Over the past two decades, deregulations of the labour market have resulted in a considerable increase in job insecurity in most European countries, even during times of economic growth.1^–7 There is an increasing interest in social epidemiology on the impact of job insecurity on employees’ health, but knowledge in this area is still limited.1 5 A meta-analysis reviewing 72 studies showed that employees with job insecurity generally had poorer health outcomes.8 However, the vast majority of studies were cross-sectional in design.

Recently, a landmark study from the Whitehall II cohort showed that job insecurity was a strong predictor of poor self-rated health and minor psychiatric disorders at 2.5 years of follow-up.9 10 Compared to employees with continued job security, employees with chronic job insecurity had the strongest decline in general and psychological health, followed by employees who lost job security during the follow-up period and employees who had job insecurity at baseline only.10 However, as the authors pointed out, the generalisability of these findings is limited, as the study population consisted exclusively of office-based white-collar civil servants in Great Britain. The associations between job insecurity and health outcomes may have been markedly different in other occupational groups. Moreover, it is widely acknowledged that factors at the societal level, such as employment protection laws, availability and amount of unemployment benefits, availability of jobs and active labour market policies can influence the impact of job insecurity on health outcomes.5

Regarding flexibility and job security, Denmark has a very special labour market system, called “flexicurity” (flexibility plus security), which is unique in Europe.6 11 12 It consists of, on the one hand, weak employment protection laws, which make it relatively easy for employers to hire and dismiss employees, and on the other hand, relatively high unemployment benefits, which can be paid for up to four years. Moreover, the Danish authorities have instituted an active labour market policy, which includes offering job training opportunities and information about job openings, but also pressure on the unemployed to seek and to accept new job offers. In addition to this flexicurity system, Denmark offers relatively generous early retirement schemes, which probably reduce fear of unemployment among older employees. The flexicurity system has been praised by economists, politicians and the public media, and has been suggested as a role model for other countries in the European Union.6 11^–13

The aim of this paper is to analyse if an association between job insecurity and a decline in health also exists under the specific Danish societal conditions. We hypothesised that job insecurity also has adverse health effects in Denmark, but that these effects would be considerably stronger among employees who were believed to have poor chances of finding another job. We further hypothesised that associations would be stronger in young and middle-aged employees, as people at the beginning or in the middle of their careers would experience job insecurity as more threatening for their future lives and as they would not have the chance to escape into early retirement in the foreseeable future.

METHODS

Study design and population

The Danish Work Environment Cohort Study (DWECS) is a longitudinal study to assess sociodemographic factors, work environment characteristics, health behaviours and health status in the Danish working population.14 We analysed the impact of job insecurity and labour market chances in 1995 on a decline in self-rated health in 2000. In 1995, a representative sample of 10 702 Danish residents were approached and 8583 participated in the survey (response rate: 80%). Among the respondents, 5344 were employed at the time of the survey and were 60 years of age or younger and therefore included in the study. Of these, 11 emigrated or died during the follow-up period and 924 did not respond to the follow-up questionnaire, yielding a follow-up sample of 4409 (follow-up response rate: 83%). For the purpose of this paper, we further excluded 204 participants with a missing value on any variable included in the analyses. Finally, we excluded 478 participants with “fair,” “poor” or “very poor” self-rated health at baseline, resulting in a study sample of 3727 participants. Mean age was 38 years (SD 11 years, range: 18–60 years) and 49% of the study sample were women.

Measurement of self-rated health

Self-rated health was measured with a question from a Danish translation of the Short Form 36 (SF-36) questionnaire.15 16 Participants were asked “In general, how would you rate your health?” with the response categories “very good,” “good,” “fair,” “poor” or “very poor.” Self-rated health has been found to be a reliable indicator of health status and a strong predictor for mortality in many countries, including Denmark.17^–21 We dichotomised responses into two categories: (1) “good health,” which included the responses “very good” and “good,” and (2) “reduced health,” which included the responses “fair,” “poor” and “very poor.” Study end point was “decline in self-rated health,” defined as moving from the “good health” category at baseline to the “reduced health” category at follow-up.

Measurement of job insecurity, labour market chances and covariates

Job insecurity was measured with the question: “Are you worried about becoming unemployed?” Participants who answered “yes” were classified as having job insecurity. Chances on the labour market were assessed with the question “Are you worried that it would be difficult for you to find another job if you became unemployed?” Participants who answered “yes” were classified as having poor chances on the labour market. We combined the two responses into the new variable “job insecurity combined with labour market chances” with the categories: (1) “no job insecurity,” (2) “job insecurity with good chances on the labour market,” and (3) “job insecurity with poor chances on the labour market.”

As covariates, we assessed gender, age, smoking, body mass index (BMI), cohabitation, having small children at home (age 6 or below), socioeconomic position (SEP), type of employment (private company or public employer) and self-rated health at baseline (“good” versus “very good”). We also recorded if a participant was unemployed at follow-up. More detailed descriptions of the covariates have been published elsewhere.14 22

Statistical analysis

All analyses were conducted with the statistical program package Stata 8.0. Differences between women and men in job insecurity at baseline and unemployment at follow-up were analysed with χ² tests. To investigate if job insecurity at baseline was predictive for unemployment at follow-up, we conducted a logistic regression analysis, adjusted for age, gender and SEP.

The impact of job insecurity on a decline in self-rated health was calculated by gender-stratified odds ratios (OR) and 95% confidence intervals (CI) with multivariate logistic regression models. We separately analysed the effects of “job insecurity” in general and of “job insecurity combined with labour market chances.” Covariates were included in four different models: model 1 was adjusted for age; model 2 was further adjusted for indicators of health behaviours (smoking, BMI); model 3 was further adjusted for cohabitation, small children at home, SEP and type of employment; and model 4 (full model) was further adjusted for self-rated health at baseline (“very good” vs “good health”). Correlation coefficients between covariates were in general low, indicating that there was no substantial colinearity in the analyses (data not shown).

We further analysed the data in two subsamples. In the first subsample (n = 3628), we excluded 99 participants who were unemployed at the follow-up survey in order to distinguish the effects of job insecurity from the effects of actual unemployment. In the second subsample (n = 3175) we analysed the data for employees 50 years or younger, in order to determine if the association between job insecurity and a decline in health was stronger among young and middle-aged employees.

RESULTS

Job insecurity at baseline and unemployment at follow-up

At baseline 3095 participants (83.0%) reported no job insecurity (2492 participants (66.9%) with good and 603 (16.2%) with poor labour market chances), 334 (9.0%) reported job insecurity combined with good labour market chances and 298 (8.0%) reported job insecurity combined with poor labour market chances. Job insecurity was higher in women than in men (18.1% vs 15.9%, p = 0.06). Women also reported job insecurity combined with poor labour market chances more often (10.0% vs 6.2%, p<0.001).

Participants with no job insecurity, job insecurity with good labour market chances and job insecurity with poor labour market chances had unemployment rates at follow-up of 2.3%, 2.7% and 6.0%, respectively. After adjustment for age, gender and SEP, participants with job insecurity with good labour market chances had an OR = 1.20 (95% CI: 0.59 to 2.44) and participants with job insecurity with poor labour market chances had an OR = 2.19 (95% CI: 1.27 to 3.76) for being unemployed at follow-up, compared to participants with no job insecurity.

Job insecurity and decline in health

At the five-year follow-up 430 participants (11.5%) showed a decline in self-rated health—that is, they moved from the “good health category” (including the responses “very good” and “good” health) to the “reduced health” category (including the responses “fair,” “poor” and “very poor” health) in the dichotomised self-rated health variable. Rates were similar for women (11.1%) and men (12.0%).

Table 1 shows the prospective associations of job insecurity at baseline with a decline in health at follow-up among women. Women with job insecurity had an age-adjusted OR of 1.77 (p = 0.001). This effect size remained virtually the same, after adjustment for smoking, BMI, cohabitation, small children living at home, SEP, type of employment and self-rated health at baseline (OR = 1.78, p = 0.002, table 1). Excluding 68 women, who had been unemployed at follow-up, did not change the effect estimate much (OR = 1.84, 95% CI: 1.27 to 2.66, data not shown in table 1).

Men with job insecurity at baseline had an age-adjusted OR of 1.37 (p = 0.08) for a decline in health (table 2). Further adjustment for covariates resulted into an attenuation of the effect estimate and in the full model, job insecurity was unrelated to a decline in health (OR = 1.06, p = 0.75). Excluding 31 men, who had been unemployed at follow-up, did not change the effect estimate much (OR = 1.09, 95% CI: 0.74 to 1.59, data not shown in table 2).

Among the covariates, higher age, smoking, obesity and “good” instead of “very good” health at baseline predicted a decline in health among both women and men in the full model. In addition, among men, being overweight and being a blue-collar worker were risk factors for a decline in health.

Job insecurity combined with labour market chances and decline in health

Table 3 shows how job insecurity combined with labour market chances was prospectively associated with a decline in health. Compared to participants with no job insecurity, both women and men with job insecurity with poor labour market chances had an increased age-adjusted risk of a decline in health (OR = 2.04, p = 0.001 and OR = 1.70, p = 0.03, respectively). Further adjustment for the other covariates did not alter effect estimates substantially among women (OR = 2.04, p = 0.001), whereas among men effect estimates were attenuated to a statistically non-significant level (OR = 1.37, p = 0.24).

Effect of job insecurity and labour market chances among employees aged 50 years or younger

Job insecurity and labour market chances were stronger predictors for a decline in health in young and middle-aged participants compared to the whole study population (table 4). Among women, job insecurity in general (OR = 1.88, p = 0.002) and job insecurity with poor labour market chances (OR = 2.13, p = 0.002) were associated with an increased risk of a decline in health in the full model. Among men, job insecurity with poor labour market chances showed a suggestive association with a decline in health (OR = 1.64, p = 0.08).

We further stratified the sample into young (18–35 years) and middle-aged (36–50 years) participants. Among women, we found similar odds ratios for job insecurity with poor labour market chances (OR = 2.13, 95% CI: 0.96 to 4.72 and OR = 2.21, 95% CI: 1.19 to 4.12 in young and middle-aged women, respectively). Among men, odds ratios for job insecurity with poor labour market chances were lower in young (OR = 1.25, 95% CI: 0.44 to 3.51) than in middle-aged employees (OR = 1.81, 95% CI: 0.91 to 3.60, data not shown in table 4).

DISCUSSION

Women who perceived their jobs as insecure had an odds ratio of 1.78 for experiencing a decline in self-rated health over the next 5 years in this study of a representative sample of the Danish workforce. As hypothesised, the health-hazardous effect of job insecurity was stronger among women who believed they had poor chances on the labour market. Also, as hypothesised, effect estimates were stronger among women at the beginning or in the middle of their careers. These results confirm findings from other countries and study populations that job insecurity is a risk factor for poor health among women.1 8 10

Among men, the associations were less clear. After adjustment for all covariates, job insecurity in general was not associated with a decline in health. Job insecurity with poor labour market chances predicted a decline in health in the age and health-behaviour adjusted model, but effect sizes were substantially attenuated in further adjustments. However, men aged 50 years or younger with job insecurity with poor labour market chances showed an odds ratio of 1.64, which approached statistical significance.

Strength and weaknesses of the study

To our knowledge this is the first prospective analyses on the effect of job insecurity combined with labour market chances on self-rated health within a representative sample of a national workforce. Owing to the prospective nature of the study, temporal order between exposure and outcome variable was established, which is an important criterion for determination of causality. Moreover, the use of the representative sample allows generalising the findings to the Danish workforce. By adding information on labour market chances we were able to include an important aspect of job insecurity, which had not been used in most previous studies.

The relatively long follow-up period of five years provided enough time to observe a sufficient number of participants experiencing a decline in self-rated health. However, the length of the follow-up is also a weakness, as job insecurity and labour market chances might have changed for a certain number of participants over time. We do not know how many participants had moved from insecure to secure jobs (and vice versa) nor when this change had happened during follow-up. Consequently, we do not know the extent of non-differential misclassification in the measurement of the exposure variable, a bias which would result into a potential underestimation of the effect size. As we measured job insecurity in both the baseline and the follow-up survey, we could have analysed if participants with continued job insecurity had a higher risk for a decline in health. However, we decided not to do this, because by using information on job insecurity and self-rated health, both measured at follow-up, we would have included a cross-sectional element.23 Instead, we are planning to merge the DWECS dataset with a national hospitalisation registry, in order to analyse if continued job insecurity increases the risk for developing specific diseases.

When both exposure and outcome variables are measured by self-report, there is the possibility that associations are caused by a third variable—for example, personality aspects, such as negative affectivity.23^–25 This is of particular a concern in cross-sectional studies, but can also apply to longitudinal analyses that are not sufficiently adjusted for baseline values of the outcome variable.23 In the present study, however, we excluded participants with less than “good” self-rated health at baseline, and we further adjusted for a variable indicating if self-rated health at baseline was “good” or “very good.” We are therefore confident that negative affectivity has not influenced our study results. Moreover, biological and behavioural risk factors (age, smoking, obesity) strongly predicted a decline in health in both women and men, further indicating that self-rated health in this study reflected first and foremost objective health status and not personality aspects.

Potential explanations of how job insecurity contributes to a decline in health

It seems reasonable to assume that job insecurity triggers negative emotions, such as worrying, anxiety, low mood or hostility, especially in employees who have poor labour market chances and who do not have the possibility to escape into early retirement. People with high levels of negative emotions tend to show poorer health behaviours, including smoking, heavy alcohol consumption, high calorie intake and low leisure time physical activity,26^–29 all of which are important risk factors for numerous diseases.30

Chronic presence of negative emotions might also increase the risk of psychological disorders, such as major depression. Research indicates that anxiety is not just a precursor but a causal factor in the aetiology of depression.31 32 Hence, it is possible that anxiety, induced by job insecurity, increased the risk of major depression in our study population. Other studies have indeed shown that job insecurity is a risk factor for depression.8–10 22 This includes one study that also used the DWECS dataset and that showed that job insecurity was associated with an increased risk in the incidence of severe depressive symptoms in men.22

Negative emotions and psychological disorders, caused by job insecurity, might also affect physical health via psycho-neuroendocrinological and psycho-immunological pathways. There is growing evidence that negative emotions, particularly depressive mood, increase the risk of coronary heart disease, independent of health behaviours.33^–35 Moreover, experimental research has shown that feelings of distress can lower immune competence, resulting into an increased susceptibility for infectious diseases.36 37

Finally, job insecurity might affect health in combination with other psychosocial working conditions. Studies have found, for example, that an imbalance between high efforts and low rewards (a construct that includes job security) at work increases psycho-physiological distress reactions and subsequent risk of disease.38^–40

Public health implication

The follow-up period of this study (1995–2000) was a time of economic prosperity with low unemployment rates in Denmark, which is still continuing.41 Only a small proportion of employees reported job insecurity in this study, a finding that is confirmed by a survey conducted in 1998, which showed that Danish employees had the lowest level of job insecurity among 16 European countries.7 Because of the relatively low percentage of job insecurity, we believe that the public health impact of job insecurity in Denmark during the time of this study was only modest. However, in a recession period with increasing job insecurity, the relative risks reported in this study will constitute a potential threat to public health in Denmark, especially among young and middle-aged employees with poor labour market chances.