The European Journal of Public Health Advance Access originally published online on March 9, 2008
The European Journal of Public Health 2008 18(5):517-521; doi:10.1093/eurpub/ckn010
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Mortality |
Mortality after long-term sickness absence: prospective cohort study
Sturla Gjesdal1, Peder R. Ringdal2, Kjell Haug3, John G. Maeland3, Stein E. Vollset3 and Kristina Alexanderson4
1 Department of Public Health and Primary Health Care, and Health Economics, University of Bergen, Norway
2 Department of Public Health and Primary Health Care, University of Bergen and National Insurance Services, Hordaland County Office, Bergen, Norway
3 Department of Public Health and Primary Health Care, University of Bergen, Bergen, Norway
4 Section for Personal Injury Prevention, Karolinska Institute, Stockholm, Sweden
Correspondence: Sturla Gjesdal, ISF UiB, Kalfarveien 31, 5018 Bergen, Norway, tel: +47 55586100/50, fax: +47 55586130, e-mail: sturla.gjesdal{at}isf.uib.no
Received August 27, 2007, accepted January 23, 2008
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
Background: The study estimated the excess mortality after long-term sickness absence (LTSA), and identified socio-demographic and diagnostic risk factors of death. Methods: Prospective cohort study during 1994–2003 in a Norwegian county with 256 654 inhabitants aged 16–62 years. A representative sample of 3386 persons with a spell of sickness absence >8 weeks was compared with the total county population with respect to all cause mortality. Comparative mortality figures (CMF) for the total sample and standardized mortality rates for diagnostic groups were calculated. Results: The CMFs were 1.5 (95% CI 1.1–1.9) for the female and 2.0 (95% CI 1.7–2.4) for the male sample. Among women, persons’ sickness certified with cancer contributed with 43% of all deaths and standardized mortality ratios (SMR) was 16.1 (11.2–23.2). The respective figure for the men was 27% and SMR was 8.0 (5.7–11.1). SMR for men with mental diagnoses was 1.7 (95% CI 1.1–2.9) and for ‘other’ (respiratory, neurological, digestive) 1.8 (95% CI 1.3–2.7). Musculoskeletal cases had not elevated SMRs. Cox proportional hazard analysis with musculoskeletal cases as reference adjusted for age and income showed very high hazard ratios (HR) for cases with cancer diagnoses. Among the men, mental and ‘other’ diagnoses had also HR above unity. Conclusion: The study verified findings from Finland and the UK of excess mortality after LTSA, also when compared with the total population of the same age. Among women, cancer cases explained all the excess mortality, whereas other cases outside the musculoskeletal group also contributed among men.
Keywords: diagnoses, mortality, sickness absence
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
Long-term sickness absence (LTSA) is an important medical and socio-political challenge1 which has been acknowledged as a powerful measure of ill health, perhaps even more than self-reported indicators.2–4 Sickness absence is related to social, occupational and medical factors,5–7 and the role of disease becomes more pronounced with increasing length of sick-leave spells.2,3 However, little is known about the consequences of being on LTSA,8 but transition to permanent incapacity is a frequent outcome.9–12
A new focus in sickness absence research has become LTSA as a predictor of future death, both all-cause3 and cause-specific mortality.13 Only two previous studies have examined the association between certified sickness absence and subsequent mortality on an individual level. Those were occupational cohort studies from the public sector in the UK3 and Finland13 and showed an increased mortality among employees with high rates of medically certified sickness absence. The study from the Whitehall II cohort analysed the mortality between 1985 and 1998,3 whereas the study from the 10-towns cohort in Finland had mortality data from 1995–2001.13 In addition, one recent study compared levels of sickness absence with morbidity and mortality within local government districts of England and Wales.14 There is still a lack of explanations for the findings in those studies, and results might vary between countries with different legislations, and different levels of sickness absence. While women generally have a higher incidence of LTSA,15–17 the mortality risk is probably higher among men, partly reflecting higher mortality rates among middle-aged men in general.
However, at a more detailed level, the sick-leave diagnoses in LTSA has not been investigated as possible predictors of future mortality, since many researchers have no access to this information.18 The aim of this study was to assess the mortality risk in a cohort on LTSA with a known sick-leave diagnosis, to compare this risk with that of the total population of the same age, and to identify socio-demographic and medical predictors of death among persons on LTSA.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
The Norwegian sickness absence benefit scheme
All employed and registered unemployed persons are covered. After 3 days, a sickness certificate is required. After 8 weeks, the certifying physician must complete an ‘eight-week sickness absence certificate’ including medical information and plans for rehabilitation.
Study samples
A population-based cohort study was carried out in the county of Hordaland, with 
10% of the population of Norway. On 1 January 1994, 106 674 men and 89 356 women in the county were eligible for sickness absence benefits. During two periods in 1994, all new ‘eight-week sickness absence certificates’ in the county (n = 4156) were collected by the NIS County Office and reviewed by one of the authors (PRR). Individuals >62 years, individuals receiving a partial disability pension, and sickness absentees with a pregnancy-related diagnosis (11.5% of the women) were excluded. After further exclusion of 155 cases due to lack of relevant information, the study sample comprised 3386 subjects, 1801 women and 1585 men. This sample has been described in a previous paper in EJPH.12
End point and mortality data
The dependent variable was all-cause mortality during 1994–2003. Data on deaths and emigration in the study sample were obtained from the National Insurance services (NIS). Census data for the population in Hordaland aged 16–62 on 1 January 1994 (256 654 persons) and deaths in 1994–2003 according to 5-year age groups were obtained from Statistics Norway.
Explanatory variables
Information regarding age, sex and the main sick-leave diagnosis, based on the Norwegian version of the International Classification of Primary Care (ICPC), were found in the ‘eight-week sickness absence certificates’. Since the majority of sickness certificates in Norway are issued by GPs, the ICPC has been used by the NIS since 1990.19,20 The ICPC is organized in chapters corresponding to organ systems. Each chapter has sub-diagnoses for cancer, which were identified. In the analyses, the cases were categorized to one of the following five diagnostic groups: musculoskeletal, psychiatric, cardiovascular, cancer and ‘other’ including non-cancer diagnoses from the respiratory, neurological, digestive, urogenital and gastrointestinal systems. The NIS supplied data on annual income before tax in 1993.
Follow-up
Each person in the study sample was followed from the date of 8 weeks’ sickness absence (in 1994) until 31 December 2003. The follow-up period varied between 3285 days (9 years) and 3679 days for surviving cases not censored by emigration. The mean follow-up was 3237 days for the men and 3337 for women.
Standardized mortality rates
The mortality rate of the county population aged 16–62 at baseline was estimated based on 25% of the deaths in 1994 and all deaths in 1995–2003 based on 5-year age groups. The mortality rate of the study sample was directly age-adjusted to the total population, and comparative mortality figures (CMFs) with 95% CI were calculated for women and men.21 Since several diagnostic subgroups had age groups with no deaths, standardized mortality ratios (SMRs) with 95% CI were calculated for musculoskeletal, mental, cardiovascular, cancer and ‘other’, separately for gender.22,23
Survival analysis
Survival analysis within the study sample was performed with Cox proportional hazard analysis with censoring at emigration or end of follow-up. In order to identify the risk of death related to diagnostic groups, adjustment for age (model 1), and for age and income (model 2) was carried out. In this analysis mental, cardiovascular and ‘other’ were collapsed into one category. The hazard ratio (HR) with 95% CI was estimated for each variable. The variables were treated as categorical because of no obvious linearity. Analyses were carried out separately for genders.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
The distribution of explanatory variables in the study sample and deaths according to diagnosis and socio-demographic variables are shown in table 1 for women and men. The cohort consisted of 53% women, reflecting the higher rate of LTSA among women, even after the exclusion of pregnancy-related cases. The income distribution differed substantially: among the men, 13% earned less than NOK 100 000 per annum (approximately EURO 12 500) compared with 27% of the women, whereas 49% of the men and 17% of the women earned more than NOK 200 000. The majority among both genders had musculoskeletal diagnoses. Among the women 3% and among the men 4% were sickness certified with a main diagnosis indicating cancer.
|
By the end of 2003, 130 men (8.2%) and 67 women (3.7%) in the study sample had died. Among the subjects with a cancer diagnosis, 56.9% of the women and 55.6% of the men had died, making up 43% and 27% of the deaths, respectively.
Table 2 presents the SMRs of the study cohort adjusted to the total county population aged 16–62 years in 1994. The mortality among the county population was 1.8 deaths per 1000 person years for women and 3.5 deaths per 1000 for men. Directly standardized to the county population, mortality rates in the study cohort were 2.7 (95% CI 2.1–3.4) per 1000 person years for women and 7.1 (95% CI 5.7–8.5) for men. CMF was, therefore, 1.5 (95% CI 1.1–1.9) for the female cohort and 2.0 (95% CI 1.7–2.4) for the male cohort. The male/female mortality rate ratio was thus 2.7 among the sickness absentees compared with 1.9 in the total county population.
|
SMRs were 1.6 (95% CI 1.2–2.0) for the women and 1.8 (95% CI 1.5–2.1) for the men. Men with mental and ‘other’ sick-leave diagnoses had significantly elevated SMRs and the same was the case for the whole non-cancer group (not reported in tables). Men sickness certified with musculoskeletal diagnoses had not elevated mortality. Among the women, no diagnostic group except for cancer had elevated SMRs, but several groups had very wide CIs, especially those with cardiovascular diagnoses. Men and women with a sick-leave diagnosis of cancer had SMRs of 8.0 and 16.1, respectively.
The results of Cox's proportional hazard analysis with the diagnosis as the main explanatory variable, adjusted for age in the first model, and for age and income in the second model, is presented in table 3 for women and men. The group diagnosed with cancer had a HR of 18.1 for men and 38.3 for women in the fully adjusted model. The group ‘all other diagnoses’ had an elevated HR among men, but not among women.
|
There was no trend effect of income on mortality risk (model 2). However, in the male sample the HR for the lowest income group was 2.0 (95% 1.2–3.3). Adjustment for income affected the HR for diagnoses and age only marginally.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
Main findings
The present study from Norway confirmed an elevated mortality risk after LTSA, with CMFs of 1.5 (1.1–1.9) for women and 2.0 (1.7–2.4) for men. However, the mortality risk varied considerably with gender, age and not least, with the sick-leave diagnosis. The small group long-term sickness certified with a cancer diagnosis, 3% of the women and 4% of the men, represented 43 and 27% of the deaths, respectively. The other diagnostic groups, with the exception of men with mental or ‘other’ diagnoses, had SMRs that did not differ from the total population of the same age and sex.
Age adjusted male/female mortality rate ratios was higher among the sickness absentees compared with the total population. The reason for this is not clear, but it indicates that men on LTSA might be a special high-risk group, perhaps caused by a higher threshold for taking sick leave. Another explanation is the fact that a larger proportion of women were sickness certified with musculoskeletal diagnoses, which had the best prognosis.
A relatively small group of 204 men with a very low income before inclusion (<100 000 NOK) had an increased mortality after adjustment for age and diagnosis. This might reflect a selection effect of health-determined income reduction and/or less access to effective medical treatment.
Previous studies have indicated that men sickness certified with a mental diagnosis often have a more serious prognosis regarding marginalization than men sickness absent with other diagnoses, and compared with women with mental sick-leave diagnoses.12,24 The elevated SMR for the men with mental sick-leave diagnoses fits this picture.
Strengths of the study
This study is the first analysis of mortality related to sickness absence with diagnosis-specific information at baseline. Since only doctor-certified absence has been associated with increased mortality,3,13 it was not surprising that the content of the medical certificates had an important role. The study was population-based in a large Norwegian county. Mortality data for the whole population and for all exposed individuals (persons on LTSA) in the two inclusion periods were available. In principle, the study included all occupations and there were large samples of both genders. There was a long follow-up period, but no dropouts since all subjects were followed to endpoint or censoring, by means of reliable public registers (NIS, Statistics Norway). In 1998, Feeney and co-workers stated that lack of information on the ‘medical reason’ for sickness absence was an important obstacle in sickness absence research.17 All cases had a main medical diagnosis attached, based on a uniform and well-known system (ICPC). Generally the validity of the diagnoses, which are not formulated as a part of a scientific investigation, may vary with different legislations and especially depending on the length of absence and level of confidentiality.25–27 In a Swedish study, diagnoses on sickness certificates were checked against diagnoses in the medical files of the patients and a high accordance was found.28 The present study was based on ‘extended sickness certificates’ issued at 8 weeks of absence, where the patient's ability to work had been assessed several times by the GP, and there was complete confidentiality towards employers.
Limitations
The main weakness of the study was the small number of deaths when cause-specific analyses were performed. This was especially pronounced among women where three deaths among cardiovascular and six deaths among the mental diagnostic group resulted in wide CIs. However, among women no diagnostic group other than cancer had elevated SMRs and when all non-cancer cases were combined, the SMR was 0.9 with a CI of 0.7–1.3. Among men, there were 15 deaths both in the cardiovascular and in the psychiatric group. The same problem was present for women in the Cox’ regressions, but here mental, cardiovascular and ‘other’ were collapsed into one group.
Several studies, including studies from Norway, have demonstrated socioeconomic gradients in mortality.28,29 Of the socioeconomic variables, only income before tax in 1993 was available. Information on family income, occupational grade or other SES indicators were not available. Several ways of categorizing the income variable, as well as income as a continuous variable, were tested in the regressions, but did not show a significant effect on mortality, except for a small group among the men with very low income. A large proportion of Norwegian women work part time, and this makes it generally problematic to use the income variable as an SES indicator among women.16
The study did not compare mortality rates among sickness absentees with the healthier section, of the employed population, as in the two previous studies.3,13 Instead, people on LTSA were compared with the total population of the same age, which also includes groups outside the labour market, like the unemployed and disability pensioners. This may explain why the majority of the long-term sickness absentees, men and women with musculoskeletal and cardiovascular sick-leave diagnoses, and women with mental sick-leave diagnoses, did not have increased SMRs.
|
Conclusions |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
LTSA is related to the health status and occurrence of diseases among the working population. Indeed, the present study confirmed previous findings of excess mortality after LTSA, also when compared with the total population of the same age. However, the mortality risk varied substantially according to diagnoses, and among women, the cases sickness certified with cancer explained all excess mortality. Among men, also those sickness certified with mental diagnoses and other cases outside the musculoskeletal group contributed, and should be considered at high risk for adverse health outcomes.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
The authors wish to thank the Norwegian National Insurance Administration for providing data. The authors are responsible for the analysis. The study was economically supported by The Swedish Council for Working Life and Social Research and AFA Insurance (Section for Injury Prevention KI), and the Norwegian Research Council (Health Economics Bergen).
Conflicts of interest: None declared.
Key points
|
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
1 Henderson M, Glozier N, Holland Elliott K. Long term sickness absence. Br Med J (2005) 330(7495):802–3.
2 Marmot M, Feeney A, Shipley M, et al. Sickness absence as a measure of health status and functioning: from the UK Whitehall II study. J Epidemiol Community Health (1995) 49(2):124–30.
3 Kivimaki M, Head J, Ferrie JE, et al. Sickness absence as a global measure of health: evidence from mortality in the Whitehall II prospective cohort study. Br Med J (2003) 327(7411):364.
4 Mansson NO, Rastam L. Self-rated health as a predictor of disability pension and death–a prospective study of middle-aged men. Scand J Public Health (2001) 29(2):151–8.
5 Melchior M, Krieger N, Kawachi I, et al. Work factors and occupational class disparities in sickness absence: findings from the GAZEL cohort study. Am J Public Health (2005) 95(7):1206–12.
6 Alexanderson K, Norlund A. Swedish council on technology assessment in health care (SBU). Chapter 12. Future need for research. Scand J Public Health (2004) 63:256–8.[CrossRef]
7 North FM, Syme SL, Feeney A, et al. Psychosocial work environment and sickness absence among British civil servants: the Whitehall II study. Am J Public Health (1996) 86(3):332–40.
8 Vingard E, Alexanderson K, 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) (63):207–15.
9 Gjesdal S, 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(4):246–54.
10 Kivimaki M, Forma P, Wikstrom J, et al. Sickness absence as a risk marker of future disability pension: the 10-town study. J Epidemiol Community Health (2004) 58(8):710–1.
11 Shiels C, Gabbay MB, Ford FM. Patient factors associated with duration of certified sickness absence and transition to long-term incapacity. Br J Gen Pract (2004) 54(499):86–91.[Web of Science][Medline]
12 Gjesdal S, Ringdal PR, Haug K, Maeland JG. 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(4):398–405.
13 Vahtera J, Pentti J, Kivimaki M. Sickness absence as a predictor of mortality among male and female employees. J Epidemiol Community Health (2004) 58(4):321–6.
14 Leijon M, Hensing G, Alexanderson K. Gender trends in sick-listing with musculoskeletal symptoms in a Swedish county during a period of rapid increase in sickness absence. Scand J Soc Med (1998) 26(3):204–13.[Web of Science][Medline]
15 Brage S, Nygard JF, Tellnes G. The gender gap in musculoskeletal-related long-term sickness absence in Norway. Scand J Soc Med (1998) 26(1):34–43.[Web of Science][Medline]
16 Gjesdal S, Bratberg E. The role of gender in long-term sickness absence and transition to permanent disability benefits. Results from a multiregister based, prospective study in Norway 1990–1995. Eur J Public Health (2002) 12(3):180–6.
17 Feeney A, North F, Head J, et al. Socioeconomic and sex differentials in reason for sickness absence from the Whitehall II Study. Occup Environ Med (1998) 55(2):91–8.
18 Brage S, Bentsen BG, Bjerkedal T, et al. ICPC as a standard classification in Norway. Fam Pract (1996) 13(4):391–6.
19 Hofmans-Okkes IM, Lamberts H. The international classification of primary care (ICPC): new applications in research and computer-based patient records in family practice. Fam Pract (1996) 13(3):294–302.
20 Breslow N, Day N. Statistical methods in cancer research. The design and analysis of cohort studies. (1987) Lyon: International agency for research on cancer.
21 Rothman KJ. Modern epidemiology. (1986) 1st edn. Boston/Toronto: Little, Brown and Company.
22 Bland M. An introduction to medical statistics. (1995) 2nd edn. Oxford, New York Tokyo: Oxford University Press.
23 Bambra C, Norman P. What is the association between sickness absence, mortality and morbidity ? Health Place (2006) 12:728–733.[CrossRef][Web of Science][Medline]
24 Hensing G, Alexanderson K, Allebeck P, Bjurulf P. Sick-leave due to psychiatric disorder: higher incidence among women and longer duration for men. Br J Psychiatry (1996) 169(6):740–6.
25 Hussey S, Hoddinott P, Wilson P, et al. Sickness certification system in the United Kingdom: qualitative study of views of general practitioners in Scotland. Br Med J (2004) 328(7431):88.
26 Soderberg E, Alexanderson K. Sickness certification practices of physicians: a review of the literature. Scand J Public Health (2003) 31(6):460–74.[CrossRef][Web of Science][Medline]
27 Stansfeld S, Feeney A, Head J, et al. Sickness absence for psychiatric illness: the Whitehall II Study. Soc Sci Med (1995) 40(2):189–97.[CrossRef][Web of Science][Medline]
28 Claussen B, Davey Smith G, Thelle D. Impact of childhood and adulthood socioeconomic position on cause specific mortality: the Oslo Mortality Study. J Epidemiol Community Health (2003) 57(1):40–5.
29 Ljungdahl LO, 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(3):148–53.[Web of Science][Medline]
30 Sihvonen AP, Kunst AE, Lahelma E, et al. Socioeconomic inequalities in health expectancy in Finland and Norway in the late 1980s. Soc Sci Med (1998) 47(3):303–15.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Melchior, J. E. Ferrie, K. Alexanderson, M. Goldberg, M. Kivimaki, A. Singh-Manoux, J. Vahtera, H. Westerlund, M. Zins, and J. Head Using Sickness Absence Records to Predict Future Depression in a Working Population: Prospective Findings From the GAZEL Cohort Am J Public Health, August 1, 2009; 99(8): 1417 - 1422. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Gjesdal, K. Haug, P. Ringdal, J. G. Maeland, J. Hagberg, T. Roraas, S. E. Vollset, and K. Alexanderson Sickness absence with musculoskeletal or mental diagnoses, transition into disability pension and all-cause mortality: A 9-year prospective cohort study Scand J Public Health, June 1, 2009; 37(4): 387 - 394. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||