1.3 Socioeconomic Status Effects on HALE of the Elderly

Previous epidemiological，demographic，and sociological researches provide persistent and almost universal findings on social differences in health and mortality. Generally，lower SES has been shown to be related to increased risk of mortality and morbidity. From a lifespan developmental perspective，it is important to explore whether the strength of this relationship varies with age. Studies examining social inequalities in health rarely considered the elderly，but the evidence is currently mounting that SES plays an important role in health in later life[77]. However，the diversity of pathways，settings，and mechanisms from SES to HALE remains overwhelming.

1.3.1 Measures of Socioeconomic Status

SES can be broadly conceptualized as one’s social capital and position in a society characterized by social inequality. Sociologists emphasize a Weberian approach that encompasses the notion of class，status，and power. However，the existing research results suggest that SES is a multifaceted construct，including at least income，education，occupation，and wealth，which is related to health in diverse ways. Each life stage seems to require different indicators to measure the SES effect on health. As people age，their life has changed，including their status in society and their own social relationships. Accordingly，the indicators that can reflect people’s SES also change，particularly for older people. In general，education，income，and occupation，or a composite of these three dimensions，were adequately examined in previous studies，as described below.

Education

Kitagawa and Hauser[78] used educational attainment as their primary indicator of SES，and since then，education has played a central role in analyses of the SES effect on health，because it is easily recorded and remains stable over an individual’s lifetime[79]. A person’s relative status within society，which can be considered as social class according to Weber’s notion，may be related to the high educational attainment of high school or college degrees. Higher education implies more knowledge about health and health behavior，and shapes the ability to inform certain lifestyle choices to promote health in later life.

Income

If education represents human and social capital at the beginning of adulthood，annual income represents only recent accumulation of material resources or the financial situation，which is thought to be closely associated with the class or economic component of Weber’s notion of social class. Relative social class is closely related to potential access to different lifestyles，a sense of security，and the opportunity to fulfill material desires. In addition to providing means for purchasing health care，higher incomes can provide better nutrition，housing，schooling，and recreation[80].

Occupation

Compared with income and educational level，occupational status is a more complex variable，and its measurement varies depending on one’s theoretical perspective about the significance of various aspects of work life. Occupation can be seen as a proxy for representing Weber’s notion of socioeconomic position，as a reflection of a person’s place in society relative to their social standing，income，and intellect[81]. Considering that most elderly individuals are not or will not for much longer be part of the labor force，occupation-based measures can only have an aftereffect，and it is unclear how strong these effects are relative to new and current living conditions.

1.3.2 Socioeconomic Status of the Old Population

In the World

Studies have established that lower SES is associated with increased levels of illness and mortality. For elderly individuals，advanced age is characterized by a withdrawal from permanent employment and a reduction in working hours. The outcome is typically a huge drop in income. Globally，older people today are significantly less likely to participate in the labor force than they were in the past. As the data from the United Nations[82] have shown，labor force participation of people age≥65 years declined by more than 40% at the global level from 1950-2000. Among men，labor force participation decreased from 55% in 1950 to 30% in 2000，while in women，the reduction was from 14% in 1950 to 10% in 2000. In addition，the female share of the older workforce is increasing，particularly in more developed countries. In general，education levels have improved in each generation over the last century，and the widespread attainment of at least primary education has been established in more developed regions. As a result，literacy among the older population is nearly universal in more developed regions；however，illiteracy remains high among older people in less developed regions，particularly among women.

The Gini coefficient is commonly used as an international measure of inequality of income distribution，in which 0 corresponds to perfect equality and 1 to perfect inequality. The Gini coefficient has risen significantly since the mid-1980s in OECD countries. In 2008，the average Gini coefficient was 0.309 for people age 18-65 years and 0.288 for people≥65 years（Figure 1.7）. The degree of income inequality for the 18- to 65-year-old group was greater than that for the≥65-year-old group at the average level，indicating that income inequality decreases in old age as a result of retirement and associated pension and other income security benefits. However，some countries showed different characteristics，such as the United States，Japan，and France. Figure 1.8 shows another international inequality indicator，the relative poverty rate（%），which is the percentage of people with an income ≤ 50% of the median income in OECD countries. Japan has one of the highest Gini coefficients among OECD countries，with all poverty rates being much higher than the average level of OECD in all age groups.

Figure 1.7 Gini coefficient of OECD countries

Source：OECD，2008[83].

Figure 1.8 Relative poverty rates（%）of OECD countries

Source：OECD Income Distribution and Poverty Database，2008[84].

In Japan

In Japan，accelerated economic growth and technological advancement have enhanced health and life expectancy，particularly after the Second World War. It seems that Japan’s health expenditure and health and welfare system organization，along with specific social and cultural particularities，translate into reduced socioeconomic differences in health outcomes. However，increases in social and health inequalities have been recently reported in Japan，which may be partially due to population aging[85].

The average gross household income and the average gross income per household member by the age bracket of the household head are shown in Figure 1.9. In 2006，the total average income per household and average income per household member were 5.67 million yen and 2.07 million yen，respectively. The average household income of all other households is higher than that of households headed by someone≥65 years old（4.32 million yen），except those headed by individuals ≤ 29 years old.

The income distribution of people≥65 years old is different from those in other age groups. Figure 1.10 shows how income is distributed for all households and aged households. Overall，78.7% of households headed by individuals age≥65 years earned an annual income of＜4 million yen in 2006. The average income of all households is evenly distributed over a range of 1 million to 8 million yen；however，the distribution of aged households becomes increasingly skewed to the lower income brackets. In 2010，the average annual income of households headed by elderly people was 4.29 million yen，which is much lower than the average 5.24 million yen income of all households. Reflecting this reality，the Gini coefficient for households headed by people≥65 years old is approximately 0.34，which is higher than the Gini coefficient of 0.31 for households headed by individuals age 18-65 years（Figure 1.7）.

Figure 1.9 Average income per household and per household member by age group of household head，2006

Source：Ministry of Health，Labor and Welfare，2006[86].

In Japan，the average amount of schooling completed is 12.3 years per person，more than 90% of the population attends high school，and approximately 40% of all upper-secondary school graduates advance to tertiary education. The education level in Japan has tended to increase over time：for example，among OECD countries，Japan is ranked in the tenth position among 55- to 64-year-olds（those who completed their education some 40 years ago）and in the third position among 25- to 34-year-olds（those who completed their education a decade ago）[87]. Due to the extremely high popularity of higher education，particularly in the last few decades，the range of levels of education attained has become larger，which has increased social inequalities in Japan. However，many aging Japanese enroll in various educational facilities，and this factor could also influence SES，although studies on this topic have not yet been performed[88].

Figure 1.10 Distribution of income by all households and aged households

Source：Ministry of Health，Labor and Welfare，2006[89].

1.3.3 Theoretical Hypothesis about SES-HALE Associations at an Older Age

The multifaceted nature of SES and HALE needs to be considered in studies on social inequality in health. The sociological background for the analysis of SES differences in HALE in elderly individuals is the question of whether social inequality increases，decreases，or remains stable during old age（Figure 1.11）.

Convergence Hypothesis-“Status Leveling Hypothesis”

Representatives of the age-as-leveler hypothesis suggest that health differentials by SES are largest in prime age and then converge at older ages due to a variety of factors[90][91]. Biological deterioration in later life could accelerate health decline of high SES，leading to an override of the significance of social factors and thus to a convergence of the status groups[92]. Another possibility is that in some cases，selective survival might eliminate SES differences in health in later life. A different assumption within the status-leveling hypothesis is that the welfare state actually reduces socioeconomic differences in old age through benefits and social security[93]. This is also called the Redistribution Hypothesis，which stresses that in many industrial countries，inequality among the elderly is less pronounced than among younger groups[94].

Figure 1.11 Hypotheses about SES-Health associations at older age

Continuity Hypothesis-“Status Maintenance Hypothesis”

The influence of SES on health in later life has also been characterized by continuity，which is indicated by little age differences or stability in the SES-HALE relationship. Status maintenance is based primarily on the influence of the working age on the retirement age through external structures in which the individual has a persisting position. Secondly，status maintenance can be based on internal dispositions such as learning behavior，habits，and one’s own self-concept[95][96].

Divergence Hypothesis-“Cumulative Advantage Hypothesis”

Other researchers have provided empirical evidence for divergence in health disparities by SES in later life，which suggests that social inequalities in health may be due to the cumulative effects of disadvantage over the life course，resulting in enhanced，rather than diminished，SES in older ages compared to middle life[97][98]. Another aspect that contributes to an accumulation of inequality is the fact that certain inequalities only become visible and effective with poor health status，for example，the exposure to one environmental hazard is likely to be combined with exposure to other hazards，and these exposures are likely to accumulate over the course of a lifetime[99].

The three hypotheses about how SES inequality could change with increasing age（leveling，maintenance，and accumulation）are not mutually exclusive. Some pathways that lead to a leveling of social inequality with age may exist together with other processes that increase inequality. For example，most of the studies conducted in the United States provided strong evidence in favor of the leveling hypothesis，because the American health care system is characterized by socially-patterned health insurance coverage up to the age of 65 years，when virtually all citizens receive basic health insurance through Medicare[100]. Furthermore，due to the large social inequalities in health during midlife in the United States，the SES disparities on health might be attenuated in old age due to a strong influence of selected mortality in earlier stages of life[101].

In summary，given the ambiguity of the extant literature，we cannot predict with confidence whether SES differentials in health will diminish，continue，or grow at older ages. Because the use of SES and health indicators may differ in various social contexts，more empirical evidence on the age patterns of SES differentials in HALE can help us evaluate the relative merit and applicability of the convergence hypothesis versus the divergence hypothesis[102].

1.3.4 Literature Review on SES-HALE Mechanisms

This book systematically reviews the empirical literature on SES effects on HALE within and outside Japan. Most of the academic researchers have shown that the SES-HALE gradient remains in older people，and also appears to influence multiple dimensions of health，including physical functioning，comorbidity，and SRH. The explanations of good health status and long life expectancy among high SES individuals were classified into conceptually distinct mechanisms，in which the impact of SES was assumed to either directly influence HALE or indirectly impact HALE.

Direct SES-HALE Mechanism

Education most likely affects health directly，indicating that the lower a person’s level of education，the more likely he/she will have a higher risk of chronic diseases and disability. Further，it is also clear that people at a low-income level have multiple health problems，leading to poor SRH.

At the national level，a study in the UK revealed that social class，unemployment rate，and the percentage of the non-white population were the primary variables associated with differences in disability-adjusted life expectancy（DALE）at birth[103]. Another similar study conducted in Spain found that illiteracy rate and the percentage of smokers in the population were the primary factors associated with the geographical variation of DALE[104]. An ecological study in Japan found that among 181 factors related to socio-demographic factors，3 factors could potentially explain the differences in DFLE of the Japanese older population：good self-rated health status，a high proportion of older workers，and the presence of a large number of public health nurses[105].

At the individual level，a cohort study among 110，790 individuals age 40-79 years in 45 areas in Japan showed that individuals with low levels of education had an increased overall risk of death（16% and 26% increased risk for men and women，respectively）[106]. The results of another cross-sectional research study on a Japanese sample of 80，899 persons＞15 years old in 1995 demonstrated that compared to people whose household income was≥5 million yen，those whose household income was＜1.5 million yen were 1.93 times more likely to perceive their own health as the worst[107]. By surveying Japanese workers age 20-40 years，Togari and Yamazaki[108] found an association between poor sense of coherence（SOC）and occupation（non-temporary＜temporary and professional＜white-collar＜blue-collar）and an association between the poorest SOC and unemployment and lower education levels.

These direct SES effects on HALE results indicate that health inequalities exist in Japan. Thus，social and political initiatives are needed to correct these social inequities.

Indirect SES-HALE Mechanism

Evidence of the socioeconomic gradient in health has pushed for explanations that go beyond the direct relationship between SES and HALE. For example，education is believed to promote good health not only by generating economic resources（income and employment）but also by providing social-psychological resources：knowledge and skills by which people are able to better self-manage illness and disease，healthy behaviors，ability to cope with stress，perceptions of control，and social support. Income has been shown to promote good health by affecting nutrition，housing quality，exposure to environmental hazards，and access to adequate physical and mental health care.

According to Mirowsky and Ross[109]，increasing educational attainment improves health by increasing individual agency，self-efficacy，and problem-solving capacity，all of which promote a healthy lifestyle. In addition，education level is less likely to be affected by health impairments that developed in adulthood compared to other indicators of SES，such as occupation or income[110]. However，recent research suggests that income is associated more strongly with progression（rather than onset）of disease than with education，suggesting that economic resources promote health in part by increasing the abilities to pay for medical care and to acquire transportation to care facilities[111][112].

In Japan，Fujino et al.[113] examined the association between educational level and major causes of death in a prospective cohort study. The findings suggest that lower levels of education can lead to insecure income，hazardous work conditions，and poor housing，which can increase the risk of death due to external causes. Previous findings from another cohort study of 1，266 Japanese male office workers age 35-59 years identified smoking as a risk factor for type 2 diabetes，and therefore，the significant inverse relationship between diabetes and education might be partly explained by the association between smoking and education[114].

A few cohort studies conducted in Japan using structural equation modeling（SEM）were conducted to explore the indirect relationships between SES and HALE. Hoshi et al.[115] discovered that health and life conditions were not determined by current dietary and lifestyle habits，which many studies showed. However，the conditions were more directly affected by three health-related dimensions three years earlier，and indirectly affected by educational attainment and previous annual income as well. Using the same data，Takahashi et al.[116] suggested that decreases in yearly income could disrupt the balance of health of older people via decreasing daily activity and poor social support. Bosako et al.[117] concluded that SES affected the duration of survival both directly and indirectly via three health factors（physical，mental，and social），and that the indirect effects were stronger than the direct effects.

SES-HALE Mechanism by Gender

Longitudinal comparative data on life expectancy indicates a central societal division between men and women（on average，women live longer than men）. Biology is only one of a group of factors that shape health outcomes，and it might not be the most important one. As Waldron[118] argues in her “Gender Role Modernization Thesis，” structural-social changes have led to social，cultural，and economic changes that are shaping a more healthy life—on average，women smoke and drink less than men and are less involved in physically demanding and risky jobs.

Gender differences in the association between income and health have also been reported in a recent cross-sectional study among 9，650 participants age 47-77 years[119]. Men tended to report more fair or poor health as household income decreased，while the results for women differed. In a study of 2，200 elderly Japanese，Liang et al.[120] reported that in contrast to Western countries，an educational crossover exists only among elderly men，which may be due to gender and SES differences in causes of death，morbidity，and health behavior. Yamazaki et al.[121] examined the association between annual household income and the eight scale scores of the Medical Outcomes Study Short Form-36 Health Survey（SF-36）as a quantifier of Health Related Quality of Life in Japan. A total of 3，395 people age≥16 years were selected from the entire population of Japan using stratified-random sampling method. Results showed that a strong association existed between annual household income and SF-36 scores among men，but there was only a limited association among women. Fukuda et al.[122] noted that the relationship between mortality and the education-income index was stronger for males than for females，and that gender differences in the association between mortality and municipal SES were due to substantially different patterns in the primary causes of death between males and females.

SES-HALE Mechanism by Age

Studies that address different health dimensions suggest that social inequalities might develop differently according to the health indicator considered. For example，as Lampert[123] has shown，using a sample age 70 to 100+ years，small socioeconomic differences in physical aspects of health up to the age of 90 years were followed by significant differences in individuals≥90 years old. A contrasting picture emerged for functional health，in which socioeconomic differences were significant at age 70-79 years and disappeared in older age groups[124]. One study using several SES and health indicators and a sample limited to individuals≥60 years old showed only slight age variation in the effect of SES on health in Germany，supporting the continuity hypothesis[125]. In Japan，a study conducted by Liang et al.[126] showed that educational differences in mortality tend to converge in the 70- to 79-year-old age group. The authors also confirmed，in the Japanese context，that the effect of SES on health was small in early adulthood，greatest in middle and early old age，and relatively small again in late old age. One of the few studies of this issue indicated that while health behavior mediated the association between education and functional health in 55- to 70-year-old people，psychosocial factors became more relevant for older people[127].

SES-HALE Mechanism by Geographic Location

Socially disadvantaged areas were reported to be associated with higher mortality，morbidity，and health related risk behaviors. A series of ecological studies using the Japanese municipal statistics on SES and mortality by Fukuda et al. suggested that a lower SES was related to higher mortality. The first study classified the municipalities across the country into quintiles according to the index of SES obtained from income and education，and found that mortality gradient by SES and excess deaths in the lower SES quintiles due to injury and suicide markedly increased from the 1973-1977 to 1993-1998 periods for both males and females[128]. Another related ecological study found that health expectancy at age 65 years was significantly positively correlated with per capita income in municipalities across Japan. The relationship was stronger in larger municipalities（i.e.，those with populations of more than 100，000 individuals）than in small- and medium-sized municipalities. The results of this study indicated that the health status of older people is substantially decreased by disadvantageous socioeconomic conditions[129].