Disparate ageing: The role of education and socioeconomic gradients in future health and disability in an international context

Population ageing is a critical policy challenge to advanced economies around the world. There were 703 million persons aged 65 years or over in the world in 2019. The number of older persons is projected to double to 1.5 billion in 2050. The share of the population aged 65 years is expected to rise from 9 percent today to 16 percent by 2050. Put another way, one in six people in the world will be aged 65 years or over (United Nations, 2015a, 2015b, 2015c). These trends pose both fiscal and population health challenges, principal among these being the persistent and large socioeconomic gradients in health. Many older persons retain overall good health and functioning well into old age, but—in the context of rapid population ageing—disparities can be exacerbated. Some of these differences are attributable to genetics, but other policy mutable factors play an important role: factors such as the natural and physical environment (air pollution and accessibility), risky behaviors (drinking, smoking and physical inactivity), and individual characteristics such as occupation and level of income. Therefore, if on one side ageing is driven by biological changes, on the other side the ageing itself reflects the accumulated effects of one's exposure to a history of external risks, and can further be influenced by social changes, such as isolation and loss of loved ones. The end result is often a complex combination of both individual characteristics and other health determinants; hence, health disparities at older age often reflect accumulated disadvantage. There is an emerging, increasing and widespread consensus about the positive role that education can have in reducing this accumulated disadvantage. This advantage is believed to occur mostly because education behaves as an enabler, which helps individuals to use more properly the inputs in the health production function (Grossman, 1972, 1975, 2000). In this way several factors contribute to the role of education in influencing health outcomes. Social and biological processes initiated in early life influence both educational achievement and adult health. Education has a direct and indirect role in driving the relationship between socio-economic status (SES) and health. Better education increases the chances to pursue personal and professional success, which in turn determine socio-economic outcomes such as access to better occupational positions and higher incomes. As such, education has an indirect influence on health by giving the possibility to improve the allocation of resources and invest more heavily in health. However, labor market participation and higher incomes that better educated individuals earn are only a partial explanation behind the indirect education-health link (Brunello et al., 2015, Grossmann and Kaestner, 1997). Education enables individuals to be more efficient in maintaining good health (Grossman, 1972) by prompting them to make better health choices (Brunello et al., 2015, Rosenzweig and Schultz, 1983), increasing their willingness and ability to access and use information (Goldman et al., 2015), and increasing their investment in social capital. The direct impact is associated with productive abilities, which help individuals act more effectively as agents by fostering generic skills such as information-gathering and decision-making. This direct aspect of education, learned effectiveness, promotes sense of control, developing habits of preventing and solving problems, regardless of available resources and prevailing conditions. Ruhm (2012) suggests that cognitive functioning and the resulting deliberative abilities contribute to a correct evaluation of long-run implications of lifestyle choices. Put differently, as suggested by Kenkel et al. (2006), better education enables individuals to obtain superior health outcomes from a fixed set of inputs, due to the better choices they make. It then follows that education improves health conditions and reduces health disparities (see Cutler and Lleras-Muney, 2008; Cutler and Lleras-Muney, 2010, Glaeser et al., 2000, for reviews) through different channels. Educational attainment is a particularly profound predictor of length of life, now surpassing both race (Harper et al., 2007; Kochanek et al., 2013) and gender (Arias, 2007; Rogers et al., 2010) in importance in the United States. Furthermore, a large literature has documented substantial associations between education and mortality, health (self-reported health, obesity, etc.) and health behaviors (smoking, excessive drinking, exercise, preventive care use, etc.). These relationships exist but vary in magnitude across countries. In the United States, those at age 25 with more than a college degree can expect to live up to seven years longer than those without a college degree (Meara et al., 2008; Hummer and Hernandez, 2013). It should be noted, however, that some studies (Clark and Royer, 2013; Behrman et al., 2011) find no causal impact of schooling on health. Educational differences in life expectancy have also widened since the 1980s, across all major race and gender groups (Goldman and Smith, 2011; Olshansky et al., 2012) and in all regions of the United States (Montez and Berkman, 2014). According to Chetty et al. (2017), inequality in life expectancy increased between 2001 and 2014 (by 2.34 years for men and 2.91 years for women in the top 5% of the income distribution, which is a good proxy of education level), but by only 0.32 years for men and 0.04 years for women in the bottom 5%. Furthermore, those with less than a high school diploma exhibit higher lifespan variability and can expect greater uncertainty in their time of death (Brown et al., 2012; Edwards and Tuljapurkar, 2005; Sasson, 2016). By contrast, college-educated Americans live longer, on average, and exhibit greater compression of mortality, with deaths narrowly concentrated at the upper tail of the age distribution—a pattern similarly observed in several European countries (van Raalte et al., 2011). According to this evidence, there is a connection between education and life expectancy; individuals with a tertiary degree tend to live longer than secondary education graduates, who already live longer than those with only primary education. These gradients are found along several dimensions of health status indicators (i.e., morbidity, mortality, and risk factors) and are documented over time as well as within and between countries. Often they are fairly large and cause many premature deaths each year. More importantly, the educational health gradients are increasing over time (partly due to population ageing). This outcome is rather striking if one considers the evolution of medical technology (and its diffusion) and the amount of effort that many high-income countries make in order to reduce health inequalities (especially in Europe). This phenomenon, which per se is execrable, remains politically unacceptable in view of declining overall mortality rates and steadily increasing life expectancy (Oeppen & Vaupel, 2002; Leon, 2011; Wang et al., 2013). In this respect, reducing health inequalities is a matter of fairness and social justice (Marmot et al., 2010). The studies in this special issue investigate the consequences of these phenomena across countries using consistent policy simulation approaches. The aim is to help developed countries design effective policy responses, and to provide a tool that could be used to fill knowledge gaps in many developing countries. Despite the existence of reliable models predicting long-term population structure by age and sex (Eurostat, 2020; United Nations, 2020a, 2020b), long-term forecasts of population health exist only in the U.S. (Goldman et al., 2013) and UK (Guzman-Castillo et al., 2017). Concerning continental Europe, the only available tool for policy makers is the one implemented by the Ageing Working Group (AWG) of the European Commission (EC, 2015), which predicts long-term trends in social security expenditure based on predictions of GDP rather than estimates of population health status. In this context, the availability of a reliable quantitative tool able to assess the impact of future demographic and epidemiological changes on population health status and healthcare demand, and on governments‘ budgets, is crucial. Health care spending accounts for a large share of public spending in all industrialized countries, to which households add a relatively large portion of private spending; health care spending is also one of the most important components of the social security expenditure. Most important, health care spending trends are expected to be upwards over the next decades. According to both the European Union and the Centers for Medicare & Medicaid Services (CMS) in the U.S., predicting the future evolution of the demand for health care services and the related health care expenditure is one of crucial challenges for all industrialized countries (Goldman et al., 2004 and Przywara, 2010). As trends in spending continue to rise, there is an increasing pressure on government budgets, health services provision and patients‘ personal finances. For a better planning of policy interventions, policy makers within OECD countries have promoted individual and collective initiatives to help forecast these trends. To obtain precise forecasts of the levels of health care spending and to establish adequate policy responses, it is paramount to have tools that allow estimating future health care expenditure and costs. Since “the complexity of the systems and multiplicity of factors affecting both total and public spending make this a highly complicated task, where results will always be surrounded by considerable uncertainties” (Przywara, 2010), fulfilling this task requires sophisticated and complex modeling methods that take into account the evolution of health, economic and demographic variables at individual and cohort levels. Microsimulation models (MSMs) have emerged as a useful tool to answer these questions (Astolfi et al., 2011, 2012). Among this class of models, the Future Elderly Model (FEM) (Goldman et al., 2004), using the Health and Retirement Study (HRS) data, has displayed the potential for microsimulations to help shape policy in the U.S. (for a recent application, see National Academies of Science, 2015), in Japan (Chen et al., 2016) and internationally, as modified versions of FEM have been employed in other countries (e.g., both the FEM end the EUFEM models have been used to study alternative policy scenarios by the OECD [Atella et al., 2017]). Typically, research on health disparities by SES focuses on narrow outcomes, usually mortality. This project will innovate by focusing also on the role of disease dynamics in producing the education-health gradient. This exercise may prove instrumental to guiding future research about the gradient as we present novel results obtained from a family of FEM like models, a multi-risk and multi-morbidity state transition dynamic micro-simulation model able to deliver long-term projections of the health status of the population in a country. Furthermore, through FEM we can retrieve the whole distribution of selected outcomes and not just the average. This approach allows, in a convenient way, to explore the several interesting aspects of how SES affects health status and all related economic effects. FEM accounts also for the multidimensional nature of health status by imposing estimated correlations between couples of individual characteristics. By implementing differential risk factors and conditional probabilities of disease incidence, disability incidence and mortality in MSMs, we will produce refined estimates of health events in the lifecycle continuum. The papers presented here draw on decades of longitudinal survey data to produce a better understanding of the dynamics linking narrow outcomes, such as unhealthy behaviors and disease incidence, and broad ones such as healthy life expectancy. The analyses have been conducted through harmonized cross-country comparisons of the education-health gradient, which is novel in this literature, given that most existing studies based on the SES-health gradient investigate a narrow component of the gradient in a single country; few assess several aspects in a single country; and even fewer scrutinize several aspects of the gradient across numerous countries, though exceptions exist (Berkman et al., 2011; Cutler et al., 2015). These analyses have been carried out by a global team of collaborators to forecast long-term trends in disease dynamics. The papers in this collection report findings from 15 countries: Austria, Belgium, Denmark, France, Germany, Italy, Netherland, Spain, Sweden, Switzerland, Canada, Japan, Korea, Mexico, and the United States. Results are organized into three sections. The first section deals with issues related to modeling health around the world and presents three contributions where FEM models are used to produce forecasts of health indicators such as mortality and morbidity on major chronic diseases in several EU countries, in Japan and in the United States. The second section explores the consequences of some health interventions in the South Korea, Singapore, and the United States, looking at the improved survival for individuals with common chronic conditions in the U.S. Medicare population, and at the role that changes in smoking can have on life expectancy and chronic disease in South Korea, Singapore, and the United States. Finally, the third section analyses the health benefits of social interventions (education programs) in Canada and in the United States. In the first paper by Atella et al. (2021), the main aim is to fill a knowledge gap in terms of future trends in mortality, disease prevalence, life expectancy and patterns of inequalities in health outcomes within the EU. In fact, in spite of the existence of several reliable models predicting the population structure by age and sex in the long-term (United Nations [UN], 2015a, 2015b, 2015c, 2019), models allowing forecasting population long-term health status at individual level are rare (see Goldman et al. (2013) for the United States and Guzman-Castillo et al. (2017) for the United Kingdom). In the EU, the only tool available to policymakers is the one implemented by the AWG of the European Commission (EC, 2015), which predicts long-term trends in social security expenditure based on predictions of GDP rather than estimates of the health status of the population. Much less is available in other OECD countries. Therefore, the availability of a reliable quantitative tool able to assess the impact of future demographic and epidemiological changes on population health status and healthcare demand, and on governments' budgets, is crucial and could offer an important support to policy makers to design and implement effective and sustainable policies. Using harmonized data from the Gateway to Global Ageing project to obtain a homogenous comparative analysis for 10 European countries, plus Korea, Mexico and United States, the paper presents reliable forecasts of the evolution of prevalence of major chronic conditions, life expectancy, disability free and quality adjusted life years, and health expenditures; the analysis provide evidence of a growing SES gradient in the health status of elderly patients. In a similar manner, the second paper by Kasajima et al. (2020) explores the same issue in Japan relying on a different methodological approach. In fact, multistate-transition MSMs such as the U.S. FEM have been developed based on panel data collection, but these data may not be always available. The authors propose a pseudo-panel method using repeated cross-sectional representative surveys as a complementary approach, and they specifically applied the model to Japan's population. Their results confirm the reliability of the approach as their estimated morbidity and mortality rates successfully replicated governmental projections of population pyramids and matched cardiovascular and cancer incidences reported in existing epidemiological studies. Furthermore, they are able to produce future projection of stroke and heart disease from which it is possible to assess lower prevalence than expected from static models, presumably because of recent declining trends in disease incidence and fatality. In the third paper by Leaf et al. (2021), the authors perform a series of validation analyses on the U.S. FEM. Given the role that these models should have in helping policy makers designing interventions, assessing their internal and external validity versus other models is of great important. In fact, when compared to traditional models, MSMs of data that may forecasting Therefore, the authors perform some validation analyses of the mortality and quality of life forecasts using a of the FEM estimated on early of data from the those estimates at they FEM mortality and projections to the mortality and observed over the same of time and to forecasts of mortality and the same they find that FEM projections are in with observed mortality rates and Finally, they a further important exercise in predicting quality of life and longitudinal outcomes, that traditional models The results of this analysis further confirm the of the FEM estimates versus the models, further the that these models can be used to provide policy makers with the correct of information they The second section of this special issue is on the estimates of the effects that could be by health interventions or The paper by et al. at the potential improved survival for individuals with common chronic conditions in the U.S. Medicare population if they could be matched with their and over from high-income countries. In the authors compared mortality trends from 2004 to this exercise, they found that the in survival for the but the for persons over which is the with access to public for the over 65 they estimated models by race and for chronic conditions and time trends. predicting survival rates for all 16 of and heart disease they found survival with as a survival improved and disparities for individuals with and The paper by et al. (2020) explores the role that changes in smoking can have on life expectancy and chronic disease in South Korea, Singapore, and the United States, three countries have a different risk factors and diseases with It is that smoking is a risk for and healthy in ageing with high rates of smoking such as in However, is about smoking interventions at to to in life expectancy and prevalence of chronic diseases and how the effects vary across Using models the authors the health effects of smoking by an of smoking among of in South Korea, Singapore, and the United States. The results how smoking interventions may have economic and social for life that vary across countries. In they find that life expectancy by to 1.5 years among and to years among that the life benefits were for those who have been Put differently, for to life expectancy and reduce the chronic disease among the future interventions should Finally, the third section analyses the health benefits of education in Canada and in the United States. In the et al. used a dynamic health model to investigate the to college in Canada in terms of health and mortality In they investigate how interventions which college among high school may impact their health health care and life expectancy. The results obtained an important role for education given that they find large both in terms of years years at age in the prevalence of health conditions in and for and health care in for Furthermore, the paper that the through which education on mortality mostly through the incidence of health conditions as well as provide a survival advantage conditional on Finally, they produce evidence on the impact of college on long-term health outcomes by the a which college The impact on mortality is found to be than those estimated from the which suggests substantial to college education in terms of healthy life which we to be around one million The and paper in this special issue is an by and that explores the role of early education on health. on by et al. (2013) and et al. the authors confirm the important role that in education can have on early in between and age The evidence across several important social outcomes, more in better and Using data from The an early to in further with data from several other they use the FEM to outcomes for a cohort most representative of specifically on health their results a in quality adjusted when is to the of the Furthermore, early education is relatively their findings that the health in papers presented in this special issue were presented in an on at the where the results were with an of and policy this special issue the learned to a they have no of

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.