Parental Unemployment and Offspring Psychotropic Medication Purchases: A Longitudinal Fixed-Effects Analysis of 138,644 Adolescents

Parental Unemployment and Offspring Psychotropic Medication Purchases: A Longitudinal... Abstract Parental unemployment is associated with worse adolescent mental health, but prior evidence has primarily been based on cross-sectional studies subject to reverse causality and confounding. We assessed the association between parental unemployment and changes in adolescent psychotropic medication purchases, with longitudinal individual-level fixed-effects models that controlled for time-invariant confounding. We used data from a large, register-based panel of Finnish adolescents aged 13–20 years in 1987–2012 (n = 138,644) that included annual measurements of mothers’ and fathers’ employment and offspring psychotropic medication purchases. We assessed changes in the probability of adolescent psychotropic medication purchases in the years before, during, and after the first episode of parental unemployment. There was no association between mother’s unemployment and offspring psychotropic purchases in the fixed-effects models, suggesting this association is largely driven by unmeasured confounding and selection. By contrast, father’s unemployment led to a significant 15%–20% increase in the probability of purchasing psychotropic medication among adolescents even after extensive controls for observed and unobserved confounding. This change takes at least 1 year to emerge, but it is long-lasting; thus, policies are needed that mitigate the harm of father’s unemployment on offspring’s mental well-being. adolescent health, confounding factors, longitudinal study, mental health, parent, population register, psychotropic drugs, unemployment Environmental exposures play an important role in the etiology of common adolescent mental health problems such as depression and anxiety (1). One potential environmental risk factor is parental unemployment, a common exposure among adolescents. In 2014, in member countries of the Organization for Economic Cooperation and Development, 10% of children aged 0–14 years lived in jobless households (2). Parental unemployment can affect the mental health of their offspring in multiple ways. First, unemployment may compromise the mental health of a parent (3, 4), which may induce mental health problems in their children (5). Second, the economic strain of unemployment may increase marital conflict as well as conflicts between parents and offspring. These, in turn, may affect the emotional well-being of the offspring (6). Lack of economic resources in the family may also affect the mental health of offspring by hindering their social participation with peers (7). According to an increasing body of evidence, parental unemployment is associated with adolescent mental health problems and worse well-being (8–21), yet such an association has not been found in all studies (22–24). Most previous studies, however, were cross-sectional and thus the potential for reverse causality (i.e., the possibility that offspring mental health problems precede parental unemployment) could not be assessed. Higher odds of suicide and attempted suicide among adolescents and young adults with a history of parental unemployment were reported in some studies in which prospective longitudinal data were used (11, 19). However, even in longitudinal studies, where exposure to parental unemployment is measured before health outcomes, unmeasured differences between exposed and unexposed adolescents may bias associations. Among the few studies in which researchers tried to address this bias, findings from a longitudinal study in which individual fixed-effects models were used indicated adolescents exposed to parental unemployment at ages 14–15 years experienced a reduction in happiness, whereas adolescents exposed at ages 11–13 years experienced an increase in happiness (25). In the present study, we aimed to disentangle the nature of the association between parental unemployment and adolescent mental health. We used nationally representative, longitudinal register data from 138,644 Finnish adolescents to assess the association between father’s and mother’s unemployment and offspring psychotropic medication purchases at ages 13–20 years. By using an individual fixed-effects design that infers effects only from within-individual changes in exposure and outcome, we attempted to control for all stable differences between adolescents and families. An important consideration is the time lag between parental unemployment and indications of offspring mental health issues. Short- and long-term effects of parental unemployment have not been differentiated in previous studies. Parental unemployment may cause immediate, short-term effects on offspring mental health, but these effects may fade as the family adapts. Conversely, it is possible that the effects are not immediate but develop gradually and can only be detected some years after parental unemployment. Therefore, we distinguished between immediate and long-term changes in psychotropic purchases. We also assessed adolescent psychotropic purchases in the years prior to parental unemployment, as well as possible mechanisms, including changes in household income, family structure, and parental mental health. Our outcome measure, annual purchases of prescribed psychotropic medication, reflects changes in the presence and severity of adolescent mental health problems, and changes in drug purchasing and prescribing. It is likely to capture the more severe end of mental health problems and problems for which pharmaceutical treatment was followed. METHODS Sample The data were obtained from individual-level linkages between administrative registers using unique personal identification codes available for all permanent residents of Finland. From these data, Statistics Finland drew a 20% random sample of households from the end of year 2000 that had at least 1 child aged 0–14 years, including all household members, supplemented with all non-coresident parents of the children who were 0–14 years old in the household sample. All parents and offspring were linked with annual information on socioeconomic position, labor market participation, and living arrangements in 1987–2012 from the registers of Statistics Finland and with information on all prescription medication purchases in 1995–2012 from the National Prescription Register of the Social Insurance Institution of Finland. For this study, we included birth cohorts for the years 1986–1997 (n = 153,179) for whom at least 2 consecutive years of follow-up data at ages 13–20 years were available during the study period. We excluded adolescents who were not living in private households (e.g., due to institutional care) throughout follow-up (n = 4,909), those who emigrated or died before age 14 years (n = 1,128), those who immigrated after age 13 years (n = 20), and those with both parents unknown (n = 311). For the remaining adolescents (n = 146,811), we linked annual information on the employment status and other sociodemographic characteristics of each parent, irrespective of coresidence with the offspring. To estimate the effect of a parent becoming unemployed, we limited our analyses to adolescents with an employed mother (n = 119,179, 81%) or employed father (n = 121,872, 83%) at age 12 years. These 2 cohorts were analyzed separately; we assessed the first unemployment of the mother and the father, respectively (Table 1). Our final data set included 138,644 adolescents because, for most adolescents, both parents were employed when their child was 12 years old. Table 1. Baseline Distribution and Yearly Prevalence of Psychotropic Medication Purchases During Follow-up Among Adolescents by Baseline Characteristics and Exposure to Parental Unemployment in Follow-upa, Finland, 1987–2012 Exposure Mother’s Unemployment During Follow-up Father’s Unemployment During Follow-up Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Yes (n = 12,115)d No (n = 107,064)e Yes (n = 12,115)d No (n = 107,064)e Yes (n = 13,418)f No (n = 108,454)g Yes (n = 13,418)f No (n = 108,454)g Total 100.0 100.0 3.8 3.1 100.0 100.0 3.9 3.1 Parental educationh  Tertiary 28.0 48.7 3.5 3.0 22.1 38.7 3.5 3.0  Upper secondary 51.7 41.5 3.8 3.2 52.3 44.3 4.1 3.1  Basic 20.2 9.7 4.2 3.3 25.7 17.1 3.9 3.2 Parental psychotropic purchasesh  No 85.5 88.8 3.3 2.8 90.4 92.9 3.7 2.9  Yes 14.5 11.2 7.0 5.6 9.6 7.1 5.9 5.3 Household income quintile  5 (Highest) 11.2 22.5 4.0 3.4 10.8 22.0 4.4 3.6  4 13.8 22.6 4.1 3.4 15.7 21.6 4.1 3.3  3 19.4 22.6 3.7 3.0 21.3 21.4 3.9 2.8  2 26.0 20.1 3.3 3.0 26.3 19.2 3.3 3.0  1 (Lowest) 29.7 12.2 3.2 3.0 26.0 15.8 3.6 2.9 No. of children in parental family  1 25.3 21.7 6.4 5.6 24.0 19.6 7.2 5.3  2 40.3 45.7 4.2 3.7 41.8 43.9 4.2 3.6  ≥3 34.1 32.5 3.9 3.0 33.9 36.3 3.9 3.0  Not in family 0.3 0.1 3.4 2.9 0.3 0.2 3.9 2.8 Family type  2 Parents 73.3 82.7 3.5 2.9 75.4 85.1 3.6 2.9  Single parent 26.5 17.2 4.6 4.4 24.3 14.7 5.0 4.3  Independent 0.3 0.1 6.4 5.6 0.3 0.2 7.2 5.3 Exposure to parental unemployment at ages 1–12 yearsh  No 33.1 69.5 3.3 3.0 44.1 80.2 3.6 2.9  Yes 66.9 30.5 4.1 3.5 55.9 19.9 4.2 3.5 Exposure Mother’s Unemployment During Follow-up Father’s Unemployment During Follow-up Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Yes (n = 12,115)d No (n = 107,064)e Yes (n = 12,115)d No (n = 107,064)e Yes (n = 13,418)f No (n = 108,454)g Yes (n = 13,418)f No (n = 108,454)g Total 100.0 100.0 3.8 3.1 100.0 100.0 3.9 3.1 Parental educationh  Tertiary 28.0 48.7 3.5 3.0 22.1 38.7 3.5 3.0  Upper secondary 51.7 41.5 3.8 3.2 52.3 44.3 4.1 3.1  Basic 20.2 9.7 4.2 3.3 25.7 17.1 3.9 3.2 Parental psychotropic purchasesh  No 85.5 88.8 3.3 2.8 90.4 92.9 3.7 2.9  Yes 14.5 11.2 7.0 5.6 9.6 7.1 5.9 5.3 Household income quintile  5 (Highest) 11.2 22.5 4.0 3.4 10.8 22.0 4.4 3.6  4 13.8 22.6 4.1 3.4 15.7 21.6 4.1 3.3  3 19.4 22.6 3.7 3.0 21.3 21.4 3.9 2.8  2 26.0 20.1 3.3 3.0 26.3 19.2 3.3 3.0  1 (Lowest) 29.7 12.2 3.2 3.0 26.0 15.8 3.6 2.9 No. of children in parental family  1 25.3 21.7 6.4 5.6 24.0 19.6 7.2 5.3  2 40.3 45.7 4.2 3.7 41.8 43.9 4.2 3.6  ≥3 34.1 32.5 3.9 3.0 33.9 36.3 3.9 3.0  Not in family 0.3 0.1 3.4 2.9 0.3 0.2 3.9 2.8 Family type  2 Parents 73.3 82.7 3.5 2.9 75.4 85.1 3.6 2.9  Single parent 26.5 17.2 4.6 4.4 24.3 14.7 5.0 4.3  Independent 0.3 0.1 6.4 5.6 0.3 0.2 7.2 5.3 Exposure to parental unemployment at ages 1–12 yearsh  No 33.1 69.5 3.3 3.0 44.1 80.2 3.6 2.9  Yes 66.9 30.5 4.1 3.5 55.9 19.9 4.2 3.5 a Adolescents aged 13–20 years with an employed mother or father at age 12 years. b Baseline distribution (%) at age 13 years. All differences in distributions between those with and without parental unemployment were significant at the 5% level. Some subgroup values do not sum to 100% because of rounding. c Prevalence (%) of psychotropic medication purchases during follow-up at ages 13–20 years. All differences in psychotropic purchases across groups were significant at the 5% level. d 10.2%; 82,093 person-years. e 89.8%; 659,056 person-years. f 11.0%; 89,413 person-years. g 89.0%; 671,378 person-years. h Parental characteristics refer to the parent whose unemployment is assessed. Table 1. Baseline Distribution and Yearly Prevalence of Psychotropic Medication Purchases During Follow-up Among Adolescents by Baseline Characteristics and Exposure to Parental Unemployment in Follow-upa, Finland, 1987–2012 Exposure Mother’s Unemployment During Follow-up Father’s Unemployment During Follow-up Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Yes (n = 12,115)d No (n = 107,064)e Yes (n = 12,115)d No (n = 107,064)e Yes (n = 13,418)f No (n = 108,454)g Yes (n = 13,418)f No (n = 108,454)g Total 100.0 100.0 3.8 3.1 100.0 100.0 3.9 3.1 Parental educationh  Tertiary 28.0 48.7 3.5 3.0 22.1 38.7 3.5 3.0  Upper secondary 51.7 41.5 3.8 3.2 52.3 44.3 4.1 3.1  Basic 20.2 9.7 4.2 3.3 25.7 17.1 3.9 3.2 Parental psychotropic purchasesh  No 85.5 88.8 3.3 2.8 90.4 92.9 3.7 2.9  Yes 14.5 11.2 7.0 5.6 9.6 7.1 5.9 5.3 Household income quintile  5 (Highest) 11.2 22.5 4.0 3.4 10.8 22.0 4.4 3.6  4 13.8 22.6 4.1 3.4 15.7 21.6 4.1 3.3  3 19.4 22.6 3.7 3.0 21.3 21.4 3.9 2.8  2 26.0 20.1 3.3 3.0 26.3 19.2 3.3 3.0  1 (Lowest) 29.7 12.2 3.2 3.0 26.0 15.8 3.6 2.9 No. of children in parental family  1 25.3 21.7 6.4 5.6 24.0 19.6 7.2 5.3  2 40.3 45.7 4.2 3.7 41.8 43.9 4.2 3.6  ≥3 34.1 32.5 3.9 3.0 33.9 36.3 3.9 3.0  Not in family 0.3 0.1 3.4 2.9 0.3 0.2 3.9 2.8 Family type  2 Parents 73.3 82.7 3.5 2.9 75.4 85.1 3.6 2.9  Single parent 26.5 17.2 4.6 4.4 24.3 14.7 5.0 4.3  Independent 0.3 0.1 6.4 5.6 0.3 0.2 7.2 5.3 Exposure to parental unemployment at ages 1–12 yearsh  No 33.1 69.5 3.3 3.0 44.1 80.2 3.6 2.9  Yes 66.9 30.5 4.1 3.5 55.9 19.9 4.2 3.5 Exposure Mother’s Unemployment During Follow-up Father’s Unemployment During Follow-up Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Yes (n = 12,115)d No (n = 107,064)e Yes (n = 12,115)d No (n = 107,064)e Yes (n = 13,418)f No (n = 108,454)g Yes (n = 13,418)f No (n = 108,454)g Total 100.0 100.0 3.8 3.1 100.0 100.0 3.9 3.1 Parental educationh  Tertiary 28.0 48.7 3.5 3.0 22.1 38.7 3.5 3.0  Upper secondary 51.7 41.5 3.8 3.2 52.3 44.3 4.1 3.1  Basic 20.2 9.7 4.2 3.3 25.7 17.1 3.9 3.2 Parental psychotropic purchasesh  No 85.5 88.8 3.3 2.8 90.4 92.9 3.7 2.9  Yes 14.5 11.2 7.0 5.6 9.6 7.1 5.9 5.3 Household income quintile  5 (Highest) 11.2 22.5 4.0 3.4 10.8 22.0 4.4 3.6  4 13.8 22.6 4.1 3.4 15.7 21.6 4.1 3.3  3 19.4 22.6 3.7 3.0 21.3 21.4 3.9 2.8  2 26.0 20.1 3.3 3.0 26.3 19.2 3.3 3.0  1 (Lowest) 29.7 12.2 3.2 3.0 26.0 15.8 3.6 2.9 No. of children in parental family  1 25.3 21.7 6.4 5.6 24.0 19.6 7.2 5.3  2 40.3 45.7 4.2 3.7 41.8 43.9 4.2 3.6  ≥3 34.1 32.5 3.9 3.0 33.9 36.3 3.9 3.0  Not in family 0.3 0.1 3.4 2.9 0.3 0.2 3.9 2.8 Family type  2 Parents 73.3 82.7 3.5 2.9 75.4 85.1 3.6 2.9  Single parent 26.5 17.2 4.6 4.4 24.3 14.7 5.0 4.3  Independent 0.3 0.1 6.4 5.6 0.3 0.2 7.2 5.3 Exposure to parental unemployment at ages 1–12 yearsh  No 33.1 69.5 3.3 3.0 44.1 80.2 3.6 2.9  Yes 66.9 30.5 4.1 3.5 55.9 19.9 4.2 3.5 a Adolescents aged 13–20 years with an employed mother or father at age 12 years. b Baseline distribution (%) at age 13 years. All differences in distributions between those with and without parental unemployment were significant at the 5% level. Some subgroup values do not sum to 100% because of rounding. c Prevalence (%) of psychotropic medication purchases during follow-up at ages 13–20 years. All differences in psychotropic purchases across groups were significant at the 5% level. d 10.2%; 82,093 person-years. e 89.8%; 659,056 person-years. f 11.0%; 89,413 person-years. g 89.0%; 671,378 person-years. h Parental characteristics refer to the parent whose unemployment is assessed. Measurements The outcome was an annual binary measure of having at least 1 purchase of psychotropic medication over a calendar year (yes/no). Psychotropic medications included antidepressants, antipsychotics, anxiolytics, sedatives, and hypnotics (codes N05A, N05B, N05C, N06A, N06BA, and N06C in the Anatomical Therapeutic Chemical Classification system (26)). In Finland, psychotropic medication is available only from authorized pharmacies by prescription from a medical doctor after clinical assessment and diagnosis. Because the propensity to seek treatment for mental health problems is driven by severity (27, 28), our measure was likely to capture the most severe end of adolescent mental health problems. Employment status was based on the main activity in the last week of the year and included the following 3 categories: 1) employed, if the parent had an ongoing employment contract or was self-employed; 2) unemployed, if the parent was registered as actively looking for employment; and 3) inactive. The employment status of each parent was assessed separately. We identified the first time between offspring ages 13 and 20 years that each parent became unemployed (i.e., was employed in 1 year and unemployed the next). The follow-up years were then coded as follows: 0, years before first unemployment; 1, the first year after unemployment; or 2, subsequent years after unemployment. All follow-up years of adolescents whose parent never became unemployed were included in the 0 category. Years after a parent’s death or emigration were coded separately as 3, parent not in population. These years were included in the analyses but the estimates are not shown. Parental education (tertiary, upper secondary, and basic), psychotropic purchase (no/yes/parent not in population), and experience of any unemployment between offspring ages 1–12 years (no/yes) were assessed for each parent separately. Disposable income of the household where the adolescent was living included the net incomes of all household members, including wages, capital income, and social benefits. To take into account the household structure, we divided the total income by the number of consumption units in the household, using the Organization for Economic Cooperation and Development equivalence scale (29). We then calculated annual quintiles across all adolescents for whom there were data in a given year. Family type was categorized as living with 2 parents, with a single parent, or living independently. The number of underaged children in the parental family was categorized as 1, 2, 3 or more, or not in family, if the offspring was living independently. Statistical analysis We predicted with 4 models the annual probability of offspring psychotropic medication purchases at ages 13–20 years according to the unemployment of each parent in the previous year. Model 1 was an ordinary least squares model controlling for offspring sex and 1-year categorical age. This model yielded the percentage-point difference in the probability of psychotropic purchases between years after first parental unemployment compared with the years before. To differentiate between immediate and long-term changes in psychotropic purchases, we categorized the exposure as first year and subsequent years after parental unemployment. In model 2, to account for observed baseline differences between offspring with and without parental unemployment, we added controls for parental education, parental psychotropic purchases, prior parental unemployment, household income, number of children in the parental family, and family type of the offspring at age 13 years. In model 3, we included individual fixed effects controlling for age. This model controlled for all observed and unobserved time-invariant differences between adolescents, because it used within-individual variation in parental unemployment to predict within-individual variation in psychotropic purchases (30). Model 4 was a fixed-effects model that also included time-varying parental psychotropic purchases, household income quintile, number of children in the parental family, and family type at ages 13–20 years. This model assessed whether changes in these characteristics could explain the association between parental unemployment and offspring mental health. We further assessed changes in the probability of psychotropic medication purchases up to 5 years before and after the first exposure to parental unemployment. This was done, first, to assess potential lagged associations with parental unemployment and, second, to inspect potential reverse causality of preexisting offspring mental health problems affecting later parental employment status. We fitted age-adjusted and fully adjusted fixed-effects models corresponding to models 3 and 4. In these models, the exposure was defined as the time (in years) to the first parental unemployment spell between offspring ages 13–20 years. The exposure variable took on values from −5 to 5, and the reference year t = 0 was the year before the first unemployment spell. We thus estimated the change in the probability of offspring psychotropic purchases in each year t = −5 to t = 5 with respect to year t = 0. All models were run separately for maternal and paternal unemployment and controlled for characteristics of the parent whose unemployment was being examined. The data included siblings whose outcomes were likely to be correlated; therefore, we calculated clustered standard errors at the level of the given parent. All analyses were performed using Stata, version 14.1 (StataCorp LP, College Station, Texas). RESULTS The annual prevalence of psychotropic purchases was 0.5–2 percentage points higher among adolescents with an unemployed mother (Figure 1A) or father (Figure 1B) compared with adolescents with employed parents. Among adolescents with an employed parent at age 12 years, 10% experienced maternal unemployment and 11% experienced paternal unemployment between ages 13–20 years (Table 1). These adolescents had a higher annual prevalence of psychotropic purchases during follow-up compared with unexposed adolescents (approximately 4% vs. 3%). Their parents were more likely to have purchases of psychotropic drugs, a low level of education, and earlier unemployment episodes. The exposed children were also more likely to live in low-income households and single-parent families. All these baseline characteristics predicted psychotropic purchases during follow-up; therefore, these differences may confound the association between parental unemployment and offspring psychotropic purchases. Figure 1. View largeDownload slide Prevalence of psychotropic medication purchases by employment status of mother (A) and father (B) and age among adolescents with employed mother or father at age 12 years, Finland, 1987–2012. Figure 1. View largeDownload slide Prevalence of psychotropic medication purchases by employment status of mother (A) and father (B) and age among adolescents with employed mother or father at age 12 years, Finland, 1987–2012. Offspring psychotropic purchases were more prevalent after the first exposure to parental unemployment than in the years before unemployment (Table 2). Controlling for age and sex (model 1), the probability of psychotropic purchases was 0.61 percentage points (95% confidence interval (CI): 0.30, 0.92) higher in the first year after maternal unemployment, and 0.83 (95% CI: 0.44, 1.22) percentage points higher in subsequent years. Corresponding figures after father’s unemployment were 0.38 (95% CI: 0.08, 0.69) and 0.96 (95% CI: 0.54, 1.37), respectively. The short-term increase in the first year after parental unemployment was substantially attenuated after controlling for observed baseline characteristics (model 2) and disappeared in the fixed-effects models that controlled for unobserved time-invariant characteristics (model 3). By contrast, the long-term increase remained significant after both mother’s (0.45, 95% CI: 0.02, 0.89) and father’s (0.71, 95% CI: 0.29, 1.12) unemployment, even after controlling for time-invariant confounding. The long-term increase was partly explained by changes in parental psychotropic purchases, household income, and family structure during follow-up (model 4), becoming nonsignificant for mother’s unemployment (0.35, 95% confidence interval: −0.09, 0.78) but remaining significant for father’s unemloyment (0.60, 95% confidence interval: 0.19, 1.02). There were no significant differences by sex of the offspring. Table 2. Percentage-Point Change in the Probability of Psychotropic Medication Purchases by Parental Unemploymenta, Finland, 1987–2012 Exposure OLS Model 1b OLS Model 2c FE Model 3d FE Model 4e P Valuef Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Mother’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.61 0.30, 0.92 0.50 0.19, 0.81 0.21 −0.19, 0.61 0.19 −0.21, 0.59 0.649  Subsequent years 0.83 0.44, 1.22 0.68 0.29, 1.07 0.45 0.02, 0.89 0.35 −0.09, 0.78 0.828 Father’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.38 0.08, 0.69 0.27 −0.04, 0.57 0.07 −0.29, 0.42 0.04 −0.31, 0.40 0.134  Subsequent years 0.96 0.54, 1.37 0.77 0.35, 1.18 0.71 0.29, 1.12 0.60 0.19, 1.02 0.922 Exposure OLS Model 1b OLS Model 2c FE Model 3d FE Model 4e P Valuef Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Mother’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.61 0.30, 0.92 0.50 0.19, 0.81 0.21 −0.19, 0.61 0.19 −0.21, 0.59 0.649  Subsequent years 0.83 0.44, 1.22 0.68 0.29, 1.07 0.45 0.02, 0.89 0.35 −0.09, 0.78 0.828 Father’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.38 0.08, 0.69 0.27 −0.04, 0.57 0.07 −0.29, 0.42 0.04 −0.31, 0.40 0.134  Subsequent years 0.96 0.54, 1.37 0.77 0.35, 1.18 0.71 0.29, 1.12 0.60 0.19, 1.02 0.922 Abbreviations: CI, confidence interval; FE, fixed-effects; OLS, ordinary least squares. a Among adolescents aged 13–20 years with employed mother or father at age 12 years. b Model 1 was an OLS model adjusted for 1-year categorical age and sex. c Model 2 was an OLS model adjusted for the variables in model 1 plus baseline characteristics of the parent whose unemployment is assessed (i.e., parental unemployment at ages 1–12 years, parental educational level, parental psychotropic purchases, household income quintile, number of children in the family, and family type at age 13 years). d Model 3 was a fixed-effects model adjusted for 1-year categorical age. e Model 4 was a fixed-effects model adjusted for the variables in model 3 plus time-varying characteristics during follow-up (i.e., parental psychotropic purchases, household income quintile, number of children in the family, family type at ages 13–20 years). fP value for sex interaction, fixed-effects model 4. g The estimates from linear probability models were multiplied by 100 to obtain the presented percentage-point changes. Table 2. Percentage-Point Change in the Probability of Psychotropic Medication Purchases by Parental Unemploymenta, Finland, 1987–2012 Exposure OLS Model 1b OLS Model 2c FE Model 3d FE Model 4e P Valuef Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Mother’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.61 0.30, 0.92 0.50 0.19, 0.81 0.21 −0.19, 0.61 0.19 −0.21, 0.59 0.649  Subsequent years 0.83 0.44, 1.22 0.68 0.29, 1.07 0.45 0.02, 0.89 0.35 −0.09, 0.78 0.828 Father’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.38 0.08, 0.69 0.27 −0.04, 0.57 0.07 −0.29, 0.42 0.04 −0.31, 0.40 0.134  Subsequent years 0.96 0.54, 1.37 0.77 0.35, 1.18 0.71 0.29, 1.12 0.60 0.19, 1.02 0.922 Exposure OLS Model 1b OLS Model 2c FE Model 3d FE Model 4e P Valuef Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Mother’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.61 0.30, 0.92 0.50 0.19, 0.81 0.21 −0.19, 0.61 0.19 −0.21, 0.59 0.649  Subsequent years 0.83 0.44, 1.22 0.68 0.29, 1.07 0.45 0.02, 0.89 0.35 −0.09, 0.78 0.828 Father’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.38 0.08, 0.69 0.27 −0.04, 0.57 0.07 −0.29, 0.42 0.04 −0.31, 0.40 0.134  Subsequent years 0.96 0.54, 1.37 0.77 0.35, 1.18 0.71 0.29, 1.12 0.60 0.19, 1.02 0.922 Abbreviations: CI, confidence interval; FE, fixed-effects; OLS, ordinary least squares. a Among adolescents aged 13–20 years with employed mother or father at age 12 years. b Model 1 was an OLS model adjusted for 1-year categorical age and sex. c Model 2 was an OLS model adjusted for the variables in model 1 plus baseline characteristics of the parent whose unemployment is assessed (i.e., parental unemployment at ages 1–12 years, parental educational level, parental psychotropic purchases, household income quintile, number of children in the family, and family type at age 13 years). d Model 3 was a fixed-effects model adjusted for 1-year categorical age. e Model 4 was a fixed-effects model adjusted for the variables in model 3 plus time-varying characteristics during follow-up (i.e., parental psychotropic purchases, household income quintile, number of children in the family, family type at ages 13–20 years). fP value for sex interaction, fixed-effects model 4. g The estimates from linear probability models were multiplied by 100 to obtain the presented percentage-point changes. To assess whether long-lasting unemployment is particularly detrimental, we further reclassified the years after first parental unemployment as 1) first year after unemployment, 2) subsequent years with long-lasting unemployment, and 3) subsequent years without long-lasting unemployment. Long-lasting unemployment referred to episodes of unemployment that lasted for 2 or more consecutive years, as opposed to unemployment spells that lasted only 1 year. We reran our original models with this more nuanced exposure measure; the results are listed in Web Table 1 (available at https://academic.oup.com/aje). These models revealed no significant differences in the association between parental unemployment and offspring psychotropic purchases according to the length of unemployment. Figure 2 shows estimates from fixed-effects models of the change in the probability of offspring psychotropic purchases 5 years before and 5 years after the first spell of parental unemployment, where the reference year t = 0 is the year before the first unemployment spell. In the 2 years before first maternal unemployment (Figure 2A), there was a 1 percentage-point increase in the probability of psychotropic purchases, but little change afterward (model 3). By contrast, in the 3–5 years before father’s unemployment (Figure 2B), there was a 1 percentage-point increase in offspring psychotropic purchases but also a steep increase in the 2–5 years after unemployment. This suggests that whereas the association between mother’s unemployment and offspring psychotropic purchases may be driven by confounding factors preceding unemployment, there appears to be a more consistent association between father’s unemployment and subsequent offspring psychotropic purchases. These associations were not explained by changes in parental psychotropic purchases, household income, or family structure (model 4). Figure 2. View largeDownload slide Change in the probability of psychotropic medication purchases at ages 13–20 years in the years before and after first unemployment of mother (A) and father (B) among adolescents with employed mother or father at age 12 years, Finland, 1987–2012. Model 3 was a fixed-effects model with 1-year categorical age; model 4 was model 3 plus time-varying characteristics in follow-up (i.e., parental psychotropic purchases, household income quintile, number of children in the family, family type at ages 13–20 years). Estimates from linear probability models were multiplied by 100. Figure 2. View largeDownload slide Change in the probability of psychotropic medication purchases at ages 13–20 years in the years before and after first unemployment of mother (A) and father (B) among adolescents with employed mother or father at age 12 years, Finland, 1987–2012. Model 3 was a fixed-effects model with 1-year categorical age; model 4 was model 3 plus time-varying characteristics in follow-up (i.e., parental psychotropic purchases, household income quintile, number of children in the family, family type at ages 13–20 years). Estimates from linear probability models were multiplied by 100. DISCUSSION We used data from a large register-based panel with extensive controls to assess the association between parental unemployment and adolescent mental health. Our results indicate that some of this association was due to confounding by unobserved differences between exposed and unexposed families. However, according to our results, there appeared to be a 15%–20% long-term increase in the probability of offspring psychotropic purchases, which emerged 1 year after father’s unemployment and was robust to a fixed-effects specification with extensive controls. For mother’s unemployment, this increase was less robust and occurred already before first unemployment. Our results thus support the hypothesis of a causal long-term effect of father’s unemployment on adolescent mental health. The present study makes several contributions to the literature. By using an individual fixed-effects design, we were able to better control for unmeasured time-invariant confounding. Approximately 25%–80% of the association between parental unemployment and adolescent psychotropic purchases was explained by preexisting and stable differences. This suggests considerable confounding in prior cross-sectional and observational studies that did not control for unobserved heterogeneity (8–24). We also distinguished between immediate and long-term changes in mental health after parental unemployment and assessed the psychotropic medication purchases of adolescents already before parental unemployment. Parental unemployment was measured cross-sectionally in previous studies, which conflated immediate and long-term associations (8–24). We showed that changes in the long term were larger and more consistent than in the short term. This may relate to our measurement of offspring mental health, because psychotropic medication purchases are likely to follow the incidence of mental health problems with some lag and only capture the most severe mental health problems treated with psychotropic medication. More consistent immediate changes could be observed with a more direct and sensitive measure of mental health. There is cross-sectional evidence that long-lasting parental unemployment is more strongly associated with adolescent mental health than shorter unemployment spells (10). In contrast, according to our longitudinal sensitivity analyses with fixed-effects modeling, there is no additional risk related to long-lasting unemployment. We examined whether changes in household income, family structure, and parental psychotropic medication purchases were mechanisms that could explain the increases in offspring psychotropic purchases after parental unemployment. These factors explained some but not all of the long-term increase in the probability of purchasing psychotropic medication after parental unemployment. In particular, the increased probability of offspring psychotropic purchases after father’s unemployment remained unexplained. One potential explanation might be that mechanisms are different for paternal and maternal unemployment and our measurements may not capture them equally well. First, women’s reaction to stressful life events such as unemployment may result in development of depression and anxiety, disorders commonly treated by psychotropics, whereas men’s reaction may be more likely to result in development of behavioral disorders such as alcohol or other substance abuse (31, 32) not captured by our register data. If changes in alcohol use could be assessed, the findings might help explain the increases in offspring psychotropic purchasing after father’s unemployment. Second, women may be more likely to seek psychotropic treatment for their mental health problems and thus more likely to be captured by our measure (33). The current empirical evidence, however, is mixed for the gendered responsivity hypothesis and the gendered treatment-seeking patterns (31–33). Although we found consistent changes in offspring psychotropic medication purchases after father’s unemployment, the results for mother’s unemployment were less consistent and potentially driven by reverse causality. This finding is consistent with prior literature examining associations of parental unemployment with outcomes other than health. For example, according to findings of prior studies, there may be a strong association between children’s poor school performance and father’s unemployment, but not maternal unemployment (34–36). A common explanation is that a mother’s unemployment has weaker effects on her own psychological well-being than father’s unemployment, because fathers are more often perceived as primary providers. Indeed, the weaker association between unemployment and own mental health among women has been found in many studies (37–39). However, unemployment among single mothers may be detrimental to children’s educational and psychological outcomes, according to findings of some studies, which suggests the association may be stronger when mothers are primary providers (40). Maternal unemployment may also have a stronger association with offspring mental health in the Nordic countries where the 2-earner family is the norm. Mixed results on the relative importance of maternal and paternal unemployment have been reported in previous cross-sectional studies on offspring mental health outcomes (10, 11, 18, 19, 23–25, 41, 42), but mother’s unemployment is more or as detrimental to offspring health as father’s unemployment mostly in Nordic studies (11, 18, 19, 25, 41). Methodological considerations In this study, we used longitudinal, nationally representative data on parents and offspring based on individual-level linkages of routinely collected administrative registers. These registers have full national coverage, practically no nonresponse or attrition, and are of good quality (26, 43). These are major strengths when assessing unemployment and mental health problems, phenomena related to selective nonresponse and loss to follow-up (44). Furthermore, the large longitudinal sample enabled us to use individual fixed-effects models, which often yield imprecise estimates due to reduced statistical power (30, 45, 46). Some limitations, however, should be acknowledged. In our models, we could not control for possible unobserved time-varying confounders that affect changes in parental employment status and offspring psychotropic purchases. In our full model, we controlled for changes in parental psychotropic purchases, household income, and family structure that could act as either confounders or mechanisms for the association between parental unemployment and offspring mental health. However, unmeasured changes in family circumstances such as parental alcohol or other substance use leading to parental unemployment and to offspring psychotropic purchases could still confound our results. In Finland, psychotropics are only available by prescription from a medical doctor after clinical assessment of mental health problems. However, not all adolescents with mental health problems seek treatment, so our measure captures a combination of mental health problems and willingness and ability to seek treatment. The propensity to seek treatment may be driven by factors such as stigma and preference for self-reliance (47). To the extent that these factors are time invariant, they were controlled for by our individual fixed-effects model. Furthermore, due to adverse effects, nonpharmacologic treatment options such as psychotherapy are often the primary option for adolescent mental health problems. This implies that adolescents with mental health problems who do not seek treatment, those who do not receive a prescription, and those who receive a prescription but do not purchase the prescribed drugs are not captured by our measure. Among Finnish adults with depression, only 25% used antidepressants in 2000 and the likelihood of treatment increased with depression severity, duration, and perceived disability (27, 28). This is probably reflected in our measure of adolescent psychotropic use, because the majority (78%) of psychotropics were antidepressants. Thus, we were likely to capture only the more severe end of adolescent mental health problems. Conclusion According to our results, father’s unemployment may lead to a significant increase in the probability of purchasing psychotropic medication among adolescents. This change is not immediate and takes at least 1 year to emerge, but the effect is long-lasting. These findings highlight the need for policies that mitigate the negative consequences of father’s unemployment on offspring’s mental well-being. ACKNOWLEDGMENTS Author affiliations: Authors’ affiliations: Population Research Unit, Faculty of Social Sciences, University of Helsinki, Helsinki, Finland (Heta Moustgaard, Pekka Martikainen); King’s College London, Department of Global Health and Social Medicine, London, United Kingdom (Mauricio Avendano); Department of Social and Behavioral Sciences, Harvard T.H. Chan School of Public Health, Boston, Massachusetts (Mauricio Avendano); Department of Public Health Sciences, Stockholm University, Stockholm, Sweden (Pekka Martikainen); and Laboratory of Population Health, Max Planck Institute for Demographic Research, Rostock, Germany (Pekka Martikainen). This work was funded by the European Union’s Horizon 2020 research and innovation programs (grant 633666 Lifepath to M.A. and grant 667661 MINDMAP to M.A., P.M., and H.M.), the National Institute on Aging (grant R01AG040248 to M.A.), Signe ja Ane Gyllenberg Foundation (grant 2654 to P.M.), NordForsk program WELLIFE (grant 83540 to P.M.), and the Academy of Finland (grants 294861 and 308247 to P.M. and H.M.). Conflict of interest: none declared. Abbreviation CI confidence interval REFERENCES 1 Uher R . The role of genetic variation in the causation of mental illness: an evolution-informed framework . Mol Psychiatry . 2009 ; 14 ( 12 ): 1072 – 1082 . Google Scholar Crossref Search ADS PubMed 2 Organisation for Economic Co-operation and Development . OECD Family Database. http://www.oecd.org/els/family/database.htm. Accessed November 28, 2016. 3 Paul KI , Moser K . Unemployment impairs mental health: meta-analyses . J Vocat Behav . 2009 ; 74 ( 3 ): 264 – 282 . Google Scholar Crossref Search ADS 4 Riumallo-Herl C , Basu S , Stuckler D , et al. . Job loss, wealth and depression during the Great Recession in the USA and Europe . Int J Epidemiol . 2014 ; 43 ( 5 ): 1508 – 1517 . Google Scholar Crossref Search ADS PubMed 5 McAdams TA , Neiderhiser JM , Rijsdijk FV , et al. . Accounting for genetic and environmental confounds in associations between parent and child characteristics: a systematic review of children-of-twins studies . Psychol Bull . 2014 ; 140 ( 4 ): 1138 – 1173 . Google Scholar Crossref Search ADS PubMed 6 Conger RD , Ge X , Elder GH Jr , et al. . Economic-stress, coercive family process, and developmental problems of adolescents . Child Dev . 1994 ; 65 ( 2 Spec No ): 541 – 561 . Google Scholar Crossref Search ADS PubMed 7 Aberg Yngwe M , Östberg V . The family’s economic resources and adolescents’ health complaints—do adolescents’ own economic resources matter? Eur J Public Health . 2013 ; 23 ( 1 ): 24 – 29 . Google Scholar Crossref Search ADS PubMed 8 Ford T , Goodman R , Meltzer H . The relative importance of child, family, school and neighbourhood correlates of childhood psychiatric disorder . Soc Psychiatry Psychiatr Epidemiol . 2004 ; 39 ( 6 ): 487 – 496 . Google Scholar Crossref Search ADS PubMed 9 Kaltiala-Heino R , Rimpelä M , Rantanen P , et al. . Adolescent depression: the role of discontinuities in life course and social support . J Affect Disord . 2001 ; 64 ( 2–3 ): 155 – 166 . Google Scholar Crossref Search ADS PubMed 10 Sleskova M , Salonna F , Geckova AM , et al. . Does parental unemployment affect adolescents’ health? J Adolesc Health . 2006 ; 38 ( 5 ): 527 – 535 . Google Scholar Crossref Search ADS PubMed 11 Agerbo E , Nordentoft M , Mortensen PB . Familial, psychiatric, and socioeconomic risk factors for suicide in young people: nested case-control study . BMJ . 2002 ; 325 ( 7355 ): 74 – 77 . Google Scholar Crossref Search ADS PubMed 12 Myklestad I , Røysamb E , Tambs K . Risk and protective factors for psychological distress among adolescents: a family study in the Nord-Trøndelag Health Study . Soc Psychiatry Psychiatr Epidemiol . 2012 ; 47 ( 5 ): 771 – 782 . Google Scholar Crossref Search ADS PubMed 13 Reinhardt Pedersen C , Madsen M . Parents’ labour market participation as a predictor of children’s health and wellbeing: a comparative study in five Nordic countries . J Epidemiol Community Health . 2002 ; 56 ( 11 ): 861 – 867 . Google Scholar Crossref Search ADS PubMed 14 Åslund C , Nilsson KW , Starrin B , et al. . Shaming experiences and the association between adolescent depression and psychosocial risk factors . Eur Child Adolesc Psychiatry . 2007 ; 16 ( 5 ): 298 – 304 . Google Scholar Crossref Search ADS PubMed 15 Reinhardt Pedersen C , Madsen M , Köhler L . Does financial strain explain the association between children’s morbidity and parental non-employment? J Epidemiol Community Health . 2005 ; 59 ( 4 ): 316 – 321 . Google Scholar Crossref Search ADS PubMed 16 Pedersen CR , Holstein BE , Köhler L . Parents’ labour market participation as predictor of children’s well-being: changes from 1984 to 1996 in the Nordic countries . Eur J Public Health . 2005 ; 15 ( 4 ): 431 – 436 . Google Scholar Crossref Search ADS PubMed 17 Fröjd S , Kaltiala-Heino R , Pelkonen M , et al. . Significance of family life events in middle adolescence: a survey on Finnish community adolescents . Nord J Psychiatry . 2009 ; 63 ( 1 ): 78 – 86 . Google Scholar Crossref Search ADS PubMed 18 Sund AM , Larsson B , Wichstrøm L . Psychosocial correlates of depressive symptoms among 12–14-year-old Norwegian adolescents . J Child Psychol Psychiatry . 2003 ; 44 ( 4 ): 588 – 597 . Google Scholar Crossref Search ADS PubMed 19 Christoffersen MN , Poulsen HD , Nielsen A . Attempted suicide among young people: risk factors in a prospective register based study of Danish children born in 1966 . Acta Psychiatr Scand . 2003 ; 108 ( 5 ): 350 – 358 . Google Scholar Crossref Search ADS PubMed 20 Forkel I , Silbereisen RK . Family economic hardship and depressed mood among young adolescents from former East and West Germany . Am Behav Sci . 2001 ; 44 ( 11 ): 1955 – 1971 . Google Scholar Crossref Search ADS 21 Fuller-Thomson E , Mehta R , Sulman J . Long-term parental unemployment in childhood and subsequent chronic fatigue syndrome . ISRN Family Med . 2013 ; 2013 : 978250 . Google Scholar PubMed 22 Harland P , Reijneveld SA , Brugman E , et al. . Family factors and life events as risk factors for behavioural and emotional problems in children . Eur Child Adolesc Psychiatry . 2002 ; 11 ( 4 ): 176 – 184 . Google Scholar Crossref Search ADS PubMed 23 Sleskova M , Tuinstra J , Madarasova Geckova A , et al. . Influence of parental employment status on Dutch and Slovak adolescents’ health . BMC Public Health . 2006 ; 6 : 250 . Google Scholar Crossref Search ADS PubMed 24 Bacikova-Sleskova M , Benka J , Orosova O . Parental employment status and adolescents’ health: the role of financial situation, parent-adolescent relationship and adolescents’ resilience . Psychol Health . 2015 ; 30 ( 4 ): 400 – 422 . Google Scholar Crossref Search ADS PubMed 25 Powdthavee N , Vernoit J . Parental unemployment and children’s happiness: a longitudinal study of young people’s well-being in unemployed households . Labour Econ . 2013 ; 24 : 253 – 263 . Google Scholar Crossref Search ADS PubMed 26 Finnish Medicines Agency Fimea & Social Insurance Institution . Finnish Statistics on Medicines 2016 [in Finnish]. Helsinki, Finland : Finnish Medicines Agency Fimea & Social Insurance Institution ; 2017 . http://www.julkari.fi/handle/10024/135599. Accessed March 26, 2018. 27 Hämäläinen J , Isometsä E , Sihvo S , et al. . Treatment of major depressive disorder in the Finnish general population . Depress Anxiety . 2009 ; 26 ( 11 ): 1049 – 1059 . Google Scholar Crossref Search ADS PubMed 28 Hämäläinen J , Isometsä E , Laukkala T , et al. . Use of health services for major depressive episode in Finland . J Affective Disord . 2004 ; 79 ( 1–3 ): 105 – 112 . Google Scholar Crossref Search ADS 29 Organisation for Economic Co-operation and Development . The OECD List of Social Indicators . Paris, France : Organisation for Economic Co-operation and Development ; 1982 . 30 Allison PD . Fixed Effects Regression Models. Quantitative Applications in the Social Sciences . Los Angeles, CA : Sage ; 2009 . 31 Hill TD , Needham BL . Rethinking gender and mental health: a critical analysis of three propositions . Soc Sci Med . 2013 ; 92 : 83 – 91 . Google Scholar Crossref Search ADS PubMed 32 Slopen N , Williams DR , Fitzmaurice GM , et al. . Sex, stressful life events, and adult onset depression and alcohol dependence: are men and women equally vulnerable? Soc Sci Med . 2011 ; 73 ( 4 ): 615 – 622 . Google Scholar Crossref Search ADS PubMed 33 Magaard JL , Seeralan T , Schulz H , et al. . Factors associated with help-seeking behaviour among individuals with major depression: a systematic review . PLoS One . 2017 ; 12 ( 5 ): e0176730 . Google Scholar Crossref Search ADS PubMed 34 Brand JE . The far-reaching impact of job loss and unemployment . Annu Rev Sociol . 2015 ; 41 : 359 – 375 . Google Scholar Crossref Search ADS PubMed 35 Rege M , Telle K , Votruba M . Parental job loss and children’s school performance . Rev Econ Stud . 2011 ; 78 ( 4 ): 1462 – 1489 . Google Scholar Crossref Search ADS 36 Kalil A , Ziol-Guest KM . Parental employment circumstances and children’s academic progress . Soc Sci Res . 2008 ; 37 ( 2 ): 500 – 515 . Google Scholar Crossref Search ADS 37 Artazcoz L , Benach J , Borrell C , et al. . Unemployment and mental health: understanding the interactions among gender, family roles, and social class . Am J Public Health . 2004 ; 94 ( 1 ): 82 – 88 . Google Scholar Crossref Search ADS PubMed 38 Strömberg R , Backlund LG , Löfvander M . Psychosocial stressors and depression at a Swedish primary health care centre. A gender perspective study . BMC Fam Pract . 2011 ; 12 : 120 . Google Scholar Crossref Search ADS PubMed 39 Bijlsma MJ , Tarkiainen L , Myrskylä M , et al. . Unemployment and subsequent depression: a mediation analysis using the parametric G-formula . Soc Sci Med . 2017 ; 194 : 142 – 150 . Google Scholar Crossref Search ADS PubMed 40 Brand JE , Thomas JS . Job displacement among single mothers: effects on children’s outcomes in young adulthood . AJS . 2014 ; 119 ( 4 ): 955 – 1001 . Google Scholar PubMed 41 Christoffersen MN . Growing up with unemployment: a study of parental unemployment and children’s risk of abuse and neglect based on national longitudinal 1973 birth cohorts in Denmark . Childhood . 2000 ; 7 ( 4 ): 421 – 438 . Google Scholar Crossref Search ADS 42 Lundborg P . Young people and alcohol: an econometric analysis . Addiction . 2002 ; 97 ( 12 ): 1573 – 1582 . Google Scholar Crossref Search ADS PubMed 43 Sund R , Gissler M , Hakulinen T , et al. . Use of health registers. In: Ahrens W , Pigeot I , eds. Handbook of Epidemiology . New York, NY : Springer ; 2014 : 707 – 730 . 44 Fischer EH , Dornelas EA , Goethe JW . Characteristics of people lost to attrition in psychiatric follow-up studies . J Nerv Ment Dis . 2001 ; 189 ( 1 ): 49 – 55 . Google Scholar Crossref Search ADS PubMed 45 Kaufman JS . Commentary: why are we biased against bias? Int J Epidemiol . 2008 ; 37 ( 3 ): 624 – 626 . Google Scholar Crossref Search ADS PubMed 46 Madsen M , Osler M . Commentary: strengths and limitations of the discordant twin-pair design in social epidemiology. Where do we go from here? Int J Epidemiol . 2009 ; 38 ( 5 ): 1322 – 1323 . Google Scholar Crossref Search ADS PubMed 47 Gulliver A , Griffiths KM , Christensen H . Perceived barriers and facilitators to mental health help-seeking in young people: a systematic review . BMC Psychiatry . 2010 ; 10 : 113 . Google Scholar Crossref Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png American Journal of Epidemiology Oxford University Press

Parental Unemployment and Offspring Psychotropic Medication Purchases: A Longitudinal Fixed-Effects Analysis of 138,644 Adolescents

Loading next page...
 
/lp/ou_press/parental-unemployment-and-offspring-psychotropic-medication-purchases-2zKNIyB60h
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0002-9262
eISSN
1476-6256
D.O.I.
10.1093/aje/kwy084
Publisher site
See Article on Publisher Site

Abstract

Abstract Parental unemployment is associated with worse adolescent mental health, but prior evidence has primarily been based on cross-sectional studies subject to reverse causality and confounding. We assessed the association between parental unemployment and changes in adolescent psychotropic medication purchases, with longitudinal individual-level fixed-effects models that controlled for time-invariant confounding. We used data from a large, register-based panel of Finnish adolescents aged 13–20 years in 1987–2012 (n = 138,644) that included annual measurements of mothers’ and fathers’ employment and offspring psychotropic medication purchases. We assessed changes in the probability of adolescent psychotropic medication purchases in the years before, during, and after the first episode of parental unemployment. There was no association between mother’s unemployment and offspring psychotropic purchases in the fixed-effects models, suggesting this association is largely driven by unmeasured confounding and selection. By contrast, father’s unemployment led to a significant 15%–20% increase in the probability of purchasing psychotropic medication among adolescents even after extensive controls for observed and unobserved confounding. This change takes at least 1 year to emerge, but it is long-lasting; thus, policies are needed that mitigate the harm of father’s unemployment on offspring’s mental well-being. adolescent health, confounding factors, longitudinal study, mental health, parent, population register, psychotropic drugs, unemployment Environmental exposures play an important role in the etiology of common adolescent mental health problems such as depression and anxiety (1). One potential environmental risk factor is parental unemployment, a common exposure among adolescents. In 2014, in member countries of the Organization for Economic Cooperation and Development, 10% of children aged 0–14 years lived in jobless households (2). Parental unemployment can affect the mental health of their offspring in multiple ways. First, unemployment may compromise the mental health of a parent (3, 4), which may induce mental health problems in their children (5). Second, the economic strain of unemployment may increase marital conflict as well as conflicts between parents and offspring. These, in turn, may affect the emotional well-being of the offspring (6). Lack of economic resources in the family may also affect the mental health of offspring by hindering their social participation with peers (7). According to an increasing body of evidence, parental unemployment is associated with adolescent mental health problems and worse well-being (8–21), yet such an association has not been found in all studies (22–24). Most previous studies, however, were cross-sectional and thus the potential for reverse causality (i.e., the possibility that offspring mental health problems precede parental unemployment) could not be assessed. Higher odds of suicide and attempted suicide among adolescents and young adults with a history of parental unemployment were reported in some studies in which prospective longitudinal data were used (11, 19). However, even in longitudinal studies, where exposure to parental unemployment is measured before health outcomes, unmeasured differences between exposed and unexposed adolescents may bias associations. Among the few studies in which researchers tried to address this bias, findings from a longitudinal study in which individual fixed-effects models were used indicated adolescents exposed to parental unemployment at ages 14–15 years experienced a reduction in happiness, whereas adolescents exposed at ages 11–13 years experienced an increase in happiness (25). In the present study, we aimed to disentangle the nature of the association between parental unemployment and adolescent mental health. We used nationally representative, longitudinal register data from 138,644 Finnish adolescents to assess the association between father’s and mother’s unemployment and offspring psychotropic medication purchases at ages 13–20 years. By using an individual fixed-effects design that infers effects only from within-individual changes in exposure and outcome, we attempted to control for all stable differences between adolescents and families. An important consideration is the time lag between parental unemployment and indications of offspring mental health issues. Short- and long-term effects of parental unemployment have not been differentiated in previous studies. Parental unemployment may cause immediate, short-term effects on offspring mental health, but these effects may fade as the family adapts. Conversely, it is possible that the effects are not immediate but develop gradually and can only be detected some years after parental unemployment. Therefore, we distinguished between immediate and long-term changes in psychotropic purchases. We also assessed adolescent psychotropic purchases in the years prior to parental unemployment, as well as possible mechanisms, including changes in household income, family structure, and parental mental health. Our outcome measure, annual purchases of prescribed psychotropic medication, reflects changes in the presence and severity of adolescent mental health problems, and changes in drug purchasing and prescribing. It is likely to capture the more severe end of mental health problems and problems for which pharmaceutical treatment was followed. METHODS Sample The data were obtained from individual-level linkages between administrative registers using unique personal identification codes available for all permanent residents of Finland. From these data, Statistics Finland drew a 20% random sample of households from the end of year 2000 that had at least 1 child aged 0–14 years, including all household members, supplemented with all non-coresident parents of the children who were 0–14 years old in the household sample. All parents and offspring were linked with annual information on socioeconomic position, labor market participation, and living arrangements in 1987–2012 from the registers of Statistics Finland and with information on all prescription medication purchases in 1995–2012 from the National Prescription Register of the Social Insurance Institution of Finland. For this study, we included birth cohorts for the years 1986–1997 (n = 153,179) for whom at least 2 consecutive years of follow-up data at ages 13–20 years were available during the study period. We excluded adolescents who were not living in private households (e.g., due to institutional care) throughout follow-up (n = 4,909), those who emigrated or died before age 14 years (n = 1,128), those who immigrated after age 13 years (n = 20), and those with both parents unknown (n = 311). For the remaining adolescents (n = 146,811), we linked annual information on the employment status and other sociodemographic characteristics of each parent, irrespective of coresidence with the offspring. To estimate the effect of a parent becoming unemployed, we limited our analyses to adolescents with an employed mother (n = 119,179, 81%) or employed father (n = 121,872, 83%) at age 12 years. These 2 cohorts were analyzed separately; we assessed the first unemployment of the mother and the father, respectively (Table 1). Our final data set included 138,644 adolescents because, for most adolescents, both parents were employed when their child was 12 years old. Table 1. Baseline Distribution and Yearly Prevalence of Psychotropic Medication Purchases During Follow-up Among Adolescents by Baseline Characteristics and Exposure to Parental Unemployment in Follow-upa, Finland, 1987–2012 Exposure Mother’s Unemployment During Follow-up Father’s Unemployment During Follow-up Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Yes (n = 12,115)d No (n = 107,064)e Yes (n = 12,115)d No (n = 107,064)e Yes (n = 13,418)f No (n = 108,454)g Yes (n = 13,418)f No (n = 108,454)g Total 100.0 100.0 3.8 3.1 100.0 100.0 3.9 3.1 Parental educationh  Tertiary 28.0 48.7 3.5 3.0 22.1 38.7 3.5 3.0  Upper secondary 51.7 41.5 3.8 3.2 52.3 44.3 4.1 3.1  Basic 20.2 9.7 4.2 3.3 25.7 17.1 3.9 3.2 Parental psychotropic purchasesh  No 85.5 88.8 3.3 2.8 90.4 92.9 3.7 2.9  Yes 14.5 11.2 7.0 5.6 9.6 7.1 5.9 5.3 Household income quintile  5 (Highest) 11.2 22.5 4.0 3.4 10.8 22.0 4.4 3.6  4 13.8 22.6 4.1 3.4 15.7 21.6 4.1 3.3  3 19.4 22.6 3.7 3.0 21.3 21.4 3.9 2.8  2 26.0 20.1 3.3 3.0 26.3 19.2 3.3 3.0  1 (Lowest) 29.7 12.2 3.2 3.0 26.0 15.8 3.6 2.9 No. of children in parental family  1 25.3 21.7 6.4 5.6 24.0 19.6 7.2 5.3  2 40.3 45.7 4.2 3.7 41.8 43.9 4.2 3.6  ≥3 34.1 32.5 3.9 3.0 33.9 36.3 3.9 3.0  Not in family 0.3 0.1 3.4 2.9 0.3 0.2 3.9 2.8 Family type  2 Parents 73.3 82.7 3.5 2.9 75.4 85.1 3.6 2.9  Single parent 26.5 17.2 4.6 4.4 24.3 14.7 5.0 4.3  Independent 0.3 0.1 6.4 5.6 0.3 0.2 7.2 5.3 Exposure to parental unemployment at ages 1–12 yearsh  No 33.1 69.5 3.3 3.0 44.1 80.2 3.6 2.9  Yes 66.9 30.5 4.1 3.5 55.9 19.9 4.2 3.5 Exposure Mother’s Unemployment During Follow-up Father’s Unemployment During Follow-up Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Yes (n = 12,115)d No (n = 107,064)e Yes (n = 12,115)d No (n = 107,064)e Yes (n = 13,418)f No (n = 108,454)g Yes (n = 13,418)f No (n = 108,454)g Total 100.0 100.0 3.8 3.1 100.0 100.0 3.9 3.1 Parental educationh  Tertiary 28.0 48.7 3.5 3.0 22.1 38.7 3.5 3.0  Upper secondary 51.7 41.5 3.8 3.2 52.3 44.3 4.1 3.1  Basic 20.2 9.7 4.2 3.3 25.7 17.1 3.9 3.2 Parental psychotropic purchasesh  No 85.5 88.8 3.3 2.8 90.4 92.9 3.7 2.9  Yes 14.5 11.2 7.0 5.6 9.6 7.1 5.9 5.3 Household income quintile  5 (Highest) 11.2 22.5 4.0 3.4 10.8 22.0 4.4 3.6  4 13.8 22.6 4.1 3.4 15.7 21.6 4.1 3.3  3 19.4 22.6 3.7 3.0 21.3 21.4 3.9 2.8  2 26.0 20.1 3.3 3.0 26.3 19.2 3.3 3.0  1 (Lowest) 29.7 12.2 3.2 3.0 26.0 15.8 3.6 2.9 No. of children in parental family  1 25.3 21.7 6.4 5.6 24.0 19.6 7.2 5.3  2 40.3 45.7 4.2 3.7 41.8 43.9 4.2 3.6  ≥3 34.1 32.5 3.9 3.0 33.9 36.3 3.9 3.0  Not in family 0.3 0.1 3.4 2.9 0.3 0.2 3.9 2.8 Family type  2 Parents 73.3 82.7 3.5 2.9 75.4 85.1 3.6 2.9  Single parent 26.5 17.2 4.6 4.4 24.3 14.7 5.0 4.3  Independent 0.3 0.1 6.4 5.6 0.3 0.2 7.2 5.3 Exposure to parental unemployment at ages 1–12 yearsh  No 33.1 69.5 3.3 3.0 44.1 80.2 3.6 2.9  Yes 66.9 30.5 4.1 3.5 55.9 19.9 4.2 3.5 a Adolescents aged 13–20 years with an employed mother or father at age 12 years. b Baseline distribution (%) at age 13 years. All differences in distributions between those with and without parental unemployment were significant at the 5% level. Some subgroup values do not sum to 100% because of rounding. c Prevalence (%) of psychotropic medication purchases during follow-up at ages 13–20 years. All differences in psychotropic purchases across groups were significant at the 5% level. d 10.2%; 82,093 person-years. e 89.8%; 659,056 person-years. f 11.0%; 89,413 person-years. g 89.0%; 671,378 person-years. h Parental characteristics refer to the parent whose unemployment is assessed. Table 1. Baseline Distribution and Yearly Prevalence of Psychotropic Medication Purchases During Follow-up Among Adolescents by Baseline Characteristics and Exposure to Parental Unemployment in Follow-upa, Finland, 1987–2012 Exposure Mother’s Unemployment During Follow-up Father’s Unemployment During Follow-up Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Yes (n = 12,115)d No (n = 107,064)e Yes (n = 12,115)d No (n = 107,064)e Yes (n = 13,418)f No (n = 108,454)g Yes (n = 13,418)f No (n = 108,454)g Total 100.0 100.0 3.8 3.1 100.0 100.0 3.9 3.1 Parental educationh  Tertiary 28.0 48.7 3.5 3.0 22.1 38.7 3.5 3.0  Upper secondary 51.7 41.5 3.8 3.2 52.3 44.3 4.1 3.1  Basic 20.2 9.7 4.2 3.3 25.7 17.1 3.9 3.2 Parental psychotropic purchasesh  No 85.5 88.8 3.3 2.8 90.4 92.9 3.7 2.9  Yes 14.5 11.2 7.0 5.6 9.6 7.1 5.9 5.3 Household income quintile  5 (Highest) 11.2 22.5 4.0 3.4 10.8 22.0 4.4 3.6  4 13.8 22.6 4.1 3.4 15.7 21.6 4.1 3.3  3 19.4 22.6 3.7 3.0 21.3 21.4 3.9 2.8  2 26.0 20.1 3.3 3.0 26.3 19.2 3.3 3.0  1 (Lowest) 29.7 12.2 3.2 3.0 26.0 15.8 3.6 2.9 No. of children in parental family  1 25.3 21.7 6.4 5.6 24.0 19.6 7.2 5.3  2 40.3 45.7 4.2 3.7 41.8 43.9 4.2 3.6  ≥3 34.1 32.5 3.9 3.0 33.9 36.3 3.9 3.0  Not in family 0.3 0.1 3.4 2.9 0.3 0.2 3.9 2.8 Family type  2 Parents 73.3 82.7 3.5 2.9 75.4 85.1 3.6 2.9  Single parent 26.5 17.2 4.6 4.4 24.3 14.7 5.0 4.3  Independent 0.3 0.1 6.4 5.6 0.3 0.2 7.2 5.3 Exposure to parental unemployment at ages 1–12 yearsh  No 33.1 69.5 3.3 3.0 44.1 80.2 3.6 2.9  Yes 66.9 30.5 4.1 3.5 55.9 19.9 4.2 3.5 Exposure Mother’s Unemployment During Follow-up Father’s Unemployment During Follow-up Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Baseline Distribution, %b Prevalence of Psychotropic Use During Follow-up, %c Yes (n = 12,115)d No (n = 107,064)e Yes (n = 12,115)d No (n = 107,064)e Yes (n = 13,418)f No (n = 108,454)g Yes (n = 13,418)f No (n = 108,454)g Total 100.0 100.0 3.8 3.1 100.0 100.0 3.9 3.1 Parental educationh  Tertiary 28.0 48.7 3.5 3.0 22.1 38.7 3.5 3.0  Upper secondary 51.7 41.5 3.8 3.2 52.3 44.3 4.1 3.1  Basic 20.2 9.7 4.2 3.3 25.7 17.1 3.9 3.2 Parental psychotropic purchasesh  No 85.5 88.8 3.3 2.8 90.4 92.9 3.7 2.9  Yes 14.5 11.2 7.0 5.6 9.6 7.1 5.9 5.3 Household income quintile  5 (Highest) 11.2 22.5 4.0 3.4 10.8 22.0 4.4 3.6  4 13.8 22.6 4.1 3.4 15.7 21.6 4.1 3.3  3 19.4 22.6 3.7 3.0 21.3 21.4 3.9 2.8  2 26.0 20.1 3.3 3.0 26.3 19.2 3.3 3.0  1 (Lowest) 29.7 12.2 3.2 3.0 26.0 15.8 3.6 2.9 No. of children in parental family  1 25.3 21.7 6.4 5.6 24.0 19.6 7.2 5.3  2 40.3 45.7 4.2 3.7 41.8 43.9 4.2 3.6  ≥3 34.1 32.5 3.9 3.0 33.9 36.3 3.9 3.0  Not in family 0.3 0.1 3.4 2.9 0.3 0.2 3.9 2.8 Family type  2 Parents 73.3 82.7 3.5 2.9 75.4 85.1 3.6 2.9  Single parent 26.5 17.2 4.6 4.4 24.3 14.7 5.0 4.3  Independent 0.3 0.1 6.4 5.6 0.3 0.2 7.2 5.3 Exposure to parental unemployment at ages 1–12 yearsh  No 33.1 69.5 3.3 3.0 44.1 80.2 3.6 2.9  Yes 66.9 30.5 4.1 3.5 55.9 19.9 4.2 3.5 a Adolescents aged 13–20 years with an employed mother or father at age 12 years. b Baseline distribution (%) at age 13 years. All differences in distributions between those with and without parental unemployment were significant at the 5% level. Some subgroup values do not sum to 100% because of rounding. c Prevalence (%) of psychotropic medication purchases during follow-up at ages 13–20 years. All differences in psychotropic purchases across groups were significant at the 5% level. d 10.2%; 82,093 person-years. e 89.8%; 659,056 person-years. f 11.0%; 89,413 person-years. g 89.0%; 671,378 person-years. h Parental characteristics refer to the parent whose unemployment is assessed. Measurements The outcome was an annual binary measure of having at least 1 purchase of psychotropic medication over a calendar year (yes/no). Psychotropic medications included antidepressants, antipsychotics, anxiolytics, sedatives, and hypnotics (codes N05A, N05B, N05C, N06A, N06BA, and N06C in the Anatomical Therapeutic Chemical Classification system (26)). In Finland, psychotropic medication is available only from authorized pharmacies by prescription from a medical doctor after clinical assessment and diagnosis. Because the propensity to seek treatment for mental health problems is driven by severity (27, 28), our measure was likely to capture the most severe end of adolescent mental health problems. Employment status was based on the main activity in the last week of the year and included the following 3 categories: 1) employed, if the parent had an ongoing employment contract or was self-employed; 2) unemployed, if the parent was registered as actively looking for employment; and 3) inactive. The employment status of each parent was assessed separately. We identified the first time between offspring ages 13 and 20 years that each parent became unemployed (i.e., was employed in 1 year and unemployed the next). The follow-up years were then coded as follows: 0, years before first unemployment; 1, the first year after unemployment; or 2, subsequent years after unemployment. All follow-up years of adolescents whose parent never became unemployed were included in the 0 category. Years after a parent’s death or emigration were coded separately as 3, parent not in population. These years were included in the analyses but the estimates are not shown. Parental education (tertiary, upper secondary, and basic), psychotropic purchase (no/yes/parent not in population), and experience of any unemployment between offspring ages 1–12 years (no/yes) were assessed for each parent separately. Disposable income of the household where the adolescent was living included the net incomes of all household members, including wages, capital income, and social benefits. To take into account the household structure, we divided the total income by the number of consumption units in the household, using the Organization for Economic Cooperation and Development equivalence scale (29). We then calculated annual quintiles across all adolescents for whom there were data in a given year. Family type was categorized as living with 2 parents, with a single parent, or living independently. The number of underaged children in the parental family was categorized as 1, 2, 3 or more, or not in family, if the offspring was living independently. Statistical analysis We predicted with 4 models the annual probability of offspring psychotropic medication purchases at ages 13–20 years according to the unemployment of each parent in the previous year. Model 1 was an ordinary least squares model controlling for offspring sex and 1-year categorical age. This model yielded the percentage-point difference in the probability of psychotropic purchases between years after first parental unemployment compared with the years before. To differentiate between immediate and long-term changes in psychotropic purchases, we categorized the exposure as first year and subsequent years after parental unemployment. In model 2, to account for observed baseline differences between offspring with and without parental unemployment, we added controls for parental education, parental psychotropic purchases, prior parental unemployment, household income, number of children in the parental family, and family type of the offspring at age 13 years. In model 3, we included individual fixed effects controlling for age. This model controlled for all observed and unobserved time-invariant differences between adolescents, because it used within-individual variation in parental unemployment to predict within-individual variation in psychotropic purchases (30). Model 4 was a fixed-effects model that also included time-varying parental psychotropic purchases, household income quintile, number of children in the parental family, and family type at ages 13–20 years. This model assessed whether changes in these characteristics could explain the association between parental unemployment and offspring mental health. We further assessed changes in the probability of psychotropic medication purchases up to 5 years before and after the first exposure to parental unemployment. This was done, first, to assess potential lagged associations with parental unemployment and, second, to inspect potential reverse causality of preexisting offspring mental health problems affecting later parental employment status. We fitted age-adjusted and fully adjusted fixed-effects models corresponding to models 3 and 4. In these models, the exposure was defined as the time (in years) to the first parental unemployment spell between offspring ages 13–20 years. The exposure variable took on values from −5 to 5, and the reference year t = 0 was the year before the first unemployment spell. We thus estimated the change in the probability of offspring psychotropic purchases in each year t = −5 to t = 5 with respect to year t = 0. All models were run separately for maternal and paternal unemployment and controlled for characteristics of the parent whose unemployment was being examined. The data included siblings whose outcomes were likely to be correlated; therefore, we calculated clustered standard errors at the level of the given parent. All analyses were performed using Stata, version 14.1 (StataCorp LP, College Station, Texas). RESULTS The annual prevalence of psychotropic purchases was 0.5–2 percentage points higher among adolescents with an unemployed mother (Figure 1A) or father (Figure 1B) compared with adolescents with employed parents. Among adolescents with an employed parent at age 12 years, 10% experienced maternal unemployment and 11% experienced paternal unemployment between ages 13–20 years (Table 1). These adolescents had a higher annual prevalence of psychotropic purchases during follow-up compared with unexposed adolescents (approximately 4% vs. 3%). Their parents were more likely to have purchases of psychotropic drugs, a low level of education, and earlier unemployment episodes. The exposed children were also more likely to live in low-income households and single-parent families. All these baseline characteristics predicted psychotropic purchases during follow-up; therefore, these differences may confound the association between parental unemployment and offspring psychotropic purchases. Figure 1. View largeDownload slide Prevalence of psychotropic medication purchases by employment status of mother (A) and father (B) and age among adolescents with employed mother or father at age 12 years, Finland, 1987–2012. Figure 1. View largeDownload slide Prevalence of psychotropic medication purchases by employment status of mother (A) and father (B) and age among adolescents with employed mother or father at age 12 years, Finland, 1987–2012. Offspring psychotropic purchases were more prevalent after the first exposure to parental unemployment than in the years before unemployment (Table 2). Controlling for age and sex (model 1), the probability of psychotropic purchases was 0.61 percentage points (95% confidence interval (CI): 0.30, 0.92) higher in the first year after maternal unemployment, and 0.83 (95% CI: 0.44, 1.22) percentage points higher in subsequent years. Corresponding figures after father’s unemployment were 0.38 (95% CI: 0.08, 0.69) and 0.96 (95% CI: 0.54, 1.37), respectively. The short-term increase in the first year after parental unemployment was substantially attenuated after controlling for observed baseline characteristics (model 2) and disappeared in the fixed-effects models that controlled for unobserved time-invariant characteristics (model 3). By contrast, the long-term increase remained significant after both mother’s (0.45, 95% CI: 0.02, 0.89) and father’s (0.71, 95% CI: 0.29, 1.12) unemployment, even after controlling for time-invariant confounding. The long-term increase was partly explained by changes in parental psychotropic purchases, household income, and family structure during follow-up (model 4), becoming nonsignificant for mother’s unemployment (0.35, 95% confidence interval: −0.09, 0.78) but remaining significant for father’s unemloyment (0.60, 95% confidence interval: 0.19, 1.02). There were no significant differences by sex of the offspring. Table 2. Percentage-Point Change in the Probability of Psychotropic Medication Purchases by Parental Unemploymenta, Finland, 1987–2012 Exposure OLS Model 1b OLS Model 2c FE Model 3d FE Model 4e P Valuef Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Mother’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.61 0.30, 0.92 0.50 0.19, 0.81 0.21 −0.19, 0.61 0.19 −0.21, 0.59 0.649  Subsequent years 0.83 0.44, 1.22 0.68 0.29, 1.07 0.45 0.02, 0.89 0.35 −0.09, 0.78 0.828 Father’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.38 0.08, 0.69 0.27 −0.04, 0.57 0.07 −0.29, 0.42 0.04 −0.31, 0.40 0.134  Subsequent years 0.96 0.54, 1.37 0.77 0.35, 1.18 0.71 0.29, 1.12 0.60 0.19, 1.02 0.922 Exposure OLS Model 1b OLS Model 2c FE Model 3d FE Model 4e P Valuef Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Mother’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.61 0.30, 0.92 0.50 0.19, 0.81 0.21 −0.19, 0.61 0.19 −0.21, 0.59 0.649  Subsequent years 0.83 0.44, 1.22 0.68 0.29, 1.07 0.45 0.02, 0.89 0.35 −0.09, 0.78 0.828 Father’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.38 0.08, 0.69 0.27 −0.04, 0.57 0.07 −0.29, 0.42 0.04 −0.31, 0.40 0.134  Subsequent years 0.96 0.54, 1.37 0.77 0.35, 1.18 0.71 0.29, 1.12 0.60 0.19, 1.02 0.922 Abbreviations: CI, confidence interval; FE, fixed-effects; OLS, ordinary least squares. a Among adolescents aged 13–20 years with employed mother or father at age 12 years. b Model 1 was an OLS model adjusted for 1-year categorical age and sex. c Model 2 was an OLS model adjusted for the variables in model 1 plus baseline characteristics of the parent whose unemployment is assessed (i.e., parental unemployment at ages 1–12 years, parental educational level, parental psychotropic purchases, household income quintile, number of children in the family, and family type at age 13 years). d Model 3 was a fixed-effects model adjusted for 1-year categorical age. e Model 4 was a fixed-effects model adjusted for the variables in model 3 plus time-varying characteristics during follow-up (i.e., parental psychotropic purchases, household income quintile, number of children in the family, family type at ages 13–20 years). fP value for sex interaction, fixed-effects model 4. g The estimates from linear probability models were multiplied by 100 to obtain the presented percentage-point changes. Table 2. Percentage-Point Change in the Probability of Psychotropic Medication Purchases by Parental Unemploymenta, Finland, 1987–2012 Exposure OLS Model 1b OLS Model 2c FE Model 3d FE Model 4e P Valuef Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Mother’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.61 0.30, 0.92 0.50 0.19, 0.81 0.21 −0.19, 0.61 0.19 −0.21, 0.59 0.649  Subsequent years 0.83 0.44, 1.22 0.68 0.29, 1.07 0.45 0.02, 0.89 0.35 −0.09, 0.78 0.828 Father’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.38 0.08, 0.69 0.27 −0.04, 0.57 0.07 −0.29, 0.42 0.04 −0.31, 0.40 0.134  Subsequent years 0.96 0.54, 1.37 0.77 0.35, 1.18 0.71 0.29, 1.12 0.60 0.19, 1.02 0.922 Exposure OLS Model 1b OLS Model 2c FE Model 3d FE Model 4e P Valuef Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Coefficientg 95% CI Mother’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.61 0.30, 0.92 0.50 0.19, 0.81 0.21 −0.19, 0.61 0.19 −0.21, 0.59 0.649  Subsequent years 0.83 0.44, 1.22 0.68 0.29, 1.07 0.45 0.02, 0.89 0.35 −0.09, 0.78 0.828 Father’s unemployment  Years before first unemployment 0 Referent 0 Referent 0 Referent 0 Referent  First year after unemployment 0.38 0.08, 0.69 0.27 −0.04, 0.57 0.07 −0.29, 0.42 0.04 −0.31, 0.40 0.134  Subsequent years 0.96 0.54, 1.37 0.77 0.35, 1.18 0.71 0.29, 1.12 0.60 0.19, 1.02 0.922 Abbreviations: CI, confidence interval; FE, fixed-effects; OLS, ordinary least squares. a Among adolescents aged 13–20 years with employed mother or father at age 12 years. b Model 1 was an OLS model adjusted for 1-year categorical age and sex. c Model 2 was an OLS model adjusted for the variables in model 1 plus baseline characteristics of the parent whose unemployment is assessed (i.e., parental unemployment at ages 1–12 years, parental educational level, parental psychotropic purchases, household income quintile, number of children in the family, and family type at age 13 years). d Model 3 was a fixed-effects model adjusted for 1-year categorical age. e Model 4 was a fixed-effects model adjusted for the variables in model 3 plus time-varying characteristics during follow-up (i.e., parental psychotropic purchases, household income quintile, number of children in the family, family type at ages 13–20 years). fP value for sex interaction, fixed-effects model 4. g The estimates from linear probability models were multiplied by 100 to obtain the presented percentage-point changes. To assess whether long-lasting unemployment is particularly detrimental, we further reclassified the years after first parental unemployment as 1) first year after unemployment, 2) subsequent years with long-lasting unemployment, and 3) subsequent years without long-lasting unemployment. Long-lasting unemployment referred to episodes of unemployment that lasted for 2 or more consecutive years, as opposed to unemployment spells that lasted only 1 year. We reran our original models with this more nuanced exposure measure; the results are listed in Web Table 1 (available at https://academic.oup.com/aje). These models revealed no significant differences in the association between parental unemployment and offspring psychotropic purchases according to the length of unemployment. Figure 2 shows estimates from fixed-effects models of the change in the probability of offspring psychotropic purchases 5 years before and 5 years after the first spell of parental unemployment, where the reference year t = 0 is the year before the first unemployment spell. In the 2 years before first maternal unemployment (Figure 2A), there was a 1 percentage-point increase in the probability of psychotropic purchases, but little change afterward (model 3). By contrast, in the 3–5 years before father’s unemployment (Figure 2B), there was a 1 percentage-point increase in offspring psychotropic purchases but also a steep increase in the 2–5 years after unemployment. This suggests that whereas the association between mother’s unemployment and offspring psychotropic purchases may be driven by confounding factors preceding unemployment, there appears to be a more consistent association between father’s unemployment and subsequent offspring psychotropic purchases. These associations were not explained by changes in parental psychotropic purchases, household income, or family structure (model 4). Figure 2. View largeDownload slide Change in the probability of psychotropic medication purchases at ages 13–20 years in the years before and after first unemployment of mother (A) and father (B) among adolescents with employed mother or father at age 12 years, Finland, 1987–2012. Model 3 was a fixed-effects model with 1-year categorical age; model 4 was model 3 plus time-varying characteristics in follow-up (i.e., parental psychotropic purchases, household income quintile, number of children in the family, family type at ages 13–20 years). Estimates from linear probability models were multiplied by 100. Figure 2. View largeDownload slide Change in the probability of psychotropic medication purchases at ages 13–20 years in the years before and after first unemployment of mother (A) and father (B) among adolescents with employed mother or father at age 12 years, Finland, 1987–2012. Model 3 was a fixed-effects model with 1-year categorical age; model 4 was model 3 plus time-varying characteristics in follow-up (i.e., parental psychotropic purchases, household income quintile, number of children in the family, family type at ages 13–20 years). Estimates from linear probability models were multiplied by 100. DISCUSSION We used data from a large register-based panel with extensive controls to assess the association between parental unemployment and adolescent mental health. Our results indicate that some of this association was due to confounding by unobserved differences between exposed and unexposed families. However, according to our results, there appeared to be a 15%–20% long-term increase in the probability of offspring psychotropic purchases, which emerged 1 year after father’s unemployment and was robust to a fixed-effects specification with extensive controls. For mother’s unemployment, this increase was less robust and occurred already before first unemployment. Our results thus support the hypothesis of a causal long-term effect of father’s unemployment on adolescent mental health. The present study makes several contributions to the literature. By using an individual fixed-effects design, we were able to better control for unmeasured time-invariant confounding. Approximately 25%–80% of the association between parental unemployment and adolescent psychotropic purchases was explained by preexisting and stable differences. This suggests considerable confounding in prior cross-sectional and observational studies that did not control for unobserved heterogeneity (8–24). We also distinguished between immediate and long-term changes in mental health after parental unemployment and assessed the psychotropic medication purchases of adolescents already before parental unemployment. Parental unemployment was measured cross-sectionally in previous studies, which conflated immediate and long-term associations (8–24). We showed that changes in the long term were larger and more consistent than in the short term. This may relate to our measurement of offspring mental health, because psychotropic medication purchases are likely to follow the incidence of mental health problems with some lag and only capture the most severe mental health problems treated with psychotropic medication. More consistent immediate changes could be observed with a more direct and sensitive measure of mental health. There is cross-sectional evidence that long-lasting parental unemployment is more strongly associated with adolescent mental health than shorter unemployment spells (10). In contrast, according to our longitudinal sensitivity analyses with fixed-effects modeling, there is no additional risk related to long-lasting unemployment. We examined whether changes in household income, family structure, and parental psychotropic medication purchases were mechanisms that could explain the increases in offspring psychotropic purchases after parental unemployment. These factors explained some but not all of the long-term increase in the probability of purchasing psychotropic medication after parental unemployment. In particular, the increased probability of offspring psychotropic purchases after father’s unemployment remained unexplained. One potential explanation might be that mechanisms are different for paternal and maternal unemployment and our measurements may not capture them equally well. First, women’s reaction to stressful life events such as unemployment may result in development of depression and anxiety, disorders commonly treated by psychotropics, whereas men’s reaction may be more likely to result in development of behavioral disorders such as alcohol or other substance abuse (31, 32) not captured by our register data. If changes in alcohol use could be assessed, the findings might help explain the increases in offspring psychotropic purchasing after father’s unemployment. Second, women may be more likely to seek psychotropic treatment for their mental health problems and thus more likely to be captured by our measure (33). The current empirical evidence, however, is mixed for the gendered responsivity hypothesis and the gendered treatment-seeking patterns (31–33). Although we found consistent changes in offspring psychotropic medication purchases after father’s unemployment, the results for mother’s unemployment were less consistent and potentially driven by reverse causality. This finding is consistent with prior literature examining associations of parental unemployment with outcomes other than health. For example, according to findings of prior studies, there may be a strong association between children’s poor school performance and father’s unemployment, but not maternal unemployment (34–36). A common explanation is that a mother’s unemployment has weaker effects on her own psychological well-being than father’s unemployment, because fathers are more often perceived as primary providers. Indeed, the weaker association between unemployment and own mental health among women has been found in many studies (37–39). However, unemployment among single mothers may be detrimental to children’s educational and psychological outcomes, according to findings of some studies, which suggests the association may be stronger when mothers are primary providers (40). Maternal unemployment may also have a stronger association with offspring mental health in the Nordic countries where the 2-earner family is the norm. Mixed results on the relative importance of maternal and paternal unemployment have been reported in previous cross-sectional studies on offspring mental health outcomes (10, 11, 18, 19, 23–25, 41, 42), but mother’s unemployment is more or as detrimental to offspring health as father’s unemployment mostly in Nordic studies (11, 18, 19, 25, 41). Methodological considerations In this study, we used longitudinal, nationally representative data on parents and offspring based on individual-level linkages of routinely collected administrative registers. These registers have full national coverage, practically no nonresponse or attrition, and are of good quality (26, 43). These are major strengths when assessing unemployment and mental health problems, phenomena related to selective nonresponse and loss to follow-up (44). Furthermore, the large longitudinal sample enabled us to use individual fixed-effects models, which often yield imprecise estimates due to reduced statistical power (30, 45, 46). Some limitations, however, should be acknowledged. In our models, we could not control for possible unobserved time-varying confounders that affect changes in parental employment status and offspring psychotropic purchases. In our full model, we controlled for changes in parental psychotropic purchases, household income, and family structure that could act as either confounders or mechanisms for the association between parental unemployment and offspring mental health. However, unmeasured changes in family circumstances such as parental alcohol or other substance use leading to parental unemployment and to offspring psychotropic purchases could still confound our results. In Finland, psychotropics are only available by prescription from a medical doctor after clinical assessment of mental health problems. However, not all adolescents with mental health problems seek treatment, so our measure captures a combination of mental health problems and willingness and ability to seek treatment. The propensity to seek treatment may be driven by factors such as stigma and preference for self-reliance (47). To the extent that these factors are time invariant, they were controlled for by our individual fixed-effects model. Furthermore, due to adverse effects, nonpharmacologic treatment options such as psychotherapy are often the primary option for adolescent mental health problems. This implies that adolescents with mental health problems who do not seek treatment, those who do not receive a prescription, and those who receive a prescription but do not purchase the prescribed drugs are not captured by our measure. Among Finnish adults with depression, only 25% used antidepressants in 2000 and the likelihood of treatment increased with depression severity, duration, and perceived disability (27, 28). This is probably reflected in our measure of adolescent psychotropic use, because the majority (78%) of psychotropics were antidepressants. Thus, we were likely to capture only the more severe end of adolescent mental health problems. Conclusion According to our results, father’s unemployment may lead to a significant increase in the probability of purchasing psychotropic medication among adolescents. This change is not immediate and takes at least 1 year to emerge, but the effect is long-lasting. These findings highlight the need for policies that mitigate the negative consequences of father’s unemployment on offspring’s mental well-being. ACKNOWLEDGMENTS Author affiliations: Authors’ affiliations: Population Research Unit, Faculty of Social Sciences, University of Helsinki, Helsinki, Finland (Heta Moustgaard, Pekka Martikainen); King’s College London, Department of Global Health and Social Medicine, London, United Kingdom (Mauricio Avendano); Department of Social and Behavioral Sciences, Harvard T.H. Chan School of Public Health, Boston, Massachusetts (Mauricio Avendano); Department of Public Health Sciences, Stockholm University, Stockholm, Sweden (Pekka Martikainen); and Laboratory of Population Health, Max Planck Institute for Demographic Research, Rostock, Germany (Pekka Martikainen). This work was funded by the European Union’s Horizon 2020 research and innovation programs (grant 633666 Lifepath to M.A. and grant 667661 MINDMAP to M.A., P.M., and H.M.), the National Institute on Aging (grant R01AG040248 to M.A.), Signe ja Ane Gyllenberg Foundation (grant 2654 to P.M.), NordForsk program WELLIFE (grant 83540 to P.M.), and the Academy of Finland (grants 294861 and 308247 to P.M. and H.M.). Conflict of interest: none declared. Abbreviation CI confidence interval REFERENCES 1 Uher R . The role of genetic variation in the causation of mental illness: an evolution-informed framework . Mol Psychiatry . 2009 ; 14 ( 12 ): 1072 – 1082 . Google Scholar Crossref Search ADS PubMed 2 Organisation for Economic Co-operation and Development . OECD Family Database. http://www.oecd.org/els/family/database.htm. Accessed November 28, 2016. 3 Paul KI , Moser K . Unemployment impairs mental health: meta-analyses . J Vocat Behav . 2009 ; 74 ( 3 ): 264 – 282 . Google Scholar Crossref Search ADS 4 Riumallo-Herl C , Basu S , Stuckler D , et al. . Job loss, wealth and depression during the Great Recession in the USA and Europe . Int J Epidemiol . 2014 ; 43 ( 5 ): 1508 – 1517 . Google Scholar Crossref Search ADS PubMed 5 McAdams TA , Neiderhiser JM , Rijsdijk FV , et al. . Accounting for genetic and environmental confounds in associations between parent and child characteristics: a systematic review of children-of-twins studies . Psychol Bull . 2014 ; 140 ( 4 ): 1138 – 1173 . Google Scholar Crossref Search ADS PubMed 6 Conger RD , Ge X , Elder GH Jr , et al. . Economic-stress, coercive family process, and developmental problems of adolescents . Child Dev . 1994 ; 65 ( 2 Spec No ): 541 – 561 . Google Scholar Crossref Search ADS PubMed 7 Aberg Yngwe M , Östberg V . The family’s economic resources and adolescents’ health complaints—do adolescents’ own economic resources matter? Eur J Public Health . 2013 ; 23 ( 1 ): 24 – 29 . Google Scholar Crossref Search ADS PubMed 8 Ford T , Goodman R , Meltzer H . The relative importance of child, family, school and neighbourhood correlates of childhood psychiatric disorder . Soc Psychiatry Psychiatr Epidemiol . 2004 ; 39 ( 6 ): 487 – 496 . Google Scholar Crossref Search ADS PubMed 9 Kaltiala-Heino R , Rimpelä M , Rantanen P , et al. . Adolescent depression: the role of discontinuities in life course and social support . J Affect Disord . 2001 ; 64 ( 2–3 ): 155 – 166 . Google Scholar Crossref Search ADS PubMed 10 Sleskova M , Salonna F , Geckova AM , et al. . Does parental unemployment affect adolescents’ health? J Adolesc Health . 2006 ; 38 ( 5 ): 527 – 535 . Google Scholar Crossref Search ADS PubMed 11 Agerbo E , Nordentoft M , Mortensen PB . Familial, psychiatric, and socioeconomic risk factors for suicide in young people: nested case-control study . BMJ . 2002 ; 325 ( 7355 ): 74 – 77 . Google Scholar Crossref Search ADS PubMed 12 Myklestad I , Røysamb E , Tambs K . Risk and protective factors for psychological distress among adolescents: a family study in the Nord-Trøndelag Health Study . Soc Psychiatry Psychiatr Epidemiol . 2012 ; 47 ( 5 ): 771 – 782 . Google Scholar Crossref Search ADS PubMed 13 Reinhardt Pedersen C , Madsen M . Parents’ labour market participation as a predictor of children’s health and wellbeing: a comparative study in five Nordic countries . J Epidemiol Community Health . 2002 ; 56 ( 11 ): 861 – 867 . Google Scholar Crossref Search ADS PubMed 14 Åslund C , Nilsson KW , Starrin B , et al. . Shaming experiences and the association between adolescent depression and psychosocial risk factors . Eur Child Adolesc Psychiatry . 2007 ; 16 ( 5 ): 298 – 304 . Google Scholar Crossref Search ADS PubMed 15 Reinhardt Pedersen C , Madsen M , Köhler L . Does financial strain explain the association between children’s morbidity and parental non-employment? J Epidemiol Community Health . 2005 ; 59 ( 4 ): 316 – 321 . Google Scholar Crossref Search ADS PubMed 16 Pedersen CR , Holstein BE , Köhler L . Parents’ labour market participation as predictor of children’s well-being: changes from 1984 to 1996 in the Nordic countries . Eur J Public Health . 2005 ; 15 ( 4 ): 431 – 436 . Google Scholar Crossref Search ADS PubMed 17 Fröjd S , Kaltiala-Heino R , Pelkonen M , et al. . Significance of family life events in middle adolescence: a survey on Finnish community adolescents . Nord J Psychiatry . 2009 ; 63 ( 1 ): 78 – 86 . Google Scholar Crossref Search ADS PubMed 18 Sund AM , Larsson B , Wichstrøm L . Psychosocial correlates of depressive symptoms among 12–14-year-old Norwegian adolescents . J Child Psychol Psychiatry . 2003 ; 44 ( 4 ): 588 – 597 . Google Scholar Crossref Search ADS PubMed 19 Christoffersen MN , Poulsen HD , Nielsen A . Attempted suicide among young people: risk factors in a prospective register based study of Danish children born in 1966 . Acta Psychiatr Scand . 2003 ; 108 ( 5 ): 350 – 358 . Google Scholar Crossref Search ADS PubMed 20 Forkel I , Silbereisen RK . Family economic hardship and depressed mood among young adolescents from former East and West Germany . Am Behav Sci . 2001 ; 44 ( 11 ): 1955 – 1971 . Google Scholar Crossref Search ADS 21 Fuller-Thomson E , Mehta R , Sulman J . Long-term parental unemployment in childhood and subsequent chronic fatigue syndrome . ISRN Family Med . 2013 ; 2013 : 978250 . Google Scholar PubMed 22 Harland P , Reijneveld SA , Brugman E , et al. . Family factors and life events as risk factors for behavioural and emotional problems in children . Eur Child Adolesc Psychiatry . 2002 ; 11 ( 4 ): 176 – 184 . Google Scholar Crossref Search ADS PubMed 23 Sleskova M , Tuinstra J , Madarasova Geckova A , et al. . Influence of parental employment status on Dutch and Slovak adolescents’ health . BMC Public Health . 2006 ; 6 : 250 . Google Scholar Crossref Search ADS PubMed 24 Bacikova-Sleskova M , Benka J , Orosova O . Parental employment status and adolescents’ health: the role of financial situation, parent-adolescent relationship and adolescents’ resilience . Psychol Health . 2015 ; 30 ( 4 ): 400 – 422 . Google Scholar Crossref Search ADS PubMed 25 Powdthavee N , Vernoit J . Parental unemployment and children’s happiness: a longitudinal study of young people’s well-being in unemployed households . Labour Econ . 2013 ; 24 : 253 – 263 . Google Scholar Crossref Search ADS PubMed 26 Finnish Medicines Agency Fimea & Social Insurance Institution . Finnish Statistics on Medicines 2016 [in Finnish]. Helsinki, Finland : Finnish Medicines Agency Fimea & Social Insurance Institution ; 2017 . http://www.julkari.fi/handle/10024/135599. Accessed March 26, 2018. 27 Hämäläinen J , Isometsä E , Sihvo S , et al. . Treatment of major depressive disorder in the Finnish general population . Depress Anxiety . 2009 ; 26 ( 11 ): 1049 – 1059 . Google Scholar Crossref Search ADS PubMed 28 Hämäläinen J , Isometsä E , Laukkala T , et al. . Use of health services for major depressive episode in Finland . J Affective Disord . 2004 ; 79 ( 1–3 ): 105 – 112 . Google Scholar Crossref Search ADS 29 Organisation for Economic Co-operation and Development . The OECD List of Social Indicators . Paris, France : Organisation for Economic Co-operation and Development ; 1982 . 30 Allison PD . Fixed Effects Regression Models. Quantitative Applications in the Social Sciences . Los Angeles, CA : Sage ; 2009 . 31 Hill TD , Needham BL . Rethinking gender and mental health: a critical analysis of three propositions . Soc Sci Med . 2013 ; 92 : 83 – 91 . Google Scholar Crossref Search ADS PubMed 32 Slopen N , Williams DR , Fitzmaurice GM , et al. . Sex, stressful life events, and adult onset depression and alcohol dependence: are men and women equally vulnerable? Soc Sci Med . 2011 ; 73 ( 4 ): 615 – 622 . Google Scholar Crossref Search ADS PubMed 33 Magaard JL , Seeralan T , Schulz H , et al. . Factors associated with help-seeking behaviour among individuals with major depression: a systematic review . PLoS One . 2017 ; 12 ( 5 ): e0176730 . Google Scholar Crossref Search ADS PubMed 34 Brand JE . The far-reaching impact of job loss and unemployment . Annu Rev Sociol . 2015 ; 41 : 359 – 375 . Google Scholar Crossref Search ADS PubMed 35 Rege M , Telle K , Votruba M . Parental job loss and children’s school performance . Rev Econ Stud . 2011 ; 78 ( 4 ): 1462 – 1489 . Google Scholar Crossref Search ADS 36 Kalil A , Ziol-Guest KM . Parental employment circumstances and children’s academic progress . Soc Sci Res . 2008 ; 37 ( 2 ): 500 – 515 . Google Scholar Crossref Search ADS 37 Artazcoz L , Benach J , Borrell C , et al. . Unemployment and mental health: understanding the interactions among gender, family roles, and social class . Am J Public Health . 2004 ; 94 ( 1 ): 82 – 88 . Google Scholar Crossref Search ADS PubMed 38 Strömberg R , Backlund LG , Löfvander M . Psychosocial stressors and depression at a Swedish primary health care centre. A gender perspective study . BMC Fam Pract . 2011 ; 12 : 120 . Google Scholar Crossref Search ADS PubMed 39 Bijlsma MJ , Tarkiainen L , Myrskylä M , et al. . Unemployment and subsequent depression: a mediation analysis using the parametric G-formula . Soc Sci Med . 2017 ; 194 : 142 – 150 . Google Scholar Crossref Search ADS PubMed 40 Brand JE , Thomas JS . Job displacement among single mothers: effects on children’s outcomes in young adulthood . AJS . 2014 ; 119 ( 4 ): 955 – 1001 . Google Scholar PubMed 41 Christoffersen MN . Growing up with unemployment: a study of parental unemployment and children’s risk of abuse and neglect based on national longitudinal 1973 birth cohorts in Denmark . Childhood . 2000 ; 7 ( 4 ): 421 – 438 . Google Scholar Crossref Search ADS 42 Lundborg P . Young people and alcohol: an econometric analysis . Addiction . 2002 ; 97 ( 12 ): 1573 – 1582 . Google Scholar Crossref Search ADS PubMed 43 Sund R , Gissler M , Hakulinen T , et al. . Use of health registers. In: Ahrens W , Pigeot I , eds. Handbook of Epidemiology . New York, NY : Springer ; 2014 : 707 – 730 . 44 Fischer EH , Dornelas EA , Goethe JW . Characteristics of people lost to attrition in psychiatric follow-up studies . J Nerv Ment Dis . 2001 ; 189 ( 1 ): 49 – 55 . Google Scholar Crossref Search ADS PubMed 45 Kaufman JS . Commentary: why are we biased against bias? Int J Epidemiol . 2008 ; 37 ( 3 ): 624 – 626 . Google Scholar Crossref Search ADS PubMed 46 Madsen M , Osler M . Commentary: strengths and limitations of the discordant twin-pair design in social epidemiology. Where do we go from here? Int J Epidemiol . 2009 ; 38 ( 5 ): 1322 – 1323 . Google Scholar Crossref Search ADS PubMed 47 Gulliver A , Griffiths KM , Christensen H . Perceived barriers and facilitators to mental health help-seeking in young people: a systematic review . BMC Psychiatry . 2010 ; 10 : 113 . Google Scholar Crossref Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

American Journal of EpidemiologyOxford University Press

Published: Sep 1, 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