SSM - Population Health 23 (2023) 101470 Contents lists available at ScienceDirect SSM - Population Health journal homepage: www.elsevier.com/locate/ssmph Educational disparities in disability-free life expectancy across Europe: A focus on the East-West gaps from a gender perspective Donata Stonkute a, b, *, Angelo Lorenti a, c, Jeroen J.A. Spijker d a Max Planck Institute for Demographic Research, Konrad-Zuse-Strasse 1, Rostock, 18057, Germany b The Faculty of Social Sciences, University of Helsinki, Finland c Max Planck – University of Helsinki Center for Social Inequalities in Population Health, Rostock, Germany and Helsinki, Finland d Centre d’Estudis Demogràfics, Centres de Recerca de Catalunya (CERCA), Carrer de Ca n’Altayó, Edifici E2, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain A R T I C L E I N F O A B S T R A C T Keywords: Education plays a crucial role in shaping the health outcomes of adults. This study examines the relationship Disability-free life expectancy between educational attainment and health across Europe. Using data from the Survey of Health, Ageing and European east-west health divide Retirement in Europe (SHARE), we estimate educational inequalities in disability-free life expectancy (DFLE) by Educational inequalities gender in seven Western European (2004–2019) and three Central and Eastern European (CEE) (2010–2019) countries. We exploit a novel approach that combines the Sullivan method and multivariate life tables to calculate DFLE using SHARE data. We find that educational differences in DFLE favoring the better-educated exist in both CEE and Western European countries, but also that the differences across countries are more pronounced among the low-educated. While the absolute gaps in DFLE between low- and high-educated in­ dividuals in CEE and Western European countries are similar, the educational disparities in DFLE impose a more significant burden on the CEE populations due to their overall lower life expectancy. Educational inequalities are larger among women than among men in CEE countries, while the results for Western European countries are mixed. Our findings further highlight the important role of the institutional context in mitigating or exacerbating educational inequalities in health. 1. Introduction While CEE countries have worse health outcomes than other Euro­ pean countries, there are also substantial health disparities within these Health outcomes have been shown to vary widely across states in the populations, particularly by gender and education (OECD, 2019). United States and across countries in Europe, with this heterogeneity Women have lower mortality rates across all ages, but report poorer likely reflecting differences not only in demographic and epidemiolog­ health and higher use of health services (Oksuzyan et al., 2008). At the ical profiles, but also in contextual settings (Jagger et al., 2020; Montez same time, low education is associated with higher morbidity and et al., 2019). However, studies on health outcomes in Europe typically mortality (Ross et al., 2012). The intersection of gender and education is focus on Western European countries, while neglecting the substantial also important, as educational returns to health are gendered (Ross variation in health across all European regions. One of the most striking et al., 2012). health disparities in Europe is the East-West divide, with Central and Both gender- and education-based inequalities in life expectancy Eastern European countries having markedly lower life expectancy than seem to be more pronounced in CEE countries than in other European the rest of the subcontinent, with a gap of almost 10 years (Eurostat, regions (Mackenbach, 2017; Pinho-Gomes et al., 2022). Distinctive 2022). However, life expectancy alone is an insufficient measure of causal mechanisms that relate gender and education to health outcomes population health. Health expectancy estimates, which consider both could be of particular importance in explaining European East-West mortality and morbidity information, indicate that CEE populations not health disparities (Geyer et al., 2006). However, studies on in­ only have shorter lifespans, but also spend fewer years in good health equalities in health expectancies by both gender and educational (Eikemo et al., 2008). attainment levels in CEE are still scarce. * Corresponding author. Max Planck Institute for Demographic Research, Konrad-Zuse-Strasse 1, Rostock, 18057, Germany. E-mail address: stonkute@demogr.mpg.de (D. Stonkute). https://doi.org/10.1016/j.ssmph.2023.101470 Received 6 June 2023; Received in revised form 13 July 2023; Accepted 20 July 2023 Available online 22 July 2023 2352-8273/© 2023 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). D. Stonkute et al. SSM - Population Health 23 (2023) 101470 Therefore, this study aims to determine the extent of gender-specific pathways: commodity and human capital. Commodity theories posit educational disparities in the duration of healthy life across European that having a higher level of education provides individuals with ad­ countries. More specifically, we estimate disability-free life expectancy vantages, such as access to better employment, higher income, and (DFLE) at age 50 by gender and education for 10 European countries. health insurance. Human capital theories posit that beyond the financial The study also attempts to identify any emerging patterns by comparing gains, education promotes health by equipping individuals with the a number of CEE countries with their Western European counterparts. knowledge and resources they need to adopt health-promoting behav­ Building on previous research, we hypothesize that women and pop­ iors. However, it is essential to recognize that as well as fostering ulations within CEE region experience shorter DFLE and spend larger favorable conditions for good health, education also plays a pivotal role proportions of their lives with disability. Furthermore, we expect that as a social determinant of inequalities in health (Ross et al., 2012). educational disparities will exhibit a more pronounced magnitude While the pathways through which gender and education act upon among women and in the context of CEE countries. health are distinct, it is also crucial to consider the intersection between Several aspects of this paper contribute to the existing literature. the two. Examining the intersection between education and gender in­ First, we combine two well-established estimation methods, the multi­ equalities in European countries requires a multi-dimensional under­ variate life table (Brown et al., 2012; Teachman & Hayward, 1993) and standing of the various social, economic, and cultural factors that shape the Sullivan method (Sullivan, 1971), to provide an innovative approach health opportunities and outcomes for individuals of different genders that exploits the information available in the Survey of Health, Ageing and educational levels. Moreover, even as gender intersects with edu­ and Retirement in Europe (SHARE) to estimate gender- and cation in shaping health, there are country-specific differences in health education-specific life tables and healthy life years for each country. To inequalities by both gender and education (Mackenbach, 2017; the best of our knowledge, this is the first paper using SHARE to estimate Romero-Ortuno et al., 2014). life tables by gender and education. In addition, estimating differential Gender differences in health vary across countries (Crimmins et al., mortality and health from a single data source enables us to study 2011). While the differences in life expectancy between men and women countries that are traditionally overlooked in the literature: namely, the are large in many Western and Southern countries, they are relatively CEE countries. Thus, we also contribute to the discussion on how small compared to those in Eastern European countries (Author 2020). educational inequalities in health influence the European East-West Due to their significant mortality disadvantage, men in CEE countries divide. are, on average, expected not only to have shorter lifespans, but also to experience fewer healthy years of life than women (Van Oyen et al., 2. Background 2013). Educational returns to health also differ across Europe (Mackenbach The differential impact of gender on mortality has been widely et al., 2018; Valverde et al., 2021), with the disparities being largest in recognized, with women exhibiting lower mortality rates than men CEE countries and smallest in Southern European countries (Mack­ across all age groups (Oksuzyan et al., 2008). Despite women generally enbach, 2017). For instance, the life expectancy difference between men having a longer life expectancy than men, it is often assumed that they with high and low education is more than 10 years in CEE countries, experience poorer overall health (Author 2020; 2022). This phenome­ while it is approximately four years in Sweden, Italy, and the non may be due, at least in part, to the tendency of women to report Netherlands (OECD, 2019). Although previous literature has primarily poorer health, a greater burden of morbidity and disability, and more focused on mortality, recent studies have observed a more pronounced frequent use of health services (Nusselder et al., 2010). The observation educational gradient in transitioning to both poor health and recovery that women have both a mortality advantage and a health disadvantage, among CEE populations (Solé-Auró & Gumà, 2022). In addition, CEE which is also referred to as the female-male health-survival paradox, is countries have been consistently shown to have the lowest health ex­ often explained by citing intricate connections between gender-related pectancies in Europe (Jagger et al., 2008, 2011), contributing to the social, behavioral, and biological factors (Oksuzyan et al., 2008). In concept of the European East-West health divide (Carlson, 2004). addition, women’s longevity advantage itself contributes to their health Both the health of the general population and disparities by gender disadvantage, as it results in a larger proportion of life spent in poor and education are associated with country-specific macro-level in­ health (Luy & Minagawa, 2014). However, the magnitude of gender dicators (Eikemo et al., 2008; Van Oyen et al., 2010). Therefore, the disparity varies significantly, and it is important to note that female persistent health divide between East and West might reflect the role of disadvantage in health does not hold true uniformly across countries or social structures, as conceptualized by the institutional theory of health health indicators selected (Crimmins et al., 2019). Moreover, despite the inequalities (Beckfield et al., 2015). The theoretical framework out­ fundamental importance of gender in health, prompting studying health lining the effects of welfare states on social health inequalities has outcomes for men and women separately, health disparities are not suggested that sociohistorical and policy contexts may configure a sce­ limited to differences by gender. nario in which health gains follow educational attainment (Ross et al., Socioeconomic factors are also considered key determinants of 2012). CEE is a particularly interesting region to focus on when studying health differences within populations. Evidence suggests that an in­ this association due to its history of state socialism, early educational dividual’s social standing within a hierarchical structure, as determined expansion, historically high levels of female employment, and atypical by factors such as income, occupation, and education, can either relationship between education and income (Haggard & Kaufman, ameliorate or exacerbate their health prospects (Siegrist & Marmot, 2009). However, these positive factors may be undermined by the lack 2004). Education, in particular, holds unique value as an indicator of of progress in the region across many health indicators, such as mortality socioeconomic status. It is a reliable and easily accessible indicator that from preventable causes (Mackenbach et al., 2015). tends to remain relatively stable after individuals complete their formal Despite the evidence of substantial gender and educational health education. Moreover, education is fundamental in establishing both disparities within CEE countries, the current research on this topic is income and occupation. Notably, the impact of education on health fragmented, and shares several limitations. Existing studies primarily outcomes persists even after adjusting for factors such as wealth and focused on population-level estimates and the period immediately occupation (Cutler & Lleras-Muney, 2010). While this is an important following the fall of communism (Carlson, 1998; Shkolnikov et al., empirical piece of evidence, theoretical support reinforces the impor­ 2006), thereby failing to account for more recent developments. tance of education when examining health disparities. Furthermore, previous research predominantly focused on mortality Higher levels of education directly contribute to better health, outcomes, which offer limited insights into the overall health status of beyond a simple correlation. Ross and Mirowsky (2010) argued that the the population (Minagawa, 2013). effects of education are embedded in the nexus between two main A growing interest has emerged in applying measures that extend 2 D. Stonkute et al. SSM - Population Health 23 (2023) 101470 beyond mortality to gain a comprehensive understanding of population European country that are known for having a shared sociohistorical health (Robine & Jagger, 2003). One such measure is disability-free life context and welfare regime (Eikemo & Bambra, 2008). The remaining expectancy (DFLE), which is defined as the average duration of a per­ analyzed countries were classified as Western European countries. son’s life without health limitations within a given population (Robine We aimed to maximize the number of countries included in our et al., 1999). DFLE is derived from disability indicators such as the study. However, some countries, particularly those in the CEE region, Global Activity Limitation Indicator (GALI), and when calculated on this participated in fewer waves. To ensure consistency, we included only basis, it is known as healthy life years. Although disparities in DFLE by those CEE countries that have participated in the survey since at least gender and education have been extensively documented (Mäki et al., wave 4, and that have not missed more than one consecutive round of 2013; Solé-Auró et al., 2015), limited research has been conducted on participation. The seven Western European countries were selected on differential health expectancy for CEE countries. Thus, whether gender the same criteria for comparative purposes. We dropped individuals and educational gaps in disability-free life expectancy display a consis­ with missing data on the variables of interest or with a baseline age tent pattern across European countries, as has been observed for overall below 50. Additional restrictions were applied to individual parts of the life expectancy, remains an open question. statistical analysis. Unfortunately, however, there is not enough data available to apply traditional methods of estimating health expectancies for a wide range 3.2. Indicators of countries by gender and educational attainment (Avendano et al., 2009; Jagger et al., 2020). Health expectancy measures rely on access to In this study, we used SHARE’s single-question measure of functional vital statistics, registries, censuses, and health prevalence data. National limitation or disability (GALI) as our health indicator. GALI is a reliable statistical offices provide data on age- and gender-specific mortality, and valid tool for measuring function and disability, and can be used for which are then standardized and disseminated by organizations like the cross-country comparisons. Specifically, this indicator refers to the European Union’s statistical office Eurostat and databases such as the observed prevalence of any long-standing activity limitations, which Human Mortality Database (HMD). However, multivariate life tables by could denote general health problems or limitations in engaging in ac­ education are scarce, except for a few countries like the Nordic European tivities people usually perform (Robine & Jagger, 2003; Van Oyen et al., countries and Spain (Klotz et al., 2017). This is because mortality re­ 2006). In the harmonized dataset, the responses were re-coded by cords linked to population databases are not broken down by educa­ identifying individuals as healthy if they reported no limitations in usual tional level, and have different time periods and stratification variables. activities, and as limited if they reported any level of limitation. We also This study seeks to address the deficits in the literature by estimating use interview status and year of death variables to obtain survival gender- and education-specific DFLE for 10 European countries. The information. added value of this study is twofold. First, we combine two established The level of educational attainment was determined using a estimation methods ‒ the multivariate life table (Brown et al., 2012; harmonized measure based on the International Standard Classification Teachman & Hayward, 1993) and the Sullivan method (Sullivan, 1971) of Education (Statistics, 2012). This measure categorized individuals – to obtain DFLE estimates. This approach allows us to use SHARE as a into three groups: 1. lower secondary education or less (ISCED 0–2); 2. single data source. To support this approach, we also generate upper secondary or vocational training (ISCED 3–4); and 3. tertiary gender-specific life tables to use for comparisons with the HMD and education (ISCED 5–6). Sample characteristics are presented in Table 1. European Health and Life Expectancy Information System (www.eurohe x.eu; henceforth EurOhex data), and to identify any discrepancies. 3.3. Analysis Second, our study contributes to the literature on health disparities across Europe, specifically between CEE and Western European coun­ In this study, we used a multivariate life table approach (Teachman tries, in two ways. First, we analyze health disparities in CEE countries, & Hayward, 1993) to create gender- and education-specific life tables. which have been under-represented in prior research. Second, we shift Following an approach similar to that used in Brown et al. (2012), we our focus from mortality to health expectancy, estimating remaining employed a set of discrete-time regression models for each country, disability-free life expectancy at age 50. stratified by gender, to estimate the probability of death by age, gender, and education. To estimate this probability, survival information was 3. Data and methods needed for at least two consecutive time points for each individual. The approach assumed that individuals are alive at age t and may die at age 3.1. Data t+1. This implied that age was the time scale we used in the models. The population discrete-time hazard was defined as the probability of death This study uses data from the Harmonized Survey of Health, Ageing occurring in the current or future time period, represented by h(tij ) = and Retirement in Europe (SHARE) developed by the Gateway to Global Pr(Ti = j |Ti ≥ j), where Ti is a random variable indicating the time Aging Data (https://g2aging.org/). SHARE is a biennial survey con­ period j when individual i dies. ducted since 2004 that provides cross-national longitudinal information The models were as follows: on demographic, socioeconomic, and health measures, as well as vital [ ( ⃒ ))] status data. The survey covers private households with members aged 50 ln − ln 1 − h(tj ⃒xi = αj + x′i β (1) or older. Our study employs comprehensive information on health and survival, such as the age at the time of death and the health status of where tj is the time interval, αj is the complementary log-log trans­ participants. Depending on when each country initially participated in formation of the baseline hazard and x′i are a set of covariates: namely, the survey, we combined data from either the 1st (2004) or the 4th wave age, age-squared, and educational level, specific to the individual i. Age (2010) to the 8th (2019) wave of SHARE. We excluded wave 3, which is was added with a second-degree polynomial; the quadratic term called SHARELIFE, because it only focuses on people’s life histories, accounted for nonlinearities in the age pattern of deaths. The models rather than relying on a regular questionnaire. Additional information were run separately for men and women, and for each country. about the SHARE project and waves used (Börsch-Supan, 2022a, 2022b, To predict death probabilities by country, gender, education, and 2022c, 2022d, 2022e, 2022f), as well as sampling details for each age, we used two-year intervals to reflect the biannual panel data country (Bergmann et al., 2019) can be found elsewhere. structure. We obtained gender- and education-specific life tables from Three out of the 10 countries studied are located in Central and estimated survival probabilities by simulating the death of a synthetic Eastern Europe (CEE): Estonia, Czechia and Slovenia. They therefore cohort of 500,000 individuals. In the simulation, each person in the age include one Baltic country, one Central European and one Southeast interval x survives to x + n if the predicted probability of death is lower 3 D. Stonkute et al. SSM - Population Health 23 (2023) 101470 Table 1 Sample characteristics and educational composition for men and women aged 50 and older. Years Sample size Proportion of men Proportion across educational levels Women Men Low Medium High Low Medium High Central and Eastern Europe Estonia 2010–2019 21 514 0.40 0.27 0.49 0.23 0.30 0.47 0.22 Czechia 2010–2019 18 141 0.41 0.44 0.46 0.11 0.39 0.44 0.17 Slovenia 2010–2019 12 111 0.43 0.42 0.43 0.15 0.25 0.58 0.17 Western Europe Austria 2004–2019 16 371 0.42 0.34 0.45 0.21 0.15 0.55 0.31 Belgium 2004–2019 24 572 0.45 0.44 0.27 0.29 0.39 0.26 0.35 Denmark 2004–2019 15 016 0.47 0.24 0.33 0.43 0.14 0.48 0.39 France 2004–2019 18 906 0.43 0.49 0.31 0.20 0.38 0.39 0.23 Italy 2004–2019 19 174 0.46 0.74 0.20 0.06 0.70 0.22 0.08 Spain 2004–2019 21 771 0.45 0.84 0.08 0.08 0.78 0.11 0.11 Sweden 2004–2019 16 664 0.46 0.39 0.30 0.31 0.43 0.30 0.27 Source: Own calculations, using pooled data from SHARE. than a random draw from a uniform distribution. We repeated this The distribution of individuals across educational levels varied by procedure iteratively from ages x to x + n for age intervals [50–52), country and gender. Most CEE countries had a large share of individuals [52–54), …, 110, and obtained life table quantities, such as the number with upper secondary or vocational training, while Western European of survivors at the corresponding ages. The period life table was con­ countries were more heterogeneous, with some having a large share of structed following standard steps (Preston et al., 2001). Life tables were individuals with low education (e.g., Spain and Italy), and others having closed at age 90+ due to limited observations at the oldest ages. an evenly distributed educational composition across all three levels (e. Finally, remaining DFLE at age 50 was estimated using the Sullivan g., Denmark and Sweden). Finally, as expected, most countries had a method (Sullivan, 1971), following that: larger proportion of women than of men in the lowest education category. 1 ∑x=90+ eDF 50 = (1− n π x )⋅n Lx (2) Table 2 displays data on the proportions of individuals experiencing l50 x=50 long-term limitations in usual activities due to health problems where n is the length of the age interval, nπ x is a proportion of in­ (indicated by the GALI indicator), categorized by education and dividuals with health limitations between ages x and x + n, l50 is a country for men and women. Women, and especially those with low number of survivors at age 50 and nLx is a number of person-years lived education, were more likely than men to report limitations. Significant between ages x and x + n. The idea underlying the method is to apply gender gaps were observed among individuals with medium education age-specific proportions of a population in an unhealthy state to the age- in Denmark and Sweden, with a 9–10% difference in the prevalence of specific person-years lived from the life table. Summing up only the health limitations. Our results showed a consistent educational pattern disability-free person-years lived across all ages yielded the DFLE. The across all analyzed countries, with a lower proportion of individuals in age-specific proportion of people with health limitations within sub- the high education group reporting limitations. With the exception of groups was determined using a sample of individuals who were still Slovenia, CEE countries had the highest prevalence of health alive. Confidence intervals for LE and DFLE were obtained using a limitations. For example, up to 76% of low-educated women in nonparametric bootstrap approach (Cameron & Trivedi, 2005) by Estonia reported limitations, while the. Southern European countries, resampling the data 1000 times. All the analyses were conducted using R and particularly Spain and Italy, had the lowest prevalence of health language (version 4.1.1). limitations. We utilized a distinctive approach that capitalized on the availability of mortality data within the SHARE dataset. However, life tables 4.1. Life and health expectancies at the population level calculated from survey data may differ considerably from those gener­ ated using high-quality data sources. To support our findings, we Fig. 1 shows estimates of remaining life expectancy at age 50 for men compared life expectancy estimates at the population level with HMD and women for all the analyzed countries compared with the HMD and and EurOhex data, with the aim of identifying any substantial differ­ ences. More details on how we constructed comparative life tables using Table 2 HMD data can be found in the supplementary material. Proportions of women and men aged 50 and older experiencing long-term limitations in usual activities due to a health problem by educational level. 4. Results Women Men Table 1 shows characteristics of the working sample, categorized by Low Medium High Low Medium High educational level and country for men and women. The sample included Estonia 0.76 0.61 0.50 0.69 0.59 0.50 Czechia 0.66 0.50 0.46 0.62 0.55 0.48 184,240 person-wave observations of individuals aged 50+ across 10 Slovenia 0.63 0.46 0.37 0.63 0.47 0.37 countries. It should be noted that the data for Central and Eastern Austria 0.62 0.48 0.43 0.63 0.47 0.43 Europe countries cover the years 2010–2019, whereas the data for Belgium 0.56 0.48 0.44 0.49 0.43 0.38 Western European countries cover the years 2004–2019. Estimates for Denmark 0.55 0.46 0.33 0.48 0.36 0.32 France 0.55 0.40 0.36 0.52 0.40 0.31 both life and disability-free life expectancies are reflective of the mid­ Italy 0.51 0.32 0.25 0.44 0.30 0.26 points of these intervals, namely 2015 and 2012, respectively. Spain 0.50 0.24 0.25 0.41 0.27 0.24 Although CEE countries joined SHARE later, the number of obser­ Sweden 0.53 0.46 0.40 0.47 0.37 0.34 vations from CEE countries was not substantially lower than that from Note: The data used for the calculations were pooled from country-specific Western European countries, except for Slovenia, which had the smallest waves of SHARE from either 2004–2019 or 2010–2019, based on country- number of observations. specific data availability. 4 D. Stonkute et al. SSM - Population Health 23 (2023) 101470 Fig. 1. Remaining life expectancy at age 50 and their 95% confidence intervals obtained from SHARE, compared to estimates obtained from HMD and EurOhex. Note: HMD estimates are based on the calendar year that represented the midpoint population given the period of the country’s participation in the survey. For Austria, Belgium, Denmark, France, Italy, Spain, and Sweden, we used life tables from 2012. For Czechia, Estonia, and Slovenia, we used life tables from 2015. In addition, we used EurOhex estimates from 2012, as it is the latest year available in the database for all countries. EurOhex data. While population-level estimates were not the main focus expect to live longer than their male counterparts across all of the of this study, we included them because they not only provided an countries studied. The differences were striking: in most of the studied overview of the health status of the population, they also allowed for countries, the highest-educated men could expect to live fewer than or comparisons with data from other reliable sources. In most cases, the the same number of years as the least-educated women. The full set of HMD and EurOhex estimates fell within the 95% confidence intervals of estimates by gender and education is provided in supplementary mate­ our estimates or were within the two-year difference, except for Danish rials Table A2. and Slovenian women. When considering life expectancy free of disability, we found that In addition, estimates at the population level revealed some impor­ women retained their advantage compared to men in the majority of tant patterns. First, we did not observe greater between-country dis­ countries, except for Belgium and Sweden. Central and Eastern Euro­ parities in disability-free life expectancy (DFLE) than in overall life pean (CEE) countries generally had a wider gender gap in life expec­ expectancy at age 50. Second, we found that individuals in CEE coun­ tancy without disability, especially among individuals with medium tries, and particularly men, spent a much smaller share of their life education. Women in CEE countries with medium education could disability-free than their Western European counterparts. Complete in­ expect to live 2–3 years longer without disability than their male formation on estimates at the population level can be found in the counterparts. supplementary material Table A1. Fig. 2 displays DFLE estimates by gender and education for the studied countries. Higher educational levels were associated with a longer disability-free life, but the differences between the highest and 4.2. Life and health expectancies by gender and education the lowest educational levels varied greatly by gender and across countries. Among women, Belgium had the smallest educational gap in Our findings indicated that women in all educational groups could Fig. 2. Remaining disability-free life expectancy at age 50 by gender and education. 5 D. Stonkute et al. SSM - Population Health 23 (2023) 101470 DFLE (3.7 years), while Slovenia and Italy had the largest differences (10 female-male health-survival paradox (Oksuzyan et al., 2008). Women’s years). Among men, Belgium had the smallest educational gap in DFLE health disadvantage may be due to limited access to resources such as (3.6 years), while Slovenia and Austria had the largest differences (9.2 income, education, and health care (Tawiah et al., 2021). On the other and 8.4 years, respectively). For Southern European and CEE countries hand, women exhibit higher levels of compliance with treatments (except for Spanish and Czech women), the health benefits of higher (Christensen et al., 2009), and greater resilience when living with education were particularly noteworthy, as there was a marked increase chronic diseases (Kiely et al., 2019). Given the complexity of gender in the differences in DFLE between individuals with medium and with differences, it is vital to approach their interpretation with caution, high educational levels compared to the differences in DFLE between recognizing that their manifestation is influenced by social, behavioral, individuals with low and medium educational levels. and epidemiological contexts (Crimmins et al., 2019). Our findings, Although the gaps in DFLE between individuals with low and high consistent with previous research, uncovered an educational gradient in educational levels were significant, the absolute disparities within CEE health outcomes. The advantage associated with high education was countries were not greater than those found in Western European first reflected in the prevalence of health limitations, as the findings countries. To illustrate, the difference in DFLE between men with low indicated that individuals with higher educational levels tended to and high. report fewer limitations than those with lower educational levels. Health educational levels was approximately six years in Denmark, Italy, disparities by education were also observed in the estimates of DFLE, but and France, which was comparable to the gap in Czechia. Furthermore, the extent of these disparities was not exclusively driven by the disad­ among men in Estonia, this gap was narrower by one year. However, it is vantage associated with low education. important to interpret these results in relative terms; specifically, in In certain countries, particularly in Southern European and Central terms of the proportion of the overall life expectancy that an individual and Eastern European countries, the differences in DFLE between in­ could expect to live disability-free. dividuals with medium and high educational levels were found to be Higher education was associated with a greater proportion of much greater than the differences in DFLE between individuals with low disability-free life, but the size of this advantage varied by gender and by and medium educational levels. The fact that the DFLE of the medium country. The estimates indicated that individuals in Italy and Spain had educated was barely higher in some instances than that of the lower the highest shares of disability-free life at age 50 across all educational educated could be because these social disadvantages extend to the groups. Notably, the estimates showed that high-educated men in Italy middle class and mainly the high educated are able to obtain good and Spain could expect to live 72–73% of their remaining years of life at employment conditions with higher income, health insurance, as well as age 50 disability-free. In contrast, the shares of disability-free life at age the knowledge and resources they need to adopt health-promoting be­ 50 were smallest for individuals in CEE countries (with the exception of haviors (Montez et al., 2018). Previous research has suggested that na­ medium- and high-educated women in Slovenia) and Austria. Across all tional contexts might be able to modify these patterns (Cambois et al., educational groups and irrespective of gender, individuals living in CEE 2016). In Southern European countries, the proportion of high-educated countries and Austria could expect to live less than 50% of their individuals was still relatively small among the oldest members of the remaining years of life at age 50 free of disability. Generally, women studied cohorts. This is likely because educational expansion occurred could expect to spend a smaller share of their life without health limi­ later in these countries than in the other European countries studied, tations. For example, while low-educated Italian women could expect to resulting in a sub-population of high-educated individuals, and partic­ live 60% (18.7 years) of their remaining years of life at age 50 disability- ularly of men, that is more selective. Meanwhile, in CEE countries, free, high-educated Estonian women could expect to live less than half where education is more evenly distributed, institutional differences (46%; 14.5 years) of their remaining years of life at age 50 free of may have been more important for the more affluent. Given the socio­ disability. historical context, individuals with low educational attainment in the The estimates indicated that Danish, Italian, and French men had a CEE countries tended to have consistently lower DFLE than their remaining DFLE at age 50 of 17.4–19.4 years. By contrast, Czech and Western European counterparts. This could be due to the cumulative Estonian men could expect to live 12.8 and 11.0 years, respectively, of effects of exposures tied to social disadvantages, such as low education, their remaining life at age 50 disability-free. These findings imply that across the life course. However, the degree to which additional levels of the six-year difference in DFLE between low and high educational education were associated with health appeared to differ across coun­ groups mentioned earlier was proportionally smaller in Western Euro­ tries. This finding is supported by the institutional theory of health in­ pean countries, and that the educational disparities in DFLE imposed a equalities (Beckfield et al., 2015) and empirical evidence of differential more significant burden on the populations in the CEE countries. educational returns to health from other geographical contexts (Montez et al., 2019). The empirical findings of this study are also consistent with 5. Discussion this theoretical contextualization, as the education-specific differences observed across the studied countries were more pronounced among the Health disparities based on gender and education in Europe have low-educated. been widely studied, but limited data has made it challenging to esti­ A particular interest of this study was investigating the differences mate these disparities in some countries. To address this issue, re­ between CEE and Western European countries. Additional insights into searchers have been compelled to adapt existing methodological the East-West divide in Europe, consistent with prior research, emerged approaches, and to develop new and original approaches. Our study from our examination of gender-specific educational disparities in DFLE emphasized the potential of using SHARE as a single source of data for across European countries. First, a substantial gender gap in DFLE fa­ differential life and health expectancy estimation, in particular for voring women was observed in CEE countries, particularly among in­ countries for which no information is available from other sources. We dividuals with medium levels of education. Second, with a few combined two established estimation techniques – discrete-time exceptions (in particular Slovenia), the three CEE countries were shown modeling and simulation – to estimate multivariate life tables, and the to have lower levels of disability-free life expectancy (DFLE). Thus, the Sullivan method to estimate disability-free life expectancy. findings indicated that regardless of their educational level or gender, The results of this paper first point to important gender differences, individuals in these CEE countries spent a smaller percentage of their in line with prior research (Pinho-Gomes et al., 2022). Across all remaining life disability-free than their counterparts in the seven educational groups, women were shown to live longer than men, with analyzed Western European countries. However, only the lowest the highest-educated men living fewer or the same number of years as educated in CEE had a lower DFLE than in any of the Western European the least-educated women in their respective countries. However, the country. Third, although the differences in DFLE between high and low differences narrowed in the DFLE estimates, which is consistent with the educational groups were not necessarily larger in CEE countries than in 6 D. Stonkute et al. SSM - Population Health 23 (2023) 101470 Western European countries, the relative disparities by educational household surveys like SHARE usually do not cover institutionalized attainment were more significant in CEE countries due to their generally populations, which might result in a selectively healthier sample. The lower level of life expectancy. proportion of institutionalized individuals varies across countries, add­ Finally, an examination of the intersection of gender and education ing complexity to cross-country comparisons (Permanyer & Spijker, also produced a distinctive pattern, with women in CEE countries 2020). Additionally, research indicates that respondents in SHARE panel exhibiting greater educational disparities in DFLE than their male are positively selected in regards to education and physical health, counterparts. Meanwhile, the findings for Western European countries factors that significantly impact sample attrition and mortality (Schulz were more mixed. Drawing upon prior research, we know that men are & Doblhammer, 2011). However, Muszynska-Spielauer and Spielauer more prone to experiencing life-threatening conditions, whereas women (2022) analyzing a larger number (and more recent) waves found only are more susceptible to impairing but rarely fatal conditions (Crimmins negligible differences in the effects of panel attrition on the et al., 2019). Moreover, it has been acknowledged that the relationship education-specific prevalence of disability and DFLE estimates. Never­ between education and health is influenced by the institutional frame­ theless, future research, especially focusing on cross-sectional estimates work in place (Beckfield et al., 2015; Ross & Mirowsky, 2010). There­ of health expectancies would benefit from including weights that ac­ fore, the combination of biological and social differences between men count for variations in educational composition between the survey and women, along with the institutional and epidemiological context of sample and the general population (Spitzer, 2020), as well as across countries, is likely to contribute to the gendered patterns of educational different countries (Sauerberg, 2021). inequalities observed. Nonetheless, while there is a clear educational The use of the Global Activity Limitation Indicator (GALI) as a gradient, the educational returns to DFLE is heterogeneous across measure of health has limitations due to potential differences across countries. For instance, compared to the low- and medium-educated cultures in self-reported disability (Permanyer & Spijker, 2020). Addi­ groups, DFLE was much higher for the high educated in Czechia (men tionally, in 2008, the translations of the question into different lan­ only) and Estonia. Conversely, in the case of Czech women DFLE was guages was revised. To address this issue, sensitivity analyses were substantially lower for the low-educated group compared to those who conducted using only waves 4–8 and an alternative health indicator were medium- and high educated. While it is difficult to find a precise (limitations in at least one basic activity of daily living) (see supple­ explanation for the fact the lowest educated in CEE have a lower DFLE mentary material Fig. A1-A3). Finally, it is important to note that the than in Western Europe, the cumulative effects of exposures tied to so­ selection of survey may significantly impact the prevalence of GALI, cial disadvantages are likely to play are role, as mentioned earlier in the educational disparities in GALI (Rubio-Valverde et al., 2019). discussion. Albeit with some exceptions, the SHARE data produced population- Notable heterogeneity was found across other findings. Slovenia was level life expectancy estimates that were similar to the Human Mortality shown to have exceptionally high DFLE among high-educated men and Database estimates. The biannual nature of the panel and the exclusion women, even compared to most Western European countries. However, of institutionalized populations should be considered when evaluating due to the small sample size for this country, the confidence intervals are the quality of these estimates. Furthermore, our findings on the associ­ wide, meaning that these estimates should be interpreted with caution. ation between education and health were consistent with the results of Austria was also found to have wide educational disparities. Especially previous research showing that longevity and health tend to be posi­ the low- and medium-educated underperformed in DFLE compared to tively associated with education (Montez & Friedman, 2014). Our other Western European countries. In line with our findings, recent re­ findings of relative differences in disability-free life expectancy by ports have also indicated a high prevalence of limitations in daily ac­ gender and education, as well as generally lower DFLE and relatively tivities and chronic conditions (European Commission, 2022; OECD, larger educational inequalities within CEE countries as compared to 2019), as well as a large proportion of life spent in an unhealthy state in other European countries fall in line with the previous literature. For Austria (Welsh et al., 2021). Denmark was also shown to have low DFLE instance, Permanyer and Bramajo (2022) showed that CEE countries values, particularly among women, which could be attributed to an have been underperforming in health-adjusted life expectancy for de­ exceptionally high prevalence of smoking among the interwar cohort in cades, while Solé-Auró and Gumà (2022) provided evidence of more that country (Lindahl-Jacobsen et al., 2016). pronounced educational inequalities in transitioning to poor health in CEE. Although the study period and the age range for reporting 5.1. Methodological strengths and limitations remaining years of disability-free life vary significantly (Majer et al., 2011; Mäki et al., 2013), s (Majer et al., 2011; Mäki et al., 2013), the Our study proposed combining two estimation approaches: discrete- relative differences in disability-free life expectancy by gender and ed­ time modeling and simulation for estimating multivariate life tables, and ucation are larger in countries considered belonging to the CEE region. the Sullivan method for estimating health expectancies using SHARE data. Using SHARE as a sole data source in estimating educational 6. Conclusions gradient in DFLE is not only an original alternative to the existing methods, it can also offer insights into DFLE trends for countries that The study found substantial health disparities by education in both currently cannot provide differential mortality statistics, which is a Central and Eastern European and Western European countries, with the crucial component of DFLE estimation. We also estimated health and life largest cross-country differences being observed among the low- expectancies for countries that are otherwise under-represented in the educated. Although the differences in disability-free life expectancy literature, which is especially relevant in light of the persisting European between the high and the low educational groups were not necessarily East-West health divide. larger in CEE countries, the relative disparities by educational attain­ Furthermore, unlike vital statistics, the SHARE survey provides ment were more significant due to the generally lower level of life ex­ comprehensive information on the health and demographic and socio­ pectancy in these countries. In addition, the educational inequalities in economic characteristics of individuals. Thus, SHARE data can be used CEE countries were found to be more pronounced for women than for to investigate dimensions other than education, which can help to men, whereas the results for Western European countries were incon­ advance our understanding of health disparities across and between CEE clusive. These findings underline the need for further research on how and Western European countries. At the same time, we acknowledge individual factors impact health that also considers the contextual several data-related limitations. Klotz et al. (2017) have extensively environment in which people live their lives. The results of this study discussed the validity concerns associated with survey data, such as the may advance our understanding of the factors that drive educational under-representation of institutionalized populations and the possible health inequalities, and the effectiveness of national policies designed to misrepresentation of differential mortality between subgroups. Indeed, address them. 7 D. Stonkute et al. SSM - Population Health 23 (2023) 101470 Authors’ contributions Aging (U01_AG09740-13S2, P01_AG005842, P01_AG08291, P30_AG12815, R21_AG025169, Y1-AG-4553-01, IAG_BSR06-11, All authors contributed to the study conception and design. Data OGHA_04-064, HHSN271201300071C, RAG052527A) and from various management: DS; Methodology: DS, AL; Formal analysis and investi­ national funding sources is gratefully acknowledged (see www. gation: DS; Visualization: DS; Writing - original draft preparation: DS; share-project.org). Writing - review and editing: DS, AL, JS; all authors read and approved This analysis uses data or information from the Harmonized Survey the final manuscript. of Health, Ageing and Retirement in Europe (SHARE) dataset and Codebook, Version F as of February 2023 developed by the Gateway to Funding Global Aging Data. The development of the Harmonized SHARE data was funded by the National Institute on Aging (R01 AG030153, RC2 Angelo Lorenti was supported by grants to the Max Planck – Uni­ AG036619, 1R03AG043052). For more information, please refer to versity of Helsinki Center from the Max Planck Society, Jane and Aatos www.g2aging.org. Erkko Foundation, Faculty of Social Sciences at the University of Hel­ sinki, and Cities of Helsinki, Vantaa and Espoo. Acknowledgements Jeroen J.A. Spijker was supported by the research project “The health of people in advanced ages: the analysis of comorbidity, multiple DS gratefully acknowledges the resources made available by the causes of death and gender and socioeconomic inequalities in health International Max Planck Research School for Population, Health and (COMORHEALTHSES)” (IP: JS), funded by the Spanish Ministry of Sci­ Data Science (IMPRS-PHDS). ence and Innovation (PID2020-113934RB-I00), the European Research The authors acknowledge Lara Bister and Sneha Sarah Mani (Inter­ Council (ERC-2019-COGagreement No 864616, HEALIN) and the national Max Planck Research School for Population, Health and Data Catalan Government under the CERCA programme. Science) for their helpful comments on early versions of this manuscript. Ethical statement regarding the use of secondary data Appendix A. Supplementary data This study was conducted with harmonized data from the Survey of Supplementary data to this article can be found online at https://doi. Health and Retirement in Europe. No ethical approval was needed for org/10.1016/j.ssmph.2023.101470. conducting the research. After filling in a user statement, every person with scientific affilia­ References tion can download the data for free as long as the data are used for purely scientific purposes. Avendano, M., Glymour, M. M., Banks, J., & Mackenbach, J. P. (2009). Health The SHARE study is subject to continuous ethics review. During disadvantage in US adults aged 50 to 74 years: A comparison of the health of rich Waves 1 to 4, SHARE was reviewed and approved by the Ethics Com­ and poor Americans with that of Europeans. American Journal of Public Health, 99(3), 540–548. mittee of the University of Mannheim. Wave 4 and the continuation of Beckfield, J., Bambra, C., Eikemo, T. A., Huijts, T., McNamara, C., & Wendt, C. (2015). the project were reviewed and approved by the Ethics Council of the An institutional theory of welfare state effects on the distribution of population Max Planck Society. In addition, the country implementations of SHARE health. Social Theory & Health, 13(3), 227–244. Bergmann, M., Kneip, T., De Luca, G., & Scherpenzeel, A. (2019). Survey participation in were reviewed and approved by the respective ethics committees or the Survey of Health, Ageing and Retirement in Europe (SHARE), Wave 1-7. Based on institutional review boards whenever this was required. The numerous Release 7.0.0. SHARE Working Paper Series 41-2019. Munich: MEA, Max Planck reviews covered all aspects of the SHARE study, including sub-projects Institute for Social Law and Social Policy. Börsch-Supan, A. M., Brandt, M., Hunkler, C., Kneip, T., Korbmacher, J., Malter, F., and confirmed the project to be compliant with the relevant legal Schaan, B., Stuck, S., & Zuber, S. (2013). Data Resource Profile: The Survey of norms and that the project and its procedures agree with international Health, Ageing and Retirement in Europe (SHARE). International Journal of ethical standards. Epidemiology. https://doi.org/10.1093/ije/dyt088 Börsch-Supan, A. (2022a). Survey of Health, Ageing and Retirement in Europe (SHARE) Wave 1; Release Version: 8.0. 0; SHARE-ERIC. Dataset. https://doi.org/10.6103/ Declaration of competing interest SHARE.w1.800 Börsch-Supan, A. (2022b). Survey of Health, Ageing and Retirement in Europe (SHARE) Wave 2; Release Version: 8.0. 0; SHARE-ERIC. Dataset. https://doi.org/10.6103/ None declared. SHARE.w2.800 Börsch-Supan, A. (2022c). Survey of Health, Ageing and Retirement in Europe (SHARE) Data availability Wave 4; Release Version: 8.0. 0; SHARE-ERIC. Dataset. https://doi.org/10.6103/ SHARE.w4.800 Börsch-Supan, A. (2022d). Survey of Health, Ageing and Retirement in Europe (SHARE) All study code is fully replicable and available in the following Wave 5; Release Version: 8.0. 0; SHARE-ERIC. Dataset. https://doi.org/10.6103/ GitHub repository: https://github.com/DonataStonkute/dfle_50. SHARE.w5.800 This paper uses data from SHARE Waves 1, 2, 4, 5, 6, and 8 (DOIs: Börsch-Supan, A. (2022e). Survey of Health, Ageing and Retirement in Europe (SHARE) Wave 6; Release Version: 8.0. 0; SHARE-ERIC. Dataset. https://doi.org/10.6103/ 10.6103/SHARE.w1.800, 10.6103/SHARE.w2.800, 10.6103/SHARE. SHARE.w6.800 w4.800, 10.6103/SHARE.w5.800, 10.6103/SHARE.w6.800, 10.6103/ Börsch-Supan, A. (2022f). Survey of Health, Ageing and Retirement in Europe (SHARE) SHARE.w8.800) see Börsch-Supan et al. (2013) for methodological de­ Wave 8; Release Version: 8.0. 0; SHARE-ERIC. Dataset. https://doi.org/10.6103/ SHARE.w8.800 tails.(1) The SHARE data collection has been funded by the European Brown, D. C., Hayward, M. D., Montez, J. K., Hummer, R. A., Chiu, C.-T., & Commission, DG RTD through FP5 (QLK6-CT-2001-00360), FP6 Hidajat, M. M. (2012). The significance of education for mortality compression in (SHARE-I3: RII-CT-2006-062193, COMPARE: CIT5-CT-2005-028857, the United States. Demography, 49(3), 819–840. Cambois, E., Solé-Auró, A., Brønnum-Hansen, H., Egidi, V., Jagger, C., Jeune, B., … SHARELIFE: CIT4-CT-2006-028812), FP7 (SHARE-PREP: GA N◦ 211909, Robine, J.-M. (2016). Educational differentials in disability vary across and within SHARE-LEAP: GA N◦ 227822, SHARE M4: GA N◦ 261982, DASISH: GA welfare regimes: A comparison of 26 European countries in 2009. Journal of N◦ 283646) and Horizon 2020 (SHARE-DEV3: GA N◦ 676536, SHARE- Epidemiology & Community Health, 70(4), 331–338. Cameron, A. C., & Trivedi, P. K. (2005). Microeconometrics: Methods and applications. COHESION: GA N◦ 870628, SERISS: GA N◦ 654221, SSHOC: GA Cambridge university press. N◦ 823782, SHARE-COVID19: GA N◦ 101015924) and by DG Employ­ Carlson, P. (1998). Self-perceived health in East and west Europe: Another European ment, Social Affairs & Inclusion through VS 2015/0195, VS 2016/0135, health divide. Social Science & Medicine, 46(10), 1355–1366. VS 2018/0285, VS 2019/0332, and VS 2020/0313. Additional funding Carlson, P. (2004). The European health divide: A matter of financial or social capital? Social Science & Medicine, 59(9), 1985–1992. from the German Ministry of Education and Research, the Max Planck Christensen, K., Doblhammer, G., Rau, R., & Vaupel, J. W. (2009). Ageing populations: Society for the Advancement of Science, the U.S. National Institute on The challenges ahead. Lancet, 374(9696), 1196–1208. 8 D. Stonkute et al. SSM - Population Health 23 (2023) 101470 Crimmins, E. M., Kim, J. K., & Solé-Auró, A. (2011). Gender differences in health: Results OECD. (2019). Health at a glance. https://doi.org/10.1787/4dd50c09-en, 2019. from SHARE, ELSA and HRS. The European Journal of Public Health, 21(1), 81–91. Oksuzyan, A., Juel, K., Vaupel, J. W., & Christensen, K. (2008). Men: Good health and Crimmins, E. M., Shim, H., Zhang, Y. S., & Kim, J. K. (2019). Differences between men high mortality. Sex differences in health and aging. Aging-Clinical & Experimental and women in mortality and the health dimensions of the morbidity process. Clinical Research, 20, 91–102. Chemistry, 65(1), 135–145. Permanyer, I., & Bramajo, O. (2022). Increased longevity in Europe: Adding years to life Cutler, D. M., & Lleras-Muney, A. (2010). Understanding differences in health behaviors or life to years? Perspectives Demogràfiques, 28, 1–4. by education. Journal of Health Economics, 29(1), 1–28. Permanyer, I., & Spijker, J. (2020). Socioeconomics and the macro-and micro-level Eikemo, T. A., & Bambra, C. (2008). The welfare state: A glossary for public health. determinants of global health inequality. Handbook of Global Health (pp. 1–36). Journal of Epidemiology & Community Health, 62(1), 3–6. Pinho-Gomes, A.-C., Vassallo, A., Carcel, C., Peters, S., & Woodward, M. (2022). Gender Eikemo, T. A., Huisman, M., Bambra, C., & Kunst, A. E. (2008). Health inequalities equality and the gender gap in life expectancy in the European Union. BMJ Global according to educational level in different welfare regimes: A comparison of 23 Health, 7(2), Article e008278. European countries. Sociology of Health & Illness, 30(4), 565–582. Preston, S. H., Heuveline, P., & Guillot, M. (2001). Demography. Measuring and modeling European Commission. (2022). State of health in the EU: Companion report 2022. population processes (Vol. 27). Blackwell Publishers. Publications Office of the European Union Luxembourg. Robine, J.-M., & Jagger, C. (2003). Creating a coherent set of indicators to monitor Eurostat. (2022). Life expectancy by age and sex [DEMO_MLEXPEC]. health across Europe: The euro-REVES 2 project. Euro-REVES Group The European Geyer, S., Hemström, Ö., Peter, R., & Vågerö, D. (2006). Education, income, and Journal of Public Health, 13(suppl_3), 6–14. occupational class cannot be used interchangeably in social epidemiology. Empirical Robine, J.-M., Romieu, I., & Cambois, E. (1999). Health expectancy indicators. Bulletin of evidence against a common practice. Journal of Epidemiology & Community Health, 60 the World Health Organization, 77(2), 181. (9), 804–810. Romero-Ortuno, R., Fouweather, T., & Jagger, C. (2014). Cross-national disparities in sex Haggard, S., & Kaufman, R. R. (2009). The Eastern European welfare state in comparative differences in life expectancy with and without frailty. Age and Ageing, 43(2), perspective. Springer. 222–228. HMD. Human Mortality Database Max Planck Institute for Demographic Research Ross, C. E., Masters, R. K., & Hummer, R. A. (2012). Education and the gender gaps in (Germany), University of California, Berkeley (USA), and French Institute for health and mortality. Demography, 49(4), 1157–1183. Demographic Studies (France). Available at: www.mortality.org. Ross, C. E., & Mirowsky, J. (2010). Why education is the key to socioeconomic Jagger, C., Crimmins, E. M., Saito, Y., Yokota, R. T. D. C., Van Oyen, H., & Robine, J.-M. differentials in health. Handbook of medical sociology, 6, 33–51. (2020). International handbook of health expectancies. Springer. Rubio-Valverde, J. R., Nusselder, W. J., & Mackenbach, J. P. (2019). Educational Jagger, C., Gillies, C., Moscone, F., Cambois, E., Van Oyen, H., Nusselder, W., & inequalities in global activity limitation indicator disability in 28 European Robine, J.-M. (2008). Inequalities in healthy life years in the 25 countries of the countries: Does the choice of survey matter? International Journal of Public Health, 64, European union in 2005: A cross-national meta-regression analysis. Lancet, 372 461–474. (9656), 2124–2131. Sauerberg, M. (2021). The impact of population’s educational composition on Healthy Jagger, C., Weston, C., Cambois, E., Van Oyen, H., Nusselder, W., Doblhammer, G., … Life Years: An empirical illustration of 16 European countries. SSM-Population Robine, J.-M. (2011). Inequalities in health expectancies at older ages in the Health, 15, Article 100857. European union: Findings from the survey of health and retirement in Europe Schulz, A., & Doblhammer, G. (2011). Longitudinal research with the second wave of share: (SHARE). Journal of Epidemiology & Community Health, 65(11), 1030–1035. Representativeness of the longitudinal sample and the mortality follow-up. Kiely, K. M., Brady, B., & Byles, J. (2019). Gender, mental health and ageing. Maturitas, Shkolnikov, V. M., Andreev, E. M., Jasilionis, D., Leinsalu, M., Antonova, O. I., & 129, 76–84. McKee, M. (2006). The changing relation between education and life expectancy in Klotz, J., Göllner, T., & Austria, S. (2017). Estimating differential mortality from EU-SILC central and eastern Europe in the 1990s. Journal of Epidemiology & Community longitudinal data: A feasibility study (Vol. 4). FACTAGE project report D. Health, 60(10), 875–881. Lindahl-Jacobsen, R., Rau, R., Jeune, B., Canudas-Romo, V., Lenart, A., Christensen, K., & Siegrist, J., & Marmot, M. (2004). Health inequalities and the psychosocial Vaupel, J. W. (2016). Rise, stagnation, and rise of Danish women’s life expectancy. environment—two scientific challenges. Social Science & Medicine, 58(8), Proceedings of the National Academy of Sciences, 113(15), 4015–4020. 1463–1473. Luy, M., & Minagawa, Y. (2014). Gender gaps-Life expectancy and proportion of life in Solé-Auró, A., Beltrán-Sánchez, H., & Crimmins, E. M. (2015). Are differences in poor health. Health Reports, 25(12), 12. disability-free life expectancy by gender, race, and education widening at older Mackenbach, J. P. (2017). Nordic paradox, southern miracle, eastern disaster: ages? Population Research and Policy Review, 34, 1–18. Persistence of inequalities in mortality in Europe. The European Journal of Public Solé-Auró, A., & Gumà, J. (2022). Healthy) aging patterns in Europe: A multistate health Health, 27(suppl_4), 14–17. transition approach. Journal of Population Ageing, 1–23. Mackenbach, J. P., Kulhánová, I., Bopp, M., Deboosere, P., Eikemo, T. A., Hoffmann, R., Spitzer, S. (2020). Biases in health expectancies due to educational differences in survey … Menvielle, G. (2015). Variations in the relation between education and cause- participation of older Europeans: It’s worth weighting for. The European Journal of specific mortality in 19 European populations: A test of the “fundamental causes” Health Economics, 21(4), 573–605. theory of social inequalities in health. Social Science & Medicine, 127, 51–62. Statistics, U. I. S. (2012). International standard classification of education: ISCED 2011. Mackenbach, J. P., Valverde, J. R., Artnik, B., Bopp, M., Brønnum-Hansen, H., UNESCO Institute for Statistics Montreal. Deboosere, P., … Martikainen, P. (2018). Trends in health inequalities in 27 Sullivan, D. F. (1971). A single index of mortality and morbidity. HSMHA Health Reports, European countries. Proceedings of the National Academy of Sciences, 115(25), 86(4), 347. 6440–6445. Tawiah, R., Jagger, C., Anstey, K. J., & Kiely, K. M. (2021). Trends in disability-free life Majer, I. M., Nusselder, W., Mackenbach, J., & Kunst, A. (2011). Socioeconomic expectancy at age 50 years in Australia between 2001 and 2011 by social inequalities in life and health expectancies around official retirement age in 10 disadvantage. Journal of Epidemiology & Community Health, 75(11), 1056–1062. Western-European countries. Journal of Epidemiology & Community Health, 65(11), Teachman, J. D., & Hayward, M. D. (1993). Interpreting hazard rate models. Sociological 972–979. Methods & Research, 21(3), 340–371. Mäki, N., Martikainen, P., Eikemo, T., Menvielle, G., Lundberg, O., Östergren, O., … Valverde, J. R. R., Mackenbach, J. P., & Nusselder, W. J. (2021). Trends in inequalities in Mackenbach, J. P. (2013). Educational differences in disability-free life expectancy: disability in Europe between 2002 and 2017. Journal of Epidemiology & Community A comparative study of long-standing activity limitation in eight European countries. Health, 75(8), 712–720. Social Science & Medicine, 94, 1–8. Van Oyen, H., Cox, B., Jagger, C., Cambois, E., Nusselder, W., Gilles, C., & Robine, J.-M. Minagawa, Y. (2013). Inequalities in healthy life expectancy in Eastern Europe. (2010). Gender gaps in life expectancy and expected years with activity limitations Population and Development Review, 39(4), 649–671. at age 50 in the European union: Associations with macro-level structural indicators. Montez, J. K., & Friedman, E. M. (2014). Educational attainment and adult health: Under European Journal of Ageing, 7, 229–237. what conditions is the association causal? Social Science & Medicine, 127, 1–7, 1982. Van Oyen, H., Nusselder, W., Jagger, C., Kolip, P., Cambois, E., & Robine, J.-M. (2013). Montez, J. K., Zajacova, A., Hayward, M. D., Woolf, S. H., Chapman, D., & Beckfield, J. Gender differences in healthy life years within the EU: An exploration of the (2019). Educational disparities in adult mortality across US states: How do they “health–survival” paradox. International Journal of Public Health, 58(1), 143–155. differ, and have they changed since the mid-1980s? Demography, 56(2), 621–644. Van Oyen, H., Van der Heyden, J., Perenboom, R., & Jagger, C. (2006). Monitoring Montez, J. K., Zhang, W., Zajacova, A., & Hamilton, T. G. (2018). Does college major population disability: Evaluation of a new global activity limitation indicator (GALI). matter for women’s and men’s health in midlife? Examining the horizontal Sozial-und Präventivmedizin, 51(3), 153–161. dimensions of educational attainment. Social Science & Medicine, 198, 130–138. Welsh, C. E., Matthews, F. E., & Jagger, C. (2021). Trends in life expectancy and healthy Muszynska-Spielauer, M., & Spielauer, M. (2022). The effect of sample attrition in the EU life years at birth and age 65 in the UK, 2008–2016, and other countries of the EU28: statistics on income and living conditions on the estimates of eurostat’s healthy life years. An observational cross-sectional study. The Lancet Regional Health-Europe, 2, Article Nusselder, W., Looman, C., Van Oyen, H., Robine, J.-M., & Jagger, C. (2010). Gender 100023. differences in health of EU10 and EU15 populations: The double burden of EU10 men. European Journal of Ageing, 7, 219–227. 9