Probing the Issue of Health Disparities:
High correlation with income and education - Policies based on the premise of homogeneity should be reconsidered

SHIMIZUTANI Satoshi Faculty Fellow, RIETI

The stereotypical image of Japanese society as homogeneous and equal is reportedly becoming a thing of the past. However, some reservation is needed when considering the recent fad to debate the discrepancies. First, many arguments take note only of the trend of disparities, not the level. Although there is an urgent need to accurately investigate developments that cannot be tracked simply with currently available data, such as the situation of families that receive public assistance and an increase in the homeless population, the level of disparities in Japan is nothing compared with the situations in Europe and the United States, even though gaps are also widening in Japan.

Second, depending on the advocate, there are a variety of opinions on how the differences should be measured. There is also no consensus among advocates in terms of whether disparity in economic, employment, education or regional status should be discussed or if more emphasis should be placed on equal opportunity as opposed to equal outcome.

Particularly, empirical investigation and discussions are not sufficient regarding the disparity in health that directly affects the quality of life in both the long and short term. In Europe and the United States, health disparities have already been recognized as an obvious fact based on data, and empirical analysis on factors affecting health discrepancies has made significant progress. In Japan also, recent studies have shown that health disparity certainly exists among individuals and regions (such as a study centering in part of Aichi Prefecture, by Professor Katsunori Kondo at Nihon Fukushi University). Health disparities are also not extraneous to socioeconomic attributes (income and education levels).

*            *            *

So, what kinds of health disparities are there specifically, and how are they related to income and education levels? I would like to illustrate these points using the Japanese Study of Aging and Retirement (JSTAR), a panel survey of global standards on middle-aged and older adults, commenced by former RIETI President Masaru Yoshitomi. In the first survey in 2007, there were a total of 4,200 participants, males and females aged 50-75, living in Takikawa city in Hokkaido; Sendai city, Adachi ward in Tokyo; Kanazawa city, and Shirakawa town in Gifu Prefecture.

This column introduces some of the results from a joint study conducted together with Professors Hidehiko Ichimura and Hideki Hashimoto of the University of Tokyo. The actual results discussed below are limited to those verified as statistically significant after prime factors such as age, sex, and region were controlled for.

First, the general physical condition indicator showed there was a tendency for both male and female respondents to report that their "self-rated present physical condition is not very good or bad," if their education level was low. Regarding activities of daily living (ADL), such as dressing, eating, and bathing, there was tendency among males with a low education level to complain of difficulty with these tasks. With respect to more sophisticated, involved daily activities (instrumental ADL), such as the preparation of meals and processing of bills, there was a tendency, among both males and females with a low education or income level, to complain of difficulty.

Such tendencies were also evident in the prevalence of certain diseases. For example, although the prevalence of hyperlipidemia and diabetes are high in cities, the prevalence of hyperlipidemia is high if the income or education level is high. However, the prevalence of diabetes is high if the income or education level is low. The prevalence of stroke is high among males with a low education level but education and income level differences were not noted in the case of cancer. In Europe, the prevalence of cancer is high if the income or educational background is low and, when contracting cancer, many cases tend to be fatal. There are also some studies showing that the higher the education level, the higher the prevalence becomes.

Chronic lung problems are more common among males with a lower education level, which may be attributable to the high smoking rate when they were young. Meanwhile, the prevalence of arthritis is high among females with a low education level. In addition, those who have a lower education level tend to complain of trouble with sensory functions (eyesight, hearing capacity, and chewing capacity).

With respect to mental health, people with depression tend to have a lower education level and have more trouble with daily activities. Such a tendency appears also in their cognitive functioning. The lower education level, the more likely they will have impairment in their cognitive function and conversely, when a person's cognitive function is impaired, they tend to have a lower income or education level. The effect of income and education levels has also emerged in the different usage of medical services. Those with a higher education level tend to undergo dental care and operations in hospitals. On the other hand, when people are treated as outpatients, even just once, the number of doctor visits tends to increase if they have a lower education level.

*            *            *

Socioeconomic attributes affect not only the present state of health and disease, they are also observed in health behaviors that increase the risk of disease. The smoking rate tends to be high if the income or education level is low. Also, if the education level is low, the amount of drinking tends to be rather low but the tendency of not exercising is strong among males, while the risk of obesity is particularly high among females.

Health checkups also show that the higher the income or education level, the higher the consultation rate becomes (54% for those with a high school education or less, and 74% for college graduates). Finally, the pattern of nutritional intake is also influenced by socioeconomic factors. If the education level is high, intake levels of salt, alcohol and cholesterol tends to be high. However, even when the fat-calorie ratio is high, those who have a high education level also tend to consume large amounts of fruits and vegetables. Alcohol intake is large even when income is high.

As mentioned, a health discrepancy clearly exists in Japan and is closely linked to socioeconomic attributes such as income and education level. This fact presents two enormous challenges.

First, despite the fact that the health insurance system in Japan regards universal care as its public stance and provides free access to medical services, there is still a health discrepancy. As also clearly shown by JSTAR, the tendency for medical services to be difficult to receive, when the level of income is low, could have an impact as one of the factors, given that the self-burden ratio of medical expenses to equivalent household income is regressive (figure). If the health discrepancy is caused by differences in the access to medical services, it is possible to think that the current health insurance system does not function sufficiently. On the other hand, if there is a health discrepancy even when access is the same and irrespective of the income level, this is a problem entirely unrelated to the design of the health insurance system.

Figure : Ratio of medical costs to household income per person over the last 12 monthsRatio of medical costs to household income per person over the last 12 months

(Note) Household income per person is adjusted for the number of family members (on an equivalent basis). Divided into quartiles by income bracket.

(Sorce) Ichimura, Hashimoto, Shimizutani "First Results from the Japanese Study of Aging and Retirement(JSTAR)"

Secondly, the suggestion that if people are in a disadvantaged position in terms of income or education they can easily fall into a disadvantageous health situation at the same time provides an important point when discussing discrepancies. This means that it is insufficient to discuss discrepancies based only on a single criterion. This also gives an essential perspective when formulating policies. As mentioned above, those with a low income, low education level, or high health risk tend not to undergo health checkups. Given this, even in the specified health examinations that started in the last fiscal year, the focus should be placed on how to raise the consultation rate of those with high health risks, rather than imploring the general public to undergo health checks.

*            *            *

So, how should policies respond to the health disparity? Before discussing this issue, it is necessary to identify the causal relationship between the state of health and income and education levels. Though there is a close correlation between the two, a causal relationship has yet to be fully analyzed. For example, regarding the relationship between the level of income and state of depression, it is not necessarily clear whether people fall into a situation where they cannot obtain sufficient income because they become depressed, or whether their low income causes their state of depression.

Naturally, the policy response will differ depending on which is determined to be the case. To identify such a causal relationship, it is essential to build up panel data that follow individuals over many years. If it is possible to conclude through continuous research that people go into depression as a result of their income decreasing due to some external factors, it will become clear that this problem could to some extent be prevented through income redistribution. Continuous research could also answer an essential question regarding discrepancy debates over whether or not gaps are fixed. In addition, even among socioeconomic attributes, the correlation with the state of health is different between income and education levels. For effective policy, it is necessary to clarify this relationship. Furthermore, even though income and education levels are linked to health discrepancy, the latter will not always be resolved by eliminating the former in the ordinary course of events.

Activities to clarify such a causal relationship are progressing in Europe and the U.S. as an area of study called "social epidemiology," a discipline of public health that has also been attracting attention lately in Japan.

Discrepancy debates tend to provoke intuitive and emotional responses. Health is influenced not only by medical policies but also socioeconomic policies. To the contrary, people's health has an impact on their socioeconomic condition. Carefully examining this mutual relationship, we should construct a solid foundation for policy planning based on scientific knowledge.

Health discrepancies confirmed at the micro level cast doubt on debates over social security which are exclusively preoccupied with revenue source discussions at the macro level. Effective policies cannot be brought out by only looking at an average image of the elderly, mechanically supposing their homogeneity. Rather it is necessary to conceive that the behavioral patterns of individuals should be examined on the premise of heterogeneity and that measures to improve policy efficiency, by exerting incentives, should be considered. For example, if the analysis results of panel data are used, it will become possible to narrow down targets for the consultation of health checkups and make policies more detailed schemes that take the characteristics of the target individuals into account.

Clearly articulating such knowledge using micro data that is internationally comparable will undoubtedly contribute not only to the future of Japan itself, but also to the world waiting for elucidation of the experiences of the country with the world's highest longevity rate.

>> Original text in Japanese

* Translated by RIETI.

July 3, 2009 Nihon Keizai Shimbun

September 8, 2009