The European Journal of Public Health Advance Access originally published online on June 10, 2008
The European Journal of Public Health 2008 18(5):454-459; doi:10.1093/eurpub/ckn049
Social factors and health |
Contextual factors and social consequences of incident disease
Ulla Christensen, Margit Kriegbaum, Charlotte O. Hougaard, Ole S. Mortensen and Finn DiderichsenDepartment of Public Health, Section of Social Medicine, University of Copenhagen, DK-1014 Copenhagen K, Denmark
Correspondence: Ulla Christensen, Department of Public Health, Section of Social Medicine, University of Copenhagen, 5 Øster Farimagsgade, P.O.Box 2099, DK-1014 Copenhagen K, Denmark, tel: +45 35327663, fax: +45 35351181, e-mail: U.Christensen{at}socmed.ku.dk
Received September 20, 2007, accepted May 2, 2008
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAcknowledgementReferences |
|---|
Background: Large geographical variations in the incidence of disability benefits have been reported, but it is unclear to what extent that is confounded by variations in disability rates and disease pattern in the population and whether local variations in rehabilitation and health insurance practice modify the employment effect of disease. We have studied risk of labour market exclusion following incident hospitalization for ischaemic heart disease (IHD), and whether this risk may be modified by contextual factors on the municipal level. Methods: A cohort design on a 10% random sample of the whole Danish population including individuals aged 43–60 years, (n = 516.454 person-years including 840 cases of IHD). The independent variable was incident hospitalization for IHD and outcome variable was defined as job loss 2 years after the event. Regional-level data included all the 275 Danish municipalities in 1996. Results: There was a strong association between incident IHD and labour market exclusion 2 years later, odds ratio (OR) = 2.8 (95% confidence intervals (CI) 2.4–3.4). Men had less risk of being excluded than women and immigrant status, low-educational attainment and co-morbidity were significantly associated with job loss. Also, regional characteristics did independently effect labour market exclusion. However, the individual relative risk of exclusion following incident IHD was not modified substantially when neither the fixed effects of the regional-level variables nor the random effect of municipality was included in the analyses. Conclusion: Geographical variation in incidence of labour market exclusion following incident disease is not primarily an effect of differential social consequences across municipal variations in labour market and socio-economic conditions.
Keywords: contextual factors, incident disease, social consequences
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementReferences |
|---|
Protection against the impoverishing effects of illness and disability was one of the earliest achievements in many welfare states. But improvements in social insurance did not necessarily protect against social exclusion. In recent years policies to promote the social inclusion of people with chronic disease (mental or somatic) or disability are therefore receiving increasing attention, particularly with respect to their integration into working life and return to work after prolonged sickness periods. National, regional and social variations in employment and social consequences of illness are therefore of growing interest, as they might tell us something about how differences in policy and practice influence employment rates among the disabled and whether inequities exist.1–5 An indicator of this, available in many countries, is rates of disability benefits. In universal welfare states such as in the Nordic countries social and regional variations in rates of disability benefits will be the outcome of a combination of variations in (i) morbidity and disability rates, (ii) work demands and access to the labour market for disabled and (iii) effectiveness of rehabilitation.6–7 Consequently, such variations cannot be interpreted directly as inequities in social consequences of ill health. Several studies have documented large geographical variations in the rates of disability pension in the Nordic welfare states. Dellve et al.8 demonstrated large regional variations in the incidence of long-term sick leave and disability pension among 102.715 Swedish home care workers. The Danish National Social Appeals Board showed 4-fold regional differences in the prevalence of disability pension in a nationwide inquiry in year 2000.9 Krokstad et al.10 showed how the prevalence of disability pension in a Norwegian county varied between the different municipalities according to their level of relative deprivation. Ydreborg and Ekberg11 showed considerable variations in the rate of rejection of applications for disability pension between Social Insurance boards in different geographical areas in a Swedish county, and similar findings have been published by Brun from Denmark.12
The effect of local variations in rehabilitation practice and insurance management will be confounded by variations in disability rates and diagnostic case mix as well as variations in local labour market structure. If only rates of disability pension are used as an indicator of employment consequences, bias will occur due to the fact that some local authorities will be more restrictive, or they may let people stay for longer periods on sickness or unemployment benefits. An alternative is therefore to study the employment consequences of specific diseases.13,14 However, to our knowledge only two studies have examined regional variations in the effect of a specific disease on the risk of labour market exclusion, and both studies examined the incidence of disability pension for psychiatric diagnosis.15,16 Neither of these studies used data on incident cases of disease, nor did they include regional-level data on socio-economic factors in the analyses.
The aim of this study is therefore to analyse the risk of labour market exclusion following incident hospitalization for ischaemic heart disease (IHD) and how this risk of exclusion may be modified by contextual factors on a municipal level.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementReferences |
|---|
Study population/material
This study is based on data from a longitudinal register maintained by the Institute of Local Government Studies (AKF) in Denmark. The register comprises 10% of the Danish population aged 15 years or older by 1 January 1981 (N = 408 000) and has been updated annually with a new cohort of 15-year olds to maintain its external validity over time. The register contains data regarding birth, sex, occupation, education, employment status and annual income. Data have been linked to the national register of hospital discharges that cover all in-patients and include data about diagnosis at discharge. IHD was classified according to the 8th Revision of the International Classification of Disease (ICD8) for the years 1981–93, admission diagnosis 410 and 411 and from 1994 according to the 10th Revision (ICD10), admission diagnosis I 20.0; I 21; I 22.
The present study used a cohort design with a series of cohorts defined by each of the years 1996–2000. Each index year (x) was defined as the year when the incident IHD took place. The study base for the single level logistic models included individuals older than 43 but younger than 60 years in the index year. Individuals who had been admitted for IHD before the index year were excluded (data before 1981 not available). The study population represented 516.454 person-years.
Measures
The outcome variable was defined as being out of employment 2 years after incident IHD, i.e. unemployment or living on transfer income (receiving disability pension, prolonged sickness benefits and welfare).
Regional-level data on the 275 Danish municipalities were obtained from Statistics Denmark covering the total Danish population aged 16–66 years in 1996–2000. Data were weighed with population size in each municipality and the variables included: average annual household income equalized by household composition; unemployment rate, employment rate; proportion receiving sickness benefits, proportion of those receiving sickness benefits who were previously unemployed; proportion of non-western immigrants. All these regional-level characteristics were categorized in quartiles. Finally, data on duration for the considerations of applications for disability pensions (dichotomized into 
6 months and >6 months) were obtained from the National Social Appeals Board in Denmark.
In addition to gender and age (dichotomized in the multilevel analysis into <55 and 
55 years), the following individual-level variables were used in the analyses: marital status—(i) married, (ii) widowed and divorced, (iii) never married; country of birth—(i) Danish born, (ii) born abroad; education—categorized as (i) low (primary education 10 years), (ii) medium (11–12 years), (iii) short further (theoretical training lower than bachelor degree), (iv) long further (equivalent of a university degree). Co-morbidity was included as any other somatic or psychiatric hospitalization in the past 2 years before the index year.
Data analysis
All analyses had to be conducted within the data system established by Statistics Denmark. While the study population for the single level logistic regression analysis represented 516.454 person-years, the size of our population sample for the multilevel modelling was restricted by Statistics Denmark. Thus, the population for the multilevel analyses represented all IHD cases and a random 30% sample of the rest of the total population in the longitudinal register maintained by the Institute of Local Government Studies (minus the IHD cases), n = 128.603 person-years. All analyses were performed in SAS, version 9.1. For the multilevel analyses the SAS Glimmix macro was used.
The logistic regression analysis was conducted in a forward stepwise manner in order to test the potential mediating role of the individual-level covariates. In order to test how the regional level influenced the risk of labour market exclusion multilevel logistic regression analyses were performed. Model 1 (table 1) was based on the final model of the single level regression analysis (table 2) with the addition of the regional level (municipality) calculated as a random intercept model. The risk of labour market exclusion at the individual level was estimated among those with IHD referred to those without IHD. Model 2 further tested the fixed effects of the regional municipal characteristics one by one. Model 3 included a full model of all regional municipal characteristics in addition to the individual level and the random intercept effect, municipality. Estimates of the random variance component were converted into median odds ratios (MOR).17,18 MOR expresses the median in the distribution of OR arising when repeatedly choosing two individuals with the same covariates from different clusters (i.e. different municipalities) and quoting the high-risk subject versus the low-risk subject. In this way, the size of the variance components may be directly compared to the effect of the individual level covariates in terms of OR. Since MOR is a median of numerous OR it is not meaningful to calculate confidence intervals (CI).
|
|
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementReferences |
|---|
Table 3 shows that among the 840 cases of incident IHD 16.1% of men and 23.9% of women were out of employment 2 years later. Only 6.1% of the men and 8.2% of the women in the total population were excluded from the labour market in the same periods (P < 0.0001). Also, there was a significant difference in the prevalence of co-morbidity 2 years prior to the index year between cases of IHD and the total population, (P < 0.0001). Unadjusted OR for labour market exclusion following incident hospitalization for IHD was 2.84 (CI 2.40–3.40) among men and women (table 2), but men had less risk of being excluded than women OR = 0.71 (CI 0.70–0.73). Marital status had an independent effect on the risk of exclusion, showing an increased risk among never married, widowed and divorced. Addition of immigrant status to the regression model attenuated the association between IHD and subsequent labour market exclusion and showed that immigrants had a double risk of labour market exclusion compared to Danish born, OR = 2.06 (CI 1.97–2.16). Likewise, low-educational level was significantly associated with unemployment as was co-morbidity 2 years before the index year. Both variables further attenuated the association between incident IHD and exclusion and the full model showed an OR of 2.52 (95% CI 2.10–3.03).
|
On the contextual level descriptive statistics for the period 1996–2000 showed huge variations among the 275 Danish municipalities for all the area-level characteristics included (table 4). However, adjusting for the proportion receiving sickness benefit and the proportion of non-western immigrants showed no effect on the estimate of labour market exclusion in the multilevel analyses and was not included.
|
Table 1 shows the multilevel logistic regression analyses of labour market exclusion in the period 1998–2002, following incident IHD 2 years previously. The individual relative risk of exclusion following incident IHD did not change substantially from the final logistic regression model in table 2 to the first multilevel model when the random effect of municipality was included. We did find independent effects of the regional municipal characteristics on labour market exclusion (Model 2). Thus, there was a graded effect along the quartiles of equalized annual household income and an effect of high-unemployment rate. Also, consideration time of more than 6 months for granting disability pension showed an OR of 1.24 (95% CI 1.17–1.32). This estimate was not attenuated in the full model. However, the relative risk of labour market exclusion following incident IHD at the individual level was not modified substantially when specific characteristics of the municipality context was included in the multilevel analyses. The random effect of municipality on labour market exclusion showed a MOR = 1.22 in Model 1, i.e. the MOR for exclusion between two individuals chosen at random from two different municipalities was 1.22. Thus, there was a significant but modest difference between the risk of being excluded from the labour market among the municipalities in this study, P < 0.0001.
The random effect of municipality was reduced when equalized household income, unemployment and employment rate was included in the model. The full model showed a MOR = 1.17, (P < 0.0001). Interaction between the effect of IHD on job-loss and the municipal characteristics, measured as departure from additivity was calculated as synergy indexes.19 The index showed an interaction effect of 1.85 (95% CI 0.97–3.49) between low-employment rate and job-loss when not adjusted for the other municipal characteristics. However, this effect was attenuated when all the characteristics were included in the model (1.53, CI 0.79–2.95). There was no interaction effect when calculating the synergy index for the other characteristics (data not shown).
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementReferences |
|---|
This study showed a strong association between incident IHD and labour market exclusion 2 years later. Men had less risk of being excluded than women. Immigrant status, low-educational attainment and co-morbidity did attenuate the effect of IHD on exclusion. These results are consistent with other studies which have shown both female sex and socio-economic indicators at the individual level to be associated with labour market exclusion, not only following longstanding illness in general,20–22 but also after specific conditions such as acute myocardial infarction and even heart transplantation.23–25 Additionally, our study also demonstrated an effect of co-morbidity, measured by somatic or psychiatric hospitalization 2 years prior to the index year. We may even have underestimated this effect as hospital discharges does not catch many disabling health problems.
The multilevel analyses did show a significant difference in the risk of being excluded between the Danish municipalities in the study. Still, the individual relative risk of exclusion following incident IHD was not modified substantially when the random effect of municipality was included. Based on an ecological deprivation index for 24 municipalities in a Norwegian county, including indicators on mean income, prevalence of disability, income and unemployment, Krokstad et al.10 found that people living in relatively deprived municipalities had an increased risk of receiving a disability pension in a 10-year follow-up period, adjusted for individual determinants. Our study also demonstrated significant variations between the municipalities in the risk of becoming either unemployed or living on transfer income. Still, our results does not correspond with the Norwegian findings as the geographical variation in incidence of labour market exclusion following disease was not found to be an effect of differential social consequences across municipal levels of affluence. One possible explanation of this divergence is the difference in the health indicators used in the studies. The indicators used in the study by Krokstad et al. were self-reported long-standing health problems. This may include a broad range of various conditions with very different symptoms and functional limitations, and the result could therefore be confounded by diagnosis. Differences between the two countries in terms of health insurance practice might also be of major importance, as illustrated by the strong variations in sickness absence between the two countries.10
In the present study we used register-based information on only one disease, incident hospitalization for IHD, and we regard specificity in disease measurement as high, and the degree of misclassification and selection bias must be seen as very limited. The hospital database has full coverage of all public hospitals (with <1% admission to private hospitals). However, there might still be some misclassification of employment status and residual confounding in the data. For example, the fact that some municipalities had no individuals registered as having received sickness benefits after unemployment is most likely due to registration error. Thus, we may have underestimated the effect of the municipal characteristics. Other factors not included in this study may have had an effect such as variations in job characteristics in terms of physical and mental demands, or differences in rehabilitation practice. There are however a lack of data available on these dimensions. These limitations should be addressed in further studies using a multi-method approach to include more information on local policies and practices.
|
Acknowledgement |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementReferences |
|---|
Financial funding for the Danish Longitudinal Study on Work, Unemployment and health was received from the Danish Research Councils (j.nr. 9801268).
Conflicts of interest: None declared.
Key points
|
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementReferences |
|---|
1 Bartley M, Owen C. Relation between socioeconomic status, employment, and health during economic change, 1973–93. Br Med J (1996) 313:445–9.
2 Arrow JO. Estimating the influence of health as a risk factor on unemployment: a survival analysis of employment duration for workers surveyed in the German Socio-Economic Panel. Soc Sci Med (1996) 42:1651–9.[CrossRef][Web of Science][Medline]
3 Van de Mheen H, Stronks K, Schrijvers CTM, Mackenbach JP. The influence of adult ill health on occupational class mobility and mobility out of and into employment in the Netherlands. Soc Sci Med (1999) 49:509–18.[CrossRef][Web of Science][Medline]
4 Lindholm C, Burström B, Diderichsen F. Does chronic illness cause adverse social and economic consequences among Swedes? Scand J Public Health (2001) 29:63–70.
5 Buchardt T. Being and becoming: Social exclusion and the onset of disability. In: CASE Report 21. (2003).
6 Jensen PH. Disability pensions: trends and Policies in Denmark. In: European disability pension policies. 11 Country trends 1970–2002.—Prinx CH, ed. (2003) European Centre Vienna: Ashgate Publishing Company. 139–64.
7 Stattin M. Retirement on grounds of ill health. Occup Environ Med (2005) 62:135–40.
8 Dellve L, Karlberg C, Allebeck P, et al. Macro-organizational factors, the incidence of work disability, and work ability among the total workforce of home care workers in Sweden. Scand J Public Health (2006) 34:17–25.
9 National Social Appeals Board. Førtidspensioner. Årsstatistik 2000 (Disability pensions, annual statistics 2000). (2001) Copenhagen: Department of Analysis.
10 Krokstad S, Magnus P, Skrondal A, Westin S. The importance of social characteristics of communities for the medically based disability pension. Eur J Public Health (2004) 14:406–12.
11 Ydreborg BA, Ekberg K. Disqualified for a disability pension – case/referent study. Disabil Rehabil (2004) 26:1079–86.[CrossRef][Web of Science][Medline]
12 Brun C, Boggild H, Eshøj P. Socioøkonomiske risikoindikatorer for førtidspension blandt erhvervsaktive danskere. [Socioeconomic risk indicators for disability pension within the Danish workforce. A registry-based cohort study of the period 1994–1998.]. Ugeskr Laeger (2003) 165:3315–9.[Web of Science][Medline]
13 Lindholm C, Diderichsen F, Burström B. Social consequences of ill health in a welfare state: differences across diagnostic groups. In: PhD Thesis. (2002) Stockholm: Karolinska Instituttet.
14 Bradley CJ, Bednarek HL, Neumark D. Breast cancer survival, work, and earnings. J Health Econ (2002) 21:757–79.[CrossRef][Web of Science][Medline]
15 Andersen L, Staland Nyman C, Spak F, Hensing G. High incidence of disability pension with a psychiatric diagnosis in western Sweden. A population-based study from 1980 to 1998. Work (2006) 26:342–52.
16 Andersen L, Wiles N, Lewis G, et al. Disability pension for psychiatric disorders: regional differences in Norway 1988–2000. Nord J Psychiatry (2006) 60:255–62.[CrossRef][Web of Science][Medline]
17 Larsen K, Petersen JH, Budtz-Jørgensen E, Endahl L. Interpreting parameters in the logistic regression model with random effects. Biometrics (2000) 56:909–14.[CrossRef][Web of Science][Medline]
18 Larsen K, Merlo J. Appropriate assessment of neighborhood effects on individual health: integrating random and fixed effects in multilevel logistic regression. Am J Epidemiol (2005) 161:81–8.
19 Rothman KJ. Epidemiology. An introduction. (2002) Oxford: Oxford University Press. 168–80.
20 Krokstad S, Westin S. Disability in society – medical and non-medical determinants for disability pension in a Norwegian total county population study. Soc Sci Med (2004) 58:1837–48.[CrossRef][Web of Science][Medline]
21 Lindholm C, Burström B, Diderichsen F. Class differences in the social consequences of illness? J Epidemiol Community Health (2002) 56:188–92.
22 Edén L, Ejlertsson G, Lamberger B, et al. Immigration and socio-economy as predictors of early retirement pensions. Scand J Soc Med (1994) 22:187–93.[Web of Science][Medline]
23 Nielsen FE, Sørensen HT, Skagen K. A prospective study found impaired left ventricular function predicted job retirement after acute myocardial infarction. J Clin Epidemiol (2004) 57:833–42.
24 Nielsen FE. Erhvervsprognosen efter akut myocardieinfarct. (Employment prognosis after acute myocardial infarction, dissertation in Danish). (2001) Copenhagen: University of Copenhagen.
25 Paris W, Woodbury A, Thompson S, et al. Returning to work after heart transplantation. Transplant (1993) 12:46–54.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||