Healthcare system and the wealth–health gradient: A comparative study of older populations in six countries

Social Science & Medicine 119 (2014) 18e26

Contents lists available at ScienceDirect

Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed

Healthcare system and the wealthehealth gradient: A comparative study of older populations in six countries Dina Maskileyson Tel-Aviv University, Faculty of Social Sciences, Labor Studies Department, P.O.B. 39040, Tel-Aviv 69978, Israel

a r t i c l e i n f o

a b s t r a c t

Article history: Received 29 May 2013 Received in revised form 31 July 2014 Accepted 13 August 2014 Available online 14 August 2014

The present study provides a comparative analysis of the association between wealth and health in six healthcare systems (Sweden, the United Kingdom, Germany, the Czech Republic, Israel, the United States). National samples of individuals fifty years and over reveal considerable cross-country variations in health outcomes. In all six countries wealth and health are positively associated. The findings also show that state-based healthcare systems produce better population health outcomes than privatebased healthcare systems. The results indicate that in five out of the six countries studied, the wealth ehealth gradients were remarkably similar, despite significant variations in healthcare system type. Only in the United States was the association between wealth and health substantially different from, and much greater than that in the other five countries. The findings suggest that private-based healthcare system in the U.S. is likely to promote stronger positive associations between wealth and health. © 2014 Elsevier Ltd. All rights reserved.

Keywords: Healthcare systems Health inequality Physical health Wealthehealth gradient Older population Cross-national comparative analysis

1. Introduction Social scientists have long demonstrated that wealthy people tend to be healthier and live longer than poor persons (Kawachi et al., 1997; Wilkinson and Pickett, 2006). The positive association between economic resources and health (usually referred to as “health gradient”) can be attributed to two reasons. First, economic resources can be used to purchase better healthcare services (e.g., Van Doorslaer et al., 2006). Second, poor health may lead to a depletion of economic resources (e.g., Smith, 2005). While both approaches are logical and quite convincing, they are by no means contradictory. To date most cross-national studies have focused on the association between economic well-being and health, showing that the average health of a population is likely to rise with economic growth (e.g., Hurd and Kapteyn, 2003) and to decline with higher inequality (e.g., Pickett and Wilkinson, 2007). However, only few studies have systematically investigated the extent to which the wealthehealth gradient differs across countries (e.g., Avendano et al., 2009; Semyonov et al., 2013) and none have examined whether the wealthehealth gradient varies in magnitude across different types of healthcare systems using individual-level data. The data for this study consist of six national samples of populations

E-mail addresses: [email protected], [email protected]. http://dx.doi.org/10.1016/j.socscimed.2014.08.013 0277-9536/© 2014 Elsevier Ltd. All rights reserved.

fifty years of age and over. The comparative analysis enables to delineate the relationship between wealth and health and to better understand whether healthcare system types affect the association between wealth and health. The contribution of this research is twofold. First, it provides a cross-national comparative study of the link between different healthcare system types and overall population health. Second, it examines, for the first time, the extent to which the association between wealth and health among older adults differs across countries, and ascertains whether the type of a nation's healthcare system is tied to this association. Thus, this research not only advances theoretical knowledge in the fields of health and gerontology, but also gives initial insights into the ways in which health policies affect wealthehealth inequality. 2. Theoretical background 2.1. Wealthehealth gradient A plethora of research on health gradient has examined the association between socioeconomic well-being of individuals and various indicators of health within specific countries. These studies have repeatedly found that individual socioeconomic statusdmeasured either by income, occupational status or educationdis positively associated with healthdmeasured by various health indicators, including self-reported health, measures of physical and

D. Maskileyson / Social Science & Medicine 119 (2014) 18e26

mental illness, long-term disabilities, functionality and longevity even after controlling for individual socio-demographic characteristics (Huisman et al., 2003; Deaton, 2008). The other body of literature includes comparative studies that focus on the relationship between structural characteristics of ecological units (i.e., nations, regions, counties etc.) and indicators of population health and mortality (Kawachi et al., 1997; Pickett and Wilkinson, 2007). In general, all cross-national ecological studies provide repeated support for the argument that a country's population, on average, benefits from greater availability of economic resources and from a more equal distribution of these resources (Kawachi et al., 1997; Eikemo et al., 2008c). That is, the population's health tends to rise and mortality rates tend to decline, with increased economic resources (measured by gross domestic product) and income equality (measured by the Gini index). The negative association between income inequality and population health is often interpreted to be the result of limited access of large segments of the population (usually the poor) to medical services and medical resources in non-egalitarian social systems (see Wilkinson, 2006 for a detailed discussion).

2.2. Framework for analyzing wealthehealth gradient Recently, welfare regime theory has gained epidemiological attention for its relevance in evaluating cross-national differences in population health and health inequality (Brennenstuhl et al., 2012). Comparative studies of health inequality have shown that countries exhibit substantial variation in population health and the health gradient, and that welfare states play an influential role in these public health outcomes (e.g., Bambra, 2006). Furthermore, countries with similar welfare and healthcare policies are likely to achieve, on average, similar population health outcomes (e.g., Chung and Muntaner, 2006). Studies have unanimously agreed that countries with social-democratic welfare regimes enhance average population health, compared to other regime types, due to their extensive social protections programs and universalism. For example, classifying 19 wealthy countries into four types of welfare regimes, Chung and Muntaner (2007) found that over a 39-year period social-democratic countries exhibit significantly better health outcomes, compared to other countries. Similarly, Eikemo et al. (2008a) reported that populations in countries with Scandinavian and Anglo-Saxon welfare regimes tend to have better health in comparison to Southern and East-European welfare regimes. Overall, this literature suggests that social-democratic welfare state regimes provide a combination of different policies (e.g., higher levels of employment and decommodification, universal access to welfare services including public health insurance etc.) which leads to better health outcomes (e.g., Chung and Muntaner, 2007). Yet, in contrast to the consistent evidence in terms of overall health, the research on welfare state regimes and the magnitude of health inequalities show quite ambiguous results. Specifically, the findings reveal that social-democratic welfare state regimes do not systematically exhibit the smallest health inequalities compared to conservative and liberal countries (e.g., Beckfield and Krieger, 2009). For instance, Eikemo et al. (2008b) found that the AngloSaxon welfare state regime had the highest income-related inequalities in health, while contrary to expectations health inequalities in Scandinavian welfare regimes were found to be significantly higher than those in the Bismarckian. Another study by Mackenbach et al. (2008) revealed significant differences among 22 European countries, in terms of the magnitude of inequalities in mortality and self-assessed health For instance, inequalities in mortality from cardiovascular disease were higher in socialdemocratic countries compared to other countries.

19

A series of recent critical reviews of empirical studies that apply welfare regime typology in comparative health research have identified issues still to be addressed in this field. Beckfield and Krieger (2009) conclude that there is no clear evidence on the relations between political systems and the magnitude of health inequities. That is, while social-democratic policies are likely to positively affect population health, the transition to capitalist economies and neoliberal reforms tend to expand health disparities. Additionally, Brennenstuhl et al. (2012) present a critical review of 33 empirical studies. They suggest that findings of significantly better health outcomes in social-democratic welfare regime are more likely to focus on overall population health, rather than on socioeconomic inequalities in health. In general, these studies have concluded that health differences by regime are not always consistent with welfare regime theory and that a broad categorization of countries according to welfare regime types is not sufficient for tackling the health inequality issue (Beckfield and Krieger, 2009; Brennenstuhl et al., 2012). The authors argue that it is important for comparative investigations on health inequality to implement healthcare system typologies that identify how healthcare systems affect not only average population health, but also the steepness of the health gradient (e.g., Brennenstuhl et al., 2012). Thus, classifying countries according to healthcare system type, as opposed to welfare regime, provides a better framework for understanding the effect of the healthcare context on the association between economic well-being and health (e.g., Wendt et al., 2009). To date, only few studies focused on the effect of the health system itself on health inequities, and those conducted revealed contradictory results (Beckfield and Krieger, 2009). Some studies provided evidence that enhanced welfare-state provisions reduce relative health disparities. For instance, the establishment of Canada's national health insurance plan led to a decline in incomebased inequality in mortality due to conditions amenable to medical treatment (e.g., Kunitz and Pesis-Katz, 2005). Other studies reported that the development of welfare-state health systems provisions did not translate into reduced health inequities (e.g., Arntzen et al., 1996). For example, the establishment of Australia's national health care system was associated with increased socioeconomic inequalities in avoidable mortality (Korda et al., 2007). It should be noted that most studies examine aggregated health indicators (e.g., infant mortality, cause-specific mortality etc.) without the use of individual-level health measures (Brennenstuhl et al., 2012). Studies that utilized individual-level indicators relied on self-assessed general health or limiting illness (see for example the studies by Bambra et al. (2009) across 13 European welfare states and by Burstrom et al. (2010) in Italy, Sweden and Britain). Furthermore, past research has not systematically examined the aforesaid issue in older populations, which are particularly vulnerable to health problems. Whereas several studies show that socioeconomic disparities in health increase throughout the life span, as individuals endure the cumulative effects on their health of earlier-life behavioral, environmental and psychosocial risk factors (e.g., Berkman and Gurland, 1998), other studies find that health differences by socioeconomic status are likely to diminish at older age (e.g., House et al., 1994). With regard to inconsistencies in these findings, more and more researchers acknowledge that studies should include socioeconomic measures such as wealth or home ownership that reflect the cumulative and dynamic nature of economic well-being as well as potential consumption which is especially relevant for the older population (e.g., Semyonov et al., 2013). Despite the evidence that wealth-based indicators are preferable to other commonly used measures, the number of studies that focus on the relationship between wealth and health is very limited, mostly due to lack of high quality data on wealth

20

D. Maskileyson / Social Science & Medicine 119 (2014) 18e26

inequality. A recent study by Semyonov et al. (2013) examined whether the wealthehealth gradient is likely to differ across sixteen countries. The findings revealed that the country's economic resources increase average population health but do not weaken the wealthehealth gradient. Additionally, greater egalitarianism does not raise the population's health levels but does weaken the wealthehealth gradient (Semyonov et al., 2013). Another study by Avendano et al. (2009) shows that health tends to increase with wealth among populations over the age 50 in the United States and Europe. Although their analysis focuses on the wealthehealth gradient, it does not classify countries by healthcare system types, and treats 10 European countries uniformly, disregarding possible country-specific effects. All findings reported above underscore the need for further investigation of wealthehealth gradient across different healthcare systems. To the best of my knowledge, no one has yet studied the extent to which the association between wealth and health varies in a comparative context and the extent to which different healthcare systems affect the relationship between wealth and health. The current study, therefore, aims to extend the results of previous studies by examining whether the wealthehealth gradient is influenced by healthcare system type across six countries when controlling for individual- and country-level variations. This research, not only advances theoretical and empirical knowledge in the fields of public health by providing a better understanding of the link between healthcare system type and the population's health, but also provides insights into the potential effect of healthcare system type on wealthehealth disparities among older populations. The contribution of the current study is in establishing which healthcare system types effectively reduce wealthehealth inequities and which tend to expand such inequities among older populations, and it might also aid in formulating new healthcare policies for the nation's population in general and for its older population in particular. 2.3. Classification of healthcare systems Because healthcare is often a major component of welfare provisions, the discussion about healthcare system typologies is inseparable from the wider welfare state debate. Accordingly, one of the most influential concepts in the development of healthcare system typologies is rooted in the concept of welfare state regimes proposed by Esping-Andersen (1990), in which he classified Western states on account of their social policies. Although this classification system was instrumental in developing healthcare system typologies, many health scholars criticized its implementation in the study of healthcare systems (e.g., Reibling, 2010). In response to these critiques, Bambra (2005) developed a typology that extends Esping-Andersen's concept to include provision of healthcare services. Her healthcare typology was based on the measures of the extent of private financing, the extent of private provision, and the general access provided by the public healthcare system (Bambra, 2005). However, the empirical indicators and range of cases selected by Bambra are, according to several researchers, “not sufficient (and not intended) to establish a robust typology of healthcare systems” (Wendt, 2009:73). Moran's (2000) study of healthcare states is one of the few that systematically includes the three dimensions of healthcare consumption, provision and technology (Wendt et al., 2009). Moran classifies healthcare states according to the state regulatory institutions that dominate in each of the three dimensions. Compared to all the above-mentioned categorizations of healthcare systems, the conceptual model proposed by Wendt et al. (2009) offers the highest level of abstraction and an essential tool for differentiating between the key features of healthcare systems.

Like Moran's, the typology proposed by Wendt et al. is also based on three crucial dimensionsdfinancing, service provision and regulation. However, Wendt et al. (2009) were the first to evaluate healthcare systems according to the presence or absence of state, private or societal actors in each dimension. Furthermore, Wendt's model takes changes in a country's healthcare system over time into account, which allows for cross-national as well as over-time comparisons. Following the Weberian tradition of ideal types, Wendt et al. (2009) developed a taxonomy of 27 distinct healthcare system types, of which three are “ideal types.” This taxonomy is based on the potential range of variation that emerges from categorizing systems according to the dimensions of financing, service provision and regulation, where each dimension is further distinguished by the level of state, societal and market involvement (Wendt et al., 2009). This classificatory framework is considered the most sufficient tool for the empirical purposes of the current study, because it provides a more detailed description of healthcare system types than earlier typologies. 2.4. Healthcare systems in six countries In line with Wendt et al.'s (2009) analytical approach, the current study classifies healthcare systems according to the roles of state, societal and private actors in the three key healthcare dimensions of financing, service provision and regulation. Systematic quantitative and qualitative indicators for each of these three dimensions were used to determine whether state, societal or private actors are dominant. The first dimension, healthcare financing, is defined as a country's health expenditure as a percentage of all its medical-care expenditure. The main question with regard to this dimension is: who is financing the healthcaredstate, societal or private agents? In this context, state healthcare expenditure refers to health financing by central, state, regional and local government authorities. Societal expenditure refers to health financing by social security funds (Wendt et al., 2009). Private expenditure refers to the healthcare spending of private insurance programs, corporations and non-profit institutions serving households as well as out-ofpocket payments. The second dimension, healthcare service provision, is defined as the provision of outpatient and inpatient healthcare (for details, see OECD, 2001). The question here is: who is providing the healthcare e state, societal or private actors? The third dimension, regulation, raises the question: who is regulating the healthcare system e state, societal or private actors? Sweden, Germany, Israel, the Czech Republic, the United Kingdom (U.K.) and the United States (U.S.) were selected as examples of the distinct healthcare system types. According to Wendt et al.'s typology, the healthcare system of Sweden is a close approximation of the ideal-type decentralized state-based system; the system in the U.K. is close to the ideal-type centralized statebased system. The healthcare systems of both the U.K. and Sweden are grouped here under the same ideal-type state-based constellation, which is consistent with Wendt et al. (2009). The main differences, between the British and Swedish healthcare systems lie in their degrees of centralization and scope of healthcare expenditure. The Israeli system is best classified as a statebased mixed-type; those in Germany and the Czech Republic represent societal-based mixed types; finally the U.S. is characterized as having a private-based mixed-type healthcare system (see Appendix 1). 3. Research questions and hypotheses The major purpose of this study is to answer the following questions: First, is household wealth positively associated with the

D. Maskileyson / Social Science & Medicine 119 (2014) 18e26

personal health of older individuals net of their education and socio-demographic attributes? Second, does household wealth contribute to health net of a household income level, education and socio-demographic characteristics? Third, is a country's healthcare system type likely to affect the health of its older population? Finally, is a country's healthcare system type likely to affect the association between household wealth and personal health? In all countries household wealth is expected to be positively associated with health, net of household income, education and socio-demographic characteristics, while the association between wealth and health is expected to be strongest in countries with private-based healthcare systems (e.g., the U.S.), and weakest in countries with state-based systems (e.g., Sweden and the U.K.).

21

95

90

90.29

90.32

Germany

Sweden

91.04

88.64 87.25

85 80.42

80

75 United States

Israel

Czech Republic

United Kingdom

Fig. 1. Average severity-weighted index values, by country.

4. Data and methods 4.1. Data 4.1.1. Healthcare dimension indicators Data for the dimension indicators were obtained from the health system reviews of the European Observatory on Healthcare Systems and the Organization for Economic Co-operation and Development (OECD) (see Appendix 1). 4.1.2. Individual-level data The data were obtained from three projects: The Survey on € rsch-Supan Health, Ageing and Retirement in Europe (SHARE; Bo and Jürges, 2005); the U.S. Health and Retirement Study (HRS; Juster and Suzman, 1995; Health and Retirement Study, 2004); and the English Longitudinal Study of Ageing (ELSA; Marmot et al., 2002). Details on each survey have been reported elsewhere (Heeringa and Connor, 1995; Juster and Suzman, 1995; Marmot €rsch-Supan and Jürges, 2005). The U.S. Health and et al., 2002; Bo Retirement Survey was approved by the institutional review board from the University of Michigan Health Services. SHARE in Europe was approved by the institutional review board at University of Mannheim, Germany. Ethical approval for ELSA was granted by the National Research and Ethics Committee in England. These projects collect information about various topics including income, assets, and health. Through in-depth interviews with nationally representative samples of adults 50 years of age or older, the projects aim to characterize the population aging of modern societies. The focus on older populations is an advantage for this study because people in advanced stages of their life cycle have had the opportunity to accumulate wealth, and their well-being and health are more dependent on economic resources. The samples for SHARE, HRS, and ELSA were drawn from 2004/2007, 2004, 2002/2003, respectively (for details see Appendix 2). The analysis was restricted to individuals who provided complete information for all relevant variables. 4.2. Variables 4.2.1. Health measure A physical health measure e severity-weighted index e was used as a dependent variable. Due to the absence of clinical records and direct biological measures in the surveys, the list of selfreported chronic diseases, symptoms, mobility limitations, arm function, fine motor function, and limitations associated with daily living activities was used as a substitute for an objective physical health measure. It should be noted that self-reported illness and physical limitations have been shown to be a useful predictor of physical health trajectories and mortality in older adults (e.g., Huisman et al., 2003). The severity-weighted index reflects the sum of 41 health problems selected by each respondent, divided by the

total number of non-missing items, multiplied by 100. Each of the health problems was weighted by the level of severity and impact on overall health (rated by the practicing physicians). Thus, the index represents a percentage of the total possible score, and ranges from 0 to 100; the better the respondent's health the higher is the value of the index. 4.2.2. Socioeconomic characteristics: wealth, income and education Net worth of a respondent's household as an important indicator of economic resources (e.g., Semyonov and Lewin-Epstein, 2011) (hereinafter referred to as wealth) is the main independent variable. Wealth is defined as the sum of the net real and net financial assets, minus debt. Financial assets reflect the sum value of accounts, bonds, stocks, mutual funds and savings. Real assets pertain to the value of the primary residence (net of mortgage), other real estate, owned businesses and owned cars. To capture the contribution of socioeconomic status to the wealth accumulation of the older population, total household income was included in the analysis. Total household income comprised all non-asset income (salary, pension, welfare, etc.) by all household members in the previous year. Total household wealth and income (PPP-adjusted, in Euro) were standardized to a percentile ranking scale, on which individuals are ranked in each country according to their relative wealth (or income) on a percentile ladder. This procedure eliminates cross-national differences in the length of the wealth (or income) ladder and allows a cross-national comparison (Mandel and Semyonov, 2005). Since existing literature has documented a large and persistent association between education and health (Ross and Wu, 1995), education level was included in the analysis (not completed secondary and lower education ¼ 1; intermediate education ¼ 0; academic education ¼ 1). 4.2.3. Socio-demographic characteristics Following previous studies (e.g., Deaton, 2008) a series of sociodemographic variables were used for control purposes, including respondent's age (in years), gender (female ¼ 0; male ¼ 1), immigrant status (not immigrant ¼ 0; immigrant ¼ 1), and whether respondent lives with partner (not living with partner ¼ 0; living with partner ¼ 1) (see Appendix 3). 4.3. Methods In addition to the descriptive statistical analysis, multiple regression equations were employed to predict health among the older population as a function of household wealth, while controlling for individual's socio-demographic and socioeconomic attributes. The regression equations were estimated, once for each country separately, and once in a pooled model with country added

22

D. Maskileyson / Social Science & Medicine 119 (2014) 18e26

as a dummy variable (Sweden serves as comparison). Sweden was selected as the omitted category, based on the assumption that state-based healthcare systems are likely to produce more egalitarian health outcomes. The Swedish state-based healthcare system is expected to have the lowest effect on the wealthehealth gradient compared to other countries. 5. Results 5.1. Cross-country mean differences in health

Table 1 OLS regression equations coefficients (s.e.) predicting the severity-weighted index in each country.

Total household wealth (%) Total household income (%) Age (centered) Male

Fig. 1 demonstrates the average severity-weighted index values by countries. The results presented in Fig. 1 reveal significant variation across countries in the physical health level. On average, the older population in the U.S. has the poorest level of physical health (80.42), about 11 points lower than in the U.K., whose population reports the highest level of physical health (91.04). The estimated average physical health in Germany (90.29) and Sweden (90.32) is quite similar. In the Czech Republic, the estimated average physical health is 88.64. In Israel, the average score on the severity-weighted index is 87.25, higher than the U.S. score but lower than that of other countries. 5.2. The association between wealth and health in six countries Next, the Pearson correlation coefficients between wealth and the severity-weighted index were calculated separately for each country. The data presented in Fig. 2 show that in all countries, without exception, wealth and health are positively associated and statistically significant, i.e., within countries, wealthier individuals generally have better health than poorer ones. The association between wealth and health is highest in the U.S. (R ¼ 0.279) and lowest in Israel (R ¼ 0.194). Such differences in the wealthehealth association might be related not only to the economic disparities within and among countries and individuals, but also to the different characteristics of healthcare systems across countries.

Low education High education Immigrant Living with partner Constant Adjusted R2 Number of observations

Germany Sweden Israel

Czech Republic

U.K.

U.S.

0.052** (0.006) 0.010 (0.007) 0.376** (0.019) 1.189** (0.351) 1.923** (0.478) 0.743 (0.407) 1.211** (0.433) 0.377 (0.420) 87.712** (0.501) 0.197 2929

0.035** (0.007) 0.025** (0.008) 0.408** (0.021) 2.148** (0.395) 1.721** (0.418) 0.797 (0.710) 0.898 (0.906) 0.005 (0.462) 85.599** (0.588) 0.208 2742

0.073** (0.003) 0.005 (0.004) 0.200** (0.009) 0.657** (0.172) 0.771** (0.210) 0.106 (0.244) 0.340 (0.308) 0.155 (0.202) 87.251** (0.278) 0.133 11,190

0.086** (0.004) 0.057** (0.004) 0.343** (0.009) 2.486** (0.184) 2.300** (0.221) 1.096** (0.242) 3.441** (0.294) 0.350 (0.220) 72.449** (0.245) 0.217 19,234

0.054** (0.006) 0.026** (0.008) 0.356** (0.018) 2.296** (0.331) 0.480 (0.407) 0.744 (0.484) 1.681** (0.605) 0.308 (0.473) 85.229** (0.535) 0.236 2967

0.044** (0.010) 0.021* (0.010) 0.512** (0.027) 1.671** (0.482) 4.786** (0.549) 0.861 (0.613) 2.568** (0.501) 2.722** (0.612) 84.107** (0.811) 0.277 2460

Note: **p < 0.01, *p < 0.05; Omitted categories: female ¼ 0; intermediate education ¼ 0; not immigrant ¼ 0; is not living with partner ¼ 0.

between wealth and health is positive and statistically significant in all countries. In addition, the analysis reveals that the severity-weighted index is positively associated with household income, and negatively associated with age. Women and single individuals tend to suffer from poorer health compared to men and those living with a partner, respectively. Low education is associated with poorer health, and high education adds significantly to health, compared to intermediate education. Immigrant status has a statistically significant positive effect on the severity-weighted index (except in the Czech Republic and the U.K.).

5.3. Explaining health disparities across countries 5.4. Explaining the wealthehealth gradient across six countries To examine whether and to what extent the wealthehealth gradient might be affected by the socioeconomic and sociodemographic composition of countries, multivariate regression models were employed for each country separately (while controlling for the socio-demographic and socioeconomic differences among individuals). The results presented in Table 1 reveal that when individual variables affect is neutralized, the association

0.30

0.25 0.218**

0.20

0.230**

0.236**

Czech Republic

Sweden

0.276**

0.279**

United Kingdom

United States

0.194**

0.15

0.10

0.05

0.00 Israel

Germany

Fig. 2. Pearson correlation estimates between severity-weighted index and wealth, by country.

To investigate whether the differences across countries are significant, the regression equation was calculated in a pooled model with countries added as a set of dummy variables. This pooled model was estimated first without the country effect, second as an additive model including country dummy variables, and, finally, with interaction terms between country and wealth (see Table 2). Model 1 in Table 2 repeats the analysis provided in the previous section for the pooled data set. The results consistently reveal that better health levels are associated with higher wealth, higher income, younger age, male, high education and living with partner. The second question addressed in this section is whether the health disparities in the six countries are statistically significant, i.e. whether healthcare system type is likely to affect the health of its older population. The additive model 2 evaluates the health gaps between each country and Sweden. Model 2 indicates considerable health disparities among countries. Interestingly, the U.K. is the only country that has better average health than Sweden. The older populations in other countries tend to have poorer health, on average, compared to Sweden. The U.S. coefficient is largest (b ¼ 10.185), meaning that the U.S. older population has about 10 points poorer health than Sweden's. The third question in this section, and the principal question of this research, concerns the disparities in the wealthehealth

D. Maskileyson / Social Science & Medicine 119 (2014) 18e26

Total household wealth (%) Total household income (%) Age (centered) Male Low education High education Immigrant Living with partner Germany Israel Czech Republic U.K. U.S. Germany*Wealth Israel*Wealth Czech Republic*Wealth U.K.*Wealth U.S.*Wealth Constant Adjusted R2 Number of observations

(1)

(2)

(3)

0.078** (0.002) 0.029** (0.003) 0.331** (0.006) 1.981** (0.121) 1.163** (0.137) 2.058** (0.163) 0.471** (0.178) 0.718** (0.145) e e e e e e e e e e e e e e e e e e e e 77.934** (0.172) 0.160 41,522

0.073** (0.002) 0.028** (0.002) 0.321** (0.006) 1.755** (0.110) 1.498** (0.130) 0.837** (0.149) 0.545** (0.173) 0.219 (0.133) 0.645* (0.287) 3.591** (0.308) 1.417** (0.288) 0.720** (0.224) 10.185** (0.217) e e e e e e e e e e 84.792** (0.258) 0.302 41,522

0.055** (0.007) 0.025** (0.002) 0.330** (0.006) 1.762** (0.110) 1.470** (0.130) 0.794** (0.149) 0.592** (0.173) 0.176 (0.133) 0.096 (0.571) 4.024** (0.603) 0.835 (0.579) 1.105* (0.451) 12.680** (0.432) 0.011 (0.010) 0.008 (0.010) 0.012 (0.010) 0.008 (0.008) 0.049** (0.007) 85.874** (0.423) 0.306 41,522

Note: **p < 0.01, *p < 0.05; Omitted categories: female ¼ 0; intermediate education ¼ 0; not immigrant ¼ 0; is not living with partner ¼ 0; Sweden ¼ 0.

association across the six countries. This question aims to determine whether the association between wealth and health is likely to be weaker in countries with state-based healthcare systems (i.e., Sweden), and stronger in countries with private-based healthcare systems (i.e., U.S.). To examine this hypothesis, a pooled model with interaction terms between country and wealth is estimated, with Sweden serving as the omitted category. Model 3 (Table 2) shows that wealth has a positive and statistically significant effect on the severity-weighted index. The coefficient estimate (b ¼ 0.055) implies that, in Sweden, every additional percent in household wealth is associated with a 0.055 point rise in health status, on average. The results also reveal that when controlling for other variables and interaction terms, the health disparities between Germany, the Czech Republic and Sweden are not statistically significant. The U.K. average physical health level is higher than Sweden's (b ¼ 1.105), while the average Israeli physical health level is lower than Sweden's (b ¼ 4.024). The older population in the U.S. suffers the poorest health (b ¼ 12.680) compared to Sweden. Model 3 shows that the associations between wealth and health in Germany, the Czech Republic, the U.K. and Israel are not different from that in Sweden. However, as expected, a considerable gap in the wealthehealth gradient was found between the U.S. and Sweden. In the U.S., every

additional percent in household wealth adds 0.104 points to physical health, approximately twice as much as in Sweden. The regression lines based on the model with interaction terms are presented in Fig. 3. Regression lines were estimated for nativeborn, non-single men, at average age, with high education levels. Fig. 3 shows that the mean physical health of the older population in the U.K. is higher than in Sweden, while physical health in Israel is, on average, lower than in Sweden. In the U.S., average physical health is lowest compared to other countries. The U.S. regression line slope is a notch steeper than the Swedish one (0.104 versus 0.055), while the Czech Republic line is less steep (0.043), which is indicated by the negative sign of the estimated interaction term coefficient. However, only the slope of the U.S. regression line is significantly different from the Swedish slope (p < 0.01). Importantly, even among the highest wealth percentile, the health gap between the U.S. older population and that of the other five countries does not converge. In other words, a man with average socio-demographic and socioeconomic characteristics can “buy” much better health in Sweden than in the U.S., at every level of the household wealth. This finding supports the assumption that the American healthcare system is an outlier among the six countries examined here. Even the rich lose in its private-based healthcare system. It should be noted that a similar picture was drawn by fitting regression lines based on the analysis presented in Table 1 separately for each country (this figure is not presented for the sake of parsimony).

6. Discussion and conclusions First, in line with previous studies, the results indicate that economic well-being is likely to increase the health of older people, net of their socio-demographic attributes and education; that is, household wealth and income exert a positive and statistically significant effect on health. Notably, however, the association between wealth and health was found to be largely independent of income. Although the association between wealth and health outcomes was slightly attenuated after controlling for income, it remained substantial and significant across countries. This supports the argument that for older populations, household wealth, which reflects the economic resources accumulated over an individual's life, is a better indicator of socioeconomic well-being than income, which reflects socioeconomic standing at a given point in time.

100 Severity-Weighted Health Index

Table 2 Pooled OLS regression models coefficients (s.e.) predicting the severity-weighted index in 6 countries.

23

95

90

85

80

75 0

25

Sweden Czech Republic

50 Total Household Wealth, % Germany United Kingdom

75

100

Israel United States

Fig. 3. Regression lines predicting the severity-weighted index by wealth, obtained from the results presented in Table 2, model 3.

24

D. Maskileyson / Social Science & Medicine 119 (2014) 18e26

Second, the findings reveal substantial health disparities across countries with different healthcare system types. This finding suggests that different systems produce different health outcomes, which supports the hypothesis that healthcare system type is likely to affect the health of older populations. Specifically, the average physical health of older populations is best in countries with statebased healthcare systems: the British centralized state-based system produced the highest health outcomes of all the six systems studied here, while the Swedish decentralized state-based system was second from the top. These findings are in line with previous studies reporting that the U.K. is observed to have better general health in comparison to Southern and East-European welfare regimes (e.g., Eikemo et al., 2008a). The findings on Sweden also support previous evidence, namely, Scandinavian countries with their highly developed welfare state and social protections, are characterized by better overall health (e.g., Chung and Muntaner, 2007). Next in line are the Czech Republic and Germany, whose societal-based systems have health outcomes that surpass those of the Israeli state-based mixed system. This also confirms previous studies. As Eikemo et al. (2008a) report, the Scandinavian and Anglo-Saxon welfare regimes are observed to have better selfperceived general health than Bismarckian, Southern and EastEuropean welfare regimes. Most notably, the U.S. private-based system produces the worst health outcomes: older Americans were likely to report worse health than their counterparts in Europe and Israel. This conclusion is supported in previous studies (e.g., Avendano et al., 2009) and by OECD data (OECD, 2009), which indicate that healthcare in the U.S., despite having the highest healthcare expenditure in the world, has comparatively poor outcomes (e.g., infant and adult mortality rates, life expectancy etc.), costs more per person, is less accessible to a large portion of its citizens, and provides healthcare services of lower quality. It should be noted that differences in health outcomes may also be attributed to other factors, including obesity, smoking, alcohol consumption, and ecological conditions. The tendency of such risk factors to affect health, however, is also dependent on the efficiency of a healthcare system. In addition to other functions, healthcare systems are supposed to promote a healthy lifestyle and discourage risky behavior. Finally, the most important findings ofdand indeed the motivation behind this studydreveal that a country's healthcare system type affects not only the health outcomes of older individuals, but also may affect the association between household wealth and health. However, contrary to the expectation that all healthcare system types affect differently the association between wealth and health, only the American private-based healthcare system was found to affect it in a significantly different, and stronger, way compared to the five other diverse systems studied here. This is contrary to previous studies that found significant variations of health inequalities by the welfare state regime among a wide range of countries (see, for example, Eikemo et al., 2008b or Mackenbach et al., 2008). That is, when focusing on wealth instead of other socio-demographic indicators, there is a considerable gap between the wealthehealth gradient in the U.S. and those in the other five countries, whose wealthehealth gradients were remarkably similar to one another. The positive association between household wealth and health is about two times stronger in the U.S. than in Sweden, Germany, the Czech Republic, the U.K. and Israel. Thus, in the U.S. every additional percent in household wealth increases physical health by about twice as much as it does in the other countries, indicating that among the six countries, household wealth is most important in predicting the health outcomes of older Americans. In the U.S., because of the large wealthehealth gradient, an individual must have a high socioeconomic standing in order to

purchase the healthcare services associated with good health, thereby compounding the health disadvantages of the poor. In comparison, the healthcare systems of Sweden, the U.K., Germany, the Czech Republic and Israel, with their less steep gradients, produce better opportunities for both their rich and poor to achieve good health. Furthermore, while it is true that in all countries the rich are, on average, healthier than the rest, due to the poor health outcomes of the U.S. system, even the wealthiest older Americans have poorer health, on average, than the poorest older individuals in the U.K., Sweden, the Czech Republic and Germany, and have comparable health to the poorest Israelis. It should be noted, that in order to determine that the findings are not exclusive to the selected severity-weighted index, a similar analysis was conducted for the unweighted index, the rarity index, and self-perceived general health. The results demonstrate a very similar pattern for all these health measures (the detailed results of this analysis are available from the author upon request). Such consistent results strengthen the conclusions that were put forth above. In addition, these findings are consistent with that of the research on middleaged adults conducted by Avendano et al. (2009). Their study indicates that U.S. adults reported worse health than English or other Europeans at every wealth level. For example, “the risk of reporting heart disease for a European with V3000 was 11%, which was equivalent to the risk for an American with V 300,000” (Avendano et al., 2009:542). The results of this study should be interpreted within the context of its limitations. First, this research focuses on relatively high-income economies. Inclusions of countries from a variety of geographic regions and levels of economic development would add to the depth of the analysis and generality of the findings. Second, the view that healthcare system influences the wealthehealth gradient should be interpreted with caution. Causality cannot be inferred from the present analysis because the data are crosssectional. Third, this research relies on self-reported measures of health. Indeed, the perception of health may differ cross-nationally, and the use of self-reported health measures may reflect cultural differences between or within countries (Corin, 1995). In the future, it would be beneficial to employ direct clinical measures and measures of amenable health conditions that might be conceptually more closely related to the actions of the health system, but were unavailable in the data used in this study. In addition, future investigations of the relationship between economic resources and health in different healthcare systems would benefit from the use of multilevel regression analysis using country-level institutional indicators, which was not possible here due to the insufficient number of countries available for second-level analysis. Direct measurement of healthcare systems features may be more promising for public health research than the use of typologies alone. Such an analysis would intimate which specific healthcare system characteristics (e.g., political-regulatory factors) may have a positive or negative effect on the overall population health and the wealthehealth gradient, and through which mechanisms wealth may affect health. Despite the limitations outlined above, the study represents an initial attempt to examine whether a country's healthcare system type produces variations not only in health outcomes, but also may affect the wealthehealth gradient. It stresses the importance of the wealth measure as a critical factor in predicting the health outcomes among older populations. It demonstrates that wealth has a more powerful effect on health in the U.S. than in the other five countries examined. The characteristics of the U.S. systems may contribute to the greater wealthehealth inequality than in other countries. The state-based and societal-based healthcare systems with their more accessible medical care and with a stronger primary care orientation (Avendano et al., 2009) than in the U.S. seem

D. Maskileyson / Social Science & Medicine 119 (2014) 18e26

to mitigate the effect of wealth on health, by providing better and more equal access to healthcare services. The study highlights the importance of using the healthcare system classification rather than welfare states regime perspective in health inequality examination. Acknowledgments I am grateful to Professor Moshe Semyonov, Professor Noah Lewin-Epstein, and the anonymous reviewers of Social Science & Medicine for their helpful comments on the earlier drafts of this paper. This work was supported by the German-Israeli Foundation for Scientific Research and Development (G.I.F.; Grant #1021-305.4/), the Israel National Institute for Health Policy and Health Services Research, Myers-JDC-Brookdale Institute of Gerontology and Human Development, and Eshel e the Association for the Planning and Development of Services for the Aged in Israel. The HRS (Health and Retirement Study) is sponsored by the National Institute on Aging (grant number NIA U01AG009740) and performed at the Institute for Social Research, University of Michigan. The English Longitudinal Study on Ageing has been supported by the NIA (grants 2RO1AG7644-01A1 and 2RO1AG017644) and several British government departmentsdthe Department for Education and Skills; Department for Environment, Food, and Rural Affairs; Department of Health; Department of Trade and Industry; Department for Work and Pensions; Her Majesty's Treasury Inland Revenue; the Office of the Deputy Prime Minister; and the Office for National Statistics. The SHARE data collection has been primarily funded by the European Commission through the 5th framework program (project QLK6CT-2001-00360 in the thematic program Quality of Life). Additional funding came from the U.S. National Institute on Aging (grants U01 AG09740-13S2, P01 AG005842, P01 AG08291, P30 AG12815, Y1-AG-4553-01, and OGHA 04-064) and the German Ministry of Education and Research as well as from various national sources is gratefully acknowledged. The SHARE data collection in Israel was funded by the U.S. National Institute on Aging (R21 AG025169), by the German-Israeli Foundation for Scientific Research and Development (G.I.F.), and by the National Insurance Institute of Israel. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.socscimed.2014.08.013. References Arntzen, A., Moum, T., Magnus, P., Bakketeig, L.S., 1996. The association between maternal education and postneonatal mortality. Trends in Norway, 1968e1991. Int. J. Epidemiol. 25 (3), 578e584. Avendano, M., Glymour, M., Banks, J., Mackenbach, J.P., 2009. Health disadvantage in US adults aged 50 to 74 years: a comparison of the health of rich and poor Americans with that of Europeans. Am. J. Public Health 99 (3), 540e548. Bambra, C., 2005. Worlds of welfare and the healthcare discrepancy. Soc. Policy Soc. 4 (1), 31e42. Bambra, C., 2006. Health status and the worlds of welfare. Soc. Policy Soc. 5 (1), 53e62. Bambra, C., Pope, D., Swami, V., Stanistreet, D., Roskam, A., Kunst, A., ScottSamuel, A., 2009. Gender, health inequalities and welfare state regimes: a crossnational study of 13 European countries. J. Epidemiol. Commun. Health 63 (1), 38e44. Beckfield, J., Krieger, N., 2009. Epi þ demos þ cracy: linking political systems and priorities to the magnitude of health inequities e evidence, gaps, and a research agenda. Epidemiol. Rev. 31 (1), 152e177. Berkman, C.S., Gurland, B.J., 1998. The relationship among income, other socioeconomic indicators, and functional level in older persons. J. Aging Health 10 (1), 81e98.

25

€rsch-Supan, A., Jürges, H. (Eds.), 2005. The Survey of Health, Ageing and RetireBo ment in Europe: Methodology. Mannheim Research Institute for the Economics of Aging, Mannheim, Germany. e, A., McDonough, P., 2012. Welfare regimes, popBrennenstuhl, S., Quesnel-Valle ulation health and health inequalities: a research synthesis. J. Epidemiol. Commun. Health 66 (5), 397e409. Burstrom, B., Whitehead, M., Clayton, S., Fritzell, S., Vannoni, F., Costa, G., 2010. Health inequalities between lone and couple mothers and policy under different welfare regimes e the example of Italy, Sweden and Britain. Soc. Sci. Med. 70 (6), 912e920. Chung, H., Muntaner, C., 2006. Political and welfare state determinants of infant and child health indicators: an analysis of wealthy countries. Soc. Sci. Med. 63 (3), 829e842. Chung, H., Muntaner, C., 2007. Welfare state matters: a typological multilevel analysis of wealthy countries. Health Policy 80 (2), 328e339. Corin, E., 1995. The cultural frame: context and meaning in the construction of health. In: Amick III, B.C., Levine, S., Tarlov, A.R., Walsh, D.C. (Eds.), Society and Health. Oxford University Press, Oxford, pp. 272e304. Deaton, A., 2008. Income, health, and well-being around the world: evidence from the Gallup world poll. J. Econ. Perspect. 22 (2), 53e72. Eikemo, T.A., Bambra, C., Judge, K., Ringdal, K., 2008a. Welfare state regimes and differences in self-perceived health in Europe: a multi-level analysis. Soc. Sci. Med. 66 (11), 2281e2295. Eikemo, T.A., Bambra, C., Joyce, K., Dahl, E., 2008b. Welfare state regimes and income-related health inequalities: a comparison of 23 European countries. Eur. J. Public Health 18 (6), 593e599. Eikemo, T.A., Huisman, M., Bambra, C., Kunst, A., 2008c. Health inequalities according to educational level under different welfare regimes: a comparison of 23 European countries. Sociol. Health Illn. 30 (4), 565e582. Esping-Andersen, G., 1990. The Three Worlds of Welfare Capitalism. Princeton University Press, Princeton, New Jersey. Health and Retirement Study, 2004. (HRS Core (Final)) Public Use Dataset. Produced and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740). Ann Arbor, MI. Heeringa, S.G., Connor, J.H., 1995. Technical Description of the Health and Retirement Survey Sample Design. University of Michigan, Ann Arbor. House, J.S., Lepkowski, J.M., Kinney, A.M., Mero, R.P., Kessler, R.C., Herzog, A.R., 1994. The social stratification of aging and health. J. Health Soc. Behav. 35 (3), 213e234. Huisman, M., Kunst, A.E., Mackenbach, J.P., 2003. Socioeconomic inequalities in morbidity among the elderly; a European overview. Soc. Sci. Med. 57 (5), 861e873. Hurd, M., Kapteyn, A., 2003. Health, wealth, and the role of institutions. J. Hum. Resour. 38 (2), 386e415. Juster, F.T., Suzman, R., 1995. An overview of the Health and Retirement Study. J. Hum. Resour. 30 (Suppl. l), S7eS56. Kawachi, I., Kennedy, B.P., Lochner, K., Prothrow-Stith, D., 1997. Social capital, income inequality and mortality. Am. J. Public Health 87 (9), 1491e1498. Korda, R.J., Butler, J.R., Clements, M.S., Kunitz, S.J., 2007. Differential impacts of health care in Australia: trend analysis of socioeconomic inequalities in avoidable mortality. Int. J. Epidemiol. 36 (1), 157e165. Kunitz, S.J., Pesis-Katz, I., 2005. Mortality of white Americans, African Americans, and Canadians: the causes and consequences for health of welfare state institutions and policies. Milbank Q. 83 (1), 5e39. Mackenbach, J.P., Stirbu, I., Roskam, A.J.R., Schaap, M.M., Menvielle, G., Leinsalu, M., Kunst, A.E., 2008. Socioeconomic inequalities in health in 22 European countries. N. Engl. J. Med. 358 (23), 2468e2481. Mandel, H., Semyonov, M., 2005. Family policies, wage structures, and gender gaps: sources of earnings inequality in 20 countries. Am. Sociol. Rev. 70 (6), 949e967. Marmot, M., Banks, J., Blundell, R., Lessof, C., Nazroo, J., 2002. Health, Wealth and Lifestyles of the Older Population in England: the 2002 English Longitudinal Study of Ageing. University College London, London, England. Moran, M., 2000. Understanding the welfare state: the case of healthcare. Br. J. Polit. Int. Relat. 2 (2), 135e160. Organization for Economic Co-operation and Development (OECD), 2001. OECD Health Data 2001: A Comparative Analysis of 30 Countries. OECD, Paris. Organization for Economic Co-operation and Development (OECD), 2009. Statistics Database: Health e Health Statistics e OECD Health Data 2009-Health Expenditure (Retrieved May 3, 2012). http://stats.oecd.org/index.aspx? DataSetCode¼HEALTH_STAT. Pickett, K.E., Wilkinson, R.G., 2007. Child well-being and income inequality in rich societies: ecological cross sectional study. Br. Med. J. 335 (7629), 1080e1085. Reibling, N., 2010. Healthcare systems in Europe: towards an incorporation of patient access. J. Eur. Soc. Policy 20 (1), 5e18. Ross, C.E., Wu, C.L., 1995. The links between education and health. Am. Sociol. Rev. 60 (5), 719e745. Semyonov, M., Lewin-Epstein, N., 2011. Wealth inequality: ethnic disparities. Soc. Forces 38 (3), 935e959. Semyonov, M., Lewin-Epstein, N., Maskileyson, D., 2013. Where wealth matters more for health: the wealthehealth gradient in 16 countries. Soc. Sci. Med. 81 (1), 10e17. Smith, J.P., 2005. Consequences and predictors of new health events. In: Wise, D.A. (Ed.), Analyses in the Economics of Aging. University Chicago Press, Chicago, pp. 213e240.

26

D. Maskileyson / Social Science & Medicine 119 (2014) 18e26

Van Doorslaer, E., Masseria, C., Koolman, X., 2006. Inequalities in access to medical care by income in developed countries. Can. Med. Assoc. J. 174 (2), 177e183. Wendt, C., 2009. Mapping European healthcare systems: a comparative analysis of financing, service provision and access to healthcare. J. Eur. Soc. Policy 19 (5), 432e445.

Wendt, C., Frisina, L., Rothgang, H., 2009. Healthcare system types: a conceptual framework for comparison. Soc. Policy Adm. 43 (1), 70e90. Wilkinson, R.G., 2006. The impact of inequality. Soc. Res. 73 (2), 711e732. Wilkinson, R.G., Pickett, K.E., 2006. Income inequality and health: a review and explanation of the evidence. Soc. Sci. Med. 62 (7), 1768e1784.