The health consequences of unemployment: the evidence

Introduction

There has been subsequent vigorous debate about the direction of causality: Does unemployment cause a deterioration in health? Are the sick more likely to become unemployed (health selection effects)? Or are both unemployment and poor health associated with another, underlying, causal factor such as socioeconomic disadvantage?

Here, we review the Australian and international evidence for an association between unemployment and adverse health outcomes and the potential causal mechanisms for this association.

We searched the literature for the past decade using MEDLINE and chose key studies evaluating current knowledge about the relationship between unemployment and ill-health for a range of health outcomes and determinants: mortality rates, mental health, lifestyle factors, and health service use. We have included some pivotal studies drawn from a longer time span, and refer to existing recent relevant reviews. We placed particular emphasis on recent Australian studies.

In assessing the evidence, it is important to keep in mind that unemployment is a complex and diverse experience rather than an easily categorised exposure variable like tobacco smoking, and that its effects will be mediated by a large number of social and individual factors.

Health outcomes and determinants

Mortality rates
Analyses of aggregated population data from the 1930s onwards have demonstrated correlations between unemployment levels and mortality rates, such as maternal mortality, infant mortality and deaths from rheumatic heart disease.³

In the 1970s, Brenner found correlations for several countries at the population level between annual unemployment levels and mortality rates over periods of up to 40 years.^10-12 However, these studies have been seriously criticised and cannot be used to infer causation.^3,13 Other studies looked at correlations between variations in unemployment levels and death rates for geographical regions and obtained generally consistent findings.^14,15 Again, such studies cannot be used to infer causation, as unemployment may be correlated at the population level with other factors causing increased mortality.

Much more convincing evidence for causality was obtained from longitudinal studies (not specifically set up to examine unemployment and health) in several countries in the 1980s. A 1% sample of census records was used to analyse mortality in unemployed men aged 15 to 64 in England and Wales in 1971 and 1981.^1,2 For both samples, employment meant lower mortality rates than the average (the "healthy worker" effect). Those unemployed who had a pre-existing illness or disability had mortality rates over three times higher than the average. Those who were unemployed but not ill at census time showed a 37% excess mortality over the following 10 years (95% confidence interval [CI], 22%-52%).

In all social classes, the mortality rate of unemployed people was higher than that of the employed, particularly for deaths from cardiovascular disease, lung cancer, accidents and suicide. If health selection were occurring among the "not-ill" unemployed, despite a strong incentive to declare illness to obtain additional benefits, then there would have been higher mortality rates at the beginning of the follow-up period that would have fallen with time. This pattern occurred among the group chronically ill at the beginning of follow-up, but the excess mortality for those "not-ill" and unemployed rose slightly over the 10-year follow-up.

These findings were confirmed and extended by other large longitudinal studies in Europe.^16-18 A census-linked study in Denmark found a 40%-50% excess death rate among the unemployed after adjustment for occupation, housing category, geographical region and marital status,¹⁶ with suicides and accidents being prominent causes of excess mortality. Excess mortality was also found among unemployed women, although the relative mortality was somewhat higher for unemployed men. There was also an inverse association between relative mortality and the level of local unemployment.

A Finnish longitudinal study of linked census data for people aged 25-59 years found a higher mortality rate in unemployed than employed people, after controlling for age, education, type of occupation and marital status.¹⁸ Mortality ratios for men and women unemployed for the first time in 1990, at a time of low national unemployment, were 2.11 (95% CI, 1.76-2.53) and 1.61 (95% CI, 1.09-2.36), respectively. The comparable ratios for those unemployed for the first time in 1992, when the national employment rate was very high, were much lower: men, 1.35 (95% CI, 1.16-1.56); women, 1.30 (95% CI, 0.97-1.75).

A recent prospective cohort study of over 6000 British men, aged 40-59 years, and continuously employed for the five years before initial screening, found that men who experienced some unemployment or retired during the five years after screening were twice as likely to die in the following 5.5 years as those who remained continuously employed.¹⁹ After excluding men who became unemployed or retired because of ill-health, and adjusting for the effects of socioeconomic status, smoking, alcohol consumption, body weight and health indicators, the mortality ratio was 1.47 (95% CI, 1.10-1.96) for the unemployed group and 1.86 (95% CI, 1.34-2.59) for the retired group.

In Australia, the serious limitations of the information recorded on death certificates have prevented analysis of mortality rates for unemployed people.⁸ However, an early Australian study²⁰ found a correlation between aggregate unemployment trends and death from ischaemic heart disease (see also Morrell et al⁶ for suicide rate and unemployment rate correlations in Australia).

Mental health
Cross-sectional and longitudinal studies^3,13 (including longitudinal studies of young Australians^4,5 -- see Morrell et al⁶ ) have consistently found poorer psychological health in unemployed compared with employed people. A prospective US study clearly showed that men aged 35-60 years who became unemployed had higher levels of depression and anxiety than those who remained employed.²¹ A study of unemployed German men over the age of 45 found higher levels of psychological distress and that these regressed with re-employment or retirement.²²

As with mortality, it is likely that the impact of unemployment on mental health is dependent on general social conditions. An earlier British study found lower levels of psychological distress among men from areas of chronically high unemployment than among men living in areas of low unemployment²³ -- perhaps reflecting better adaptation through networks, community solidarity and lower costs of living in areas with higher unemployment.

In Australia, several longitudinal studies have shown poorer psychological health in unemployed compared with employed young people and that these differences first emerge after entry into the labour market. Furthermore, when unemployed young people find jobs their mental health improves.^4,5,24

Disease and disability
Cross-sectional population studies have documented more illness and poorer self-reported health in unemployed people after adjusting for the effects of social status and other variables.^8,25-27 An analysis of population survey data for Britain in 1991-92 found that, after controlling for education level and type of occupation, unemployed men and women had over twice the odds of having a limiting chronic illness compared with employed men and women, and a 60%-80% higher odds of reporting poor health.²⁷

Factory-closure studies, which minimise the effects of health selection, have found increased levels of medically diagnosed health problems, particularly cardiovascular disease and its risk factors, including high serum cholesterol levels and high blood pressure.^28-30 Despite occasional studies finding no association between unemployment and ill-health,³¹ the balance of evidence suggests that unemployment, at least among adult men, has an association with physical health, and in particular with cardiovascular disease.

In Australia, unemployed men and women aged 25-64 years were found to be about twice as likely to report being in poor or fair health (as opposed to good or excellent health); they also reported 30%-40% more serious chronic illnesses and 20%-30% more recent health problems than their employed counterparts. Differences in levels of smoking, risk drinking, physical inactivity and overweight did not account for these health differences (Figure).⁸

Lifestyle risk factors
Several studies have found higher rates of smoking and alcohol use and poorer diet among unemployed people,^32-35 although a causal link has not been confirmed.^36,37 A prospective study of British men aged 40-59 years found no evidence that they increased their smoking or drinking on becoming unemployed, but they were more likely to gain weight.³⁶ The men who became unemployed had higher levels of smoking and alcohol consumption at the initial screen, emphasising the importance of controlling for lifestyle factors (and socioeconomic status).

In Australia, the 1989-90 National Health Survey found that unemployed Australians were about 40%-50% more likely to be smokers, but were less inactive than the employed;^8,38 unemployed men were 57% more likely to have had measured hypertension than employed men (95% CI, 13%-100%).⁸

Health service use
Cross-sectional studies and factory-closure studies have documented higher levels of hospital admissions, doctor visits and outpatient visits among the unemployed.^28,39 These are usually interpreted as an indicator of poorer health. There has been little attempt to determine whether the higher level of health service use by the unemployed is commensurate with increased need.

In Australia, the 1989-90 National Health Survey found that unemployed men visited the doctor significantly more often, unemployed women reported significantly more hospital outpatient visits, and unemployed people used more pharmaceutical agents.^8,40 A multivariate analysis suggested that reported health status largely accounted for reported differentials in health service use.⁸

Differences between demographic groups
Unemployment falls disproportionately on younger and older workers and on already disadvantaged groups such as low income earners, recent migrants, indigenous people and workers with few skills.^26,27,41,42 In one study, the difference in levels of ill-health between unemployed and employed men was found to be greater for those in lower socio economic groups (as defined by education level and last occupation).²⁷ This may be a result of lower re-employment prospects among the unskilled or their fewer financial resources to cushion the effects of unemployment, although health selection may also play a greater role in lower socio economic groups.

Excess mortality caused by unemployment is highest for middle-aged men, and there is also some evidence that unemployment affects mental health more for this group than others. Health status reported by unemployed women tends not to be as low as for unemployed men, and similarly unemployed women report fewer health risk factors.^8,27,38

All these studies emphasise that the health consequences of unemployment are not the same for all groups in the population. This point is also underlined by the few studies that have found groups of employed people whose health is worse than that of unemployed people.⁵

Interpersonal effects
There have been surprisingly few studies of the effects of unemployment on the health of other family members. The British Office of Population Censuses and Surveys longitudinal study found a 20% excess mortality (95% CI, 4%-41%) among wives of unemployed men.² Other adverse effects of unemployment on family life include higher risk of separation and divorce, domestic violence, unwanted pregnancy, increased perinatal and infant mortality, poorer infant growth and increased health service use.^3,13,37,43

In Australia, children whose parents are unemployed, compared with those with one or more of their parents employed, were reported to have around 26% more serious chronic illnesses (95% CI, 3%-55%), 20%-30% more visits to the doctor, and around twice as many outpatient visits.⁴⁴

Aetiology
Few studies of unemployment and health have advanced beyond questions of causality and the magnitude of health effects to investigate aetiological issues. At the individual level, poverty, psychological impacts and health-related behaviours and lifestyle changes have been suggested (Box 1).

Conclusions

Acknowledgements

References

1. Moser KA, Goldblatt PO, Fox AJ, Jones DR. Unemployment and mortality: comparison of the 1971 and 1981 longitudinal study census samples. BMJ 1987; 1: 86-90.
2. Moser K, Goldblatt P, Fox J, Jones D. Unemployment and mortality. In: Goldblatt P, editor. Longitudinal study: mortality and social organisation. London: OPCS, 1990. (Series LS No. 6.)
3. Smith R. Unemployment and health: a disaster and a challenge. Oxford: Oxford University Press, 1987.
4. Morrell S, Taylor R, Quine S, et al. A cohort study of unemployment as a cause of psychological disturbance in Australian youth. Soc Sci Med 1994; 38: 1553-1564.
5. Graetz B. Health consequences of employment and unemployment: longitudinal evidence for young men and women. Soc Sci Med 1993; 36: 715-724.
6. Morrell S, Taylor R, Kerr C. Unemployment and young people's health. Med J Aust 1998; 168. In press.
7. National Health Strategy. Enough to make you sick: how income and environment affect health. Melbourne: National Health Strategy, 1992. (Research Paper No. 1.)
8. Mathers CD. Health differentials among adult Australians aged 25-64 years. Canberra: Australian Institute of Health and Welfare, 1994. (Health Monitoring Series No. 1.)
9. Mathers CD. Health differentials among young Australian adults. Canberra: Australian Institute of Health and Welfare, 1996. (Health Monitoring Series No. 4.)
10. Brenner MH. Health costs and benefits of economic policy. Int J Health Serv 1977; 7: 581-623.
11. Brenner MH. Mortality and the economy: a review, and the experience of England and Wales, 1936-1976. Lancet 1979; 2: 568-573.
12. Brenner MH. Economic change, alcohol consumption and heart disease mortality in nine industrialised countries. Soc Sci Med 1987; 25: 119-132.
13. Shortt S. Is unemployment pathogenic? A review of current concepts with lessons for policy planners. Int J Health Sci 1996; 26: 569-589.
14. Junankar PN. Unemployment and mortality in England and Wales: a preliminary analysis. Oxford Economic Papers 1991; 43: 305-320.
15. Starrin B, Larsson G, Brenner S-O. Regional variations in cardiovascular mortality in Sweden -- structural vulnerability in the local community. Soc Sci Med 1988; 27: 911-917.
16. Iversen L, Andersen O, Andersen PK, et al. Unemployment and mortality in Denmark, 1970-80. BMJ 1987; 295: 878-884.
17. Stefansson C-G. Long-term unemployment and mortality in Sweden, 1980-1986. Soc Sci Med 1991; 32: 419-423.
18. Martikainen P, Volkonen T. Excess mortality of unemployed men and women during a period of rapidly increasing unemployment. Lancet 1996; 348: 909-912.
19. Morris JK, Cook DG, Shaper AG. Loss of employment and mortality. BMJ 1994; 308: 1135-1139.
20. Bunn AR. Ischaemic heart disease mortality and the business cycle in Australia. Am J Public Health 1979; 69: 772-781.
21. Linn M, Sandifer R, Stein S. Effects of unemployment on mental and physical health. Am J Public Health 1985; 75: 502-506.
22. Frese M, Mohr G. Prolonged unemployment and depression in older workers: a longitudinal study of intervening variables. Soc Sci Med 1987; 25: 173-178.
23. Jackson P, Warr P. Mental health of unemployed men in different parts of England and Wales. BMJ 1987; 295: 525.
24. Banks MH, Jackson PR. Unemployment and the risk of minor psychiatric disorder in young people: cross-sectional and longitudinal evidence. Psychol Med 1982; 12: 789-798.
25. Arber S, Lahelma E. Inequalities in women's and men's ill-health: Britain and Finland compared. Soc Sci Med 1993; 37: 1055-1068.
26. Bartley M, Owen C. Relation between socioeconomic status, employment and health during economic change, 1973-93. BMJ 1996; 313: 445-449.
27. Arber S. Integrating nonemployment into research on health inequalities. Int J Health Serv 1996; 26: 445-481.
28. Beale N, Nethercott S. The health of industrial employees four years after compulsory redundancy. J Roy Coll Gen Pract 1987; 37: 390-394.
29. Iverson L, Sabroe S. Participation in a follow-up study of health among unemployed and employed people after a company closedown: drop outs and selection bias. J Epidemiol Community Health 1988; 42: 396-401.
30. Mattiasson I, Lindgarde F, Nilsson JA, Theorell T. Threats of unemployment and cardiovascular risk factors: longitudinal study of quality of sleep and serum cholesterol concentrations in men threatened with redundancy. BMJ 1990; 301: 461-466.
31. Van der Horst FF, Muris JN, Philipsen H, van der Grinten R. Causality in the relation between health and long-term unemployment. In: Verhaar C, Jansma L, editors. On the mysteries of unemployment. Studies in operational regional science. Volume 10. Dordrecht: Kluwer Academic Publishers, 1992: 225-252.
32. Hammarstrom A. Health consequences of youth unemployment -- review from a gender perspective. Soc Sci Med 1994; 38: 699-709.
33. Lee AJ, Crombie IK, Smith WCS, Tunstall-Pedoe HD. Cigarette smoking and employment status. Soc Sci Med 1991; 33: 1309-1312.
34. Bartley M. Unemployment and ill health: understanding the relationship. J Epidemiol Community Health 1994; 48: 333-337.
35. Power C, Estaugh V. Employment and drinking in early adulthood: a longitudinal perspective. Br J Addiction 1990; 85: 487-494.
36. Morris JK, Cook DG, Shaper AG. Non-employment and changes in smoking, drinking and body weight. BMJ 1992; 304: 536-541.
37. Wilson SH, Walker GM. Unemployment and health: a review. Public Health 1993; 107: 153-162.
38. Schofield D. Unemployment and health risk indicators. Paper presented at the third National Conference on Unemployment. Brisbane: Queensland University of Technology, 13-16 June 1996. (Available from the authors on request.)
39. Yuen P, Balarajan R. Unemployment and patterns of consultation with the general practitioner. BMJ 1989; 298: 1212-1214.
40. Schofield D. The impact of employment and hours of work on health status and health service use. Canberra: National Centre for Social and Economic Modelling, 1996. (NATSEM Discussion Paper No. 11.)
41. Australian Bureau of Statistics. Australian social trends 1997. Canberra: ABS, 1997. (Catalogue No. 4102.0.)
42. McClelland A. Long-term unemployment: costs and responses. Australian Economic Quarterly 1993; April-June: 26-30.
43. Dooley DJ, Fielding L, Levi L . Health and unemployment. Ann Rev Public Health 1996; 17: 449-464.
44. Mathers CD. Health differentials among Australian children. Canberra: Australian Institute of Health and Welfare, 1995. (Health Monitoring Series No. 3.)
45. Marmot MG, Bosma H, Hemingway H, et al. Contribution of job control and other risk factors to social variations in coronary heart disease incidence. Lancet 1997; 350: 235-239.
46. Harris E, Webster I, Harris M, Lee P. Unemployment and health: the healthcare system's role. Med J Aust 1998; 168. In press.
47. Saunders P. Employment growth and poverty: an analysis of Australian experience 1983-1990. Sydney: University of New South Wales, Social Policy Research Centre, 1990. (Discussion Paper No. 25.)