Single transitions and persistence of unemployment are associated with poor health outcomes

The relevance of health for economic and community participation has been confirmed by many scholars [1, 2]. The phrase health is wealth, and wealth is health clearly summarises the reciprocal character of this relationship. With respect to economic participation, evidence suggests that employment is beneficial for health [3‐7], in particular for depression and mental health [8, 9]. Unemployment, on the other hand, has been linked to poorer self-rated health [10‐12], mental illness and depression [8, 10, 12], as well as increased morbidity and mortality [10]. Although the underlying mechanisms are unclear, three possible explanations are mentioned in the literature for the relation between becoming unemployed involuntary and poorer health. Given the consistent association between wealth and health, a decline of income (sometimes into poverty) because of becoming unemployed is one explanation. Another explanation is offered by stress induced by unemployment e.g. by loss of self-esteem and a time structure to their days. A third explanation stems from the possibility that unemployed individuals have been found to show more self-destructive behaviour, ranging from increased levels of smoking and drinking to self-destructive behaviour like (attempted) suicide [4, 13, 14].

Furthermore, although some studies suggest the relationship between job loss and health and health behaviour to be causal, this issue is still debated [3, 15‐17]. For instance, while some evidence exists for that unemployment is causally related to an increase in smoking, this association is poorly understood, which also applies for other health behaviours, such as obesity [18]. There is also support that the association between unemployment and health reflects a selection effect, which is implying that individuals with poorer health are more likely to become and stay unemployed [19‐21]. The latter is underlined by findings of studies differentiating between the reasons for entry into involuntary unemployment. Some studies report no negative health effects after job loss due to exogenous reasons (such as the closure of business), which suggests that a selection effect is likely to explain the association between unemployment and health [22, 23].

Studying the relation between employment and health is further complicated by the fact that, because of the increase in flexibility at the labour market, people can experience multiple unemployment spells during their working life, leading to long-term unemployment. Although a link between persistence in unemployment and poorer health seems plausible [24‐26], these studies are limited by the relatively small sample size and the use of self-reported data on exposures and outcomes. Furthermore, it is unclear how the persistence of unemployment over time might influence health outcomes [24]. Three different possibilities of development have been proposed in the literature. The first model relates to a dose-response relation, where increasing persistence of unemployment is followed by a further deterioration in health. The second model suggest the development of a steady-state situation- i.e. after a certain level of effect additional unemployment will not add anything more. The final model suggests immunity to repeated exposure of unemployment – i.e. after sometime a maximum is reached and that any more unemployment will result in lower effect estimates [24].

It has also been suggested that the health impact of unemployment may depend on macro-level economic circumstances, such as an economic recession [27‐30]. On the one hand, involuntary job loss during an economic recession may be less detrimental to health, because unemployment is less rare and the stigma of unemployment that could potentially harm (mental) health decreases [31]. On the other hand, the impact of unemployment on health may be increased during an economic recession as it may lead to increased insecurities about the chances for reemployment and therefore to increased psychological distress [32]. Previous studies have suggested that the post-2008 economic recession intensified the negative impact of unemployment on general and mental health [31, 33, 34] and that differences with respect to unhealthy behaviour, such as smoking and drinking, increased between the employed and unemployed [34]. The existing evidence regarding the relationship between the economic recession and health is most consistently found for suicides and mental health [27]. However, until now, few studies have investigated whether the associations between the degree in persistence of unemployment and health (behaviour) differed during and after the economic recession that started in 2008. E.g. one study reported that the decline in health is in particular steep for people who have been unemployed for several years [35], while another study found job-loss and long-term employment to be risk factors for substance use in times of economic recession [28].

This paper aims to provide more insight in the complex relationship between employment status and health, without trying to make causal statements. In this observational study, we examine the associations between single transitions in employment status and persistence of unemployment and health and health-related behaviour. Subsequently, we examine whether the associations between single transitions in employment status and persistence of unemployment and health (behaviour) are affected by the latest recession, which began in 2008.

Of 57,904, 40,401, 57,911 and 57,906 persons who had data on employment status (at t1 and t2) and outcomes (at t3), respectively, persons with missing data on covariates were excluded. The final sample sizes in the analyses on single transitions in employment status were n = 47,734 (82.4%; general health), 32,919 (81.5%; mental health), 47,739 (82.4%; obesity) and 47,734 (82.4%; smoking). Similarly, of 39,168, 26,368, 39,174 and 39,171 persons who had data on employment status (between t1 and t5) and outcomes (at t6), respectively, data from 29,866 (76.3%; general health), 19,593 (74.3%; mental health), 29,870 (76.2%; obesity) and 29,867 (76.2%; smoking) were included in the analyses on persistence of unemployment. Individuals who were excluded due to incomplete data in the analysis on single transitions in employment status and general health were older, less likely to be lower educated, but generally less healthy than those who were included in the analyses (Appendix 1). Similar differences were observed for the other outcomes (results not shown).

In 2005, 52% of our study population were men with a mean age of 42.5 years. Compared with individuals who remained employed between 2004 and 2005, those who found a job, became unemployed or remained unemployed between 2004 and 2005 were less educated, more likely to be a current smoker and to have a poor perceived and mental health at the subsequent year. Those who remained unemployed between 2004 and 2005 were more likely to be female and obese than those who stayed employed and to be the least healthy, with almost 50% reporting being in poor general health and 43% reporting poor mental health (Table 1).

In this observational study, we found that employment transitions are associated with individuals’ health and health related behaviour. Compared with individuals who were persistently employed, individuals who lost their job and especially those who were persistently unemployed for up to 5 years were more likely to report poor mental and general health. The associations with obesity and smoking followed the same pattern, although less pronounced, especially for smoking. These associations seemed to be partly, but not fully attributable to a range of confounders and personal income.

Our results on single transitions showed that the unemployed who transited to work scored significantly more favourable on general and mental health as well as on smoking and obesity than those who either became or remained unemployed. Furthermore, the likelihood of poor mental and general health and obesity became higher with increasing degree of persistence of unemployment in the previous 5 years, indicating that chains of job loss-events are important for health as well. This is in line with previous studies indicating that individuals who are more persistently unemployed are in poorer health than the stable employed [24‐26, 40]. Odds of smoking, however, became higher up until 4 years of unemployment, but no longer thereafter. Although these findings are possibly explained by the development of a steady-state situation after years of unemployment [24], our observational data do not allow for definitive conclusions.

Our study dealt with the association between health and both exit from and entry into employment simultaneously; a similar approach was used in a few studies so far [21, 40, 41]. Our findings provide additional evidence for that gaining or regaining employment is beneficial for health. Hence, reemployment might be an important strategy to improve the health of unemployed individuals. This is in line with previous research [4, 8]. Due to our study design, however, we were unable to identify causative relationships between unemployment and health. Therefore, the directions of the associations remain unclear. For instance, it is not clear if unemployment is associated with poor health or if good health is a prerequisite for getting and keeping a job. For both directions, support is available in the literature [19]. Further research is required to study the possible influence of this ‘healthy worker effect’ on the observed associations.

Our finding that the associations between single transitions and persistence in unemployment and poor health are partly accounted for by income is in keeping with findings of a recent longitudinal study showing that income is not a mediator of the relationship between unemployment and health [29]. Further research is required to study whether other intermediates, such as financial strain, as suggested by Tøge [29] may explain the associations.

Our finding that the associations between work transitions and health did not vary by times of an economic recession is novel. This is not in keeping with results of several previous studies conducted in Europe and North America, which reported an intensification of the health damaging effects of job loss during economic recessions [31, 33, 34, 38, 42, 43]. Other studies conducted in Finland and Sweden, however, concluded that in times of an economic recession the negative health impact of work transitions was less severe [44, 45] or absent [46]. Prior evidence suggests that possible detrimental effects of economic recessions can be mitigated by social welfare programs [47‐50]. Hence, it is possible that social benefits play a protective role for health during economic changes. Perhaps the Dutch social security system, which is relatively extensive compared to other countries, buffered the negative effects of the economic recession on the association between unemployment and health and related behaviour. Another explanation for the non-existent impact of the recession in our study is that the recession may change the character of employment transitions. Unfortunately, the window of observation in our study was too short to study this properly. A secondary comparison of characteristics across employment transitions over time showed that people who remained unemployed became slightly higher educated over time. Other characteristics and health outcomes, however, remained stable over time (results not shown), suggesting that changing characters of the transitions over time is unlikely to explain our findings.

Our study has a number of strengths. The longitudinal availability of work status data, as well as other variables such as income, allowed us to study single transitions in employment status and persistence in unemployment over multiple years. Our study also benefits from the use of linkage with register data to objectively ascertain the work status. A limitation, however, is that persons not included in our analyses were generally less healthy than included participants, which means we cannot exclude some selection bias. Furthermore, selection bias may be present because we were not able to distinguish unemployed persons who gave up their job search from those still trying to get a job. It is plausible that discouraged non-workers have lower career chances and worse health, which may have led us to underestimate any association between single transitions and persistence in unemployment and health. Another limitation is that health data was measured only once. We therefore do not know if persons were in worse health before their health was measured and if their work status changed because of it. Indeed, findings from a secondary analysis in our study showed that individuals with better health were more likely to be (persistently) employed and vice versa (results not shown). Hence, replication of our study using registered longitudinal health data would strengthen our findings and may also provide insight in the development of health outcomes over time. Studies taking the main reason for entry into unemployment into account would further disentangle the existence of possible selection effects [16]. Unfortunately, based on our data no further differentiation with respect to this aspect was possible. Furthermore, although the study population was rather large, the decreasing number of persons with increased persistence of unemployment resulted in estimates with larger confidence intervals for these groups. In addition, although we adjusted for a range of confounders, residual confounding due to unmeasured factors, such as ethnicity and financial strain cannot be excluded. Finally, it has earlier been suggested that the health status of employees may be negatively affected by job insecurity [10, 51‐53]. Unfortunately, we did not have data available on types of contracts (i.e. temporarily or permanent) and whether job loss was involuntary or not. It is likely that voluntary job loss has attenuated the observed findings toward the null, leading to an underestimate of the estimates. Research focusing on precariousness of the new employment and health status would provide valuable information concerning this topic.

Summarising, our observational findings suggest that single transitions and persistence in unemployment are associated with higher likelihood of poor mental and general health, obesity, and to a lesser extent smoking. These associations seemed to be partly, but not fully attributable to a range of confounders and personal income. In line with previous studies, this study implies that re-employment might be an important strategy to improve health of unemployed individuals. This holds true, in particular, for individuals who have been unemployed for a longer duration in time. The relatively extensive Dutch social security system may explain our findings that the economic recession did not modify the observed associations between transitions in employment status and poorer health. Nevertheless, findings from other prospective studies with longitudinal measures of health status are needed to confirm our findings.