Objective To determine the impact of ill health on exit from paid employment in Europe among older workers.
Methods Participants of the Survey on Health and Ageing in Europe (SHARE) in 11 European countries in 2004 and 2006 were selected when 50–63 years old and in paid employment at baseline (n=4611). Data were collected on self-rated health, chronic diseases, mobility limitations, obesity, smoking, alcohol use, physical activity and work characteristics. Participants were classified into employed, retired, unemployed and disabled at the end of the 2-year follow-up. Multinomial logistic regression was used to estimate the effect of different measures of ill health on exit from paid employment.
Results During the 2-year follow-up, 17% of employed workers left paid employment, mainly because of early retirement. Controlling for individual and work related characteristics, poor self-perceived health was strongly associated with exit from paid employment due to retirement, unemployment or disability (ORs from 1.32 to 4.24). Adjustment for working conditions and lifestyle reduced the significant associations between ill health and exit from paid employment by 0–18.7%. Low education, obesity, low job control and effort–reward imbalance were associated with measures of ill health, but also risk factors for exit from paid employment after adjustment for ill health.
Conclusion Poor self-perceived health was strongly associated with exit from paid employment among European workers aged 50–63 years. This study suggests that the influence of ill health on exit from paid employment could be lessened by measures targeting obesity, problematic alcohol use, job control and effort–reward balance.
- Self-perceived health
- mental health
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
What this paper adds
The current study showed that self-perceived poor health was more strongly associated with exit from paid employment during follow-up compared with other health measures such as ‘chronic diseases’, ‘mobility problems’ and ‘instrumental limitations in daily activities’.
Lifestyle factors as well as working conditions explained 0% to 19% of health related exits from paid employment.
The population attributable fraction of less than good health for exit from paid employment was largest for disability (61%), followed by unemployment (27%) and retirement (9%).
In many industrialised countries the population is ageing due to increasing life expectancy and falling birth rates.1 A rather paradoxical development is that, despite increases in life expectancy, the average time people spend in paid employment has decreased in most European countries. Although part of this decrease is explained by prolonged education among younger cohorts, a more important contributor is the higher rate of exit from the labour market at older ages.2 As a consequence, many countries are developing policies to encourage older workers to remain longer in the labour market and delay retirement.3 Clearly, the success of these policies will depend on a better understanding of ageing in the workforce and the particular role of health and work characteristics in continuing work or exiting from the labour market. Recent evidence suggests that work can be good for health, reversing the harmful effects of long-term unemployment and prolonged sickness absence.4 However, the current assumption seems to be that illness is incompatible with being in work.4
It is obvious that ill health plays a role in exit from the labour market through work disability. Although less consistent, there is also evidence that ill health may cause exit through unemployment and early retirement. 3 5–10 In several European countries it was shown that ill health was a risk factor for transitions between paid employment and various forms of non-employment, including retirement, unemployment and taking care of the household.5 9 The strength of the current study is the exploration of the three pathways simultaneously. Second, four different measures of health were used to study the role of ill health in exit from paid employment.
In order to minimise the negative effect of ill health on work participation, it is of interest to study factors that explain the relationship between ill health and exit from paid employment and may therefore be important targets for primary preventive interventions in occupational populations. Research on occupational health has shown the negative influence of poor working conditions on workers' health.11–13 Lifestyle factors, such as lack of physical activity, smoking, alcohol use and overweight, are well-established determinants of poor health, and thus, may be important targets of intervention in order to decrease health related exit from paid employment.
The following research questions were formulated. First, which measures of health are predictive for exit from paid employment? Second, how much of the observed associations between ill health and future exit can be explained by work related factors and lifestyle?
The study population consisted of participants of the Survey on Health and Ageing in Europe (SHARE study). SHARE is a longitudinal survey that aims to collect medical, social and economic data on the population aged over 50 years in 11 European Union countries (Sweden, Denmark, the Netherlands, Belgium, Germany, Austria, Switzerland, France, Italy, Spain and Greece).14 15 In the participating SHARE countries, the institutional conditions with respect to sampling were so different that a uniform sampling design for the entire project was not feasible. Different registries at national or local level were used that permitted stratification by age. The sampling designs varied from simple random selection of households to complicated multistage designs.
The first wave of data was collected by interviews between April and October 2004. The overall household response across the 11 SHARE countries in which data collection took place in 2004 was 57.4%, although substantial differences among countries were observed.15 The available dataset from the first wave of data collection (SHARE Release 2.0) contains information on 28 517 participants, with 12 965 subjects (45%) aged between 50 and 63 years. Individuals aged 63 years and older were excluded from the current study, since it was assumed that workers normally retired when they reached 65 years of age at the end of follow-up. While this assumption certainly has limitations, given the complexity of defining retirement at the individual level and the small proportion of workers above the age of 63 years in the study population (about 2%), it was considered to be the definition that was most comparable across countries. Employment status was unknown for 93 persons, resulting in a study population of 12 872 subjects, of whom 7119 (55%) were in paid employment. After 2 years, 8729 subjects participated again in the questionnaire survey (SHARE Release 1.0), resulting in a response rate of 67%. Complete information on employment status in 2006 was available for 8568 subjects. For the longitudinal analysis of the influence of ill health on exit from the labour market, a cohort of 4611 subjects was available with paid employment in 2004 and complete information on individual and work related characteristics at baseline and work status at follow-up in 2006.
Labour force participation
The outcome of this study is work status, which was based on self-reported current economic status that best described each respondent's situation based on four mutually exclusive categories: paid work, retired, unemployed or disabled. According to the SHARE definition, employed individuals were those who had declared they had carried out any kind of formal paid work in the last 4 weeks, including self-employed work in a family business. The unemployed were those who had been laid off from their last job but could not claim normal pension benefits, and had therefore spent some time in unemployment before effectively being retired. Sickness or disability insurance applied to people who exited the labour force because of recognised health problems.14 The category of disabled participants predominantly includes those whose health problems at work were an eligibility criterion for receiving a disability pension. Total exit from the workforce was defined as exit either through early retirement, unemployment or disability.
The European version of Self-Perceived Health, a 5-point scale questionnaire with answers ranging from very good to (very) bad, was used to define poor health (less than good). This frequently used questionnaire has been shown to be a good indicator of general physical and mental health.16 17 A second health measure was having at least one of the following chronic diseases diagnosed by a doctor during one's lifetime: heart disease, stroke, diabetes, lung disease, asthma, arthritis or rheumatism, and osteoporosis. Functional limitations, reflecting the ability of individuals to perform normally in society, were characterised with two dichotomous measures of health. The first measure of interest, mobility problems, reflects limitations with mobility, arm or fine motor functions. Mobility problems were defined as one or more affirmative answers on a list of 10 mobility problems, such as walking 100 m and reaching or extending arms above shoulder level. The second measure, instrumental limitations, was positive for subjects with limitations in one or more of the 13 instrumental activities of daily life, such as preparing meals and making phone calls.
The highest education successfully completed was coded according to the 1997 International Standard Classification of Education (ISCED-97) and categorised into low (pre-primary, primary and lower secondary education), intermediate (upper secondary education) and high (post secondary education). Body mass index (BMI) was calculated by dividing body weight in kilograms by the square of body height in metres. BMI was recoded into normal (<25 kg/m2), overweight (≥25 and <30 kg/m2) or obese (≥30 kg/m2). Marital status was used to categorise individuals into those who were living with a spouse or a partner in the same household (reference category) and those living alone. Smokers were subjects who were currently smoking; all others were categorised as non-smokers. Problematic alcohol use was defined as alcohol consumption of two or more glasses of alcoholic beverage on at least 5 days a week in the last 6 months. Physical activity was measured with single questions on regular participation in moderate and vigorous activities, both on a 4-point scale ranging from ‘more than once a week’ to ‘hardly ever, or never’. Those who reported moderate or vigorous activity less than once a week were considered to lack leisure time physical activity.15
Work related characteristics
Work related characteristics were assessed by a short battery of items derived from (i) the Job Content Questionnaire assessing the demand–control balance18 and (ii) the effort–reward imbalance questionnaire.12 All items were on a 4-point scale ranging from 1 ‘strongly agree’ to 4 ‘strongly disagree’. A single item measured high time pressure (‘I'm under constant time pressure due to a heavy work load’). Lack of job control was measured by the sum score of two items (‘I have very little freedom to decide how I do my work’; ‘I have an opportunity to develop new skills’). Country-specific median values were used to define the presence of high time pressure and lack of job control.
Effort–reward imbalance was measured by two items on ‘effort’ (‘physically demanding’ and ‘time pressure’) and five items on ‘reward’ (‘receive adequate support’, ‘receive recognition’, ‘adequate salary’, ‘job promotion prospects’, ‘job security’). Effort–reward imbalance was defined by the ratio of the sum score of the ‘effort’ items and the sum score of the ‘reward’ items, adjusted for the number of items.19 Effort–reward imbalance was defined as a score within the upper tertile of this ratio per country.19
A high physical work demand was measured with one item (‘My job is physically demanding’). A country-specific median value was used to define the presence of high physical work demand.
Logistic regression was used to evaluate cross-sectional associations at baseline between four measures of ill health as dependent variables and individual and work characteristics as independent variables, adjusting for country.
Risk factors for exit from paid employment during the 2-year follow-up period were evaluated by means of multinomial logistic regression analysis. The study population consisted of subjects with paid employment at baseline and ORs were calculated for the likelihood of transition to early retirement, unemployment or disability during the 2-year follow-up period. The results for homemakers were not shown as this group consisted mainly of women, but subjects who exited paid employment through becoming a home worker remained in the sample. The first step in the analysis was to establish univariate associations between the dependent variable work status, and health measures, sociodemographic factors, lifestyle factors and work characteristics as independent factors, including country as fixed effect. In the second step, multivariate analyses were conducted to model employment status at the end of follow-up as a function of four measures of health. For the initial selection of potential covariates for the multivariate model, univariate associations with a significance level of p<0.05 were considered. For each independent variable measure of health, we calculated ORs for the dependent variable exit from work adjusted for age, gender and education (reference model) and further adjusted for lifestyle factors and work related characteristics separately, and in combination. For each regression model the percentage change in the OR of each pathway of exit was calculated (100×(ORreference model−OR+explanatory factors)/(ORreference model−1).20 One of the main advantages of this method is that it can be used to estimate the direct and indirect contributions of explanatory factors. One limitation is that the percentage change can be similar for different absolute changes in ORs. However, all contributions were calculated relative to the same ORs, which were also presented. Therefore, we believe that this limitation has a minor effect on our results.
Population attributable fractions (PAF) were calculated for significant determinants of exit from paid employment, using the formula PAF=Pe(OR−1)/(1+PE(OR−1)),21 where Pe represents the prevalence of exposure in the study population.
All statistical models were based on the number of people with complete data available. The statistical analyses were carried out with SPSS v 15.0.22
Approximately 17% of the employed workers reported less than good health (table 1). Interrelations of the four health measures were moderate, with Spearman correlations varying from 0.06 to 0.33. In total, 55% of the subjects with poor health had a chronic disease, 57% mobility problems and 9% instrumental limitations. The chronic diseases with the highest prevalences were depression (17.7%, n=814), arthritis/osteoporosis (12.3%, n=565) and respiratory diseases (5.7%, n=265) (data not shown). About 61% of subjects with a chronic disease perceived their health as being good.
Table 2 shows that important determinants at baseline for all four health measures were lack of physical activity in leisure time (ORs from 1.24 to 1.87) and effort–reward imbalance at work (ORs from 1.25 to 1.64). A high BMI was also associated with most measures of health.
During the 2-year follow-up period, 17% (n=794) of employed workers exited the workforce, primarily because of retirement (11%) (table 3). Considerable differences in the prevalences of exit from paid employment and pathways of exit were found among countries. Table 4 shows that self-perceived poor health was the measure of health most predictive of transition to unemployment (OR 2.49), retirement (OR 1.50) and work disability (OR 5.04). All four health measures were associated with any exit from work (ORs from 1.56 to 2.08). The role of ill health on exit from paid employment was comparable for workers with a full-time or part-time contract (data not shown). All lifestyle factors except smoking were associated with exit from paid employment through retirement (ORs from 1.23 to 1.40). Among work related factors, lack of job control showed the highest increased risks for all three exit pathways (ORs from 1.23 to 2.68). Table 5 shows the observed associations between different measures of ill health and transitions to non-participation, after adjustment for lifestyle factors and work characteristics. Significant ORs between ill health and exit from paid employment decreased by 0% to 10% after adjustment for lifestyle factors, 4% to 9% after adjustment for working conditions, and 4% to 19% after adjustment for lifestyle factors and working conditions simultaneously. Adjustment for lifestyle factors and work related characteristics had a smaller influence on the association between ill health and work disability compared to the other pathways of exit from paid employment. In the fully adjusted models for each of the four health measures, the lifestyle factors obesity and problematic alcohol use remained significant in at least one of the models. Regarding work related characteristics, lack of job control and effort–reward imbalance at work remained significant after full adjustment in at least one of the four models.
The population attributable fractions of less than good self-perceived health for transition into unemployment, retirement and disability were 27%, 9% and 61%, respectively.
During a 2-year follow-up, 17% of workers employed at baseline left paid employment, primarily because of early retirement. Controlling for individual and work related characteristics, poor self-perceived health was strongly associated with exit from paid employment due to retirement, unemployment or disability (ORs from 1.32 to 4.24). In order of decreasing importance, chronic diseases, mobility problems and instrumental limitations also influenced exit from paid employment, most notably through disability. Significant associations between ill health and exit from paid employment changed from 0% to 19% after adjustment for lifestyle and work characteristics.
Some limitations must be taken into account in this study. First, the attrition rate between baseline and follow-up was high (68%).23 However, in our analyses among subjects initially employed at baseline, no differences were found between responders and non-responders during follow-up for all health measures at baseline.
Second, there are large variations between European countries in the association between ill health and various forms of exit from paid employment.5 These variations may reflect differences between countries in institutional arrangements (eg, availability of disability benefit schemes for those with health problems), or other factors (eg, more or less selectivity of unemployment dependent on overall levels of unemployment). All analyses were therefore adjusted for country. Due to small numbers, country-specific or region-specific analyses were not feasible. The analyses stratified for Scandinavian (Sweden, Denmark), Bismarckian (Austria, Belgium, France, Germany, the Netherlands, Switzerland) and Southern European regions (Greece, Italy, Spain) showed that the conclusions drawn from the total population were also valid within the regions. That is, in each region the health measure self-perceived health was the measure most predictive for exit from paid employment, most notably through disability.
Third, all variables were based on self-reported data, which could have caused reporting bias. The problem with using self-reported health in an empirical analysis of labour force participation is that it may be an endogenous explanatory variable.16 24 25 According to the justification hypothesis, individuals justify their non-participation by claiming that they are have ill health. Subjects with intentions at baseline to leave paid employment in the near future may also have been more prone to report high work demands or a less beneficial effort–reward balance in order to justify their future exit from paid employment.17
Fourth, the current study used a follow-up period of 2 years and, therefore, had limited discriminatory power and does not afford insight into the long-term effects of poor health on exit from paid employment or the relevant time windows for these effects. A European study showed that poor health had the strongest effects on leaving the workforce in the year before the transition.8 Thus, it is expected that the reported influence of ill health on exit from paid employment is a fair reflection of the effects of ill health on work participation. The influence of ill health on exit from paid employment decreased for older workers, as the decision to continue work after 60 years of age is more influenced by other factors, such as eligibility criteria for early retirement and the labour market.
Several studies have analysed the effects of health on exit from paid employment in older workers. 3 9 5–7 26–29 The results of this study support the selection hypothesis, whereby people with poor health are more likely to leave paid employment.30 The influence of type of health measure differs by route of exit, but an overall effect on total exit was consistently present for all measures of ill health.
The relationship between poor health and exit from paid employment may be explained by a mismatch between an individual's capacities and the requirements of the job.2 Functional limitations might therefore be more important than self-perceived poor health for future loss of paid employment. However, the analyses showed that poor self-perceived health was a stronger predictor for pathways of exit than functional limitations, expressed by either mobility problems or instrumental limitations in daily activities. An explanation could be that self-perceived health also includes mental health, whereas functional limitations primarily concern physical health. The high prevalence of depression in the cohort may have contributed to the association between self-perceived health and future exit from paid employment.
The analyses also showed that having ever being diagnosed with a chronic disease played a less important role in exit from paid employment. This may be explained by the fact that people diagnosed with chronic conditions who remained in paid employment are a selection of the fittest survivors,31 while those who had already left paid employment due to these diseases before baseline were not included in our sample. Analyses on the role of onset of disease during the follow-up period was not feasible as only 12 subjects reported that the onset of their chronic disease had been diagnosed during the follow-up period.
The direct influence of ill health on exit from paid employment had ORs varying between 1.37 and 5.04. The corresponding population attributable fractions of less than good self-perceived health for transition into unemployment, retirement and disability were 27%, 9% and 61%, respectively. Under the assumption that the observed associations represent a causal process, these associations and population attributable fractions indicate that good health is an important factor in maintaining paid employment. Based on this finding, interventions aimed at discouraging exit from paid employment should aim to prevent or minimise ill health. Given the strong associations at baseline between obesity and lack of leisure time physical activity with several measures of ill health, health promotion interventions should be considered that increase physical activity and support a healthy diet.32 33 The consistent associations at baseline between lack of job control, high physical work demands and effort–reward imbalance with several measures of ill health, outline the importance of improving both working conditions and work organisation.
We observed that adjustment for lifestyle factors and work related characteristics resulted in reasonable changes in health related exit from paid employment. The change was only important for statistically significant associations because a small difference in OR could otherwise result in a high proportion of change. The influence of lifestyle factors and work characteristics on the impact of ill health on labour force exit highlights the importance of helping workers with health problems continue working, for example by empowering workers with chronic diseases.34
In the fully adjusted model, obesity, problematic alcohol use, low job control and effort–reward imbalance remained statistically significant for at least one of the pathways of exit in at least one of the models.
Different studies support the association between unhealthy lifestyles, such as lack of physical activity, obesity and problematic alcohol use, and exit from paid employment. 5 29 35–37 In the fully adjusted multinominal models, problematic alcohol use was consistently associated with entering work disability with ORs varying from 1.84 to 1.88. However, it has been suggested that this may be explained by problems with working times, work output, concentration, occupational safety and cooperation, irrespective of health status.8 In the multivariate model, obesity was associated with becoming unemployed (OR 1.67). This is in agreement with a French study that reported obesity as a risk factor for unemployment after controlling for self-reported health.36
Smoking was not associated with early exit. Previous studies have shown contradictory results, with significant associations for smoking 8 38–41 as well as non-significant associations with different forms of exit from paid employment.36 42 43
In the fully adjusted models, lack of job control remained a significant predictor of exit through retirement and disability, whereas effort–reward imbalance predicted unemployment. Several studies have corroborated the observed direct influence of strenuous working conditions on exit from paid employment.7 29 38 44 45In a cross-sectional analysis of the SHARE survey at baseline, a high imbalance between efforts and rewards was also associated with intended early retirement after controlling for poor self-perceived health.19 Hence, preventive measures against problematic alcohol use, obesity, job control and effort–reward imbalance, will help reduce health related early exit from paid employment.
This study only focused on exit from paid employment, but poor health could have an additional impact in terms of changing jobs and stalled careers. The Health and Retirement Survey27 showed that workers often changed jobs within several years after the onset of health problems. This might also be true for the onset of poor health in earlier phases of their careers (younger workers). Poor health may also have adverse effects on performance at work, as observed regarding the influence of poor health on sickness absence46 and loss of productivity at work.47 48 The duration of employment contracts could also influence the maintenance of paid employment. However, only 7% of the subjects with paid employment had a temporary employment contract, and thus this parameter could not be evaluated in this study.
The health status of older European workers has a major influence on the likelihood of maintaining paid employment. Self-perceived poor health and, to a lesser extent, having a chronic disease, perceived mobility problems and limitations seem predictive for future work participation. There is consistent evidence that social inequalities in health depend on work related factors as well as lifestyle behaviours.49 50 The results of this study suggest that labour market participation by older workers with ill health may be sustained by interventions that promote a healthier lifestyle and healthier working conditions. As exit from paid employment is often irreversible at an older age, prevention of work loss by improving the worker's health or improving an ill worker's work circumstances and lifestyle should be a key priority. Important entry points for policy could be lifestyle interventions, improvements in job control and effort–reward balance, and social policies to encourage employment among older people with health problems.
Funding This paper uses data from release 2 of SHARE 2004 and release 1 of SHARE 2006. The SHARE data collection has been primarily funded by the European Commission through the 5th framework programme (project QLK6-CT-2001-00360 in the thematic programme Quality of Life). Additional funding came from the US National Institute on Ageing (U01 AG09740-13S2, P01 AG005842, P01 AG08291, P30 AG12815, Y1-AG-4553-01 and OGHA 04-064). Data collection in Austria (through the Austrian Science Foundation, FWF), Belgium (through the Belgian Science Policy Office) and Switzerland (through BBW/OFES/UFES) was nationally funded. The SHARE data collection in Israel was funded by the US National Institute on Ageing (R21 AG025169), by the German-Israeli Foundation for Scientific Research and Development (G.I.F.) and by the National Insurance Institute of Israel. Further support by the European Commission through the 6th framework program (projects SHARE-I3, RII-CT-2006-062193, and COMPARE, CIT5-CT-2005-028857) is gratefully acknowledged. Mauricio Avendano is supported by a grant from the Netherlands Organisation for Scientific Research (NWO, grant no. 451-07-001) and a Bell Fellowship from the Harvard Center for Population and Development Studies.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.