You are here

  • Home
  • Archive
  • Volume 70, Issue 4
  • Educational differentials in disability vary across and within welfare regimes: a comparison of 26 European countries in 2009

Article Text

Article menu

Download PDFPDF

Disability
Educational differentials in disability vary across and within welfare regimes: a comparison of 26 European countries in 2009
  1. Emmanuelle Cambois1,
  2. Aïda Solé-Auró2,
  3. Henrik Brønnum-Hansen3,
  4. Viviana Egidi4,
  5. Carol Jagger5,
  6. Bernard Jeune6,
  7. Wilma J Nusselder7,
  8. Herman Van Oyen8,
  9. Chris White9,
  10. Jean-Marie Robine10
  1. 1Department of Mortality, Health and Epidemiology, Institut National d'Etudes Démographiques (INED), Paris, France
  2. 2Department of Political and Social Science, Universitat Pompeu Fabra (UPF), Barcelona, Spain
  3. 3Department of Public Health, University of Copenhagen, Faculty of Health Sciences, Copenhagen, Denmark
  4. 4Department of Statistical Science, Sapienza University of Rome, Roma, Italy
  5. 5Newcastle University Institute for Ageing and Institute of Health & Society, Campus for Ageing and Vitality, Newcastle upon Tyne, UK
  6. 6Department of Epidemiology, Institute of Public Health, and Danish Ageing Research Center, University of Southern Denmark, Odense, Denmark
  7. 7Department of Public Health, Erasmus MC, University Medical Center Rotterdam, Rotterdam, The Netherlands
  8. 8Department of Public Health and Surveillance, Scientific Institute of Public Health, Brussels, Belgium
  9. 9Government Statistical Service, Office for National Statistics, Government Buildings, Newport, UK
  10. 10Institut National de la Santé et de la Recherche Médicale (INSERM), Ecole Pratique des Hautes Etudes (EPHE) and Institut National d'Etudes Démographiques (INED), Montpellier, France
  1. Correspondence to Dr Emmanuelle Cambois, Department of Mortality, Health and Epidemiology, Institut National d'Etudes Démographiques (INED), 133 Bd Davout, Paris 75980, Cedex 20, France; cambois{at}ined.fr

Abstract

Background Social differentials in disability prevalence exist in all European countries, but their scale varies markedly. To improve understanding of this variation, the article focuses on each end of the social gradient. It compares the extent of the higher disability prevalence in low social groups (referred to as disability disadvantage) and of the lower prevalence in high social groups (disability advantage); country-specific advantages/disadvantages are discussed regarding the possible influence of welfare regimes.

Methods Cross-sectional disability prevalence is measured by longstanding health-related activity limitation (AL) in the 2009 European Statistics on Income and Living Conditions (EU-SILC) across 26 countries classified into four welfare regime groups. Logistic models adjusted by country, age and sex (in all 30–79 years and in three age-bands) measured the country-specific ORs across education, representing the AL-disadvantage of low-educated and AL-advantage of high-educated groups relative to middle-educated groups.

Results The relative AL-disadvantage of the low-educated groups was small in Sweden (eg, 1.2 (1.0–1.4)), Finland, Romania, Bulgaria and Spain (youngest age-band), but was large in the Czech Republic (eg, 1.9 (1.7–2.2)), Denmark, Belgium, Italy and Hungary. The high-educated groups had a small relative AL-advantage in Denmark (eg, 0.9 (0.8–1.1)), but a large AL-advantage in Lithuania (eg, 0.5 (0.4–0.6)), half of the Baltic and Eastern European countries, Norway and Germany (youngest age-band). There were notable differences within welfare regime groups.

Conclusions The country-specific disability advantages/disadvantages across educational groups identified here could help to identify determining factors and the efficiency of national policies implemented to tackle social differentials in health.

  • AGEING
  • Functioning and disability
  • SOCIAL INEQUALITIES
  • POLICY
  • DEMOGRAPHY

https://doi.org/10.1136/jech-2015-205978

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.

    Introduction

    While longevity increased in most European countries, a significant part of life is still lived with diseases and disability, with large variations across Europe1 ,2 and between socioeconomic groups.3–5 Actions to reduce these disability differentials as a means to increase healthy ageing have become important public health goals.6–10

    Disability results from health events, such as chronic diseases, that worsen bodily functions and hamper the performance of activities, thereby challenging social participation and quality of life.11 Disability arises from exposure to health events as well as from the (non)availability of resources to individuals to help them adapt to declining functions (assistive devices, care giving, adapted environment).12 Therefore, differentials by socioeconomic status (SES) in disability prevalence stem from complex interactions between the individual, and their household and country characteristics.7 ,13 Individuals vary in their exposures to harmful life and work conditions or behaviours, their ability to adjust to functional disorders, and their access to environmental adaptations and assistive devices or care. National contexts modify the impact of individual characteristics on disability risks7 ,14 generally through the: (1) availability and quality of care, primary prevention and protection programmes throughout the nation; (2) social welfare context,15 ,16 defining the level of social transfers, access to education and to care (child, medical, elderly), and the priority given to disability policies to facilitate adjustments.

    Comparing disability differentials across countries highlights (un)favourable country contexts. Several studies have discussed the SES differentials in health across Europe, based on classifications of welfare regimes:4 ,17–20 they showed large variations in the magnitude of the differentials across countries, but the relationship with welfare regimes was unclear. A high level of social transfers is expected to reduce exposure to deprivation and related disability risks, translating into reductions in the health disadvantage of low-SES groups. However, whether such regimes consistently reduce this disadvantage, and whether other regimes do not, is uncertain. Moreover, to what extent the different regimes benefit high-SES groups is also unknown.21 These questions suggest a need to further disentangle how each extreme of the SES scale is affected by country-specific circumstances, for instance by enabling the high-SES groups to gain the most and/or the low-SES groups to gain the least (than the average SES effect).

    Welfare regimes can be defined by the degree to which people rely on the labour market, family support, or social transfers to get resources and cover basic needs.18 Four aggregated groups of welfare regimes were considered.4 ,19 ,20 The social democratic regimes of Nordic countries have high levels of social transfers, which should lower health risks associated with deprivation and thereby reduce the disability disadvantage in the low-SES groups. In contrast, the Beveridgian and Bismarkian regimes of Western and Southern European countries correspond to a larger dependency on the labour market, with different levels of social transfers and healthcare systems.22 Low social transfers and uneven care access could exacerbate inequalities in exposures, and in ability to afford care; they could increase the disadvantage of low-SES groups and/or the advantage of high-SES groups. Western and Southern European countries are examined separately; the higher reliance on family support in the latter might lower disability disadvantages, if informal care giving compensates for unmet needs. Finally, the move of Eastern European and Baltic countries from centralised state control of production to a market economy has resulted in improved healthcare systems, but with an increased share of private expenses;23 these changes are likely to increase advantages as well as increase disadvantages of high-SES and low-SES groups, but in what proportions is unclear.

    In this context, we analysed the variation across 26 European countries in the extent of excess disability prevalence of low-SES groups (disability disadvantage) and reduced disability prevalence of high-SES groups (disability advantage). The aim was to identify country-specific patterns deviating from the average pattern. We refer to the welfare regimes in line with previous studies,4 ,19 ,20 to highlight similarities and differences.

    Methods

    Data

    The ‘European Union Statistics on Income and Living Conditions’ (EU-SILC) is a database monitored by the national statistical offices, designed to provide comparable data across the EU. We used the 2009 EU-SILC cross-sectional data. In most countries, data are collected by ad hoc interview surveys, providing self-reported health and SES variables. Elsewhere, sociodemographic variables are collected through population registers; self-reported health being collected by a complementary survey, often using telephone interviews. We examined sample selection, survey designs, collection mode and question wording to ensure comparability (see online supplementary box S1 and table S1). Owing to varying response rates, we assessed the representativeness of country samples regarding the distributions of age, occupation and education. We subsequently excluded Iceland, Luxembourg and Malta, and recommend caution for a number of other countries (see Discussion section). We excluded individuals aged 80 years and over due to missing information. Our study comprises 290 521 individuals aged 30–79 years from 26 countries (table 1).

    View this table:
    Table 1

    Country description: European Statistics on Income and Living Conditions (EU-SILC) participation rate, sample sizes for ages 30–79 years, distribution across educational groups, prevalence of activity limitation (AL) and differentials (2009)

    Disability and education measurements

    Disability is based on the Global Activity Limitation Indicator measuring health-related activity limitation (AL) with a single question: ‘For at least the past 6 months, to what extent have you been limited because of a health problem in activities people usually do?’ (severely limited; limited but not severely vs not limited). AL is self-reported and so varies across European countries, partly due to varying propensity to report health problems.24 ,25 However, AL is consistently correlated with more detailed measurement instruments for disability,24 ,26 ,27 and it is predictive of mortality28 and of consumption of care services.29

    Education, a common proxy for SES, is strongly related to health and disability risk through a variety of pathways; specifically early life circumstances, household circumstances, job opportunities, and the development of skills to maintain health and to adjust to health problems.30 ,31 We considered three groups based on the level of education achieved, using the International Standard Classification of Education:i low (0–2 primary and lower secondary education), middle (3–4 upper secondary education) and high (5–6 tertiary education).

    Data analysis

    We examined the prevalence of AL by country and education, across the 30–79 years of age range as a whole and in three age-bands (30–49, 50–64 and 65–79 years of age) to highlight changes between birth cohorts. Prevalence is standardised to the pooled weighted sample population by the 5-year age group.

    The relative AL advantages and disadvantages of the high-educated and low-educated groups are assessed using logistic regression models, pooling the data from the 26 countries.ii We estimated ORs for AL using a country by education interaction term, with the middle-educated group as reference. The model is adjusted for age, sex and country (to account for the country variation in the level of AL). From this model, we derived the 26 country-specific predictive margins for the three educational groups, and estimated the (unweighted) predictive margins for the all-countries average (see online supplementary table S2); we obtained the country-specific and all-countries average ORs of AL for the high-educated and low-educated groups, related to the middle-educated group. The country-specific ORs represent the relative disability advantage of being in a high-educated group and disadvantage of being in a low-educated group in a country, which can then be compared to the average pattern.iii The model was run for the 30–79 years of age range, and then separately for the three age-bands.

    Results

    Disability prevalence across educational groups

    The age-standardised prevalence of AL varies across the 26 countries, both within the four welfare regime groups (table 1) and by age-band (figure 1). In figure 1, we represented the relative proportion of the educational groups within the populations by the size of the circles. Thus in Nordic and Western countries, low-educated groups are generally larger in the oldest age-band than in the youngest, in contrast to Eastern and Southern countries, justifying the analyses for each age-band.

    Figure 1
    Figure 1

    Age-standardised* prevalence of activity limitation (AL) in low-educated, middle-educated and high-educated groups and the size of the educational group by age group (represented by the size of the circles)—26 European countries by welfare regime (Nordic, Western, Southern, Eastern and Baltic countries) in 2009, and 95% CIs.

    *Standardised by the 5-year age group distribution of the pooled weighted sample, ordered by AL in middle-educated group in the age group of 30–79 years age-band. Note: Country labels—Austria (AT), Belgium (BE), Bulgaria (BG), Cyprus (CY), the Czech Republic (CZ), Germany (DE), Denmark (DK), Estonia (EE), Finland (FI), France (FR), Greece (GR), Hungary (HU), Ireland (IE), Italy (IT), Latvia (LV), Lithuania (LT), the Netherlands (NL), Norway (NO), Poland (PL), Portugal (PT), Romania (RO), Slovakia (SK), Slovenia (SI), Spain (ES), Sweden (SE), United Kingdom (UK).

    Low-educated groups consistently show the highest AL prevalence and high-educated groups the lowest prevalence, although the gap differs between countries and age-bands. There is no evidence of a systematic relationship between the relative size of the groups and the size of the differentials. Based on prevalence, no consistent pattern was found within welfare regime groups regarding the magnitude of the advantage/disadvantage across educational groups, partly due to the different levels of AL.

    Disability advantages and disadvantages

    Figure 2 shows the country-specific ORs of AL, with their 95% CIs and the average AL advantage/disadvantage plotted in dotted lines (0.65 and 1.48, respectively, in the 30–79 years of age group). Figure 2 shows where the AL-advantage/disadvantage is greater or smaller compared to that expected based on the average effect of being in a high-educated or low-educated group (controlling for country-specific level of AL, age and sex). A number of countries deviated from the average pattern. More specifically, two of the four Nordic countries, Sweden and Finland, showed a significantly smaller AL-disadvantage relative to the average for the low-educated groups, in the 30–79 years of age group (and across age-bands although not statistically significant). However, the disadvantage was significantly larger in Norway and Denmark, compared to Sweden and Finland (see also online supplementary table S3). In addition, the relative AL-advantage for the high-educated groups was smaller in Denmark, but larger than average in Norway, and more pronounced than in most other countries (see online supplementary table S3).

    Figure 2
    Figure 2

    Country-specific and average ORs of activity limitation (AL) associated to high-educated and low-educated groups, compared to middle-educated group (after adjustment for country, sex and age)+ by age group and welfare regime (Nordic, Western, Southern, Eastern and Baltic countries).

    +Model: Embedded Image with X1i for Country i=(1-26); X2i for Age#Country i=(1-26); X3ij for Sex#Country i=(1-26) and j=(1-2); X4iz for Education#Country i=(1-26) and z=(1-3)

    Average ORs using average predicted margins [pHigh/(1−pHigh)]/[pMiddle/(1−pMiddle)]. CIs for the average ORs are computed based on the sum of the country-specific variance (see online supplementary table S2).

    Note: Country labels—Austria (AT), Belgium (BE), Bulgaria (BG), Cyprus (CY), the Czech Republic (CZ), Germany (DE), Denmark (DK), Estonia (EE), Finland (FI), France (FR), Greece (GR), Hungary (HU), Ireland (IE), Italy (IT), Latvia (LV), Lithuania (LT), the Netherlands (NL), Norway (NO), Poland (PL), Portugal (PT), Romania (RO), Slovakia (SK), Slovenia (SI), Spain (ES), Sweden (SE), United Kingdom (UK).

    The relative AL-advantage/disadvantage patterns were generally similar across Western and Southern countries, and were close to the average. However, there are some exceptions. In the 30–79 years of age group, low-educated Belgians and Italians experience a larger AL-disadvantage compared to the average. In the youngest age group (30–49 years), the low-educated Italians and Spaniards experienced a smaller AL-disadvantage (only significant for Spaniards). The youngest high-educated Germans had a larger advantage.

    In Baltic and Eastern European countries, a similar advantage/disadvantage pattern is found across half of the countries. There was a larger AL-advantage for the high-educated group in the Czech Republic, Estonia, Hungary, Lithuania and Romania; but smaller among oldest Poles. The AL-disadvantage was also larger among low-educated Czechs and Hungarians, but smaller among Romanians and the oldest Bulgarians.

    Discussion

    Summary of findings and comparison with other studies

    Controlling for the variation in the prevalence of AL across countries, our results identified countries where the AL-advantage and disadvantage across educational groups deviate from the average pattern; we found intuitive links to specific welfare regimes, although not systematically.

    In line with previous studies, the Nordic countries did not conform to a common pattern, with Sweden generally showing small health differentials and Norway, large ones.4 ,19 ,32 ,33 The small disability differential in Sweden results from an AL-disadvantage of the low-educated group (relative to middle-educated), which is smaller than the average; the same is found in Finland. The larger differential in Norway results from the larger AL-advantage of the high-educated group and rather large AL-disadvantage of the low-educated group. In Denmark, the rather large AL-disadvantage for the low-educated group combined with an unexpectedly smaller advantage of the high-educated group (relative to the middle-educated group). Beyond the common protective and redistributive policies in these countries, various features of the national contexts (such as health systems, income variation or health-related practices) are likely to affect the level of self-reported AL as well as the protective and hazardous exposures for the low-educated and high-educated groups. Further focus on the specific cases that deviate from the average is therefore justified, in order to better understand how these differentials arise and are maintained. For instance, it would be of interest to explore the reduced relative AL-advantage of the high-educated Danes since this might arise from greater similarity in the health advantages of middle-educated and high-educated groups while the low-educated group lag behind, or to a higher level of unfavourable exposures relative to other high-educated Europeans. Interestingly, this could be related to the small difference in tobacco consumption between educational groups (which is linked to disability) in Denmark.4 ,34–36 Additionally, the increased disability advantage of high-educated Norwegians may be related to the larger income inequalities that affect health outcomes in that country32 (relative to other countries) as well as a larger private share of health expenses.22 Regarding disability, the greater advantage of high-educated groups may be related to differences in individual resources for adjusting life and work conditions to mitigate the disabling effects of functional limitations.

    Although the patterns were more homogeneous within the three other welfare regime groups, there were, again, exceptions. Low-educated groups experience rather large relative AL-disadvantage in Belgium, Italy, the Czech Republic and Hungary. This pattern may be influenced by a combination of lower social transfers and private grounded health systems that act to increase unmet needs.18 The high-educated groups experience larger relative disability advantage in half of the Baltic and Eastern European countries and for the youngest age-band in Germany: more selected access to healthcare, protection and prevention programmes for the better-off groups may be driving this imbalance.23 In the Czech Republic and Hungary, the educational gradient in disability is stretched at both ends. How this situation arose in these countries, where the AL prevalence is often high, is important for understanding current (and future) SES health differentials.

    Our study showed varying patterns across birth cohorts, in line with earlier findings.32 ,37 The rather small health differentials found in Germany in previous studies4 ,19 could be disproportionately affected by the oldest generations, since in the youngest age group a larger advantage for the high-educated group compared to the average was found. We also found a reduced AL-disadvantage in Spain in the youngest cohorts (almost significant in Italy). It may be that family support and informal care giving in these countries possibly limit the effects of deprivation, reducing unmet needs. However, this pattern was not found in the oldest cohort, which had a higher prevalence of AL and related need for care. Assessing cohort variation is not straightforward, especially when due to mortality selection effects, but our results confirm the need to investigate them further.

    The variation in the AL-advantage/disadvantage across educational groups indicates where national contexts differently affect the various SES groups. Whether this variation is due to specific policy actions that result in uneven health returns for the different SES groups is of interest; particularly those policies that reduce SES health gaps (universal policies, proportional universalism, targeted policies).38 Such policies can modify the level of access to care, health-related practices or socioeconomic circumstances of low-educated and high-educated groups, contributing to reducing SES differentials.20 ,32 Socioeconomic variables from EU-SILC (such as income, occupation or employment status) could be further used to assess their contribution to the AL advantage/disadvantage of educational groups—although their level of comparability is also challenging. In parallel, macroeconomic variables could be considered as possibly modifying the extent of the AL-advantage/disadvantage33 (economic development, social transfers, health funding systems and disability policies).

    Limitations of the data

    The EU-SILC data set provides disability data for a very large sample of European countries. However, there are a number of limitations to consider. After stratification by age group and education, a country sample is limited in size, and for this reason we did not repeat analyses by sex; this lack of precision limits the scope to detect statistical significance and, therefore, inferential interpretation of country patterns. Furthermore, the large number of estimates implies a risk of type I statistical errors among our results; although part of our findings are consistent with the literature, others need to be explored in new studies and using other data sets, to deepen the explanatory part of the analysis.

    The comparability of the data sets is generally an issue in international studies and we addressed this issue. Facing a varying response rates, we highlighted where there was good representativeness of sample, then excluded a number of countries (see online supplementary box S1 and table S1). We included Slovakia, the UK and Sweden, despite a slight under-representation of the low-educated group in these samples; this carries a risk of underestimation of their AL-disadvantage (poor health being associated with non-participation).39 Regarding comparability of the wording, AL in EU-SILC is harmonised for most countries, however, some differences persist (see online supplementary box S1); for instance, the Bulgarian question refers to ‘activity limitations at work’, which might orient the respondent's answer and induce different patterns compared to other countries.

    More generally, AL is self-reported, which may result in variations in the propensity to report disability. The wording of the question, the mode of data collection and the cultural perception of health might affect the reported prevalence. Disability indicators are usually less sensitive to health perception than self-perceived health or diseases, but we cannot distinguish cultural differences although our model adjusts for country levels.27 ,40

    The comparison of health differentials between educational groups requires caution due to the varying meaning (and coding) of educational levels and the changing relative size of the groups (figure 1). We did not account for the proportion of the groups in the models, to control for possible selection effects (the smaller a group, possibly the more it was selected on health-related and socioeconomic-related characteristics) as there was no obvious systematic pattern between the size of the group and the level of advantage/disadvantage. As suggested earlier, further adjusting for a number of socioeconomic characteristics of the groups might be more effective in explaining the different meanings of educational levels.

    Conclusion

    Despite these limitations, this paper brings novel results through exploration of the extent to which disability patterns for low-educated and high-educated groups vary across countries. We found some unsurprising results: a reduced disability disadvantage of low-educated groups in two Nordic countries, probably benefitting from the protective policies and publicly grounded health systems; a larger advantage of high-educated groups in Baltic and Eastern European countries, where more privately grounded policies and health systems might translate into a general advantage of the high-educated group. But we also found inconsistency within welfare regime groups. These results confirm the need for refining policy contexts in countries to better understand the role of specific schemes on SES differentials in disability. Repeated with other data sets, and further enriched by qualitative indicators on the country context, our findings could contribute to the debate on which policy responses are needed to reduce disability inequalities. Depending on whether high-educated groups progress faster and/or low-educated groups lag behind, our approach could help policy makers to make decisions on the relative benefits of increasing social and health protection and prevention actions.

    What is already known on this subject

    • Wide educational disparities in health and disability exist; but their magnitude across countries varies markedly.

    • To what extent the country contexts and welfare regimes modify the size of the health differentials is not yet fully explained.

    What this study adds

    • The variation in the extent of social differentials across countries compared with the European average, in terms of the relative excess/reduced prevalence of disability at each extreme of the educational gradient, is put forth.

    • The results give new insight into the countries where differentials result from the higher social groups being relatively more advantaged compared to average regarding disability prevalence and the lower social groups being relatively more disadvantaged.

    • Departures from the average pattern within welfare regime groups were found in the Nordic countries, suggesting the need for further exploration of the greater protective and hazardous effects of country contexts on disability prevalence across social groups.

    References

    1. ↵
      1. Jagger C,
      2. Gillies C,
      3. Moscone F, et al
      . Inequalities in healthy life years in the 25 countries of the European Union in 2005: a cross-national meta-regression analysis. Lancet 2008;372:2124–31. doi:10.1016/S0140-6736(08)61594-9
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Solé-Auró A,
      2. Crimmins EM
      . The oldest old: health in Europe and the United States. In: Robine J-M, Jagger C, Crimmins EM, eds. Healthy longevity, a global approach. Annual Review of Gerontology and Geriatrics. Springer Publishing Company, 2013:3–33.
    3. ↵
      1. Crimmins E,
      2. Cambois E
      . Social inequalities in health expectancy. In: Robine JM, Jagger C, Mathers C, et al., eds. Determining health expectancies. Chichester: John Wiley & Sons, Ltd, 2003:111–26.
    4. ↵
      1. Mackenbach JP,
      2. Stirbu I,
      3. Roskam AJ, et al
      . Socioeconomic inequalities in health in 22 European countries. N Engl J Med 2008;358:2468–81. doi:10.1056/NEJMsa0707519
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      1. Mäki N,
      2. Martikainen P,
      3. Eikemo T, et al
      . Educational differences in disability-free life expectancy: a comparative study of long-standing activity limitation in eight European countries. Soc Sci Med 2013;94:1–8. doi:10.1016/j.socscimed.2013.06.009
      OpenUrlCrossRefPubMed
    6. ↵
      1. Jagger C,
      2. McKee M,
      3. Christensen K, et al
      . Mind the gap—reaching the European target of a 2-year increase in healthy life years in the next decade. Eur J Public Health 2013;23:829–33. doi:10.1093/eurpub/ckt030
      OpenUrlAbstract/FREE Full Text
    7. ↵
      1. Marmot M,
      2. Friel S,
      3. Bell R, et al
      . Closing the gap in a generation: health equity through action on the social determinants of health. Lancet 2008;372:1661–9. doi:10.1016/S0140-6736(08)61690-6
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Lagiewka K
      . European innovation partnership on active and healthy ageing: triggers of setting the headline target of 2 additional healthy life years at birth at EU average by 2020. Arch Public Health 2012;70:23. doi:10.1186/0778-7367-70-23
      OpenUrlCrossRefPubMed
    9. ↵
      1. Rechel B,
      2. Grundy E,
      3. Robine JM, et al
      . Ageing in the European Union. Lancet 2013;381:1312–22. doi:10.1016/S0140-6736(12)62087-X
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      1. Robine JM,
      2. Cambois E,
      3. Nusselder W, et al
      . The joint action on healthy life years (JA: EHLEIS). Arch Public Health 2013;1:1–5.
      OpenUrl
    11. ↵
      World Health Organization. International classification of functioning, disability and health: ICF. WHO, 2001.
    12. ↵
      1. Verbrugge LM,
      2. Rennert C,
      3. Madans JH
      . The great efficacy of personal and equipment assistance in reducing disability. Am J Public Health 1997;87:384–92. doi:10.2105/AJPH.87.3.384
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Eikemo TA,
      2. Bambra C,
      3. Judge K, et al
      . Welfare state regimes and differences in self-perceived health in Europe: a multilevel analysis. Soc Sci Med 2008;66:2281–95. doi:10.1016/j.socscimed.2008.01.022
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Mackenbach JP,
      2. Karanikolos M,
      3. McKee M
      . The unequal health of Europeans: successes and failures of policies. Lancet 2013;381:1125–34. doi:10.1016/S0140-6736(12)62082-0
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      1. Navarro V,
      2. Shi L
      . The political context of social inequalities and health. Int J Health Serv 2001;31:1–21. doi:10.2190/1GY8-V5QN-A1TA-A9KJ
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Lundberg O,
      2. Yngwe MA,
      3. Stjärne MK, et al
      . The role of welfare state principles and generosity in social policy programmes for public health: an international comparative study. Lancet 2008;372:1633–40. doi:10.1016/S0140-6736(08)61686-4
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. Eikemo TA,
      2. Bambra C
      . The welfare state: a glossary for public health. J Epidemiol Community Health 2008;62:3–6. doi:10.1136/jech.2007.066787
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. Brennenstuhl S,
      2. Quesnel-Vallee A,
      3. McDonough P
      . Welfare regimes, population health and health inequalities: a research synthesis. J Epidemiol Community Health 2012;66:397–409. doi:10.1136/jech-2011-200277
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Eikemo TA,
      2. Huisman M,
      3. Bambra C, et al
      . Health inequalities according to educational level in different welfare regimes: a comparison of 23 European countries. Sociol Health Illn 2008;30:565–82. doi:10.1111/j.1467-9566.2007.01073.x
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Avendano M,
      2. Jurges H,
      3. Mackenbach JP
      . Educational level and changes in health across Europe: longitudinal results from SHARE. J Eur Soc Policy 2009;19:301–16. doi:10.1177/1350506809341512
      OpenUrlCrossRefWeb of Science
    21. ↵
      1. Huijts T,
      2. Eikemo TA
      . Causality, social selectivity or artefacts? Why socioeconomic inequalities in health are not smallest in the Nordic countries. Eur J Public Health 2009;19:452–3. doi:10.1093/eurpub/ckp103
      OpenUrlFREE Full Text
    22. ↵
      OECD. Health care systems: Getting more value for money. OECD Economics Department Policy Notes 2, 2010:1–12.
    23. ↵
      1. Rechel B,
      2. McKee M
      . Health reform in central and eastern Europe and the former Soviet Union. Lancet 2009;374:1186–95. doi:10.1016/S0140-6736(09)61334-9
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      1. Berger N,
      2. Van Oyen H,
      3. Cambois E, et al
      . Assessing the validity of the Global Activity Limitation Indicator in fourteen European countries. BMC Med Res Methodol 2015;15:1. doi:10.1186/1471-2288-15-1
      OpenUrlCrossRefPubMed
    25. ↵
      1. Jürges H
      . True health vs response styles: exploring cross-country differences in self-reported health. Health Econ 2007;16:163–78. doi:10.1002/hec.1134
      OpenUrlCrossRefPubMedWeb of Science
    26. ↵
      1. Cabrero-Garcia J,
      2. Julia-Sanchis R
      . The Global Activity Limitation Index mainly measured functional disability, whereas self-rated health measured physical morbidity. J Clin Epidemiol 2014;67:468–76. doi:10.1016/j.jclinepi.2013.10.005
      OpenUrlCrossRefPubMed
    27. ↵
      1. Jagger C,
      2. Gillies C,
      3. Cambois E, et al
      . The Global Activity Limitation Index measured function and disability similarly across European countries. J Clin Epidemiol 2010;63:892–9. doi:10.1016/j.jclinepi.2009.11.002
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Van der Heyden J,
      2. Berger N,
      3. Van Oyen H
      . Comparison of self-rated health and activity limitation as predictors of short term mortality in the older population. Public Health 2015;129:283–5. doi:10.1016/j.puhe.2014.12.005
      OpenUrlCrossRefPubMed
    29. ↵
      1. Van der Heyden J,
      2. Van Oyen H,
      3. Berger N, et al
      . Activity Limitation predict health expenditures in the general population in Belgium. BMC Public Health 2015;15:267. doi:10.1186/s12889-015-1607-7
      OpenUrlCrossRefPubMed
    30. ↵
      1. Baker D,
      2. Leon J,
      3. Smith Greenaway E, et al
      . The education effect on population health: a reassessment. Popul Dev Rev 2011;37:307–32. doi:10.1111/j.1728-4457.2011.00412.x
      OpenUrlCrossRefPubMed
    31. ↵
      1. Montez JK,
      2. Hayward MD
      . Cumulative childhood adversity, educational attainment, and active life expectancy among U.S. adults. Demography 2014;51:413–35. doi:10.1007/s13524-013-0261-x
      OpenUrlCrossRefPubMed
    32. ↵
      1. Huijts T,
      2. Eikemo TA,
      3. Skalicka V
      . Income-related health inequalities in the Nordic countries: examining the role of education, occupational class, and age. Soc Sci Med 2010;71:1964–72. doi:10.1016/j.socscimed.2010.09.021
      OpenUrlCrossRefPubMed
    33. ↵
      1. Kangas O,
      2. Blomgren J
      . Socio-economic differences in health, income inequality, unequal access to care and spending on health: a country-level comparison of Finland and 16 other European countries. Res Finnish Soc 2014:51–63.
    34. ↵
      1. Van Oyen H,
      2. Berger N,
      3. Nusselder W, et al
      . The effect of smoking on the duration of life with and without disability, Belgium 1997–2011. BMC Public Health 2014;14:723. doi:10.1186/1471-2458-14-723
      OpenUrlCrossRefPubMed
    35. ↵
      1. Brønnum-Hansen H,
      2. Juel K
      . Smoking expands expected lifetime with musculoskeletal disease regardless of educational level. Eur J Epidemiol 2004;19:195–6. doi:10.1023/B:EJEP.0000017829.62827.a2
      OpenUrlPubMed
    36. ↵
      1. Cavelaars A,
      2. Kunst A,
      3. Geurts J, et al
      . Educational differences in smoking: international comparison. BMJ 2000;320:1102–7. doi:10.1136/bmj.320.7242.1102
      OpenUrlAbstract/FREE Full Text
    37. ↵
      1. Bambra C,
      2. Netuveli G,
      3. Eikemo TA
      . Welfare state regime life courses: the development of western European welfare state regimes and age-related patterns of educational inequalities in self-reported health. Int J Health Serv 2010;40:399–420. doi:10.2190/HS.40.3.b
      OpenUrlAbstract/FREE Full Text
    38. ↵
      1. Marmot, M
      . Fair Society, Healthy Lives. Strategic review of health inequalities in England post-2010. The Marmot Review, 2010:1–242.
    39. ↵
      1. Lorant V,
      2. Demarest S,
      3. Miermans PJ, et al
      . Survey error in measuring socio-economic risk factors of health status: a comparison of a survey and a census. Int J Epidemiol 2007;36:1292–9. doi:10.1093/ije/dym191
      OpenUrlAbstract/FREE Full Text
    40. ↵
      1. Tubeuf S,
      2. Jusot F,
      3. Devaux M,
      4. Sermet C
      . Social heterogeneity in self-reported health status and measurement of inequalities in health. IRDES Working paper 12, 2008:1–24.

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Contributors EC designed the study, performed the literature search, data analysis and interpretation, and wrote the manuscript. ASA performed the literature search, data analysis and interpretation, and wrote the manuscript. J-MR contributed to the study design, data analysis and interpretation, and wrote the manuscript. HB-H, VE, CJ, BJ, WJN, HVO and CW critically revised the study design and the interpretation, and substantially contributed to the writing of the manuscript.

    • Funding This research was part of the European Joint-Action on European Health and Life Expectancy Information System (JA-EHLEIS), funded by the Executive Agency for Health and Consumer of the European Commission (agreement number 2010 23 01).

    • Competing interests None declared.

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

    • Data sharing statement The research is based on data which are publically available.

    • ↵i http://www.uis.unesco.org/education/pages/international-standard-classification-of-education.aspx

    • ↵ii Logistic regression equation (models included EU-SILC baseline weights):Formula with X1i for Country i=(1-26); X2i for Age#Country i=(1-26); X3ij for Sex#Country i=(1-26) and j=(1-2); X4iz for Education#Country i=(1-26) and z=(1-3)

    • ↵iii The model was repeated with each of the 26 countries as reference, to test differences between countries (see online supplementary table S3).