You are here

Article Text

Article menu

Download PDFPDF

Health inequalities
Trends in inequalities in premature mortality: a study of 3.2 million deaths in 13 European countries
Free
  1. Johan P Mackenbach1,
  2. Ivana Kulhánová1,
  3. Gwenn Menvielle2,3,
  4. Matthias Bopp4,
  5. Carme Borrell5,
  6. Giuseppe Costa6,
  7. Patrick Deboosere7,
  8. Santiago Esnaola8,
  9. Ramune Kalediene9,
  10. Katalin Kovacs10,
  11. Mall Leinsalu11,12,
  12. Pekka Martikainen13,
  13. Enrique Regidor14,
  14. Maica Rodriguez-Sanz5,
  15. Bjørn Heine Strand15,
  16. Rasmus Hoffmann1,
  17. Terje A Eikemo1,16,
  18. Olof Östergren17,
  19. Olle Lundberg17,18
  20. for the Eurothine and EURO-GBD-SE consortiums
  1. 1Department of Public Health, Erasmus MC, Rotterdam, The Netherlands
  2. 2Centre for research in Epidemiology and Population Health (CESP), U1018, INSERM, Villejuif, France
  3. 3University of Versailles Saint Quentin, Versailles, France
  4. 4Institute of Social and Preventive Medicine, University of Zurich, Zurich, Switzerland
  5. 5Agència de Salut Pública de Barcelona, Barcelona, Spain
  6. 6Department of Clinical Medicine and Biology, University of Turin, Turin, Italy
  7. 7Department of Sociology, Vrije Universiteit Brussel, Brussels, Belgium
  8. 8Department of Public Health, Basque Government, Vitoria, Spain
  9. 9Lithuanian University of Health Sciences, Kaunas, Lithuania
  10. 10Demographic Research Institute of the Central Statistical Office, Budapest, Hungary
  11. 11Stockholm Centre on Health of Societies in Transition, Södertörn University, Huddinge, Sweden
  12. 12Department of Epidemiology and Biostatistics, National Institute for Health Development, Tallinn, Estonia
  13. 13Department of Sociology, University of Helsinki, Helsinki, Finland
  14. 14Department of Preventive Medicine and Public Health, Universidad Complutense de Madrid, Madrid, Spain
  15. 15Division of Epidemiology, Norwegian Institute of Public Health, Oslo, Norway
  16. 16Department of Sociology and Political Science, NTNU, Trondheim, Norway
  17. 17Center for Health Equity Studies, Stockholm, Sweden
  18. 18Department of Health Sciences, Mid Sweden University, Östersund, Sweden
  1. Correspondence to Professor J P Mackenbach, Department of Public Health, Erasmus MC, P.O. Box 2040, Rotterdam 3000 CA, The Netherlands; j.mackenbach{at}erasmusmc.nl

Abstract

Background Over the last decades of the 20th century, a widening of the gap in death rates between upper and lower socioeconomic groups has been reported for many European countries. For most countries, it is unknown whether this widening has continued into the first decade of the 21st century.

Methods We collected and harmonised data on mortality by educational level among men and women aged 30–74 years in all countries with available data: Finland, Sweden, Norway, Denmark, England and Wales, Belgium, France, Switzerland, Spain, Italy, Hungary, Lithuania and Estonia.

Results Relative inequalities in premature mortality increased in most populations in the North, West and East of Europe, but not in the South. This was mostly due to smaller proportional reductions in mortality among the lower than the higher educated, but in the case of Lithuania and Estonia, mortality rose among the lower and declined among the higher educated. Mortality among the lower educated rose in many countries for conditions linked to smoking (lung cancer, women only) and excessive alcohol consumption (liver cirrhosis and external causes). In absolute terms, however, reductions in premature mortality were larger among the lower educated in many countries, mainly due to larger absolute reductions in mortality from cardiovascular disease and cancer (men only). Despite rising levels of education, population-attributable fractions of lower education for mortality rose in many countries.

Conclusions Relative inequalities in premature mortality have continued to rise in most European countries, and since the 1990s, the contrast between the South (with smaller inequalities) and the East (with larger inequalities) has become stronger. While the population impact of these inequalities has further increased, there are also some encouraging signs of larger absolute reductions in mortality among the lower educated in many countries. Reducing inequalities in mortality critically depends upon speeding up mortality declines among the lower educated, and countering mortality increases from conditions linked to smoking and excessive alcohol consumption such as lung cancer, liver cirrhosis and external causes.

  • Mortality
  • Inequalities
  • Time-Series

https://doi.org/10.1136/jech-2014-204319

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

    During the 1980s and 1990s, socioeconomic inequalities in mortality have widened in many countries. This has been well documented for relative inequalities in mortality between educational or occupational groups in Northern, Western and Southern Europe.1–5 For absolute inequalities in mortality, findings have been less consistent, with increases reported for some countries such as Finland1–3 ,6 and Norway,1 ,3 ,4 but not for others such as Sweden1–3 and Italy.1 ,2 Inequalities in life expectancy between socioeconomic groups have generally increased as well.6–12 In Eastern Europe, inequalities in mortality have increased too, on a relative and on an absolute scale.13 ,14 Widening inequalities have also been reported for non-European high-income countries, such as the USA15–18 and New Zealand.19 ,20

    Although some studies have started to look at trends into the 21st century, a comprehensive analysis of recent trends, allowing a comparison of these trends between countries, is lacking. We therefore studied changes in mortality by education in all European countries with available data between the 1990s and 2000s, and focused on whether relative and absolute inequalities in mortality have widened or narrowed, and if so, which causes of death have contributed to these developments. We consider relative and absolute inequalities to be relevant: while the first are independent from average rates of mortality and therefore useful for analytical purposes, the second indicate the real impact that these inequalities have on the mortality risks of people with a lower socioeconomic position, which is useful for informing public health decisions.21

    Data and methods

    Data

    The data were obtained from official mortality registers and cover 13 countries: 4 countries in Northern Europe (Finland, Sweden, Norway, Denmark), 4 countries in Western Europe (England and Wales, Belgium, France and Switzerland), 2 countries in Southern Europe (Spain and Italy), and 3 countries in Eastern Europe (Hungary, Lithuania and Estonia). These are all European countries with comparable data for the 1990s and 2000s which, together, are likely to represent the full range of inequality trends in the subcontinent. Most data cover complete national populations, as in the case of the national longitudinal mortality registers of the Nordic countries. The exceptions are England and Wales (a 1% representative sample of the population; known as the Office of National Statistics Longitudinal Mortality Study), France (a 1% representative sample of the population; the so-called Echantillon Démographique Permanent created by the French National Institute of Statistics (INSEE)), Spain (Barcelona, Madrid and Basque Country only; urban and regional mortality registers) and Italy (Turin only; known as the Turin Longitudinal Study). A full overview of data sources is given in table 1.

    View this table:
    Table 1

    Mortality data sources.

    For each country, numbers of deaths by 5-year age group, sex and educational group were obtained for the 1990s and the 2000s. These data were mostly collected in the framework of a longitudinal mortality follow-up of population censuses carried out around 1991 and around 2000, respectively. Although around 2010, a more recent census has been held in many countries, mortality follow-up will only be complete after 5–10 years, and at the moment of writing (2014), our data, therefore, represent the most recent data available for those countries. Data for Hungary, Lithuania and Estonia derive from unlinked cross-sectional studies in which numerator data (deaths by educational group, as counted in the mortality register) and denominator data (persons by educational group, as counted in a census) have been collected at approximately the same point in time. In most countries, linkage between the population and death registries was more than 95% complete; for countries where linkage failure exceeded 5% (Madrid, Barcelona and the Basque Country) weights were used in the analysis to account for the difference. A total of around 3.2 million deaths are included in this study, deriving from around 360 million person-years of observation.

    Socioeconomic status was indicated by highest level of completed education. We focused on educational inequalities in mortality (instead of, eg, occupational inequalities in mortality) because comparable data on educational attainment are available for men and women in all European populations under study. Additionally, education is the most stable measure of socioeconomic position because it is normally completed early in adulthood which avoids reverse causation problems (ie, health outcomes at older ages cannot change a person's level of education).22 Education was classified according to the International Standard Classification of Education (ISCED-97). The categories used in this analysis were ‘no, primary or lower secondary education’ (ISCED 0, 1, 2; ‘low’), ‘upper secondary and post-secondary non-tertiary education’ (ISCED 3, 4; ‘mid’) and ‘tertiary education’ (ISCED 5, 6; ‘high’). In the 1990s dataset for England and Wales only two levels of education could be distinguished (‘low and mid’ vs ‘high’), and we therefore also classified the English data for the 2000s in this way. In order to reduce problems with the educational classification of elderly people, the main analysis is limited to the age group 30–74 years. Because average life expectancy at birth exceeded 74 years in most European countries during the study period, we label mortality in this age-range as ‘premature’.

    In addition to all-cause mortality, we analysed four broad cause-of-death groups that together account for all deaths (cardiovascular diseases, cancer, other diseases and external causes), and five specific causes of death each that signal potential explanatory factors (ischaemic heart disease (smoking, diet, hypertension, hypertension detection and treatment, myocardial infarction treatment), cerebrovascular disease (hypertension, hypertension detection and treatment, stroke treatment), lung cancer (smoking), liver cirrhosis (excessive alcohol consumption), and road traffic accidents (excessive alcohol consumption, road safety)). Cause-of-death information was not available for mortality by education in France. Causes of death were classified according to the ninth or tenth revision of the International Classification of Diseases (ICD). ICD-code numbers are given in web appendix table A1.

    Data for the 1990s were collected in the framework of the Eurothine study, and have been used for international comparative purposes before.23 Data for the 2000s were collected in the framework of the EURO-GBD-SE study.24 Data were centrally harmonised to enhance comparability between countries and over time.

    Methods

    We calculated age-standardised mortality rates and their 95% Confidence Intervals (CIs)25 by period, country, level of education, and sex, using the European Standard Population as defined by WHO. Age-adjusted rate ratios comparing the ‘low’ and ‘mid’ educated with the ‘high’ educated and their 95% CIs were calculated with Poisson regression using STATA statistical software V.12.0.

    The age-standardised mortality rates were used to calculate proportional and absolute changes in mortality by country, level of education and sex, and subtraction of these changes among the ‘high’ educated from those among the ‘low’ educated allowed us to evaluate differences in proportional and absolute mortality change between educational groups.

    The age-standardised mortality rates were also used to calculate population-attributable fractions of education for mortality for the two points in time. This is a straightforward measure of the population impact of inequalities in mortality.21 The population-attributable fraction takes into account inequalities in mortality between educational groups and the size of educational groups, and quantifies the proportion of all deaths in the population that could be avoided or postponed if everyone had the mortality rates of the high educated.21 ,26

    Most countries with longitudinal data classified person-years and deaths by age at baseline, and as length of follow-up differed between studies (table 1) adjustment was necessary. This adjustment procedure was developed and validated within the EURO-GBD-SE study,27 and was used wherever deaths were classified by age at baseline. Further details can be found in web appendix table A2.

    Results

    All-cause mortality

    Age-standardised mortality rates by education for the 1990s and 2000s can be found in figure 1 and online supplementary appendix table A3, whereas rate ratios are presented in table 3. Mortality declined between the 1990s and 2000s in all educational groups, except the low educated in Hungary, Lithuania and Estonia, among whom mortality remained stable or increased. Mortality among the high educated in Hungary, Lithuania and Estonia was similar to that among the low educated in many Western European countries (figure 1).

    Figure 1
    Figure 1

    All-cause mortality among the low and high educated, by country and sex, 1990s and 2000s Notes: Mortality standardised to the European Standard Population. Low = ‘no, primary or lower secondary education’ (ISCED 0,1,2). High = ‘tertiary education’ (ISCED 5, 6). For full details, including 95% Confidence Intervals, see online supplementary appendix table A3.

    Among men, relative inequalities in mortality increased in most populations, with the exceptions of England and Wales, France, Barcelona, the Basque Country, Madrid and Turin, where no clear changes were found. Among women, about half the countries had widening and half had stable inequalities. The rise in relative inequalities was very strong in Hungary, Lithuania and Estonia, particularly for men (table 2).

    View this table:
    Table 2

    Changes between the 1990s and 2000s in rate ratios by education for all-cause mortality, by country and sex

    As a result of these changes, the variation in magnitude of inequalities in mortality between European countries has increased considerably. In the 1990s, the South (represented by Barcelona, the Basque Country, Madrid and Turin) already tended to have smaller inequalities in mortality than the North and West, and the North and West than the East (represented by Hungary, Lithuania and Estonia), particularly among men. In the 2000s, the same pattern is still found, but with substantially larger differences between countries (table 2).

    These changes in relative inequalities in mortality between the 1990s and 2000s are the result of large differences between educational groups in proportional mortality change (see the column for all causes in tables 3A, B). For example, in the case of Finnish men, mortality declined 10.7%-points more among the high than among the low educated, as mortality declined by 35.9% among the high and 25.2% among the low educated. Light grey shading in table 3 indicates that mortality declined less among the low than among the high educated; dark grey shading indicates that mortality actually increased among the low educated. The latter is seen for all-cause mortality in Lithuania and Estonia only. In other countries, widening relative inequalities in all-cause mortality are due to less declines among the low than among the high educated. Among men in England & Wales and Barcelona and among women in England & Wales, Madrid and Turin, there was more proportional decline in all-cause mortality among the low than among the high educated.⇓⇓⇓

    View this table:
    Table 3A

    Differences between low and high educated in proportional and absolute mortality decline between the 1990s and 2000s, by cause of death, country and sex

    View this table:
    Table 3B
    View this table:
    Table 3C
    View this table:
    Table 3D

    However, tables 3C, D show that differences between educational groups in absolute mortality change have often favoured the low educated. For example, in the case of Finnish men, mortality declined by 75.3 deaths per 100 000 more among the low than among the high educated, as it went down by 361.9 and 286.6 deaths per 100 000, respectively. In many populations, including Finland (men only), Sweden, Norway, England and Wales, Belgium (men only), France, Switzerland, Barcelona, Basque Country (men only), Madrid, and Turin, absolute reductions in mortality have been larger among the low than among the high educated. Please note that there is no contradiction between smaller proportional reductions and larger absolute reductions in mortality: the latter also depend on starting levels of mortality which are always higher among the low educated.

    Cause-specific mortality

    A pattern of generally smaller proportional reductions in mortality among the low educated is found for many specific causes of death, including all cardiovascular disease, ischaemic heart disease, cerebrovascular disease, all cancer, lung cancer (men only), and all other diseases (tables 3A, B). Among men, the main exceptions are liver cirrhosis and external causes, for which several countries have experienced increases in mortality among the low educated. Among women, mortality has increased among the lower educated for liver cirrhosis and external causes, and also for lung cancer in many countries. Rising all-cause mortality among the low educated in Lithuania and Estonia is due to rising mortality from a wide range of causes of death.

    As mentioned above, in Northern, Western and Southern Europe, absolute reductions in all-cause mortality have been larger among the low than among the high educated in many countries. This was often due, in large part, to larger absolute reductions for cardiovascular disease (tables 3C, D). Among Finnish men, for example, the difference in decline of mortality between low and high educated was 75.3 deaths per 100 000 for all-cause mortality, of which 61.3 deaths per 100 000 are accounted for by cardiovascular disease mortality. In many countries, other causes of death, mainly cancer and all other diseases, contributed as well, but less so among women, which partly explains why larger absolute reductions in all-cause mortality among the low educated are more clearly and more often seen among men than among women.

    Population-attributable fractions

    Over time, the proportion of higher educated has increased in all populations (table 1), and larger inequalities in mortality between educational groups, therefore, do not necessarily imply that the population impact of these inequalities has become larger. However, as figure 2 shows, the population-attributable fractions have risen somewhat in many countries, among men and among women, particularly in Hungary, Lithuania and Estonia.

    Figure 2
    Figure 2

    Changes between the 1990s and 2000s in population-attributable fraction of education for all-cause mortality, by country and gender. Note: Population-attributable fractions were calculated as (M—Mhigh)/M, in which M=Age-standardised mortality rate in the whole population, and Mhigh=Age-standardised mortality rate among the high educated.26

    Discussion

    Summary of findings

    Relative inequalities in premature mortality increased in most populations in the North, West and East of Europe, but not in the South. This was mostly due to smaller proportional reductions in mortality among the lower than the higher educated, but in the case of Lithuania and Estonia mortality rose among the lower and declined among the higher educated. Mortality among the lower educated rose in many countries for conditions linked to smoking (lung cancer, women only) and excessive alcohol consumption (liver cirrhosis and external causes). In absolute terms, however, reductions in premature mortality were larger among the lower educated in many countries, mainly due to larger absolute reductions in mortality from cardiovascular disease and cancer (men only). Despite rising levels of education, population-attributable fractions of lower education for mortality rose in many countries.

    Strengths and limitations

    This is the most comprehensive analysis of trends in inequalities in mortality ever conducted, covering 13 countries with different socioeconomic and health conditions. However, its broad international scope inevitably raises issues of data comparability. Despite extensive harmonisation efforts, our comparisons between countries and over time may be biased by differences in data collection, for example, with regard to study design, population coverage, time-periods covered, and data classification. For example, in Hungary, Lithuania and Estonia, mortality data were collected in a cross-sectional unlinked design. A study comparing cross-sectional unlinked and longitudinal mortality data in Lithuania found that the former overestimated inequalities in mortality, but the bias was larger at older ages and significantly attenuated when the age limits were set to 30–74 years as in our study.28

    There are also differences between countries in population coverage. The most notable relate to Spain and Italy, for which only a few urban and relatively prosperous populations could be included that are not necessarily representative of the whole of Spain or Italy. However, studies of the situation in the 1980s, which used national data for Spain and Italy from methodologically less-refined sources, also found smaller inequalities in mortality there than in other European countries,2 ,29 as did recent national-level studies from Spain30 and Italy,31 ,32 so there is no reason to think that our study misrepresents the situation in these Southern European countries.

    Information on education was fairly complete in all countries, and the proportion of individuals with missing education ranged between 2% and 6% only. We reclassified national educational levels into the ISCED scheme, and this should have removed most differences in classification between countries. The main problem in our study concerns England and Wales, for which we had to pool together the ‘low’ and ‘mid’ educated because of the crude educational classification used in the 1990s mortality data. However, the sensitivity analysis reported in online supplementary appendix table A4, in which a two-group classification was used in all countries, confirms that whereas many other countries experienced an increase in relative inequalities in all-cause mortality, England and Wales did not, and its absolute inequalities therefore also went down more than elsewhere. It would have been useful to add another indicator of socioeconomic position to our analysis, such as occupational class or income, but income is generally unavailable in mortality registers, whereas data on occupational class that are comparable between countries and over time are available for a small number of countries only.23

    As our analysis has focused on changes over time within countries, between-country variations in data collection do not pose a major risk of bias. Potentially more important are changes over time in data collection, but these have been minimal (table 1). The main change has been in the length of follow-up and in the classification of person-years and deaths by age (see online supplementary appendix table A2). As our adjustment method for the latter problem could not be validated for the 1990s, and some uncertainty therefore remains concerning the comparability of absolute levels of mortality between the 1990s and 2000s, we have restricted analyses of absolute changes to within-country comparisons between the low and high educated, which are unlikely to be substantially biased. Still, our results require confirmation from national-level studies that are more strictly harmonised over time than is feasible in an international comparative study.

    Our analysis covers the age group 30–74 years at baseline, and thereby excludes younger age groups, and also under-represents older age groups which currently carry the largest share of the burden of mortality, particularly in Western Europe where average life expectancy at birth in the 2000s exceeded 75 years for men and 80 years for women. This implies that our results cannot necessarily be generalised to mortality in the whole population, and this may also explain discrepancies with the few national studies which have followed trends in inequalities in life expectancy into the 2000s and which cover a wider age range.8 ,11 In our data, educational inequalities in life expectancy from the age of 30 years have increased in most countries, whereas inequalities in partial life expectancy between age 30 and 75 years have been stable or declined (results not shown).

    Interpretation

    This is the first study to systematically document and compare trends since the 1990s in inequalities in mortality for a larger number of countries. It shows that after the turn of the century, relative inequalities have continued to widen in most European countries, but that in contrast to previous periods1 absolute reductions of mortality seem to have been larger among the low educated in many countries in the North, West and South of Europe. This may announce a future narrowing of the gap in mortality between educational groups, but for the time being, rising population-attributable fractions indicate an increasing, not a diminishing public health problem.

    As these larger absolute declines were often partly due to larger absolute reductions in mortality from cardiovascular disease, changes in ‘proximate’ determinants of cardiovascular disease, such as health-related behaviours (eg, smoking, diet, exercise) and/or healthcare interventions (eg, hypertension detection and treatment, thrombolytic therapy), must have been larger among the low educated. Smoking-related inequalities are widening in many European countries,33 ,34 so smoking is an implausible explanation. A recent narrowing of absolute inequalities in cardiovascular disease mortality has previously been reported for England,33 ,35 and has more likely been caused by an even distribution of treatment benefits35 than of risk factor changes.36 It is unlikely, on the other hand, that England's more favourable trends are related to its national policy to tackle health inequalities because the latter has not had clear population-wide effects.6 ,37 ,38

    Southern Europe presents a partly different pattern. Inequalities in mortality are smaller than elsewhere, and the relative gap has also widened less over time, as has been shown before for Barcelona.39 Part of the explanation for smaller inequalities in mortality in Southern Europe is that inequalities in mortality from smoking-related and alcohol-related conditions, and from conditions amenable to medical intervention are relatively small, particularly among women.23 Despite rising mortality from lung cancer among low educated women, and from liver cirrhosis among low educated men, Barcelona, the Basque Country, Madrid and Turin have been able to strengthen their favourable positions because differences in mortality decline between the low and high educated often were smaller than in other countries (tables 3A, B). Further study of downstream and upstream determinants of inequalities in mortality in Southern Europe may help to better exploit this ‘good practice’ for mutual learning.

    Inequalities in mortality in Eastern Europe have exploded during the study period, and in Lithuania and Estonia this was not due to differential mortality decline, as in other parts of Europe and in Hungary, but to a combination of rising mortality among the low, and declining mortality among the high educated.13 The same has been observed in Russia.40 It is likely that these unfavourable developments were caused by the economic crisis and the subsequent economic reforms that followed the collapse of the Soviet Union. In the early 1990s, Estonia and Lithuania experienced large declines in national income and large increases in unemployment rates, and more so than Eastern European countries that had not been part of the Soviet Union.13 Our results point to the important contribution of mortality from external causes to the widening gap in all-cause mortality in these two countries (table 3), which suggests a mediating role of excessive alcohol consumption. Migration patterns (eg, selective emigration of higher educated and relatively healthier persons towards the West) may also have played a role.

    Conclusions

    Relative inequalities in premature mortality have continued to rise in most European countries, and since the 1990s, the contrast between the South (with smaller inequalities) and the East (with larger inequalities) has become stronger. While the population impact of these inequalities has further increased, there are also some encouraging signs of larger absolute reductions in mortality among the lower educated in many countries. Reducing inequalities in mortality critically depends upon speeding up mortality declines among the lower educated, and countering mortality increases from conditions linked to smoking and excessive alcohol consumption such as lung cancer, liver cirrhosis and external causes.

    References

    1. ↵
      1. Mackenbach JP,
      2. Bos V,
      3. Andersen O,
      4. et al
      . Widening socioeconomic inequalities in mortality in six Western European countries. Int J Epidemiol 2003;32:830–7.
      OpenUrlAbstract/FREE Full Text
    2. ↵
      1. Kunst AE,
      2. Bos V,
      3. Andersen O,
      4. et al
      . Monitoring of trends in socioeconomic inequalities in mortality: experiences from a European project. Demogr Res 2004;Special collection 2, article 9:229–54.
    3. ↵
      1. Shkolnikov VM,
      2. Andreev EM,
      3. Jdanov DA,
      4. et al
      . Increasing absolute mortality disparities by education in Finland, Norway and Sweden, 1971–2000. J Epidemiol Community Health 2012;66:372–8.
      OpenUrlAbstract/FREE Full Text
    4. ↵
      1. Strand BH,
      2. Groholt EK,
      3. Steingrimsdottir OA,
      4. et al
      . Educational inequalities in mortality over four decades in Norway: prospective study of middle aged men and women followed for cause specific mortality, 1960–2000. BMJ 2010;340:c654.
      OpenUrlAbstract/FREE Full Text
    5. ↵
      1. Borrell C,
      2. Plasencia A,
      3. Pasarin I,
      4. et al
      . Widening social inequalities in mortality: the case of Barcelona, a southern European city. J Epidemiol Community Health 1997;51:659–67.
      OpenUrlAbstract/FREE Full Text
    6. ↵
      Department of Health. Fair Society, Healthy Lives (the Marmot review). London: Department of Health, 2010.
    7. ↵
      1. Martikainen P,
      2. Valkonen T,
      3. Martelin T
      . Change in male and female life expectancy by social class: decomposition by age and cause of death in Finland 1971–95. J Epidemiol Community Health 2001;55:494–9.
      OpenUrlAbstract/FREE Full Text
    8. ↵
      1. Deboosere P,
      2. Gadeyne S,
      3. Van,
      4. et al
      . The 1991–2004 Evolution in Life Expectancy by Educational Level in Belgium Based on Linked Census and Population Register Data. Eur J Popul 2009;25:175–96.
      OpenUrlCrossRef
    9. ↵
      1. Bronnum-Hansen H,
      2. Baadsgaard M
      . Increasing social inequality in life expectancy in Denmark. Eur J Public Health 2007;17:585–6.
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Bronnum-Hansen H,
      2. Baadsgaard M
      . Widening social inequality in life expectancy in Denmark. A register-based study on social composition and mortality trends for the Danish population. BMC Public Health 2012;12:994.
      OpenUrlCrossRefPubMed
    11. ↵
      1. Steingrimsdottir OA,
      2. Naess O,
      3. Moe JO,
      4. et al
      . Trends in life expectancy by education in Norway 1961–2009. Eur J Epidemiol 2012;27:163–71.
      OpenUrlCrossRefPubMed
    12. ↵
      1. Tarkiainen L,
      2. Martikainen P,
      3. Laaksonen M,
      4. et al
      . Trends in life expectancy by income from 1988 to 2007: decomposition by age and cause of death. J Epidemiol Community Health 2012;66:573–8.
      OpenUrlAbstract/FREE Full Text
    13. ↵
      1. Leinsalu M,
      2. Stirbu I,
      3. Vagero D,
      4. et al
      . Educational inequalities in mortality in four Eastern European countries: divergence in trends during the post-communist transition from 1990 to 2000. Int J Epidemiol 2009;38:512–25.
      OpenUrlAbstract/FREE Full Text
    14. ↵
      1. Leinsalu M,
      2. Vagero D,
      3. Kunst AE
      . Estonia 1989–2000: enormous increase in mortality differences by education. Int J Epidemiol 2003;32:1081–7.
      OpenUrlAbstract/FREE Full Text
    15. ↵
      1. Ma J,
      2. Xu J,
      3. Anderson RN,
      4. et al
      . Widening educational disparities in premature death rates in twenty six states in the United States, 1993–2007. PloS ONE 2012;7:e41560.
      OpenUrlCrossRefPubMed
    16. ↵
      1. Krieger N,
      2. Rehkopf DH,
      3. Chen JT,
      4. et al
      . The fall and rise of US inequities in premature mortality: 1960–2002. PLoS Med 2008;5:e46.
      OpenUrlCrossRefPubMed
    17. ↵
      1. Hadden WC,
      2. Rockswold PD
      . Increasing differential mortality by educational attainment in adults in the United States. Int J Health Serv 2008;38:47–61.
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Jemal A,
      2. Ward E,
      3. Anderson RN,
      4. et al
      . Widening of socioeconomic inequalities in U.S. death rates, 1993–2001. PloS ONE 2008;3:e2181.
      OpenUrlCrossRefPubMed
    19. ↵
      1. Blakely T,
      2. Tobias M,
      3. Atkinson J
      . Inequalities in mortality during and after restructuring of the New Zealand economy: repeated cohort studies. BMJ 2008;336:371–5.
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. Fawcett J,
      2. Blakely T,
      3. Kunst A
      . Are mortality differences and trends by education any better or worse in New Zealand? A comparison study with Norway, Denmark and Finland, 1980–1990s. Eur J Epidemiol 2005;20:683–91.
      OpenUrlCrossRefPubMedWeb of Science
    21. ↵
      1. Mackenbach JP,
      2. Kunst AE
      . Measuring the magnitude of socio-economic inequalities in health: an overview of available measures illustrated with two examples from Europe. Soc Sci Med 1997;44:757–71.
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Daly MC,
      2. Duncan GJ,
      3. McDonough P,
      4. et al
      . Optimal indicators of socioeconomic status for health research. Am J Public Health 2002;92:1151–7.
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. Mackenbach JP,
      2. Stirbu I,
      3. Roskam AJ,
      4. et al
      . Socioeconomic inequalities in health in 22 European countries. N Engl J Med 2008;358:2468–81.
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      1. Lundberg O,
      2. Ostergren O,
      3. Menvielle G,
      4. et al
      . Inequalities in mortality across Europe: An international comparative study. In: Eikemo T, Mackenbach JP (eds.). EURO-GBD-SE. The potential for reducing health inequalities in Europe. Final Report. Rotterdam: Department of Public Health Erasmus MC, 2012.
    25. ↵
      1. Keyfitz N
      . Sampling variance of standardized mortality rates. Hum Biol. 1966;38:309–17.
      OpenUrlPubMedWeb of Science
    26. ↵
      1. Levin ML
      . The occurrence of lung cancer in man. Acta—Unio Internationalis Contra Cancrum 1953;9:531–41.
      OpenUrl
    27. ↵
      1. Ostergren O,
      2. Menvielle G,
      3. Lundberg O
      . Adjustment method to ensure comparability between populations reporting mortality data in different formats in the EURO-GBD-SE project. In: Eikemo T, Mackenbach JP. (eds.) EURO-GBD-SE. The potential of reduction of health inequalities in Europe. Final report. Rotterdam:Department of Public Health Erasmus MC, 2012:113–21.
    28. ↵
      1. Shkolnikov VM,
      2. Jasilionis D,
      3. Andreev EM,
      4. et al
      . Linked versus unlinked estimates of mortality and length of life by education and marital status: evidence from the first record linkage study in Lithuania. Soc Sci Med 2007;64:1392–406.
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. Kunst AE,
      2. Groenhof F,
      3. Mackenbach JP
      . Mortality by occupational class among men 30–64 years in 11 European countries. EU Working Group on Socioeconomic Inequalities in Health. Soc Sci Med 1998;46:1459–76.
      OpenUrlCrossRefPubMedWeb of Science
    30. ↵
      1. Regidor E,
      2. Kunst AE,
      3. Rodriguez-Artalejo F,
      4. et al
      . Small socio-economic differences in mortality in Spanish older people. Eur J Public Health 2012;22:80–5.
      OpenUrlAbstract/FREE Full Text
    31. ↵
      1. Federico B,
      2. Costa G,
      3. Mackenbach JP,
      4. et al
      . Inequalities in mortality in Italy. in press.
    32. ↵
      1. Marinacci C,
      2. Grippo F,
      3. Pappagallo M,
      4. et al
      . Social inequalities in total and cause-specific mortality of a sample of the Italian population, from 1999 to 2007. Eur J Public Health 2013;23:582–7.
      OpenUrlAbstract/FREE Full Text
    33. ↵
      1. Martikainen P,
      2. Ho JY,
      3. Preston S,
      4. et al
      . The changing contribution of smoking to educational differences in life expectancy: indirect estimates for Finnish men and women from 1971 to 2010. J Epidemiol Community Health 2013;67:219–24.
      OpenUrlAbstract/FREE Full Text
    34. ↵
      1. Giskes K,
      2. Kunst AE,
      3. Benach J,
      4. et al
      . Trends in smoking behaviour between 1985 and 2000 in nine European countries by education. J Epidemiol Community Health 2005;59:395–401.
      OpenUrlAbstract/FREE Full Text
    35. ↵
      1. Bajekal M,
      2. Scholes S,
      3. Love H,
      4. et al
      . Analysing recent socioeconomic trends in coronary heart disease mortality in England, 2000–2007: a population modelling study. PLoS Med 2012;9:e1001237.
      OpenUrlCrossRefPubMed
    36. ↵
      1. Scholes S,
      2. Bajekal M,
      3. Love H,
      4. et al
      . Persistent socioeconomic inequalities in cardiovascular risk factors in England over 1994–2008: a time-trend analysis of repeated cross-sectional data. BMC Public Health 2012;12:129.
      OpenUrlCrossRefPubMed
    37. ↵
      Department of Health. Reducing health inequalities: an action report. London: Department of Health, 1999.
    38. ↵
      1. Mackenbach JP
      . Can we reduce health inequalities? An analysis of the English strategy (1997–2010). J Epidemiol Community Health 2011;65:568–75.
      OpenUrlAbstract/FREE Full Text
    39. ↵
      1. Borrell C,
      2. Azlor E,
      3. Rodriguez-Sanz M,
      4. et al
      . Trends in socioeconomic mortality inequalities in a southern European urban setting at the turn of the 21st century. J Epidemiol Community Health 2008;62:258–66.
      OpenUrlAbstract/FREE Full Text
    40. ↵
      1. Shkolnikov VM,
      2. Leon DA,
      3. Adamets S,
      4. et al
      . Educational level and adult mortality in Russia: an analysis of routine data 1979 to 1994. Soc Sci Med 1998;47:357–69.
      OpenUrlCrossRefPubMedWeb of Science

    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.

      Files in this Data Supplement:

    Footnotes

    • Collaborators Other members of the Eurothine and EURO-GBD-SE consortiums who have contributed to this study, in addition to the named coauthors of this paper, are Chris White and Lynsey Brown (Office for National Statistics, Newport, Wales, UK), and Otto Andersen and Anita Lange (Statistics Denmark, Copenhagen, Denmark).

    • Contributors JPM had the original idea for the study, performed some the analyses, and wrote and revised the paper. IK performed most of the analyses. GM harmonised the data. OO and OL developed an adjustment method. GM, MB, CB, GC, PD, SE, RK, KK, ML, PM, ER, MR-S, B-HS, RH, OO and OL each prepared the data for one of the populations included in the study. TE coordinated the study. All authors contributed to the interpretation of the results, commented on the first version of the paper, and approved the final version of the paper. JPM is guarantor of the paper.

    • Funding The analysis reported in this paper was based on data collected in the Eurothine and EURO-GBD-SE studies, both financially supported by the Health and Consumer Protection Directorate (DG SANCO) of the European Commission (contract numbers 2003125 and 20081309, respectively). The EURO-GBD-SE study was also supported financially by the Netherlands Organisation for Health Research and Development (ZonMw, project number 121020026).

    • Competing interests None.

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

    • Data sharing statement No additional unpublished data are available, because of data protection regulations.

    Linked Articles