The healthy population–high disability paradox Jane E Ferrie, 1,2 Marianna Virtanen, 3 Mika Kivimaki 2,3,4 Scandinavia, in common with much of Europe and the USA, is facing a conun- drum: despite better health, more auto- mation and less manual work, an increasing proportion of the working-age population is economically inactive and in receipt of disability benefits. 12 The diag- nosis associated with much of this increase is mental disorders, depression in particular, although the prevalence of mental disorders has not increased signifi- cantly among the general population. 2 Increasing rates of disability pension in the young are particularly alarming; in Sweden, for example, newly granted pen- sions for mental disorders rose over five- fold (approximately 50/100 000 to over 250/100 000) in women aged 20–29 between 1990 and 2005. 3 The systematic review and meta-analyses of Van Rijn et al in this edition of OEM examine associations between three health indicators ( poor self-perceived health, poor mental health and self-reported chronic disease) and three subsequent routes of labour force exit; unemploy- ment, disability pension and early retire- ment. The majority (29/44) of studies in this review are from Scandinavia; the rest from Europe and the USA. 4 The results of this meta-analysis shed some light on the healthy population– high disability paradox. While the pooled relative risk of a disability pension was 80% higher among employees with pre- existing mental ill-health compared with those without, the mean population attributable fraction (PAF) was 11%, indi- cating that poor mental health made only a modest contribution to the decision to grant a disability pension. Similarly, the mean PAF for chronic disease, much of it musculoskeletal, at 21%, was still far from the proportion one might expect given the primacy of ill-health to the disability pension decision. Even the mean PAF for self-perceived health, which incorporates cognitive appraisal of both mental and physical health, as well as parameters such as work characteristics, was only 37%. Follow-up time (range 1–25 years) com- bined with the imprecision inherent in self-reported health measures will provide a partial explanation of these low PAF values. 4 Eight out of 10 people aged 55–64 have at least one chronic condition. Thus, although poor health, by definition, is the primary reason for award of a disability pension, most employees with chronic dis- eases continue to work. This suggests that other factors, in addition to health, influ- ence labour force exit, including disability pension. Interestingly, Van Rijn et al 4 found more than moderate heterogeneity between the studies (I 2 >50% for disabil- ity pension), but failed to identify deter- minants of this heterogeneity. Research on moderating factors, such as workplace- level factors and policy context, is now needed. Contextual factors will play a role. In the UK, for example, median labour force participation rates decline rapidly from age 50 in women and age 55 in men. 5 Data from Denmark and Norway show considerable differences by socioeconomic status within this overall pattern. A higher percentage of workers in lower grade occupations exit early on a disability pension or die before retirement than workers in higher grade occupations. 6 Although women have a higher overall risk of disability pension, the combination of fewer educational qualifications and a lower grade occupation is particularly problematic for men. 6 A reasonable hypothesis might be that this socio- economic gradient in disability pension is largely the result of the strong positive health gradient running up the occupa- tional hierarchy. However, recent observa- tions from the Norwegian Hordaland Health Study, which show health pro- blems explain little of the association between socioeconomic status and disabil- ity pension, would appear to refute this. 7 One of the problems that beset labour market research is that of outcome defini- tions. Unemployed workers in receipt of unemployment benefits have to fulfil set criteria. However, the label ‘unemployed’, as evidenced by the current review, can cover a wide range of self-reported out- comes, from <15 h work a week through short-term government schemes to unemployment lasting over 2 years. 4 Given the stigma attached to the label ‘unemployed’, social desirability may lead some participants to adopt the ‘sick role’ to ‘legitimise’ their status or older workers to classify themselves as early retirees. Another problem is the change in meaning of labour market exit outcomes over time. The literature suggests that until the 1980s early retirement was almost always connected to a breakdown in health. In 1982, Kingson 8 claimed that the health of men whose application for disability benefit was refused was as poor as that of those granted the benefit. Myers challenged Kingson’s findings. Showing that Kingson had overestimated the level of ill-health among those not granted disability benefits, he proposed that individual-level economic factors had greater salience in the early retirement decision. 9 Subsequent work sup- ports Myer’s conclusions. 10 Factory closure studies from the 1970s, 1980s and 1990s, for example, show that adverse effects on health begin from the time when people sense that their jobs are no longer secure and redundancy is a possibility. 11 In con- trast, a study of shipyard workers showed no ill-health among men aged >58 threa- tened with redundancy but eligible for a relatively generous early retirement package. 11 Economic booms increase job vacancies, drive up wages and seem to suck those with health problems into the labour force. Recession appears to have the opposite effect. When jobs are scarce, even minor health problems may be a barrier to employment. Wages can also be so low that people cannot afford to relin- quish benefits to take up employment. The Global Financial Crisis, which started in late 2007, saw the rate of unemploy- ment rise sharply in most countries included in the current meta-analysis. 4 However, rates of labour force participa- tion and unemployment tell us little about wages or conditions. Secure employment appears to provide protection against unemployment among women with a per- manent contract, even in the case of high sickness absence, whereas similar levels of sickness absence are strong risk factors for job loss and unemployment among tem- porary workers. 12 The current recession in many countries has witnessed further casualisation, falls in real earnings and cuts to welfare benefits. In this context, the pattern of labour force exit by health is likely to change again. While 1 School of Social and Community Medicine, University of Bristol, Bristol, UK; 2 Department of Epidemiology and Public Health, University College London, London, UK; 3 Finnish Institute of Occupational Health, Helsinki, Tampere and Turku, Finland; 4 Institute of Behavioural Sciences, University of Helsinki, Helsinki, Finland Correspondence to Dr Jane Ferrie, Department of Epidemiology and Public Health, University College London, London WC1E 6BT, UK; j.ferrie@ucl.ac.uk 232 Ferrie JE, et al. Occup Environ Med April 2014 Vol 71 No 4 Commentary group.bmj.com on August 7, 2014 - Published by oem.bmj.com Downloaded from