12 Schmitz H . Why are the unemployed in worse health? The causal effect of unemployment on health . Lab Econ 2011 ; 18 : 71 – 8 . Google Scholar CrossRef Search ADS 13 Steele F , French R , Bartley M . Adjusting for selection bias in longitudinal analyses using simultaneous equations modeling: the relationship between employment transitions and mental health . Epidemiology 2013 ; 24 : 703 – 11 . Google Scholar CrossRef Search ADS PubMed 14 Helgesson M , Johansson B , Nordqvist T , et al. Unemployment at a young age and later sickness absence, disability pension and death in native Swedes and immigrants . Eur J Public Health 2013 ; 23 : 606 – 10 . Google Scholar CrossRef Search ADS PubMed 15 Helgesson M , Johansson B , Nordqvist T , et al. Sickness absence at a young age and later sickness absence, disability pension, death, unemployment and income in native Swedes and immigrants . Eur J Public Health 2015 ; 25 : 688 – 92 . Google Scholar CrossRef Search ADS PubMed 16 Torvik FA , Ystrom E , Czajkowski N , Tambs K , Røysamb E , Ørstavik R , Knudsen GP , Kendler KS , Reichborn-Kjennerud T . Socioeconomic status and sick leave granted for mental and somatic disorders: a prospective study of young adult twins . BMC Public Health 2015 ; 15 : 13 . Google Scholar CrossRef Search ADS PubMed 17 Helgesson M , Tinghög P , Niederkrotenthaler T , et al. Labour-market marginalisation after mental disorders among young natives and immigrants living in Sweden . BMC Public Health 2017 ; 17 : 593 . Google Scholar CrossRef Search ADS PubMed 18 Hultin H , Lindholm C , Möller J . Is there an association between long-term sick leave and disability pension and unemployment beyond the effect of health status? A cohort study . PLoS ONE 2012 ; 7 : e35614 . Google Scholar CrossRef Search ADS PubMed 19 Harkko J , Kouvonen A , Virtanen M . Educational attainment, labour market position and mental ill health as pathways from adversities in adolescence to disability pension in early adulthood: a Finnish cohort study using register data . Scand J Public Health 2016 ; 44 ; 678 – 87 . Google Scholar CrossRef Search ADS 20 Vancea M , Utzet M . How unemployment and precarious employment affect the health of young people: a scoping study on social determinants . Scand J Public Health 2017 ; 45 : 73 – 84 . Google Scholar CrossRef Search ADS PubMed 21 Bartley M , Blane D , Montgomery S . Health and the life course: why safety nets matter . BMJ 1997 ; 314 : 1194 – 6 . Google Scholar CrossRef Search ADS PubMed 22 Gissler M , Haukka J . Finnish health and social welfare registers in epidemiological research . Norsk Epidemiologi 2004 ; 14 : 113 – 20 . 23 Iacus SM , King G , Porro G . Causal inference without balance checking: coarsened exact matching . Polit Anal 2012 ; 20 : 1 – 24 . Google Scholar CrossRef Search ADS 24 Aittomäki A , Martikainen P , Laaksonen M , et al. Household economic resources, labour-market advantage and health problems – a study on causal relationships using prospective register data . Soc Sci Med 2012 ; 75 : 1303 – 10 . Google Scholar CrossRef Search ADS PubMed 25 Virtanen P , Vahtera J , Nakari R , et al. Economy and job contract as contexts of sickness absence practices: revisiting locality and habitus . Soc Sci Med 2004 ; 58 : 1219 – 29 . Google Scholar CrossRef Search ADS PubMed 26 Laaksonen M , Blomgren J , Tuulio-Henriksson A . Sickness allowance histories among disability retirees due to mental disorders: a retrospective case–control study . Scand J Public Health 2016 ; 44 : 291 – 9 . Google Scholar CrossRef Search ADS PubMed 27 Stewart JM . The impact of health status on the duration of unemployment spells and the implications for studies of the impact of unemployment on health status . J Health Econ 2001 ; 20 : 781 – 96 . Google Scholar CrossRef Search ADS PubMed 28 Butterworth P , Leach LS , Pirkis J , Kelaher M . Poor mental health influences risk and duration of unemployment: a prospective study . Soc Psychiatry Psychiatr Epidemiol 2012 ; 47 : 1013 – 21 . Google Scholar CrossRef Search ADS PubMed 29 Dooley D , Prause J . The Social Costs of Underemployment: Inadequate Employment as Disguised Unemployment . Cambridge : Cambridge University Press , 2004 . 30 House JS , Lantz PM , Herd P . Continuity and change in the social stratification of aging and health over the life course: evidence from a nationally representative longitudinal study from 1986 to 2001/2002 (Americans' Changing Lives Study) . J Gerontol B Psychol Sci Soc Sci 2005 ; 60 : S15 – 26 . Google Scholar CrossRef Search ADS 31 Haas SA . Health selection and the process of social stratification: the effect of childhood health on socioeconomic attainment . J Health Soc Behav 2006 ; 47 : 339 – 54 . Google Scholar CrossRef Search ADS PubMed 32 Böckerman P , Hyytinen A , Maczulskij T . Devil in disguise: does drinking lead to a disability pension? Prev Med 2016 ; 86 : 130 – 5 . Google Scholar CrossRef Search ADS PubMed 33 Böckerman P , Maczulskij T . The Education-health Nexus: fact and fiction . Soc Sci Med 2016 ; 150 : 112 – 6 . Google Scholar CrossRef Search ADS PubMed 34 Harkonmäki K , Korkeila K , Vahtera J , et al. Childhood adversities as a predictor of disability retirement . J Epidemiol Community Health 2007 ; 61 : 479 – 84 . Google Scholar CrossRef Search ADS PubMed 35 Kestilä L . Pathways to health: Determinants of health, health behaviour and health inequalities in early adulthood. Doctoral Thesis, Univ. Helsinki. 2008 . 36 Stuart EA . Matching methods for causal inference: a review and a look forward . Stat Sci 2010 ; 25 : 1 – 21 . Google Scholar CrossRef Search ADS PubMed 37 Backhans MC , Hemmingsson T . Unemployment and mental health-who is (not) affected? Eur J Pub Health 2012 ; 22 : 429 . Google Scholar CrossRef Search ADS 38 Tøge AG , Blekesaune M . Unemployment transitions and self-rated health in Europe: a longitudinal analysis of EU-SILC from 2008 to 2011 . Soc Sci Med 2015 ; 143 : 171 – 8 . Google Scholar CrossRef Search ADS PubMed 39 Reine I , Novo M , Hammarström A . Unemployment and ill health – a gender analysis: results from a 14-year follow-up of the Northern Swedish Cohort . Public Health 2013 ; 127 : 214 – 22 . Google Scholar CrossRef Search ADS PubMed 40 Karlsson N . Prospective cohort studies of disability pension and mortality in a Swedish county. Doctoral Thesis, Karolinska University Press, Stockholm. 2008 . © The Author(s) 2018. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved.

Journal

The European Journal of Public HealthOxford University Press

Published: Mar 3, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off