You are here

  • Home
  • Archive
  • Volume 66, Issue 2
  • Diagnosis-specific sick leave as a long-term predictor of disability pension: a 13-year follow-up of the GAZEL cohort study

Article Text

Article menu

Download PDFPDF

Research report
Diagnosis-specific sick leave as a long-term predictor of disability pension: a 13-year follow-up of the GAZEL cohort study
  1. K Alexanderson1,
  2. M Kivimäki2,
  3. J E Ferrie2,
  4. H Westerlund1,3,
  5. J Vahtera4,
  6. A Singh-Manoux2,5,
  7. M Melchior5,
  8. M Zins5,6,
  9. M Goldberg5,6,
  10. J Head2
  1. 1Division of Insurance Medicine, Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden
  2. 2Department of Epidemiology and Public Health, University College London Medical School, London, UK
  3. 3Stress Research Institute, Stockholm University, Sweden
  4. 4Finnish Institute of Occupational Health, Helsinki, Finland
  5. 5INSERM, U1018 Occupational and Social Determinants of Health, Villejuif, France
  6. 6Centre for research in Epidemiology and Population Health, Versailles Saint Quentin University, Villejuif, France
  1. Correspondence to Professor, Kristina Alexanderson, Division of Insurance Medicine, Department of Clinical Neuroscience, Berzelius väg 3, Karolinska Institutet, SE-171 77 Stockholm, Sweden; kristina.alexanderson{at}ki.se

Abstract

Background Factors that increase the risk of labour market exclusion are poorly understood. In this study, we examined the extent to which all-cause and diagnosis-specific sick leave predict subsequent disability pension (DP).

Methods Prospective cohort study of 20 434 persons employed by the French national gas and electric company (the GAZEL study). New sick-leave spells >7 days in 1990–1992 were obtained from company records. Follow-up for DP was from 1994 to 2007.

Results The HR, adjusted for age and occupational position, for DP was 3.5 (95% CI 2.7 to 4.5) in men and 2.6 (95% CI 1.9 to 3.5) in women with one or more sick-leave spells >7 days compared with those with no sick leave. The strongest predictor of DP was sick leave with a psychiatric diagnosis, HR 7.6 (95% CI 5.2 to 10.9) for men and 4.1 (95% CI 2.9 to 5.9) for women. Corresponding HRs for sick leave due to circulatory diagnoses in men and women were 5.6 (95% CI 3.7 to 8.6) and 3.1 (95% CI 1.8 to 5.3), for respiratory diagnoses 3.9 (95% CI 2.6 to 5.8) and 2.6 (95% CI 1.7 to 4.0), and musculoskeletal diagnoses 4.6 (95% CI 3.4 to 6.4) and 3.3 (95% CI 2.2 to 4.8), respectively.

Conclusions Sick leave with a psychiatric diagnosis is a major risk factor for subsequent DP, especially among men. Sick leave due to musculoskeletal or circulatory disorders was also a strong predictor of DP. Diagnosis-specific sick leave should be recognised as an early risk marker for future exclusion from the labour market.

  • Sick leave
  • sick-leave diagnoses
  • disability pension
  • ill-health retirement
  • psychiatric
  • disabling disease
  • occupational health
  • psychiatric
  • sickness absence

https://doi.org/10.1136/jech.2010.126789

Statistics from Altmetric.com

    Request Permissions

    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.

    Background

    Although health and longevity are improving over time,1 the proportion of people out of the workforce due to health disorders has risen.2 Similarly, the rates of disability pension (DP) are increasing in many European countries.2 These trends put a strain on the economy, employers and insurance companies.3 There is a need to identify early risk markers for future DP in order for healthcare providers, employers and others to take preventive actions.

    With the exception of very young people, DP is generally preceded by long-term sickness absence. Long-term sick leave has therefore been seen to be a good risk marker for subsequent DP,4–8 although it is recognised that the majority of both long- and short-term sick-leave spells do not lead to DP. A recent population-based prospective cohort study from Sweden found elevated risks of future DP among persons with sick-leave spells exceeding 7 days and also observed that the risk differed according to sick-leave diagnosis.9 However, the extent to which sick leave is useful in determining long-term risk of DP remains unclear as the follow-up time in most studies has been relatively short.3 Moreover, very few studies have considered diagnosis-specific sick leave as an early marker for DP because most registers of sickness absence lack data on medical diagnoses of the sick-leave spells.

    The aim of this study was, therefore, to investigate all-cause and diagnosis-specific sick leave as long-term risk markers for DP.

    Methods

    Study design and study population

    This 13-year prospective study uses company register data from the GAZEL cohort study, France. The GAZEL study began in January 1989 when 44 922 employees of France's national gas and electric company, Electricité de France-Gaz de France (EDF-GDF), women aged 35–50 years and men aged 40–50 years, were asked to participate in a long-term observational study. Of these, about 45% participated yielding a sample of 20 625 subjects (27% women). Cohort participants hold civil servant-like status, have benefited from job stability and have had opportunities for upward occupational mobility during their career. Typically, employees were hired in their 20s and have stayed with the company until old age retirement, usually between 55 and 60 years of age with <1% in the study lost to follow-up.10

    The present analysis included 20 434 GAZEL participants who were employed on 1 January 1990 and still alive on 1 January 1994. We used records from EDF-GDF company registers to obtain baseline characteristics (sex, age and type of occupation in 1990), sick leave (1990–1992) and the date of DP, old age pension or death until the end of 2007.

    Ethical approval for the GAZEL study was provided by the French Commission Nationale Informatique et Libertés, which oversees ethical aspects of data collection in France.

    Sick-leave data

    All new sick-leave spells exceeding 7 days that began between 1 January 1990 and 31 December 1992 were included. Diagnoses for medically certified sick-leave spells were coded by company physicians using an abridged version of the International Classification of Diseases 9th version (ICD-9).11 The following five diagnostic categories were used in the present study: musculoskeletal (ICD-9 Chapter XIII), psychiatric (V), circulatory (VII), respiratory (VIII) and injuries (XVII). Sick-leave spells in all other diagnostic categories were classified as ‘all other diagnoses’. To be included in a particular diagnostic category, participants had to have at least one sick-leave spell >7 days with that diagnosis during the 3-year exposure window. Dichotomous variables were derived for each diagnostic category (0 spells, ≥1 spells). Sick-leave diagnoses were missing for 20% of sick-leave spells during 1990–1992.

    Disability pension

    At the time of this study, all EDF-GDF employees were covered by a company-run insurance scheme, which entitled them to full pay during periods of sickness absence and 80% of their salary during retirement or when on DP. Company policy with regard to long-term sick leave was to grant a DP after 5 years of sickness absence. However, in order to stay on sickness benefit rather than DP, some workers interrupted their sick leave for a short period, before claiming sickness benefits again. To overcome this problem in the data, DP due to health problems was classified as temporary or permanent DP or >600 sick-leave days during two consecutive years between 1 January 1994 and 31 December 2007. In order to separate the exposure measure (a new sick-leave spell in 1990–1992) from the outcome measure (DP), a ‘washout’ period was used and outcome events in the 12 months (1 January to 31 December 1993) immediately following the sickness absence exposure window were not included in the analyses.

    Covariates

    Age in 1990 was grouped into three categories: 36–39, 40–44 and 45–50 years. Occupational grade in 1990 was available from EDF-GDF company records and coded into three groups: manual worker or clerk (low), technician or administrative associate professional (intermediate) and engineer or manager (high).

    Statistical analysis

    Differences in incidence of DP were tested with Cox proportional hazards models. Each individual in the study was followed from 1 January 1994 to date of DP or censored at the date of old age retirement, death or at the end of the study, 31 December 2007.

    In Cox regression models, participants with no new sick-leave spells >7 days in 1990–1992 served as the reference category. HRs for the risk of a DP during 1994–2007 were adjusted for age and occupational position. We illustrated the effect of sick leave on the probability of incident DP by psychiatric sick leave, all other types of sick leave and no new sick-leave spells >7 days in the exposure period, using Kaplan–Meier curves, tested with the log-rank p value. All analyses were stratified by sex as there are sex differences in sickness absence, DP and possibly in the underlying mechanisms.3 12–17 We tested the proportional hazards assumption by including an interaction of each diagnosis-specific absence variable with the log of follow-up time. For women but not men, these interactions with time were statistically significant for any sick-leave spell (test for interaction, p=0.03) and for all diagnostic categories (tests for interaction, psychiatric diagnoses p=0.006, circulatory p=0.004, respiratory p=0.02, musculoskeletal p=0.002 and injuries p=0.03). The interaction results indicated that hazards ratios for women tended to attenuate over follow-up time. To illustrate this, we used a piecewise Cox regression model to estimate separate HRs for <6 years of follow-up and ≥6 years of follow-up. The analyses were conducted using the SAS V.9.1 program.

    Results

    A higher percentage of women than men had one or more sick-leave spell >7 days in 1990–1992 (table 1). During the 13-year follow-up from 1994, 529 employees were granted a DP, 4.4% of the women and 1.9% of the men. The risk of DP was tripled among men who had a new sick-leave spell >7 days (HR 3.5, 95% CI 2.7 to 4.5) in 1990–1992, and more than doubled among women in the corresponding category (HR 2.6, 95% CI 1.9 to 3.5), compared with those with no sick leave (table 2). The HRs for DP varied by sick-leave diagnosis and were high (HRs from 3 to 8) for all the five diagnostic groups studied. The HR for DP was higher for men than for women in all diagnostic groups, with the exception of injuries, and was especially high for men who had at least one new sick-leave spell with a psychiatric diagnosis: 7.56 compared with 4.14 for women (test for sex interaction, p=0.019). There was also a high likelihood of DP among men with sick leave due to a circulatory disease (HR 5.6) or musculoskeletal disease (HR 4.6). For all diagnostic groups, the HRs for DP in women were higher in the first 6 years of follow-up than in the later years of follow-up (table 3).

    View this table:
    Table 1

    GAZEL cohort participants (N=20 434), incidence of sick-leave spells >7 days (SL) 1990–1992 and disability pension (DP) 1994–2007, in women and men

    View this table:
    Table 2

    HRs and 95% CIs of incident disability pension (DP) in 1994–2007 in women and men by sick leave in 1990–1992

    View this table:
    Table 3

    HRs and 95% CIs of incident disability pension (DP) in 1994–2007 in women by sick leave in 1990–1992

    There were large differences in the cumulative probability of DP associated with psychiatric sick leave as compared with all other sick leave and no sick leave during the exposure period (figure 1). The risk was highest among those who initially had a sick-leave spell due to a psychiatric diagnosis and lowest among those with no new sick-leave spell in 1990–1992. Probability of DP was similar in men and women in the first 4–6 years; in the period that followed, women had a higher probability of DP than men for psychiatric sick leave, other sick leave and no sick leave.

    Figure 1
    Figure 1

    Cumulative hazard function for disability pension in 1994–2007 among participants of the GAZEL cohort who in 1990–1992 had no new sick-leave spells >7 days, those who had such sick-leave spells due to psychiatric disorders and those who had sick-leave spells with other diagnoses, respectively, for women and men.

    Discussion

    This 13-year follow-up of DP showed high relative risks among employees who had been sickness absent compared with those with no sick leave. Although the association between sick leave and DP was strong in both sexes, the HRs were higher among men than among women. This was especially noticeable for psychiatric diagnoses for which the HR was 7.6 in men, even though a washout period of 1 year was used. These data suggest that sickness absence of >7 days should be regarded as a risk marker for future DP, especially sick leave due to psychiatric or circulatory diagnoses.

    Although sick leaves are common, there are hardly any studies of the long-term consequences of being sickness absent.18 19 To date, most studies have focused on risk factors for sickness absence rather than sickness absence as a risk factor for DP.20–22 However, risk factors for sickness absence are not necessarily the most important information needed by doctors, employees or employers when considering sick leave at an individual level.3 Earlier research on the GAZEL cohort has found a remarkably persistent association between sickness absence and future long-term self-rated health status for the majority of diagnostic categories for sickness absence23 and mortality.24 The results of the present study indicate that more attention should also be paid to sickness absence as a marker of future risk of DP.

    Our results are in line with those of the largest study on this topic to date, the population-based cohort study of all 176 629 persons of working age living in a Swedish county.9 In that study, 18% of participants had a new sick-leave spell of >7 days during 1985. Over a follow-up of 11 years, 7% were granted DP with an overall fourfold hazard for DP in the first 5 years and a threefold hazard in the following 5 years. Moreover, there was great variation in risk by diagnosis with risk highest for psychiatric diagnoses. An important difference is that the Swedish investigation9 found musculoskeletal sick-leave diagnoses (rather than circulatory diagnoses) to be the second strongest predictor of future DP, with HRs of 7 and 4, respectively, for the two follow-up periods. Further comparisons are hampered by the population-based design of the Swedish study in that levels of ill health are likely to be higher, given that unemployed and inactive individuals, who tend to have high rates of disease and DP, were included. In the present study, we adjusted for occupational position, known to be a predictor of both rates of sick leave and rates of DP, whereas data on occupational position were not available in the Swedish study. Also, in the Swedish study, analyses were not stratified by sex. Sickness absence studies have repeatedly shown that women have higher rates of sickness absence and DP than men25–27 Nevertheless, it seems as if the future consequences of sickness absence, in terms of the RR of DP, might be higher for men, an observation that confirms findings from a Scandinavian study.28

    Psychiatric disorders are very common in the population and rates of sickness absence due to psychiatric disorders have increased in many Western nations over the last decade.29–32 In agreement with our findings, several studies have indicated that long-term sick leave with psychiatric diagnoses is a risk factor for DP. Karlsson and colleagues4 and Gjesdal and colleagues5 6 observed this for sick-leave spells that exceeded 8 weeks and Vaez and colleagues31 for spells >3 months. The present study extends these findings by showing that diagnosis-specific sick leave >7 days may also represent an important risk marker for DP.

    Methodological considerations

    In recent decades, the level of sickness absence has fluctuated considerably in some countries, for example, the Netherlands, Sweden and Norway.20 33 However, this is not the case in France, which makes it a good setting for this type of study. A further strength of this study is the low turnover among employees at EDF-GDF. However, this also means that the results might not be generalisable to worksites where the turnover is higher. Other strengths are the unusually long follow-up period (13 years), good data quality and no missing data for the outcomes. A limitation is that the study is based on volunteers. However, the extent of healthy worker bias is likely to be comparable to that observed in other occupational cohort studies.34 35

    The validity of sick-leave diagnoses has been questioned, yet seldom studied. The few studies conducted conclude that the validity is in fact high, especially when using high ICD-chapter levels, as was the case in our analysis.36 In the present study, the diagnoses were confirmed by an occupational health physician, further strengthening the validity. Of course, the high rate of sick-leave spells lacking a diagnosis is a limitation of this study. However, psychiatric sick-leave diagnoses probably have especially high validity. This is because they are considered to be most stigmatising and so would not be assigned if not accurate.30 37

    The use of a ‘washout’ period of 1 year could, if anything, lead to an underestimation rather than overestimation of the risk.

    Conclusions

    These data suggest that diagnosis-specific sickness absence should be recognised as a marker of future DP and considered as an indicator of need for preventive measures. In particular, psychiatric sickness absence should be considered as a risk factor for future early exclusion from the workforce.

    What is already known on this subject

    • Rates of disability pension or ill-health retirement are increasing in many European countries.

    • Sick leave is associated with risk of disability pension in the short term, but few studies have examined the long-term risk of disability pension.

    What this study adds

    • Sick leave of >7 days was a strong predictor of disability pension over 13 years of follow-up.

    • Associations were particularly strong for sick leave with a psychiatric diagnosis, especially among men.

    • Also short-term sick leave could be considered as a risk marker for future exclusions from the labour market.

    Acknowledgments

    The authors express their thanks to EDF-GDF, especially to the Service Général de Médecine de Contrôle, and to the ‘Caisse centrale d'action sociale du personnel des industries électrique et gazière’. We also wish to acknowledge the Cohortes team of the Unité 1018 Inserm responsible for the GAZEL data base management.

    References

    1. ↵
      1. Marmot M
      . The Status Syndrome. How Social Standing Affects Our Health and Longevity. New York: Time Books, Henry Holt and Company, LLC, 2004.
    2. ↵
      The Organisation for Economic Co-operation and Development (OECD). Sickness, Disability and Work: Breaking the Barriers; A Synthesis of Findings across OECD Countries. OECD Publishing, 2010.
    3. ↵
      1. Alexanderson K,
      2. Norlund A
      , eds. Sickness absence: causes, consequences, and physicians' sickness certification practice. A systematic literature review by the Swedish Council on Technology Assessment in Health Care. Scand J Public Health Suppl 2004;32:1–263.
      OpenUrlCrossRefWeb of Science
    4. ↵
      1. Karlsson NE,
      2. Carstensen JM,
      3. Gjesdal S,
      4. et al
      . Risk factors for disability pension in a population-based cohort of men and women on long-term sick leave in Sweden. Eur J Public Health 2007;18:224–31.
      OpenUrlWeb of Science
    5. ↵
      1. Gjesdal S,
      2. Bratberg E
      . Diagnosis and duration of sickness absence as predictors for disability pension: results from a three-year, multi-register based and prospective study. Scand J Public Health 2003;31:246–54.
      OpenUrlAbstract/FREE Full Text
    6. ↵
      1. Gjesdal S,
      2. Ringdal PR,
      3. Haug K,
      4. et al
      . Predictors of disability pension in long-term sickness absence: results from a population-based and prospective study in Norway 1994–1999. Eur J Public Health 2004;14:398–405.
      OpenUrlAbstract/FREE Full Text
    7. ↵
      1. Kausto J,
      2. Virta L,
      3. Luukkonen R,
      4. et al
      . Associations between partial sickness benefit and disability pensions: initial findings of a Finnish nationwide register study. BMC Public Health 2010;10:361.
      OpenUrlCrossRefPubMed
    8. ↵
      1. Virtanen M,
      2. Kivimaki M,
      3. Vahtera J,
      4. et al
      . Sickness absence as a risk factor for job termination, unemployment, and disability pension among temporary and permanent employees. Occup Environ Med 2006;63:212–17.
      OpenUrlAbstract/FREE Full Text
    9. ↵
      1. Kivimäki M,
      2. Ferrie J,
      3. Hagberg J,
      4. et al
      . Sick leave as a risk marker of disability pension: 11-year prospective cohort study in Sweden. J Epidemiol Community Health 2007;61:915–20.
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Goldberg M,
      2. Leclerc A,
      3. Bonenfant S,
      4. et al
      . Cohort profile: the GAZEL Cohort Study. Int J Epidemiol 2007;36:32–9.
      OpenUrlFREE Full Text
    11. ↵
      WHO. International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). 2010. http://www.cdc.gov/nchs/icd/icd9cm.htm.
    12. ↵
      1. Hensing G,
      2. Andersson L,
      3. Brage S
      . Increase in sickness absence with psychiatric diagnosis in Norway: a general population-based epidemiologic study of age, gender and regional distribution. BMC Med 2006;4:19.
      OpenUrlCrossRefPubMed
    13. ↵
      1. Hensing G,
      2. Alexanderson K
      . The association between sex segregation, working conditions, and sickness absence among employed women. Occup Environ Med 2004;61:1–6.
      OpenUrlFREE Full Text
    14. ↵
      1. Barmby TA,
      2. Ercolani MG,
      3. Treble JG
      . Sickness absence: an international comparison. Econ J 2002;112:F315–31.
      OpenUrlCrossRef
    15. ↵
      1. Hensing G,
      2. Alexanderson K
      . The relation of adult experience of domestic harassment, violence, and sexual abuse to health and sickness absence. Int J Behav Med 2000;7:1–18.
      OpenUrlCrossRefWeb of Science
    16. ↵
      1. Mastekaasa A,
      2. Dale-Olsen H
      . Do women or men have the less healthy jobs? An analysis of gender differences in sickness absence. Eur Sociol Rev 2000;16:267–86.
      OpenUrlAbstract
    17. ↵
      1. Ala-Mursula L,
      2. Vahtera J,
      3. Kouvonen A,
      4. et al
      . Long hours in paid and domestic work and subsequent sickness absence: does control over daily working hours matter? Occup Environ Med 2006;63:608–16.
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. Vingård E,
      2. Alexanderson K,
      3. Norlund A
      . Swedish Council on Technology Assessment in Health Care (SBU). Chapter 9. Consequences of being on sick leave. Scand J Public Health Suppl 2004;32:207–15.
      OpenUrlCrossRefWeb of Science
    19. ↵
      1. Floderus F,
      2. Göransson S,
      3. Alexanderson K,
      4. et al
      . Self-estimated life situation in patients on long-term sick leave. J Rehabil Med 2005;37:291–9.
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Alexanderson K,
      2. Norlund A
      . Swedish Council on Technology Assessment in Health Care (SBU). Chapter 1. Aim, backround, key concepts regulations, and current statistics. Scand J Public Health Suppl 2004;32:12–30.
      OpenUrlCrossRefWeb of Science
    21. ↵
      1. Alexanderson K,
      2. Hensing G
      . More and better research needed on sickness absence. Scand J Public Health Suppl 2004;32:321–3. Editorial.
      OpenUrlCrossRefWeb of Science
    22. ↵
      1. Alexanderson K,
      2. Norlund A
      . Swedish Council on Technology Assessment in Health Care (SBU). Future need for research. Scand J Public Health Suppl 2004;32:256–8.
      OpenUrlCrossRefWeb of Science
    23. ↵
      1. Vahtera J,
      2. Westerlund H,
      3. Ferrie J,
      4. et al
      . All-cause and diagnosis-specific sickness absence as a predictor of sustained sub-optimal health: a 14-year follow-up in the GAZEL cohort. J Epidemiol Community Health 2010;64:311–17.
      OpenUrlAbstract/FREE Full Text
    24. ↵
      1. Ferrie JE,
      2. Vahtera J,
      3. Kivimäki M,
      4. et al
      . Diagnosis-specific sickness absence and all-cause mortality in the GAZEL study. J Epidemiol Community Health 2009;63:50–5.
      OpenUrlAbstract/FREE Full Text
    25. ↵
      1. Festin K,
      2. Alexanderson K
      . Changes in sick-leave diagnoses over eleven years in a cohort of young adults initially sick-listed due to low back, neck, or shoulder diagnoses. J Rehabil Med 2009;41:423–8.
      OpenUrlCrossRefPubMed
    26. ↵
      1. Allebeck P,
      2. Mastekaasa A
      . Swedish Council on Technology Assessment in Health Care (SBU). Chapter 5. Risk factors for sick leave - general studies. Scand J Public Health Suppl 2004;63:49–108.
      OpenUrlPubMed
    27. ↵
      1. Lidwall U,
      2. Marklund S
      . What is healthy work for women and men? - A case-control study of gender- and sector-specific effects of psycho-social working conditions on long-term sickness absence. Work 2006;27:153–63.
      OpenUrlPubMed
    28. ↵
      1. Hensing G,
      2. Brage S,
      3. Nygård J,
      4. et al
      . Sickness absence with psychiatric disorders - an increased risk for marginalisation among men? Soc Psychiatry Psychiatr Epidemiol 2000;35:335–40.
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. Järvisalo J,
      2. Anderson B,
      3. Boedeker W,
      4. et al
      . Mental disorders as a major challenge in prevention of work disability. Helsinki: Kela, 2005.
    30. ↵
      1. Hensing G,
      2. Wahlström R
      . Swedish Council on Technology Assessment in Health Care (SBU). Chapter 7. Sickness absence and psychiatric disorders. Scand J Public Health 2004;32:152–80.
      OpenUrlAbstract/FREE Full Text
    31. ↵
      1. Vaez M,
      2. Rylander G,
      3. Nygren A,
      4. et al
      . Sickness absence and disability pension in a cohort of employees initially on long-term sick leave due to psychiatric disorders in Sweden. Soc Psychiatry Psychiatr Epidemiol 2007;42:381–8.
      OpenUrlCrossRefPubMedWeb of Science
    32. ↵
      1. Waddell G,
      2. Burton K
      . Is Work Good for Your Health and Well-being? London: TSO, 2006.
    33. ↵
      1. Bergendorff S
      . Sickness absence in Europe - a comparative study. 4th International Research Conference on Social Security. Antwerp National Social Insurance Board, 2003.
    34. ↵
      1. Goldberg M,
      2. Chastang JF,
      3. Leclerc A,
      4. et al
      . Socioeconomic, demographic, occupational, and health factors associated with participation in a long-term epidemiologic survey: a prospective study of the french GAZEL cohort and its target population. Am J Epidemiol 2001;154:373–84.
      OpenUrlAbstract/FREE Full Text
    35. ↵
      1. Goldberg M,
      2. Chastang JF,
      3. Zins M,
      4. et al
      . Health problems were the strongest predictors of attrition during follow up of the GAZEL cohort. J Clin Epidemiol 2006;59:1213–21.
      OpenUrlCrossRefPubMedWeb of Science
    36. ↵
      1. Ljungdahl L,
      2. Bjurulf P
      . The accordance of diagnoses in a computerized sick-leave register with doctor's certificates and medical records. Scand J Soc Med 1991;19:148–53.
      OpenUrlPubMedWeb of Science
    37. ↵
      1. Hensing G,
      2. Alexanderson K,
      3. Akerlind I,
      4. et al
      . Sick-leave due to minor psychiatric morbidity: role of sex integration. Soc Psychiatry Psychiatr Epidemiol 1995;30:39–43.
      OpenUrlCrossRefPubMedWeb of Science

    Footnotes

    • Funding This work was supported by an ESRC Research Seminar Series Competition 2007/8 (RES-451-26-0491). KA was supported by the Swedish Council for Working Life and Social Research. MK and JV are supported by the Academy of Finland (grants #117604, #124271, #124322 and 129262) and the EU New OSH ERA research programme, JEF, AS-M and JH are supported by the National Institutes on Aging (NIA RO1AG013196). AS-M is supported by a EUYRI award from the European Science Foundation. HW is supported by the Swedish Council for Working Life and Social Research (2004-2021, 2007-1143). MM is supported by a “Young researcher” program from France's National Research Agency (ANR). The GAZEL cohort study was funded by EDF-GDF and INSERM and received grants from the ‘Cohortes Santé TGIR Program’, Agence nationale de la recherché (ANR) and Agence française de sécurité sanitaire de l'environnement et du travail (AFSSET).

    • Competing interests None declared.

    • Patient consent Obtained.

    • Ethics approval This study was conducted with the approval of The French Commission Nationale Informatique et Libertés.

    • Provenance and peer review Not commissioned; externally peer reviewed.