University of Southern California Dissertations and Theses

The effects of time and space on health status of older adults in China

(USC Thesis Other)

The effects of time and space on health status of older adults in China

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content THE EFFECTS OF TIME AND SPACE ON HEALTH STATUS OF OLDER ADULTS IN CHINA by Yawen Li A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (SOCIAL WORK) August 2009 Copyright 2009 Yawen Li ii Dedication To my grandparents, who raised me, loved me, taught me, and aroused my curiosity in understanding the life of older adults. iii Acknowledgements I would like to thank the members of my dissertation committee for their valuable guidance and continuous support: Dr. Lawrence Palinkas, Dr. Merril Silverstein, and Dr. Iris Chi. I am grateful to each of them for their mentorship and generosity. I am honored to have had the opportunity to learn from such remarkable scholars. Special thanks to Dr. Iris Chi, my mentor and dissertation chair, for her untiring support for me to work on the topic which is genuinely interesting to me. I am also fortunate to have worked with her closely over the past six years, observing and learning from her. I am enlightened by not only her research abilities, but also her abilities to communicate, survive, and thrive as a female and a minority scholar on campus. I would not be able to complete my doctoral degree without the love and support of my family. I would like to thank my husband, Inghwee Chok, for his constant encouraging, supporting, and cheering me along the way. I am also thankful to my parents, Changqing Li and Ruiqiong Wu, who value my education more than anything in their lives and always encourage me to pursue my dream without any worries. My brother, Shuwen Li, accompanies and takes care of our parents most of time, which greatly reduces my worries while living thousands miles away from them. Lastly, I would like to thank a special person who ignited the changes in my life. It is exactly 10 years ago when I met Veronica Pearson for the first time. She introduced me to a profession called social work, which I had never heard before at that time. Amazed by her illustration on the social problems and issues that social workers can fix, I was drawn to the concept that I would be part of that profession some day. It is at that iv moment that my wonderful journey from Kunming to Hong Kong, then from Hong Kong to the United States started. It is also that moment, which defined the professional trajectory I have taken: from an English translator, a community health worker, a social worker, and then to a social work professor. I enjoy every minute of it so far. v Table of Contents Dedication ii Acknowledgements iii List of Tables vi List of Figures vii Abstract viii Chapter 1: Introduction, Background, and Conceptual Framework 1 Chapter 2: Data Description 17 Chapter 3: The Impact of Socioeconomic Status on Health Trajectories among Older Adults in China 22 Chapter 4: The Changing Social Environment and its Influence on Health among Older Adults in China 55 Chapter 5: Studying the Effects of Social Environment on Health among Older Adults in China 92 Chapter 6: Discussion and Conclusions 111 References 117 vi List of Tables Table 3.1: Sample Characteristics 35 Table 3.2: Two-Level Models of Functional Health Status Change 37 Table 3.3: Two-Level Models of Self-Rated Health Status Change 42 Table 3.4: Two-Level Models of Cognitive Function Change 46 Table 3.5: Logistic Regression of Status in 2006: Participant versus Dead or Lost to Attrition in 2006 49 Table 4.1: Sample Characteristics 68 Table 4.2: Outcome Distribution 70 Table 4.3: Correlation between Community-Level Variables 71 Table 4.4: Multi-Level Growth Curve Model on Instrumental Activity of Daily Living 75 Table 4.5: Multi-Level Growth Curve Model on Self-rated Health 80 Table 4.6: Multi-Level Growth Curve Model on Cognitive Function 85 vii List of Figures Figure 2.1 Map of Surveyed Provinces in CHNS 18 Figure 4.1: The Conceptual Framework of Ecological Effects on Health Development 58 Figure 5.1: The Conceptualization of Social Environment 98 Figure 5.2: The Integrated Life Course and Ecological Perspective 110 viii Abstract This dissertation consists of two independent studies and one conceptual paper, addressing one overarching question: how changes in socioeconomic status (SES) based social stratification and changes in social environment affect the health status of older adults in China across time and space. In the two empirical studies, analyses were performed using the four waves of data from the China Health and Nutrition Survey (CHNS), a longitudinal survey of nine provinces in China. Latent growth curve analyses from the first study (2,345 individuals aged 55 and above) show that the health of older Chinese adults continued to be shaped by individual education, occupational class, and income. Cumulative advantage hypotheses are supported by the findings that higher education predicts both better self-rated health and cognitive function at baseline and slower declines over time. Higher income is associated with better functional health at baseline as well as a more gradual decline over time. In addition, this study reveals that the relationship between occupational class and cognitive function as well as between income and self-rated health do not support the cumulative advantage hypothesis. In the second study, multi-level growth curve analyses (2,345 individuals age 55 and above from 191 communities) show that social environment exerted independent effects on health outcomes over and beyond the effects of individual-level factors. Social environment, particularly household median income, communication development, and time needed for accessing healthcare affected initial health status but not the rate of change in health status. The effects of social environment on health differ by individual education and income. This study suggests that continuing improvement on economic ix and social environment could potentially improve the health status of older adults in China. Following the two empirical papers is one conceptual paper, addressing the conceptual and methodological challenges of studying the effects of social environment on health. Adopting the political economy perspective as its theoretical framework, this paper describes social environment as being composed of economic development and social development. It is suggested in this paper that changes in social environment should be taken into account while studying the ecological effects of social environment on health. In the end, an integrated life course and ecological perspective is proposed to study the dynamic relationship between social environment and health. In summary, this dissertation demonstrates that changes in individual socioeconomic status as well as in macro-socioeconomic development had substantial influences on health status among older Chinese adults. 1 Chapter 1: Introduction, Background, Conceptual Framework Introduction This dissertation includes two independent studies and one conceptual paper, addressing how changes in socioeconomic status (SES) based social stratification and changes in social environment affect the health status of older adults in China across time and space. Specifically, this dissertation investigates health status across time, focusing on how the effect of SES on health status changes with age. In addition, this dissertation examines how the changes in social environment across space affect the health status of older adults over time. Background Since the economic reforms in the late 1970s, China has embarked on one of the greatest social experiments of our time. The continuing expansion of its economy over the past three decades has been unprecedented. The expansion, however, has been accompanied by a widening development gap between urban and rural areas, as well as between coastal and inland regions. One of the detrimental consequences of market- oriented economic reforms is an increase in socioeconomic inequality and a rise in health disparities. There is an almost 20 years’ gap in life expectancies between those lived in urban developed areas and those living in rural deprived areas (Zhao, 2006). In addition, the gaps in infant mortality and morbidity have also increased dramatically between urban and rural areas (Liu, Hsiao, and Eggleston, 1999; Zhao, 2006). Rising health disparities among older adults have also been documented in terms of high mortality rates (Liang , McCarthy , Jain , Krause , Bennett & Gu, 2000; Zimmer, Kaneda, and 2 Spess, 2007), incidence of functional disabilities (Beydoun & Popkin, 2005; Liang, Liu, and Gu, 2001; Yi & Vaupel, 2002), incidence of chronic illness (Zimmer & Kwong, 2004), and poor self-rated health (Liu & Zhang, 2004; Yip , Subramanian , Mitchell , Lee , Wang & Kawachi, 2007; Zeng, Vaupel, Xiao, Zhang, and Liu, 2002). At the same time, China is experiencing rapid population aging. China has the largest number of older adults in the world with 114 million older adults aged 60 and above in 2005 (National Bureau of Statistics of China., 2006). Among the older population, the number of those who are aged 80 and above and who need the most care and support is expected to reach 100.5 million, or 25.5% of the oldest old in the world by 2050 (United Nations, 2005). As elsewhere, the existence of an aging population carries important implications for the need to meet increasing demands for support and healthcare. It also implies the needs to restructure service delivery to meet those demands. Therefore, understanding factors related to the causes of health disparities is crucial. Most of previous research on the health disparities of the Chinese population has been focused on examining the patterns of health status and related determinants of health at particular times, whereas the changes in health status and the reasons leading to the changes have remained largely unexplored (Gao, Qian, Tang, Eriksson, and Blas, 2002; Zeng et al., 2002; Zimmer & Kwong, 2004). In addition, despite the dramatic societal changes in China since the 1970s, their effects on individual health have been left unexplored empirically. Previous studies have focused more on microlevel factors such as individual SES (Anson & Sun, 2004; Gao et al., 2002; Liu & Zhang, 2004), urban and rural residency (Lawton, Silverstein, and Bengtson, 1994; Raudenbush, 2000), health 3 insurance coverage (Yu, 2005), gender (Yu & Sarri, 1997), and health knowledge and behaviors (Feng, Ren, Shaokang, and Anan, 2005; Hesketh, Ding, and Tomkins, 2003). Less is known about the effects of social environment on individual health outcomes. The picture of health disparities among older Chinese populations as a whole is therefore incomplete. This dissertation addresses these gaps with the hope of contributing insights into the roles of time and space in influencing the health of older adults in China by capitalizing on the panel data from the Chinese Health and Nutrition Survey (CHNS), a longitudinal survey of nine provinces conducted by the Population Center at the University of North Carolina in collaboration with the Center for Disease Control in China from 1989 to 2006. Conceptual Frameworks The two studies in this dissertation are informed by the life course perspective, the integrated life course perspective and the ecological approach. The conceptual paper uses the political economy perspective as its theoretical framework to understand the important aspects of social environment in relation to health for older adults in China. The Life Course Perspective The life course perspective is integral to gerontology research. According to Bengston, Elder, and Putney (2005), the five principal ideas which define the life course perspective are: 1) that lives are linked, referring to interdependence and interconnectedness of lives across the generations provided by bonds of kinship; 2) that lives are shaped by historical time and place; 3) that life transitions and their timing 4 relative to the social contexts in which individuals make choices are important; 4) that individuals take an active role in constructing their lives; and 5) that aging and human development are lifelong processes in which the relationships, events, and behaviors during earlier life stages have consequences for later life relationships, status, and well- being. Researchers have adopted the life course perspective to bridge two rival approaches to the study of chronic disease etiology, namely the perspective of biological programming in utero, and the perspective of effects of adult lifestyle (Kuh, Ben-Shlomo, Lynch, Hallqvist, and Power, 2003). The biological programming approach hypothesizes that the structure or function of organs, tissues, or body systems may have been formed during critical periods of growth and development as early as the perinatal or in utero period, influencing future risk of adult chronic disease. An alternative approach focuses on adult behaviors, examining how those behaviors influence the onset and progression of disease in adulthood. The life course perspective connects both the perinatal and adult periods through the central premise that various factors, both biological and social, influence health and disease in adult life independently, cumulatively, and interactively (Kuh et al., 2003). Two of the five principal ideas of the life course perspective, namely that lives are shaped by historical time and place, as well as that life is a long developmental process, are emphasized in the study of health disparities. The primary foci of the life course perspective are human biological and social developmental trajectories over time and how those trajectories are shaped by social context (Krieger, 2001b). Particular attention 5 has been given to the study of health inequalities, especially inequalities associated with socially patterned exposures during childhood, adolescence, and early adult life (Kuh et al., 2003). Guided by the life course perspective, Wadsworth (1997) reviewed age related psychosocial factors and examined how they affected health outcomes at later stages of life. These factors included socioeconomic circumstances such as occupation, education attainment, financial security, health behaviors, housing conditions, nutrition habits, and parental relationships. Such factors are more likely to operate in a cumulative fashion than solely or mainly at critical developmental periods. Accumulation of risk may be due to clustering of exposures or to forming chains of risk such that one negative exposure increases the subsequent risk of another negative exposure (Lynch & Smith, 2005). For example, poor socioeconomic circumstances in early childhood may predict both poor biological conditions such as those resulting from poor nutrition, slow rate of growth, and increased risk of childhood infections and illnesses. It may also predict detrimental social effects such as decreased chances for education, the high possibility of living in unsafe communities, and the increased likelihood of chronic psychological stress resulting from lack of money. All of these unfavorable conditions add up and create a tenuous situation for physical and mental health in the long run. The life course perspective has been gaining increasing popularity in study of health disparities in recent years. However, most research on the effects of social factors on health has been conducted by associating childhood factors with adult health outcomes. 6 Much less investigated are the effects of adult midlife experience on health in later life (Wadsworth, 1997). In the first study of this dissertation, the life course perspective was adopted to examine how SES affects health status over time and how that effects vary with age. Through this study, the knowledge of how SES influences health status at various stages of life could be gained. The Ecological Approach Following earlier eras in which research focused primarily on sanitation, infectious diseases, and chronic diseases, epidemiology research has moved to embrace eco-epidemiology since the 1990s (Susser & Susser, 1996) and has experienced a surge in publications (Diez Roux, 2002; Entwisle, 2007). The concept of ecology has been used as a metaphor to refer to interactive systems in which individuals and their surrounding environment are connected in a coherent manner. Many terms such as eco-epidemiology, ecosocial theory,ecological approach, area or place effects, neighborhood effects, contextual effects, and multilevel frameworks have been used interchangeably in epidemiology literature though sometimes the terms connote slightly different emphases or implications. In this dissertation, the term ecological approach was used to refer to the framework guiding the analyses of the social environment and its influences on individual health. Macintyre and Elleway (1993) have stressed that epidemiology studies should take into account human “habits, modes of life, and [human] relationships to their surroundings” (p. 36) and have suggested that there are potential influences on human health which result both from the infrastructure and the social environment. In other 7 words, the location where people live and work and their situation there affect health outcomes. Later Macintyre et al. (2002) have developed an organizing framework based on the concept of general human needs. They identified five features of local areas which might influence health: physical features of the environment shared by all residents in a locality; availability of healthy environments at home, work, and play; services provided to support people in their daily lives; sociocultural features of neighborhoods based on the political, economic, ethnic and religious history of given communities; and the reputation of areas, referring to how the areas are perceived both by their residents, by outsiders, or by service or amenity planners and providers, bankers, and investors. Such perceptions may influence the infrastructure of an area. This general human needs framework provides an intuitive typology of possible ecological factors but does not specify how these factors permeate their localities to affect health outcomes at the individual level. Of the many challenges Macintyre et al. (2002; 1993) have mentioned, the need to differentiate conceptual constructs about ecological effects and the need to consider a plausible time interval between environmental exposure and health outcomes are particularly noteworthy. The conceptual constructs often confused are compositional effects and contextual effects. The former refers to characteristics of individuals residing in given areas while the latter refers to communities’ global features that are irreducible to individual-level characteristics. In response, Cummins et al. (2005) have advocated the direct measurement of environmental features that might influence health. The need to consider a plausible time interval suggests that the effects of living environment imposed on health is cumulative over time rather than instantaneous, therefore analyzing different 8 health outcomes in relation to contemporary measures of environment using a cross- sectional design seems inappropriate. Longitudinal data and testable hypotheses on plausible time lag from environment exposure to manifestation of health outcomes are required to better extrapolate ecological effects. An alternative perspective on community effects on health was proposed by Starfield and Shi (2004; 1999). They organized factors that affect health by their proximity to an individual’s core health status. This framework begins with individual- level factors, which are the most proximate determinants, such as the physiology, including biological and genetic states, illnesses and injuries; material resources, supporting life’s needs to which individuals have access, such as food and shelter; social connectedness including relationships with others, involvement in social events, and the ability to influence controllable events; chronic stress such as that created by racism and classism; and health services received. More distal factors operate at the level of the community in which individuals reside and work. These include occupational/environmental exposures such as exposure to noise, pollution, and weather conditions; material resources such as food markets, schools, banks, and means of transportation; social resources such as community centers, playgrounds, parks, and crime prevention programs; psychological factors such as the experience of institutionalized racism; and health system characteristics such as the availability of healthcare resources and their distribution as governed by availability of funds. 9 The most indirect and distal factors in Starfield’s analysis are social policies which directly influence the community. Policies included are occupational/environmental policies, social policies, economic policies, and health policies. Economic policies and social policies promote and regulate the production, consumption, reproduction, distribution and redistribution of resources. Together with health policies, they also decide the organization and financing of health services. This framework proposed by Starfield et al. (2004; 1999) distinguishes itself from others by considering the linkage between contextual characteristics and individual experiences, both as direct effects (e.g., the impact of healthcare service delivery systems on individual health status) and as cross-level interaction effects (e.g., the impact of healthcare service delivery systems on the effects of aging). However, this framework does not consider the time dimension when specifying the relationship between individual experiences and the context. How individual and context have changed, and how the changing context affects individual health over time is left unexplored. The Integrated Life Course Perspective and Ecological Approach Wheaton and Clarke (2003) have extended current ecological studies by integrating it with the life course perspectives, adding a time dimension to their conceptualization of ecological effects. In their study of contextual effects on early adult mental health, they hypothesized that the effects of neighborhood could be either concurrent or lagged. Concurrent effects are the combination of both early and current contextual effects. The effects of the early context on current health are mediated by the present context, assuming that there is sufficient stability in neighborhood context over 10 time. On the other hand, contextual effects are lagged when the effects of early neighborhood are mediated by stressors experienced in early stage of life. They reasoned that stress proliferation and cumulative burden effects may lead to the emergence of early adulthood mental health problems. This framework suggests a new way to understand relationships between individual changes as well as the changes in social contexts at the same time. In this dissertation, the integrated thinking of the life course perspective and ecological approach was adopted to guide conceptualization of how changes in social environment affect individual health over time. As shown in Figure 5.2, the time dimension, as emphasized in the life course perspective is represented by the horizontal line, referring to both changes in health status and changes in social context over time. The contextual dimension, as stressed in the ecological perspective is depicted as the vertical line on the right side. It implies that individuals are nested within the social context and are influenced by the surrounding social environment. In the first study, the time effects on health status changes were examined, testing how relationship between SES and health status change with time (age). In the second study, the space effects on health status change were investigated, testing how the social environment in a geographic region affects health development as well as how the effects of social environment on individual health differ by individual SES. The Political Economy Perspective Emerging from the context of the economic crisis in the U.S. in the 1970s, the political economy perspective has been applied to study a variety of social issues 11 including aging and related economic and social policies in advanced capitalist societies. Drawing from the theory of state, Estes (1984) argued that assuring the survival of economic systems is the primary function of any capitalist state such as the United States. The secondary function of the state is to maintain the legitimacy and operation of the social order to avoid destructive social unrest. The implication of the two major functions of the state in terms of social spending is that social spending has to ensure a supply of productive labor and to create effective demand for goods and services so that favorable conditions for economic growth can be maintained. At the same time, social spending needs to ameliorate conditions that might be politically threatening; it must promote the legitimacy of the state and the economic system, and promote control of social behaviors. Therefore, social spending is used to assist business and industry in their accumulation of wealth through favorable tax treatment and government subsidies, at the same time creating social and welfare programs of education, job training, nutrition, shelter, health care, and unemployment compensation to ameliorate the displacement costs resulting from the operation of the economic system. There always exists a tension between the two competing demands for social spending: favorable treatment for business on the one hand and social and welfare programs on the other. According to Estes (1984), the resources in the society are distributed depending on who the dominating organization or groups are. She used a conflict approach with the elite and class theories to explain that public policy and the allocation of resources result from conflict and power struggles, in which certain groups and organizations dominate resources and control outcomes. In relation to health and health related issues, the social groups and organizations struggle for ideological 12 dominance, economic resources, and political resources such as control of appropriate government bureaus and non-governmental political organizations. In the study of health, the political economy perspective addresses economic and political determinants of health and the distribution of disease within and across societies (Doyal & Pennell, 1979). Its core assumption is that the fundamental causes of social inequalities in health lie in the economic and political institutions shaping the decisions that create, enforce, and perpetuate economic and social privilege and economic and social inequality (Krieger, 2001a). The perspective of the political economy of health, pioneered by Estes (1979; 1984; 2001) has been extensively developed to examine the broad implications of economics, politics, society, and culture l on the health of the aged population. Central to these analyses are questions related to the social determinants of health and illness; the social creation of dependency and the management of that dependency status through public policy and health services; medical care as an ideology and as an industry in the control and management of the aging; the consequences of public policies for the elderly as a group and as individuals; the role and functions of the state in relation to aging and health; and the social construction of the realities of both old age and health (Estes, 1984). Though this perspective has been used extensively in studies of advanced capitalist societies, it is less applied to study health and aging issues in non-capitalist societies, or those which incorporate features of both socialism and capitalism. In the conceptual paper of this dissertation, the political economy perspective was applied to 13 conceive and understand important aspects of social development in relation to health in China. Research Questions The first paper addresses the effects of SES on health over time. Specifically, following questions are asked: Do the effects of SES on health vary over time? Will SES- based health disparities continue and increase over time? The second paper of this dissertation addresses the effects of the social environment on health over time. Research questions asked are: Will social environment, including economic development and social development, affect the health trajectory of old adults above and beyond individual factors? Will social environment modify the relationship between individual SES and health outcomes? In the third paper of this dissertation, the questions asked are: How should one conceive of the social environment in China? What aspects of social environment are important to the health of older adults? What challenges exist in current research that should be addressed in future research? How to address those challenges in a conceptual model? Contributions This dissertation contributes to the existing literature by empirically testing the time-related and place-related effects on the health status of older adults. Using longitudinal datasets enables the examination of time-related changes in health status and the effects of time-variant or time-invariant covariates on health over the long run. It is hoped that this dissertation may contribute to knowledge in the following areas. 14 Age variation in SES-based health disparities This life course perspective posits that various factors, both biological and social, influence health and disease independently, cumulatively, and interactively (Kuh et al., 2003). It provides a general theoretical framework to understand health trajectories across the life span. In the first paper, guided by the life course perspective, the cumulative advantage hypothesis and the age-as-lever hypothesis were tested to determine how SES- based health disparities vary with age. Western literature usually shows that education and income-based health disparities diverge over time and the effects of occupation weaken as the individual ages (Marmot & Shipley, 1996; Willson, Shuey, and Elder, 2007). However, such study has rarely been conducted among populations in developing countries where socioeconomic status may affect individual health differently because of the varied emphasis or meanings given to SES under different macro-socioeconomic structures. It is not certain whether education brings the same health benefits to older adults who live in an area where the marginal return for education is lower than expected. It is not certain whether occupation brings the same health benefits to older adults whose occupation may not entitle them to the same health access. Neither is it certain whether income means the same for the health of older adults whose incomes are largely dependent on their children and other family members. Studies conducted among older Chinese adults have established the relationship between SES and various health outcomes using cross-sectional designs (Zeng et al., 2002; Zimmer & Kwong, 2004), but they did not address the issue of over-time 15 relationship between SES and health, leaving the assumption that those relationships are static, instead of changing over time. This dissertation complements previous studies and takes our understanding of that topic one step further. Changes in social environment in relation to health outcomes Current ecological frameworks (Macintyre et al., 2002; Starfield, 2004) for understanding the impact of social environments on health have rarely considered the changing nature of social environment. In the second study of this dissertation, an integrated ecological and life course perspective (Wheaton & Clarke, 2003) was adopted to conceptualize the effect of changes in social environment on health development. Accompanying the longitudinal data collected at the individual level is longitudinal community data in the CHNS, which documents the changes in communities over the years. Using this data enables the test of time-variant factors in the social environment in relation to changes in health; thus moving current research beyond making unrealistic assumptions that communities do not change or the magnitude of that change is negligible. Instead, this dissertation takes changes in communities into account and models the concurrent and lag effects of changing communities on health outcomes. Finally, this study expands our knowledge about older Chinese adults, a population few studies have examined on a longitudinal basis. In particular, studying the effects of contextual factors on older adults mitigates a potential problem in ecological analysis, that is, that community indicators become less predictive of highly mobile populations. This is because older adults tend to be far less geographically mobile than younger adults, so they tend to have long-term exposure to their communities and more 16 frequent contact with community resources and services (Balfour & Kaplan, 2002). Therefore, community environments are likely to exert a greater influence on them (Diez Roux, 2002). In summary, this dissertation may provide useful information for policy- makers and program managers. In the following chapters, data used for this dissertation was presented in Chapter 2. Then Chapters 3 and 4 describe the two empirical studies which address the overarching question of how social stratification, accompanied by the rapid social changes in China, at both the individual-level and community-level, affects health outcomes of older adults in China. Chapter 5 is a conceptual paper addressing the theoretical and methodological issues in current research. Finally, Chapter 6 is the discussion and conclusion. 17 Chapter 2: Data Description General Description of the Survey This dissertation examined the roles of time and space in influencing the health of older adults in China by capitalizing on the panel data from the Chinese Health and Nutrition Survey (CHNS), a longitudinal survey of nine provinces conducted by the Population Center at the University of North Carolina in collaboration with the Center for Disease Control in China from 1989 to 2006. This survey collected information on health, nutrition, family planning policies, and programs implemented by national and local governments in China to investigate how the social and economic transformations of Chinese society have affected the health and nutritional status of the population. Sampling The nine provinces (Guangxi, Guizhou, Heilongjiang, Henan, Hubei, Hunan, Jiangsu, Liaoning, and Shandong) selected for this survey vary substantially in geography, economic development, public resources, and health indicators. A multistage random cluster process was used to draw the sample surveyed in each of the provinces. Counties in the nine provinces were stratified by income (low, middle, and high) and a weighted sampling scheme was used to randomly select four counties in each province. In addition, the provincial capital and a lower income city were selected when feasible. In two provinces, other large cities had to be selected. Villages and townships within the counties and urban and suburban neighborhoods within the cities were selected randomly. In 1989-1993 there were 190 primary sampling units, and a new province and its sampling units were added in 1997. Response rate are high (80.4% for 1997, 82.9% for 2000, 80.2% for 2004, and 81% for 2006), but families that migrate from one community 18 to a new one were not followed. Attempts were made to follow those respondents who moved within the primary sampling units and some larger urban communities. The final sample lost due to death is 124(23.5%) in 2000, 308 (31.8%) in 2004, and 408 (36.7%) in 2006. Figure 2.1 Map of Surveyed Provinces in CHNS Questionnaires The survey questionnaires are composed of three parts, household, individual, and community questionnaire. In household questionnaire, questions are asked on economic activities of each person in the households, household living conditions, household assets, and health services available for the household. For health services, detailed information including insurance coverage, medical providers, and health facilities that the household might use under selected circumstances, accessibility, time and travel costs, and perceived quality of care are collected. In individual questionnaire, individual activities, life style, health status, marriage and birth history, body shape and mass media exposure, etc, are collected. In this survey, activities of daily living and memory test were 19 conducted among adults aged 55 and older. For community questionnaire, one representative from each of the primary sampling units filled it out. Information on community infrastructure (water, transport, electricity, communications, and so on), services (family planning, health facilities, and retail outlets), population, prevailing wages, and related variables were collected. Samples There are 13,888 individuals interviewed in the 1997 survey. In 2000, all newly- formed households who resided in the sampling areas as well as additional households were added to replace those no longer participating in the survey. New communities were also added to replace communities no longer participating. Thus, a total of 14,911 individuals were interviewed in 2000. The same procedures were repeated in both 2004 and 2006, resulting in a total sample of 11,555 individuals in 2004 and 11,742 individuals in 2006. The working samples selected for the two studies in this dissertation are those who age 55 and above in 1997, which results in a total of 2,345 respondents. This dissertation took full advantage of all the available data each respondent had no matter if he/she had completed all the four waves of data collection. Those 2,345 respondents surveyed resided in 191 sampling units which included 36 urban neighborhoods, 36 suburban neighborhoods, 36 towns and 108 villages. Information on the 191 sampling units collected using the community questionnaires each wave was used in the second study to test how changes in social environment have affected health status of older adults. 20 Uniqueness of the Dataset The CHNS dataset provides advantages not found in most other studies of health in later life. The first advantage is that the survey is one of the few longitudinal studies in China covering a crucial period of time when economic reforms began producing large- scale structural changes (1989-2006). A second advantage is the great degree of heterogeneity between localities in terms of their SES and other related health and demographic features. A third advantage is that multiple waves of community data were collected as part of the longitudinal household surveys which include direct measurements of community features such as infrastructure, social resources and investment, healthcare infrastructure, and health services. This kind of practice is not the norm in data collection (Entwisle, 2007). The advantages of this survey data allow the exploration of dynamic trends in age-related changes among older adults; it also enables the investigation of dynamic changes in social environment in relation to individual health. Limitations of the Dataset However, it is important to acknowledge that this dataset is not without its limitations. First, the provinces sampled in the survey are mostly from eastern part of China. Many in-land and western provinces were not included. Therefore, the national representativeness of this sample is comprised. Second, the target population for this survey is all-age groups instead of older adults specifically. Hence, the sample size available for this dissertation is not as big as that of other surveys which focus mainly on 21 older population (China Research Center on Aging, 2003; Zeng et al., 2002). Third, the design of community-level questionnaire is focused on nutrition-related community characteristics, such as the distribution and distance of grocery or liquor stores. Other broader social environmental factors, including community built environment, quantity and quality of social services, are not measured. Finally, some measurements of community characteristics have changed over the years, which prevent us from using some variables when modeling changes in communities. In the following chapters, more detailed information on the advantages and disadvantages of the data are discussed. 22 Chapter 3: The Impact of Socioeconomic Status on Health Trajectories among Older Adults in China Chapter 3 Abstract Few studies have examined the over-time relationship between socioeconomic status and health among the older Chinese adults. This study examines whether the effects of education, income, and occupational class vary over time. Employing data from the four waves (1997, 2000, 2004, and 2006) of the China Health and Nutrition Survey (CHNS), this study estimates the latent growth curve models of functional health, self- rated health, and cognitive function. The results show that the health of older Chinese adults continued to be shaped by individual education, occupational class, and income. Evidence was found to support cumulative advantage hypotheses: Higher education is associated with better self-rated health and cognitive function at baseline as well as slower declines over time. Higher income is associated with better functional health at baseline as well as a more gradual decline over time. At the same time, it is found that lower occupational class predicts worse cognitive function at baseline; however, it is associated with a less steep decline later on. The finding that higher income is associated with a steeper decline in self-rated health is also unexpected. Taken together, this study demonstrates the effects on SES on health status at baseline, which are in consistent with findings from cross-sectional studies. In addition, the findings of this study reveal a more complicated relationship between SES and health development over time. 23 Introduction The Chinese society has experienced a dramatic social stratification over the past 30 years since it embarked on a journey of economic reform in the late 1970s. The growing economy has brought wealth, which has lifted the country from absolute poverty. At the same time, the growth has also widened the gaps between the rich and poor, the rural and urban areas, as well as coastal regions and in-land provinces. These gaps not only exist in income distribution, but in the whole spectrum of social resources and opportunities, including education resources and opportunities for work. One of the dire consequences resulting from this social stratification is manifest in health, particularly in old age when most health issues begin to emerge. Previous studies have demonstrated a strong relationship between SES and a series of health outcomes, including mortality (Liang et al., 2001), self-rated health (Liu & Zhang, 2004; Luo & Wen, 2002), chronic conditions (Zimmer & Kwong, 2004), and cognitive function (Zhang, Gu, and Hayward, 2008). However, less clear is whether the effects of SES persist into old age and whether these effects differ over time. The life course perspective provides a general theoretical framework to understand health trajectories across the life span. This perspective posits that various factors, both biological and social, influence health and disease independently, cumulatively, and interactively (Kuh et al., 2003). Studies conducted in western countries have generally adopted this perspective. Within this framework, the cumulative advantage hypothesis has been developed to explain the widening patterns of SES-based disparities across the life span. 24 The cumulative advantage hypothesis posits that SES-based health disparities diverge over time. It suggests that individuals with advantaged socioeconomic status are able to maintain and gain increasing health benefits over time as they age. More and more empirical studies using longitudinal designs have supported this hypothesis and demonstrated that the health advantages of the well-educated are larger in older age groups than in younger ones (Mirowsky & Ross, 2008; Ross & Wu, 1996; Willson et al., 2007) . On the other hand, the age-as-leveler argument has been proposed to explain the pattern of converging health disparities in later life after a divergence during adulthood (Beckett, 2000; House, Lantz, and Herd, 2005). The age-as-leveler hypothesis posits that the relationship between SES and health at younger ages is stronger compared to older ages. It is because health begins to be more closely related to age due to the universal biological frailty or the existence of social programs, which compensate the biological frailty (House et al., 2005). The life course perspective provides a good start to understand the varying relationship between SES and health outcome. However, the observed relationship may not be universal nor may it be able to explain health differentials among older Chinese adults. Most of these studies were conducted in Western developed nations where macro societal structures on education, income distribution, and occupation classification are vastly different from that of developing nations. It is possible that in a different cultural context, SES may be evaluated differently in terms of its importance, its marginal return to health, or its varied importance in people’s life. For example, in rural China where illiteracy is high among older adults, other measures of SES such as income, instead of 25 education, may have a better predictive power to health outcomes. It is also possible that the human capital of children, such as financial or instrumental support, would be the key later in the life-span instead of the solely measure of individual income. Relatively few studies conducted among Chinese population have used longitudinal design, which is particularly necessary in detecting the long term effects of SES on health. Results from cross-sectional studies could not disentangle the aging effects from the cohort effects, meaning the observed relationships between SES and health could be differences from generation to generation, rather than intrinsic differences resulting from aging development within individuals. Guided by the theory of life course perspective, this study tested the cumulative advantage hypothesis and the age-as-lever hypothesis using the longitudinal data (1997- 2006) from CHNS. This study uses latent growth curve modeling to examine whether SES-based health disparities increase or diminish over time. This study complements previous cross-sectional studies conducted among older Chinese adults and has theoretical implications regards the age variation of SES-based health disparities. Education and Health The relationship between education and health among older adults has been consistently confirmed in both cross-sectional and longitudinal studies conducted in the U.S. as well as other European countries (McMunn, 2006). An increasing amount of evidence has indicated that education-based health disparities increase over time. Ross and Wu (1996) found that the gap in self-reported health, physical functioning, and physical well-being among people with high and low educational attainment increases with age. Lynch (2003) found that the effect of education on self-rated health strengthens 26 across age and the effect becomes stronger across cohorts. He argued that it is probably because of the development in education and the changes of the distribution in disease and mortality. Studies conducted later used large-scale longitudinal data over a longer period of time, and these studies took selection bias into account. By doing so, these studies have offered more convincing evidence of educational effects over time. Both Willson et al. (2007) and Mirowsky and Ross’ (2008) studies found that education produces greater divergence in self-rated health with age, and this effect becomes stronger with each cohort. Haas (2008) found that both respondents’ own as well as his/her father’s education level are related to baseline functional limitation as well as the trajectory of functional health later. Findings from studies conducted among older Chinese adult revealed mixed results on the relationship between education and health. Using the Chinese Longitudinal Healthy Longevity Survey (CLHLS), Liu and Zhang (2004) found that older adults (80 and older) with some education tend to evaluate their health more positively compared to those without education. In another study, education lost its significance on both self- rated heath and functional health once other socioeconomic factors were controlled for (Zimmer & Kwong, 2004). In regard to cognitive function, Zhang et al. (2008) found that lower education level was associated with higher chance of cognitive impairment in old age among the older adults in China, which is similar to findings in the U.S (Turrell , Lynch , Kaplan , Everson , Helkala , Kauhanen et al., 2002). Based on the findings from literature, this study intended to test the following hypotheses: H1: education is related to increasing disparities on functional health over time H2: education is related to increasing disparities on self-rated health over time 27 H3: education is related to increasing disparities on cognitive function over time Occupational Class and Health Findings from previous studies on the relationship between occupation and health remain equivocal. One group of researchers found weakened or no relationship between occupation and health (Amaducci , Maggi , Langlois , Minicuci , Baldereschi , Di Carlo et al., 1998; Miech & Hauser, 2001). One possible explanation is that after retirement, occupation becomes less relevant to older adults because the meditational paths from occupation to health such as exposure to hazard working environment no longer exist. Another explanation is that social programs such as Social Security, Medicare and Medicaid available in those developed countries may, to certain extent, compensate those in lower occupational class after retirement, which leads to the convergence of health status. Another group of researchers, instead, found a smaller but lasting effect of occupation on health in old age. In a study conducted among Spanish elderly, Daminan et al. (1999) found that social class as measured by occupation is one of the main determinants of self-rated health among the young old (65-74). Breeze (2001) compared men in clerical or manual (low-grade) jobs in middle age with senior administrators and found that they had quadrupled the odds of poor physical performance in old age, tripled the odds of poor general health, and doubled the odds of poor mental health and disability. These cross-sectional studies supported the association between occupation class and health, but did not verify whether that effect becomes weakened or strengthened over time. Marmot and Shipley (1996) used 25 year follow-up survey data from the British 28 civil servants survey in Whitehall study. They showed that although SES- based health inequalities persist into old age, the power of occupational class to predict mortality declines quite dramatically after retirement. No studies have examined the long-term relationship between occupational class and health among older Chinese adults. A cross-sectional study conducted among the old adults aged 80 and above using CLHLS data showed that non-agricultural professionals and homemakers reported significantly better self-rated health compared to agricultural workers (Zeng et al., 2002). In another cross-sectional study, Zimmer and Kwong‘s study (2004) used pension as an indicator to differentiate those working in professional activities or for state-owned enterprises from others. They found that pension is a strong predictor of physical limitation among urban older adults. Based on these findings, following hypotheses were generated: H4: occupational class is related to increasing disparities on functional health over time H5: occupational class is related to increasing disparities on self-rated health over time H6: occupational class is related to increasing disparities on cognitive function over time Income and Health The long-term effects of income on health outcomes have been investigated using longitudinal survey data, which enables the testing of the cumulative advantage hypothesis. Maddox et al. (1994) examined the dynamics of functional status and of income among 11,000 adults aged initially 58-63 surveyed over a decade in the 29 Longitudinal Retirement History Study. Their findings demonstrated that income is an independent and relatively better predictor of functional impairment than chronological age. Willson et al. (2007) analyzed the longitudinal data from the Panel Study of Income Dynamics and showed that wealth, income, and their life course patterns have significant effects on self-rated health net of education effects. Income produces the most dramatic health trajectories, compared to wealth’s more gradual relationship with health over time. Findings from studies conducted among Chinese population generally confirmed a similar pattern of association between income and health. A study conducted by Luo and Wen (2002) using CHNS data revealed income affects self-rated health through the meditational paths of living conditions, access to health care services, and community development. Ho et al. (2001) found income dependency increased the risk of cognitive impairment by four folds among older Chinese adults. Based on these findings, following hypotheses were tested: H7: income is related to increasing disparities on functional health over time H8: income is related to increasing disparities on self-rated health over time H9: income is related to increasing disparities on cognitive function over time Method Sample Data for this study came from the China Health and Nutrition Survey (CHNS), a longitudinal survey of nine provinces between 1989 and 2006. The Population Center at the University of North Carolina conducted the whole process of sampling, interviewing, and data managing for the surveys in collaboration with the Center for Disease Control in China. Face-to-face interviews at individual, household, and community level were 30 conducted to obtain information. This study analyzed four waves (1997, 2000, 2004, and 2006) of individual data. The samples consisted of 2,345 adults aged 55 and above who were interviewed in 1997. Of that total, 1,092 (46.6%) older adults were observed for four times. The others were interviewed one to three times, resulting in 1,817 (77.5%), 1,376 (58.7%), and 1,233 (52.6%) older adults interviewed in the following survey years (2000, 2004, and 2006) respectively. Measures Health outcomes The dependent variables chosen for this study were functional health, self-rated health, and cognitive function. Starting in 1997, the CHNS research team has started collecting functional health data among those aged 55 and above. Functional health was measured by the degree of difficulty in performing the instrumental activities of daily living. Following items: shopping, cooking, using public transportation, managing money, and using the telephone were measured. Respondents indicated the level of difficulty performing each task on a 4 point scale: 1 (no difficulty), 2 (with some difficulty), 3 (need help), 4 (unable to do it). These items had a high reliability score (alpha=.89), therefore the sum of the five items was used. The score ranged from 0 (no difficulties) to 20 (unable to do any tasks). Good reliability and validity of the scale were also reported in a study by Beydoun and Popkin (2005). Self-rated health was measured by a single question, “Right now, how would you describe your health compared to that of other people of your age?” The respondent answered on a 4-point scale ranging from 1(excellent) to 4(poor). This item has been 31 used in previous studies (Chen & Meltzer, 2008; Pei & Rodriguez, 2006) and has demonstrated its close association with income and income inequality. Cognitive function was measured by items selected from Mini-Mental Status Exam (MMSE) (Folstein, Folstein, and Mchugh, 1975). Five domains of cognitive function including orientation, registration, attention, calculation, and recall were measured in this study. Cronbach’s alpha of reliability test for these five domains is .73. Scores range from 0 to 32 and the higher the score, the better cognitive function. SES measures Education attainment was measured by number of years of formal education received. In the following analyses, education was divided by 10 to improve the scaling of coefficients. Education was centered at grand mean to facilitate easy interpretation. Individual annual income was calculated by the CHNS research team (detailed information on calculation can be found at http://www.cpc.unc.edu/projects/china/longitudinal/datasets/master_income.html). Individual annual income for each wave is the sum of seven sources of income including business, farming, fishing, gardening, livestock, non-retirement wages, and retirement income for each wave. The value for each wave was inflated to 2006 currency values. In order to avoid extreme values, the log function was applied to transform individual annual incomes. Missing value on income was replaced using the mean individual annual income of the community where the respondents lived. Individual annual income was treated as time-varying covariate in the analysis. This study used Lu’s (2002) definition of five major occupational classes in contemporary China based on social, cultural and economic resources that workers in 32 each occupation could command. In this study, one question was asked in the survey: “What is your primary occupation?” Based on the answers to that question, each respondent was assigned to either of the following classes. Upper class (class 1), referring to government officials, administrators, executives, managers, and senior professionals; Upper middle class (class 2), including junior professionals and technical workers, skilled workers, athletes, actors, musicians, army officers, and policy officers; Middle class (class 3), referring to office staff, ordinary soldiers, and policemen; Lower middle class (class 4), including non-skilled workers and service workers such as chefs, janitors, and laundry workers; Low class (class 5), referring to farmers, fishermen, and hunters. For the question about primary occupation, there are 211 (9%) respondents reported as being retired. Their occupation was replaced by the primary occupation they had reported in the previous surveys. In addition, 131 (5.6%) respondents reported being homemakers. Their occupation was replaced using their husband’s primary occupation if the respondents were female. Controlled variables include: age (in years), gender (1=male), marital status (1=married), health insurance (1=yes), obsess (1=yes, BMI>30), overweight (1=yes, 25<BMI<=30), underweight (1=yes, BMI<=18.5), smoke (1=yes), heavy alcohol use (1=yes, drink more than 3 or 4 times per week), and chronic condition (1=yes). Chronic condition was assessed by asking respondents if they had any hypertension, heart disease, stroke, diabetes, or bone fracture. Analysis This study used the latent growth curve modeling method to examine the effects of SES on IADL, self-rated health, and cognitive function after controlling for 33 demographic characteristics, health behaviors, and other health conditions. The latent growth curve models not only describe changes in health status over time while taking account of the correlations between successive measurements, they also capture variations in baseline measurements and in the rate of change over time. In the level-one model of this study, time-varying health scores were regressed on linear and quadratic age terms (age was centered at 55, the youngest age among the respondents at the first wave of survey) for each respondent across as many as four measurements. These regressions generated random intercept and slope estimates that described person-specific growth curves. In the level-two model, the effects of fixed explanatory variables, including SES and control variables, on the three random effects (intercept, initial rate of change, and accelerated speed of change) were examined. At this level, individuals were considered to be random factors. SES and control variables were introduced to explain the observed variability in intercepts and slopes between individuals. Coefficients are interpreted as deviation from the average underlying trajectory. Positive coefficients on the intercept increase the baseline level of limitation while negative ones reflect lower initial levels. Covariates with positive effects on the slope are those that are associated with steeper increases in limitations over time while those covariates with negative effects on the slopes are associated with more gradual increases. In this study, the focus is on the rate of change (age) and the accelerated speed of change (age-square term) in health outcomes. If the covariates on the slope have a positive value, it suggests health diverges over time. If the value is negative, then health convergence is predicted. 34 For other variables, grand mean centering was used in the model to facilitate the interpretation of the results. Analyses were conducted using Mplus, which allows a large number of covariance structures, the inclusion of cases with incomplete data as well as non-constant times at which data values were obtained. Loss of participants in longitudinal studies is inevitable, which is especially true in studies among older adults. Attrition has induced a mortality selection bias, which means ignoring those who drop from the study may lead to underestimating the effects of SES on health because those from lower SES drop from the study because they have already worse health status in the first place. In order to assess how attrition affects conclusions about over-time effects of SES on health, logistic regressions were conducted to examine predictors of whether older adults participated, died or were missing at the final wave (2006). Results Sample characteristics Frequencies, means and standard deviations of variables used in the study are presented in Table 3.1. Sample characteristics are based on the total sample of 2,345 surveyed at baseline (1997). Of the three indicators of SES, it was assumed that education and occupation-based class did not change over time, whereas individual income changed across waves. Income was converted into U.S. dollars based on the exchange rate of 1:7. Also found in Table 1 are the means and standard deviations of the three outcome variables for each wave. 35 Table 3.1: Sample characteristics M/% SD Age 65.9 7.9 Gender (male) 47.0 Marital Status (married) 75.8 Education (high school) 70.2 Class1 5.7 Class2 5.5 Class3 10.2 Class4 20.0 Class5 58.7 Income_97 (in $) 494.0 516.3 Income_00 (in $) 746.7 896.9 Income_04 (in $) 1060.0 1059.0 Income_06 (in $) 1266.0 1311.1 Health Insurance (yes) 23.6 Under-weight 11.2 Over-weight 17.1 Obese 3.4 Smoking 35.0 Drinking 31.3 Chronic Illness (yes) 41.3 Self-rated health_97 2.50 0.77 Self-rated health_00 2.72 0.72 Self-rated health_04 2.79 0.77 36 Table 3.1, Continued Self-rated health_06 2.81 0.75 IADL_97 7.02 3.51 IADL_00 7.01 3.32 IADL_04 7.24 3.76 IADL_06 7.43 3.85 Cognition_97 14.18 5.66 Cognition_00 14.13 5.48 Cognition_04 10.86 6.20 Cognition_06 9.71 6.03 Table3.2 presents the estimates from the latent growth model. The model is shown in four progressive steps. First is the unconditional model (Model 1), a model with only random age effects at Level 1. Second is the conditional model (Model 2) with SES (education, occupational class, and income) added to predict variation in the random age effects (only initial level and the rate of change). Third is the conditional model (Model 3) with the control variables added, testing whether the effects of SES on intercept and the rate of change remain. Finally, the conditional model (Model 4) with explanatory variables expanded to the acceleration (age-square term) is shown. Model 1 in Table 3.2 shows the overall mean of IADL of 5.519 for older adults at age 55. The negative coefficient for age (-.023) and the positive coefficient for age squared (.015) indicate that IADL first improves and then declines at an accelerated rate over time. 37 Model 2 tests whether education, occupational class, and income affect both initial level of functional health and its following rate of change. Education and occupational class are associated with initial level of functional health but do not affect its following development. Income is associated with worse initial level of functional health (.266), but with a less steep decline (-.004). Table 3.2 Two-Level Models of Functional Health Status Change Model 1 Model 2 Model 3 Model 4 Initial status Intercept 5.519 *** 7.097 *** 7.366 *** 7.28 *** Education/10 -1.029 *** -0.850 *** -0.852 *** Class 5 0.306 -0.070 -0.078 Class 4 -0.344 * -0.478 *** -0.494 *** Class 3 0.337 * 0.160 0.210 Class 2 0.220 0.099 0.088 Income 0.266 *** 0.226 *** 0.252 *** Male 0.267 *** 0.257 ** Married -1.396 *** -1.455 *** Insurance -0.079 -0.092 Chronic illness 0.576 *** 0.566 *** Underweight 0.637 *** 0.643 *** Overweight 0.064 0.073 Obesity 0.007 0.062 Smoke -0.109 -0.092 Drink -0.312 *** -0.299 *** Poor health 1.162 *** 1.247 *** 38 Table 3.2, Continued Rate of change Intercept -0.023 *** 0.161 * 0.188 ** 0.353 Education 0.003 -0.012 -0.006 Class 5 -0.017 0.007 0.037 Class 4 -0.034 -0.029 0.015 Class 3 0.038 0.069 -0.05 Class 2 0.027 0.037 0.073 Income -0.004 ** -0.034 ** -0.098 * Male -0.023 -0.003 Married 0.003 0.163 * Insurance -0.012 0.020 Chronic illness 0.069 *** 0.089 Underweight -0.066 ** -0.079 Overweight -0.010 -0.031 Obesity 0.064 -0.056 Smoke 0.054 * 0.014 Drink -0.003 -0.027 Poor health -0.109 *** -0.315 *** Accelerated speed of change (age-square term) Intercept 0.015*** 0.012 0.009 -0.032 Education -0.002 Class 5 -0.009 Class 4 -0.013 Class 3 0.034 Class 2 -0.011 39 Table 3.2, Continued Income 0.017 Male -0.006 Married -0.045 * Insurance -0.009 Chronic illness -0.005 Underweight 0.003 Overweight 0.006 Obesity 0.033 Smoke 0.011 Drink 0.006 Poor health 0.057 *** Variance 2.8675 3.2082 3.0963 3.0952 Number of observations (n) 10174 18335 18335 18335 AIC 83636.3 188588.6 175517.1 175266.9 BIC 83657.9 188705.9 175790.7 175665.6 * p<.05; ** p<.01; ***p<.001 (2-tailed tests). Note: Functional health was measured by IADL, including five items on a 4-point scale, from 1 (no difficulty) to 4 (unable to do it), the higher the score, the worse functional health is. Model 3 adds control variables and shows that adding these demographic and health controls does not significantly change the association between SES and both initial status of functional health and its rate of change. The coefficient of education on initial status reduces from -1.029 to -.85 after introducing these control variables. 40 Model 4 expands the explanatory variables to the age-square term. Education, class, and income remain to be significant in their effects on initial status and the rate of change; however, none of the three factors affect the accelerated speed of change. An increase of one year education is associated with .0852 (-.852/10) score lower on initial level of functional health, but education is not related to either the rate of change or the accelerated speed of change in functional health. Therefore, hypothesis 1 (H1) is not supported. Being in class 4 predicts -.494 score lower on initial level of functional health compared to those in class 1; however, occupational class does not affect the following rate of change. Hypothesis 4 is not supported either. Those with higher income had worse initial functional health, but they experienced a less steep decline compared to those with lower income. While older adults experienced an accelerated decline after certain age, income is no longer a significant factor in predicting their functional health development. Hypothesis 7, income is related to increasing disparities in functional health over time, is partially therefore supported. In addition to the effects of SES on functional health status and change, it is interesting to note that those being married had better initial level of functional health (-1.455), then they experienced a short-term decline (.163), but over time, their decline (-.045) was less steep compared to those who were not married. Model 4 also shows that being male, having chronic illness, or being underweight is associated with worse functional health. Drinking alcohol 3-4 times a week is related to better initial level of functional health. The most potent predictor of functional health is self-reported health. Those who reported poor health had a significantly higher level of functional disability, though they 41 first experienced a less steep decline compared to those who reported good health, their decline accelerated faster over time. Self-rated Health Same model building procedures were applied in modeling development in self- rated health. The overall mean of self-rated health is 2.44 at age 55, the rate of change is .027, and the accelerated rate of change is -.001, without controlling for any covariates. It suggests that as respondents got older, they rated their health more negatively, then to a certain point (85 years old, calculated using the coefficient), they began to evaluate their health less negatively. Model 2 shows that those with higher education or those in higher occupational class had better self-rated health, but they did not differ from others in their rate of change. Income is not associated with either the initial status or the following rate of change. Control variables were added into Model 3. Education lost its significant in predicting the initial level of self-rated health, but it is related to the rate of change over time. One year’s increase in education is related to .001(.01/10) score lower in the rate of change. The effects of explanatory variables on acceleration were taken into account in Model 4. Education became no longer significant in predicting either initial level or the following change. Hypothesis 2, testing whether education is related to increasing disparities in self-rated health, is only partially supported. Occupation remains significant in predicting the initial level of self-rated health, but it is not associated with the following rate of change. Hypothesis 5 is not supported. Model 4 also shows that higher income is predictive of better initial self-rated health, however, unexpectedly; it is related 42 to a greater accelerated rate of decline over time. Hypothesis 8, income is related to increasing disparities on self-rated health over time, is not supported by the data. Model 4 also reveals that those being married or having chronic illness had worse self-rated health at the beginning. Those who had health insurance or those who drank alcohol 3-4 times a week had better initial self-rated health. The most potent predictor of self-rated health is functional health. Those who reported higher level of functional disability had significantly worse self-rated health, witnessed a less steep decline for a short time, and then experienced the decline at a greater accelerated rate over time. Table 3.3 Two-Level Models of Self-Rated Health Status Change Model 1 Model 2 Model 3 Model 4 Initial status Intercept 2.440 *** 2.667 *** 2.089 *** 2.035 *** Education/10 -0.087 *** -0.003 -0.002 Class 5 0.166 *** 0.139 *** 0.141 *** Class 4 0.147 *** 0.136 *** 0.138 *** Class 3 0.146 *** 0.092 ** 0.091 * Class 2 0.202 *** 0.151 *** 0.144 *** Income -0.017 -0.042 *** -0.033 ** Male -0.015 -0.014 Married 0.083 *** 0.090 *** Insurance -0.057 *** -0.059 *** Chronic illness 0.205 *** 0.210 *** Underweight 0.034 0.029 Overweight -0.035 -0.047 Obesity -0.072 -0.093 43 Table 3.3, Continued Smoke -0.008 -0.014 Drink -0.134 *** -0.130 *** IADL 0.074 *** 0.076 *** Rate of change Intercept 0.027 *** 0.036 * 0.075 *** 0.203 *** Education/10 -0.002 -0.010 *** -0.014 Class 5 0.000 -0.001 -0.003 Class 4 0.001 -0.004 -0.007 Class 3 0.002 0.005 0.009 Class 2 0.002 0.004 0.023 Income -0.001 0.004 -0.016 Male 0.004 0.001 Married -0.001 -0.018 Insurance -0.008 -0.002 Chronic illness 0.004 -0.009 Underweight 0.005 0.015 Overweight 0.003 0.027 Obesity 0.012 0.053 Smoke -0.001 0.013 Drink 0.003 -0.008 IADL -0.006 *** -0.012 *** 44 Table 3.3, Continued Accelerated speed of change (age-square term) Intercept -0.001 *** -0.004 * -0.005 ** -0.041 ** Education/10 0.001 Class 5 0.001 Class 4 0.001 Class 3 -0.001 Class 2 -0.005 Income 0.005 * Male 0.001 Married 0.005 Insurance -0.002 Chronic illness 0.004 Underweight -0.003 Overweight -0.006 Obesity -0.011 Smoke -0.004 Drink 0.003 IADL 0.002 ** Variance 0.7599 0.7575 0.7196 0.7195 Number of observations (n) 11590 19203 19203 19203 AIC 6696.8 11039.0 9996.4 10017.0 BIC 6718.4 11156.0 10270.0 10416.0 * p<.05; ** p<.01; ***p<.001 (2-tailed tests). Note: Self-rated health was measured on a 4-point scale, 1 (excellent) to 4 (worse), the higher the score, the worse self-rated health is. 45 Cognitive Function Table 3.4 shows the estimates of changes in cognitive function from four models using the same model building strategy as described above. Model 1 in Table 3.4 shows that the overall mean of cognitive function is 15.797 at age 55, the average rate of change is -.227, and the accelerated rate of change is -.004. It suggests an accelerating downward trend with increasing age. Model 2 adds individual SES. Education and occupational class are associated with initial level of cognitive function, but are not related to any growth parameters. Income does not predict either the initial status or the rate of change. After control variables were introduced to Model 3. The effects of education (1.587/10) and occupational class (-.0875 for class5, -.636 for class4) on the initial status cognitive function remain but with reduced strength. Model 3 also shows that education and class, which are not significant in predicting the rate of change in Model 2, are associated with a less steep decline in cognitive function after control variables were added. Those with higher education had a less steep decline in cognitive function over time, supporting hypothesis 3. Compared to those in class 1 (higher class), older adults who were in class 4 (lower class) had a lower rate of decline over time. Therefore, hypothesis 7 is not supported. 46 Table 3.4 Two-Level Models of Cognitive Function Change Model 1 Model 2 Model 3 Model 4 Initial status Intercept 15.797 *** 11.078 *** 15.616 *** 15.996 *** Education/10 2.414 *** 1.587 *** 1.579 *** Class 5 -1.353 *** -0.875 *** -0.933 *** Class 4 -0.769 ** -0.636 ** -0.731 ** Class 3 -0.930 ** -0.324 -0.439 Class 2 -0.361 -0.122 -0.201 Income -0.120 -0.006 -0.024 Male 0.520 *** 0.542 *** Married 0.553 *** 0.472 ** Insurance 0.713 *** 0.686 *** Chronic illness -0.062 -0.067 Underweight -0.355 -0.459 Overweight 0.043 0.081 Obesity 1.330 *** 1.244 ** Smoke -0.059 -0.068 Drink -0.125 -0.123 Poor health -1.874 *** -2.030 *** IADL -0.612 *** -0.624 *** Initial rate of change Intercept -0.227 *** -0.443 *** -0.805 *** -1.908 *** Education/10 0.008 0.059 ** 0.088 Class 5 0.104 0.095 0.243 Class 4 0.135 0.146 * 0.390 47 Table 3.4, Continued Class 3 -0.011 -0.038 0.252 Class 2 0.073 0.068 0.261 Income 0.043 0.004 0.050 Male -0.047 -0.103 Married 0.008 0.226 Insurance 0.099 ** 0.160 Chronic illness -0.087 ** -0.063 Underweight 0.035 0.274 Overweight -0.014 -0.089 Obesity -0.058 0.121 Smoke -0.005 0.024 Drink 0.045 0.047 Poor health 0.271 *** 0.650 *** IADL 0.026 *** 0.079 *** Accelerated speed of change (age-square term) Intercept -0.004 * -0.007 0.014 0.326 ** Education/10 -0.009 Class 5 -0.043 Class 4 -0.070 Class 3 -0.081 Class 2 -0.055 Income -0.013 Male 0.016 Married -0.060 48 Table 3.4, Continued Insurance -0.016 Chronic illness -0.007 Underweight -0.060 Overweight 0.019 Obesity -0.046 Smoke -0.008 Drink -0.001 Poor health -0.105 *** IADL -0.016 * Variance 5.678 5.4579 5.1045 5.1016 Number of observations (n) 9420 15479 15479 15479 AIC 303604.7 460680.0 402435.5 401558.6 BIC 303626.4 460799.3 402724.8 401980.7 * p<.05; ** p<.01; ***p<.001 (2-tailed tests). Note: Cognitive function was measured by items taken from MMSE, the score ranges from 0 to 32, the higher the score, the better cognitive function is. Model 4 tested the effects of explanatory variables on the acceleration. The significant effects of education and occupational class on growth parameters from Model 3 disappeared. Only self-reported poor health (-.105) and IADL (-.016) remain significant in predicting the accelerated declines over time. 49 Table 3.5 Logistic Regression of Status in 2006: Participant versus Dead Or Lost to Attrition in 2006 (n=2,345) Attrition Dead Age 1.03 *** 1.09 *** Male 0.91 2.07 *** Married 0.73 ** 0.92 Education 1.00 0.97 *** Class 2 0.95 0.70 Class 3 0.89 0.85 Class 4 0.87 0.68 Class 5 0.48 *** 0.92 Income 1.17 * 0.93 -2log likelihood 2770.67 1925.302 Pseudo R 2 0.06 0.16 In order to assess whether attrition either due to death or loss of participants affects the conclusions about over-time effects of SES on health outcome, logistic regressions were conducted to identify the predictors of death and attrition. Results are presented in Table 3.5. The odds ratios show that being in the lowest occupational class (class 5) is associated with a less likelihood of attrition whereas higher income predicts a higher probability of attrition. The contributions of class and income to attrition suggest that the underestimation of the effects of class and income on health change over time among participating survivors is small. Higher education predicts a less likelihood of 50 being dead, suggesting there is a possible underestimation of the effects of education on health over time. Discussion Previous cross-sectional studies have established a strong relationship between SES and health among older Chinese adults, but few studies have examined whether SES affects health trajectories at old ages using longitudinal designs. This study took advantage of the longitudinal nature of the CHNS by modeling the trajectories of health over multiple time points and testing the effects of SES on the rate of change in health outcomes. It complemented cross-sectional studies by showing that how the various components of SES affect initial levels as well as the following development of health status among older Chinese adults. In this study, evidence was found to support the cumulative effects of education on both self-rated health and cognitive function. This study echoes the findings from previous cross-sectional studies (Zeng et al., 2002; Zimmer & Kwong, 2004) showing that education is correlated with self-rated health status. In addition, it also shows that education exaggerates the health disparities in self-rated health as older adults grow older. Similar to a previous study (Zhang et al., 2008), this study also confirmed the protective effect of education on cognitive function. Higher educational level was associated with better cognitive function as well as a slower rate of decline over time. This finding is also in consistent with western literature (Turrell et al., 2002). Despite the fact that it is out of the scope of this study to examine the specific mechanisms through which education affects cognitive function, it is reasonable to suspect that they may be related to healthy behaviors and less hazardous exposure within working environments. In addition, 51 education also represents intellectual stimulation, and it creates a cognitive reserve to better protect against age-associated losses (Satariano, 2006). Similar to a study conducted among American elderly (Scherr , Albert , Funkenstein , Cook , Hennekens , Branch et al., 1988), this study also shows that lower occupational class is related to worse cognitive function at age 55. However, the finding in this study also shows an unexpected twist in the over-time relationship between occupation and cognitive function. It is found that before cognitive function picks up the accelerated deterioration in the later stage, those in the lower class (class 4, comprised mostly of non-skilled workers and service workers such as chefs, janitors, and laundry workers) actually experienced a slower decline compared to those in higher class (class 1, compose mostly of government officials, administrators, executives, managers, and senior professionals), despite the fact that they had worse cognitive function at 55. It is suspected that the majority of older Chinese adults, especially those in lower occupational classes tend to depend on their land or children for old age support due to the lack of social programs such as pensions or social security. Older Chinese adults usually have to work till they are no longer able to. Occupation continues to be part of their life rather than stop at age 65. On the contrary, for those who are in the higher class, they are more likely to retire as age 55 (for women) or 60 (for men) as stipulated by the state retirement policy. It is possible that continuing engagement through work might be a stimulus to keep the cognition functioning, at least temporarily before it declines in an accelerated speed. Though the effects of occupation on the accelerated rate of change are not significant, the direction of relationships shows that those in the lower classes had a decline in the accelerated stage. A possible explanation is that older adults from lower 52 occupational classes usually work in more strenuous and dangerous work environments. Such environments produce a high probability of disabilities, chronic conditions, and injuries, which in the end put a toll on their cognitive function at a later stage. The hypothesis that income-based health disparities in physical function increase over time has been supported by previous studies (Kim & Durden, 2007; Willson et al., 2007). Recent evidence has suggested that, compared to education, which was more related to the onset of health problems, income had better predictive power on the course or progression of health problems once they existed (House et al., 2005). This study confirmed this finding in that even though those with higher incomes had worse functional health at the beginning, their functional decline later was more gradual compared to those with lower income. This is not particularly surprising considering that the fact that health insurance coverage is low, particularly among older adults, and medical cost has been increasing rapidly over the years, it is likely that older adults with lower income may delay or forgo their healthcare or treatment, resulting in the development of more serious health conditions or disabilities at later time. However, an unexpected finding in this study is that those with higher income had significantly better initial self-rated health, but they experienced a faster acceleration of decline in self-rated health later. It is not clear whether high income group also has a higher expectation toward their health and then rated their heath more negatively when they face more physical challenges in the later stage of life. Future research looking into the psychological comparison with past health may help answer this question. It should be noted that the majority of older adults, especially older adults in rural China rely on their children for financial support (Zeng et al., 2002). Financial 53 support from children, instead of income may be a more relevant factor for older adults’ health. A study conducted by Lou and Wen (2002) indicated that the mediational paths from income to health were living condition and access to health care. Older adults who had no medical insurance coverage may have to rely on children for financial support to access healthcare. This speculation is partially supported by the robust effects found in the control variable, health insurance in the model. This study is not without limitations. One limitation is the possibility of mortality selection bias. Logistic regression was conducted to identify the predictors of attrition, which informed us the possibilities of underestimation or overestimation of the effects of SES on health development in the results. Despite the fact that the results showed a low probability of underestimating the effect of occupational class and income on health and possible underestimation of education effect, further steps should be taken to address the possible biased estimation in the model. A few solutions such as the use of propensity scores should be explored in the future studies. Another limitation is that the measure of cognitive function was based on a few questions selected from a Mini-Mental State Exam. It is effective as a screening tool to differentiate respondents in this study. A clinical evaluation, however, would provide more accurate information and opportunities for cross-comparisons with other groups. An additional issue of concern is that the income measures, which included seven potential sources of income including business, farming, fishing, gardening, livestock, non-retirement wages, and retirement, income may still miss some other possible incomes of older adults. One major income source for older Chinese adults could be money from children, including remittance from children who migrated to other cities for work. Another possibility is that income does not capture all 54 the means by which older adults have their needs meet. For example, older adults may receive healthcare services without paying the bill out of their pockets because their children pay for them, or they receive goods and gifts instead of money from children and friends. Future study should consider social support in relation to health change and delve into more detailed income components for older Chinese adults. Despite these limitations, this study contributes to the on-going discussion on over-time effects of SES on health. This study does not support the cumulative advantage hypothesis in general. From a policy perspective, this study suggests that interventions should focus on early education, which can provide life-long benefits on health. 55 Chapter 4: The Changing Social Environment and its Influence on Health among Older Adults in China Chapter 4 Abstract Macro social environment in China has dramatically changed and affected many aspects of people’s life. However, little empirical evidence is available to demonstrate the existence of those effects. This study investigates how community economic development and social development over time affect individual health trajectories. Employing data from the four waves (1997, 2000, 2004, and 2006) of the China Health and Nutrition Survey (CHNS), this study estimates multi-level growth curve models of functional health, self-rated health, and cognitive function. The results show that: 1) the aspects of community social environment, particularly community household median income, communication development, and time cost for healthcare access exert independent effects on health outcomes over and beyond the individual-level factors; 2) these social environment factors affect the initial health status, but not the consequent development of health; 3) the effects of social environment on health outcomes differ by individual education and income. Taken together, this study demonstrates that social environment affects health among older Chinese adults independently. Continual progress in economic and social environment could potentially improve health. 56 Introduction China has been undergoing dramatic changes since the economic reform in the 1970s. However, little empirical research has been conducted to examine the effects of the changes in socioeconomic environment on individual health. Previous studies have focused more on individual-level factors and much less are known about the effects of social environment on individual health outcomes, leaving an incomplete understanding of health disparities among older Chinese populations as a whole (Liang et al., 2000; Liu & Zhang, 2004; Liu et al., 1999). The accumulating evidence documented in the western literature has informed us how communities affect individual health in general terms. It is, however, difficult to apply these findings to the Chinese population. One of the reasons is that few researchers have considered the changes in social environment in relation to health. Instead, communities are assumed to be static, or if they do change, the magnitude of change is negligible. Such assumptions do not apply well to the situation in contemporary China. Since the economic reforms in the late 1970s, China has witnessed rapid economic growth over the past 30 years. The continuing expansion in its economy has been accompanied by growing disparities across populations and areas in wealth and resources distribution. One of the detrimental disparities is shown in health outcomes. There is an almost a two-decade gap in life expectancies between those who lived in urban developed areas and those who lived in rural deprived areas (Zhao, 2006). In addition to the gap in life expectancies, the gaps in infant mortality and morbidity have also increased dramatically between urban and rural areas (Liu et al., 1999; Zhao, 2006). 57 Rising health disparities among older populations have also been documented in terms of high mortality rates (Liang et al., 2000), incidence of functional disabilities (Beydoun & Popkin, 2005; Yi & Vaupel, 2002), incidence of chronic illness (Zimmer & Kwong, 2004), and poor self-rated health (Liu & Zhang, 2004; Yip et al., 2007; Zeng et al., 2002). Understanding macro-level factors related to the causes to health disparities are crucial because social determinants at individual level such as education, occupation, and income are largely shaped by the social environment in which an individual is situated. Therefore, interventions targeted at social environment could lead to effective changes at individual level. In this study, one research question was addressed: How have social environment changes in China between 1997 and 2006 affected health among adults aged 55 and above? Community data from four waves (1997, 2000, 2004, and 2006) of the China Health and Nutrition Survey (CHNS) were matched to the corresponding individual data. Then the relationships between changes in social environment and health outcomes among older adults in China were tested with the complied data. The conceptualization of ecological effects on health has gone through several stages of development: from the categorization of the features of environment based on basic human needs (Hillemeier, Lynch, Harper, and Casper, 2003; Macintyre et al., 2002), to the conceptualization of relationships between individual, community, and policy-level factors and health (Starfield, 2004; Starfield & Shi, 1999). Following that, the understanding of ecological effects has been further extended by integrating it with the life course perspective (Wheaton & Clarke, 2003). This study builds on these previous thoughts on ecological effects and adopted the integrated life course perspective and 58 ecological approach as its guiding framework. As shown in Figure 4.1, the time dimension, as emphasized in the life course perspective, is represented by the vertical line, referring to the changes in health status as well as the changes in social environment over time. The contextual dimension, as stressed in the ecological perspective, is represented by the horizontal line on the right side. It implies that individuals are nested within the social environment, and are influenced by their surrounding social environment. It is assumed that the social environment evolves as individuals experience personal changes. In addition to testing the direct effects of social environment on health, possible cross- level effects were explored, testing whether the effects of social environment on health differ by individual SES. Figure 4.1 The Conceptual Framework of Ecological Effects on Health Development 59 Social Environment and Health There is no agreed-upon definition of social environment despite its pervasive use in ecological studies. The aspects of social environment which are often investigated in relation to health are community socioeconomic status, social structure, and quality of the environment (Yen & Syme, 1999). Others have conceived social environment differently. For example, McNeill et al. (2006) conceived it as a five- dimension construct including social support and social networks; socioeconomic position and income inequality; racial discrimination; social cohesion and social capital; and neighborhood factors. Considering its relevance and importance in the Chinese context, social environment is regarded as being composed of two aspects: economic development and social development. One of the notable changes in China has been its economic development since the 1970s. The most obvious impact accompanied by economic development is the reduced poverty level. Reduced poverty has consistently shown to be related to positive health outcomes at both societal and individual levels (World Health Organ, 1999). Zimmer (2007) used the average wage within a community as the community-level SES measure and found it important in predicting mortality. In addition, Pei and Rodriguez (2006) and Li and Zhu (2006) investigated income inequality in relation to self-rated health. They found that different measures of income inequalities were all related to the increased risk of reporting poor health for people living in provinces with greater income inequalities. In this study, community median household income was used as the indicator of economic development, and it is hypothesized that community median household income is positively associated with health outcomes. 60 Compared to the comparatively straight-forward relationships between economic development and health, the relationships between social development and health have remained equivocal, largely because social development has been conceived differently and measured differently across various studies. Considering their relevance to the Chinese social context, three essential aspects of social environment are considered in this study. The first aspect is communication which refers to facilities and methods which facilitate the transfer and exchange of information. Zimmer et al. (2007) used number of amenities in a community, including telegraphs, telephones, post offices, newspapers, movie theaters, paved roads, and 24-hour electricity, and found that older Chinese adults living in communities with a large number of amenities have a lower risk of mortality. The second aspect of the social environment is community education level. Education is usually measured by the percentage of adults without high school completion or median education level. Most often, community education level is measured together with other general community SES indicators instead of being regarded as an independent measure as its own. For example, education, as measured by percentage of adults aged 25 who had completed high school or college, was used as one indicator to construct a neighborhood summary score, which was found to be related to incidence of coronary heart disease among Americans (Diez-Roux , Kiefe , Jacobs , Haan , Jackson , Nieto et al., 2001). It has been suggested that other aspects of education deserve a focused assessment, such as the levels of funding, characteristics of school systems, curricula, and learning-related aspects of community life. These aspects indicate the emphasis given to education and the investment committed by a community, which itself may be related to health outcomes (Hillemeier et al., 2003). It is hypothesized in 61 this study that higher levels of community education investment are associated with better health outcomes. The final aspect of social environment considered in this study is healthcare. Relatively little attention has been paid to healthcare system despite the fact that evidence has suggested the important role of health and social service availability in determining health outcomes (Macinko, Shi, Starfield, and Wulu, 2003). Shi and Starfield (2000) found that individuals living in states with a higher ratio of primary care physicians relative to the population are more likely to report good health than those living in states with a lower ratio. The current study examines the macro-level determinants for health disparities among the Chinese population by using data from a representative sample of older adults in China, by considering changes at the community level, and by testing the community effects on several measures of health. The primary hypothesis of this study is that community-level factors are independently associated with health for older adults in China. Four measures of community-level factors, including community median household income, communication, education, and healthcare accessibility are included and hypothesized to be independently associated with health outcomes. Three measures of health, including functional health, self-rated health, and cognitive function, were investigated. 62 Method Sample Data came from the CHNS, which covered nine provinces with the various levels of development in economy and social services. The Population Center at the University of North Carolina, in collaboration with the Center for Disease Control in China conducted sampling, interviewing, and data managing for the surveys. Face-to- face interviews at the individual, household, and community level were conducted to obtain information. A multistage random cluster process was used to draw the sample surveyed in each of the provinces. Counties in the nine provinces were stratified by income (low, middle, and high), and a weighted sampling scheme was used to randomly select four counties in each province. In addition, the provincial capital and a lower income city were selected when feasible, except that other large cities rather than provincial capitals had to be selected in two provinces. Villages and townships within the counties and urban and suburban neighborhoods within the cities were selected randomly. The working sample for this study consisted of 2,345 adults aged 55 and above which were interviewed in 1997. Of that total, 1,092 (46.6%) older adults were observed four times. The others were interviewed one to three times, thus resulting in 1,817 (77.5%), 1,376 (58.7%), 1,233 (52.6%) older adults followed in the following survey years (2000, 2004, and 2006) respectively. These respondents resided in 191 primary sampling units, including 32 urban neighborhoods, 30 suburban neighborhoods, 32 towns (county capital city), and 96 rural villages. The community questionnaires for each of the primary sampling units were completed by one respondent who was considered 63 knowledgeable about community infrastructure and services such as water, transport, electricity, communications, family planning, health services and facilities, retail stores, population characteristics, and prevailing wages. Measures Health outcomes The dependent variables for this study were functional health, self-rated health, and cognitive function. Starting in 1997, the CHNS research team started collecting functional health data among those aged 55 and above. Functional health was measured by the degree of difficulty in performing the instrumental activities of daily living (IADL). The following items: shopping, cooking, using public transportation, managing money, and using the telephone, were measured. Respondents indicated the level of difficulty performing each task on a 4 point scale: 1 (no difficulty), 2 (with some difficulty), 3 (need help), 4 (unable to do it). These items had a high reliability score (alpha=.89). Beydoun and Popkin (2005) reported good reliability and validity of the scale using the same dataset. The sum of the five items was used in this study. The scores ranged from 0 (no difficulty) to 20 (unable to do any tasks). Self-rated health was measured by a single question, “Right now, how would you describe your health compared to that of other people of your age?” The respondent answered on a 4-point scale ranging from 1 (excellent) to 4 (worst). Similar measure has been used in previous studies (Chen & Meltzer, 2008; Pei & Rodriguez, 2006; Poortinga, 64 Dunstan, and Fone, 2007) and reported as being closely associated with income and income inequality. Cognitive function was measured by items selected from the Mini-Mental Status Exam (MMSE) (Folstein et al., 1975). Five domains of cognitive function including orientation, registration, attention, calculation, and recall were measured in this study. Cronbach’s alpha of reliability test for these five domains is .729. Scores range from 0 to 32. The higher the score, the better cognitive function is. Community-level Variables Community socioeconomic status was represented by household annual median income. Household income was calculated by the CHNS research team, and detailed information on the calculation can be found at http://www.cpc.unc.edu/projects/china/longitudinal/datasets/master_income.html). Household income was the sum of all nine potential sources of income and revenue minus expenditures. These sources of income included business, farming, fishing, gardening, livestock, non-retirement wages, retirement income, subsidies, and other income. The annual household income was aggregated to each community by median. All the individuals in a community shared the same annual household income. The value at each wave was then inflated to 2006 currency values to facilitate comparability across time. In order to avoid extreme values, the log function was applied to transform annual median household income (+1). 65 Community social development was represented by community communication, education, and healthcare. Communication was assessed by four questions regarding the availability of the following services in local communities, including telegraph (1=yes), telephone (1=yes), postal service (1=yes), and fax (1=yes). The sum of the four items was used to reflect the development of communicative facilities in a community. Education was measured by the existence of educational institutions including primary schools (1=yes), middle schools (1=yes), high schools (1=yes). The sum of the three items was used to reflect the educational developmental level of a local community. Two aspects of healthcare were examined: physical accessibility and financial accessibility of healthcare. Physical accessibility was assessed by the time needed to travel to the nearest health clinic in the community (in minutes) and the time spent on waiting to be seen by a doctor in a clinic (in minutes). The sum of the two items was calculated and used in the analysis. Financial accessibility was assessed by the money spent on traveling to the nearest health clinic (in RMB) and the average cost of treating a cold/flu (in RMB). The total amount of money spent on both transportation and treatment of a cold/flu was calculated and inflated to 2006 currency values. These items were assessed in the household survey and aggregated to the community level using the mean. Respondents in each community shared the same scores for the two items. Individual-level variables Individual-level variables included age (in years), gender (1=male), marital status (1=married), education (years), health insurance (1=yes), and chronic condition (1=yes). 66 Chronic condition was assessed by asking respondents if they had any hypertensions, heart diseases, strokes, diabetes, or bone fractures. Individual annual income was calculated by the CHNS research team (detailed information on the calculations can be found at http://www.cpc.unc.edu/projects/china/longitudinal/datasets/master_income.html). Individual annual income for each wave consisted of the sum of seven sources of income, including business, farming, fishing, gardening, livestock, non-retirement wages, and retirement income for each wave. The value for each wave was inflated to 2006 currency values. In order to avoid extreme values, the log function was applied to transform individual annual incomes. Missing value on income was replaced using the mean individual annual income of the community from where respondents lived. Analysis Three-level growth-curve models were used to describe the changes and patterns of individual health outcomes and their variations across individuals and communities. The modeling method was a combination of approaches for modeling the development of individual health outcomes over time and individuals nested within communities. Specifically, individual growth trajectories of health with time-invariant covariates comprised the level-1 model; the variation in growth parameters among the older adults within a community was captured in the level-2 model; and the variation among communities was represented in the level-3 model. This approach not only enabled us to determine the random variability of the growth parameters of health outcomes among older adults and average individual effects among communities by estimated variance 67 and covariance, it also allowed us to examine the separated effects of the contextual and individual indictors. Additionally, cross-level interactions between individual SES and community-level factors (the effect of community-level predictors on the coefficients of individual-level SES on health outcome) were tested. Analyses were conducted using Mplus, which allows a large number of covariance structures, the inclusion of cases with incomplete data as well as non-constant times at which data values were obtained. Results Table 4.1 presents frequencies, means and standard deviations of variables used in the study. All the descriptive statistics presented in the table were based on the total sample of 2,345 individuals surveyed at baseline (1997) and 191 communities surveyed for four times. Income was converted into U.S. dollars based on the 2006 exchange rate of 1:7. In Table 4.2, means and standard deviations of the three outcome variables for each wave are presented. 68 Table 4.1: Sample characteristics (n=2,345) M/% SD Individual-level factors Age 65.9 7.9 Gender (male) 47.0 Marital Status (married) 75.8 Education (college) 19.9 Class1 5.7 Class2 5.5 Class3 10.2 Class4 20.0 Class5 58.7 Income_1997 (in $) 494.0 516.3 Health Insurance (yes) 23.6 Chronic Illness (yes) 41.3 Community-level factors Median household income_1997 (in $) 525.33 229.34 Median household income_2000 (in $) 679.22 340.35 Median household income_2004 (in $) 909.59 552.43 Median household income_2006 (in $) 1056.07 692.16 Communication_1997 2.71 1.2 Communication_2000 2.72 1.13 Communication_2004 2.53 1.25 Communication_2006 2.57 1.13 Education_1997 1.25 0.91 69 Table 4.1, Continued Education_2000 1.24 0.89 Education_2004 1.19 0.89 Education_2006 1.19 0.97 Time spent for care_1997 33.48 15.22 Time spent for care_2000 28.27 13.62 Time spent for care_2004 24.63 11.80 Time spent for care_2006 24.19 12.61 Cost for care_1997 (in $) 3.96 3.28 Cost for care_2000 (in $) 5.50 4.19 Cost for care_2004 (in $) 6.48 4.97 Cost for care_2006 (in $) 7.63 6.15 70 Table 4.2: Outcome distribution M/% SD Outcome variables IADL_97 7.02 3.51 IADL_00 7.01 3.32 IADL_04 7.24 3.76 IADL_06 7.43 3.85 Cognition_97 14.18 5.66 Cognition_00 14.13 5.48 Cognition_04 10.86 6.20 Cognition_06 9.71 6.03 Self-rated health_97 2.50 0.77 Self-rated health_00 2.72 0.72 Self-rated health_04 2.79 0.77 Self-rated health_06 2.81 0.75 For community-level factors, it is noted that median household income almost doubled from 1997 to 2000; at the same time, the cost of healthcare increased steadily over the years. A steady decrease in time spent on accessing healthcare was also observed. Less obvious patterns, which fluctuated over time, were found in community communication development and education. Bivariate correlation between community- level factors showed that the five factors are significantly correlated (Table 4.3). In the analysis, averaged contextual values over the four waves were used to represent overall 71 changes of the areas over time. For health outcomes, there is a consistent decline in functional health, cognitive function, and self-rated health from 1997 to 2006. Table 4.3 Correlation between community-level factors Variable (1) (2) (3) (4) (5) Median household income (1) 1 Communication (2) .356*** 1 Education investment (3) -.113*** .335*** 1 Time spent for care (4) .494*** .042* -.169*** 1 Cost for care (5) .701*** .098*** -.187*** .665*** 1 Tables 4.4-4.6 show the results from the three models for each dependent variable. Model 1 is the unconditional growth model. The parameter estimates indicate the initial status at mid-point of the 9 year interval and average rate of changes in outcome. Model 2 includes individual- level factors, testing whether the changes in outcomes differ between individuals on both the initial status and their rates of change. Model 3 examines the effects of community-level factors on outcome variables after controlling for individual-level factors. Model 4 further tests the cross-level interaction effects between community-level factors and individual SES. The information on model fit is in the footnotes to the tables. The criteria for acceptable model fit used in the study are three important indices, the comparative fit index (CFI), root mean-square error of approximation (RMSEA), and standardized root mean-square residual (SRMR) (Bentler, 2003). In Model 4, random effects on slopes were tested. There is no longer a single 72 covariance matrix structure, but the covariance matrix varies as a function of the covariates that have random slopes. Therefore, abovementioned fit indexes are not available. Instead, AIC and BIC were reported. Functional Health Table 4.4 presents model fit indices and parameter estimates for functional health. Model 1 shows that the overall mean of IADL is 7.561, and the average rate of change is .112. The significant variance components suggest that individuals and communities differ significantly in their initial status of IADL and subsequent development in therein. The next step is to look at individual and community-level predictors in explaining the variances. Individual-level factors were added to Model 2 to test how individuals differed in initial status and the rate of change according to their individual characteristics. The significant age and age square term indicate that the model assumes a linear and quadratic functional form, suggesting that functional health declines as individual ages and the decline accelerates as individual reaches certain age. Being male, married, having a higher education, being in lower class, or having health insurance was associated with better initial functional health. Those who were older, in lower occupational class, or who had chronic illness had a steeper decline in functional health, whereas those who had a higher educational level at baseline had a less steep decline. The relationship between reporting poor health and IADL is less straightforward: reporting poor health at baseline predicts worse functional health at baseline but a less steep decline over time. One possible explanation is that those who reported poor health had already reached the 73 plateau of functional health decline, therefore, the functional health began to flatten afterwards instead of continue decline in a linear form. In Model 2, variance components remained significant, but the value decreased significantly. This suggests that adding individual-level factors into the model has helped explain the 71% ((8.115-2.326)/8.115) of variance observed between individuals and 52% ((.824-.553)/.824) of variance between communities in the initial status of IADL. Individual-level factors also explained the 63% of variance in rate of change between individuals and the 10.5% of variance between communities. The significant variance components between communities suggest the need to test whether the differences in functional outcomes between communities could be explained by the community-level factors specified in this study. Model 3 shows the results of IADL regressed on community-level factors after controlling for individual- level factors. Among the four community-level factors entered in the model, only communication is associated with the initial status of IADL. Specifically, better communication development over the years on average is associated with better functional health. Adding community-level factors into the model helped to explain 42% of variance in the initial status of IADL and 13% of variance in the rate of change between communities. In Model 4, cross-level effects were explored to test whether community resources modify the relationship between individual SES and functional health. Significant individual SES factors found in Model 3 were selected to interact with community communication, the only significant community-level factor found in Model 74 3. The results show that living in a community with better communication development is associated with worse functional health among the low educated. On the other hand, those with high education level benefit more in less developed communities compared to those with low education. 75 Table 4.4: Multi-level growth curve model on IADL Variable Model 1 Model 2 Model 3 Model 4 Constant Change Constant Change Constant Change Constant Change Intercept 7.561 *** 0.112 *** 6.423 *** -0.038 6.549 *** -0.035 6.301 ** -0.064 Individual-level factors Age 0.077 0.155 ** 0.157 0.163 ** 0.127 0.162 ** Age 2 0.802 *** -0.02 0.792 *** -0.022 0.809 *** -0.02 Male -0.297 * 0.041 -0.366 ** 0.028 -0.342 ** 0.027 Married -0.571 ** 0.001 -0.582 ** -0.003 -0.562 ** 0.001 Education -0.050 *** -0.003 * -0.044 *** -0.002 Class5 0.005 -0.019 -0.231 -0.038 -0.034 -0.018 Class4 -0.621 ** -0.039 -0.719 ** -0.032 Class3 0.263 0.108 * 0.12 0.078 0.31 0.102 * Class2 0.113 0.077 0.028 0.06 0.198 0.079 Income 1 (lower 20%) -0.040 0.065 -0.189 0.053 -0.179 0.055 Income 2 (top 20%) -0.080 0.019 -0.135 0.01 -0.127 0.013 Health insurance -0.521 *** 0.002 -0.301 * 0.043 -0.305 * 0.039 Reporting poor health 0.800 *** -0.065 ** 0.828 *** -0.064 * 0.805 *** -0.064 * Chronic illness 0.556 *** 0.098 *** 0.584 *** 0.099 *** 0.597 *** 0.098 *** 76 Table 4.4, Continued Community-level factors Median household income -0.668 -0.141 -0.638 -0.115 Communication -0.298 ** -0.009 -0.281 * -0.024 Education investment -0.03 -0.011 -0.038 -0.008 Time spent for care -0.006 0.001 -0.006 0.001 Cost for care -0.002 -0.001 -0.002 -0.001 Cross-level interaction Education X Communication 0.022 ** 0.00 Class4 X Communication -0.174 0.04 Variance (within) 8.115 *** 0.052 *** 2.326 *** 0.045 ** 3.392 *** 0.038 ** 3.346 *** 0.039 ** Variance (between) 0.824 *** 0.019 *** 0.553 *** 0.017 ** 0.321 ** 0.015 *** 0.296 ** 0.014 *** Model X 2 (df) 29.241 (15) 89.11 2(43) 97.885 (53) CFI 0.979 0.978 0.978 26541.352 (AIC) TLI 0.984 0.965 0.964 26850.599 (BIC) RMSEA 0.021 0.020 0.020 SRMR .022/.083 .014/.029 .011/.032 ***P<.001; **P<.01; *P<.05. 77 Self-rated Health The same model building strategy was used to model the effects of individual- and community-level factors on self-rated health. In Table 4.5, Model 1 shows that the overall mean of self-rated health is 2.661, and by each wave, self-rated health increases by .036. Though the variance components are significant, the variance on the rate of change is quite small (.001), suggesting a similar pattern of trajectory on self-rated health between individuals and communities. The distribution of self-rated health assumes a linear and quadratic form as indicated by the significant age square term. Having better education and health insurance are associated with better initial self-rated health. On the other hand, chronic illnesses predict worse self-rated health. Having health insurance is associated with a gradual decline in self-rated health, whereas having chronic illness predicts a steeper decline in self-rated health. Community-level factors were added into Model 3. The results show that after controlling for individual-level factors, median household income is associated better self-rated health at baseline, whereas spending longer time in obtaining care is associated with worse self-rated health. Adding community-level variables into the equation helps explain 50% of variance between individuals and 30% of variance between communities. Cross-level interaction effects were tested by adding the interaction terms between individual SES (education and top 20% income group) with median household income as well as time spent for care. Results are presented in Model 4. Only one interaction turns out to be significant: living in communities where longer time is needed 78 for seeking healthcare is associated with worse self-rated health for those at the top of the income distribution. 79 Table 4.5: Multi-level growth curve model on self-rated health Variable Model 1 Model 2 Model 3 Model 4 Constant Change Constant Change Constant Change Constant Change Intercept 2.661 *** 0.036 *** 2.344 *** 0.032 *** 2.369 *** 0.037 * 2.37 *** 0.044 ** Individual-level factors Age 0.206 *** 0.01 0.184 ** 0.008 0.176 ** 0.006 Age 2 -0.051 * -0.005 -0.041 -0.005 -0.037 -0.004 Male -0.029 -0.002 -0.054 * -0.005 -0.053 * -0.007 Married 0.063 0.012 0.043 0.007 0.037 0.005 Education -0.006 *** 0.000 -0.004 * 0.000 Class5 0.099 -0.009 0.094 -0.016 0.106 -0.017 Class4 0.048 -0.002 0.070 -0.004 0.077 -0.005 Class3 0.049 -0.011 0.086 -0.001 0.094 0.002 Class2 0.077 -0.009 0.124 0.004 0.129 0.007 Income 1 (lower 20%) -0.005 -0.007 -0.011 -0.010 -0.012 -0.012 Income 2 (top 20%) 0.018 0.016 * -0.011 0.016 * Health insurance -0.109 ** -0.019 ** -0.040 -0.011 -0.024 -0.005 80 Table 4.5, Continued Chronic illness 0.232 *** 0.012 * 0.230 *** 0.015 ** 0.219 *** 0.014 * Community-level factors Median household income -0.482 *** 0.004 -0.456 ** -0.004 Communication 0.030 -0.003 0.032 -0.004 Education investment 0.023 -0.007 0.017 -0.008 Time spent for care 0.004 * 0.000 0.003 0.000 Cost for care -0.001 0.000 -0.001 0.000 Cross-level interaction Education X Median household income 0.012 Education X Time spent for care 0.000 Income2 X Median household income -0.121 Income2 X Time spent for care 0.006 * Variance (within) 0.091 *** 0.001 * 0.157 0.005 0.078 *** 0.000 0.075 *** 0.000 Variance (between) 0.071 *** 0.001 *** 0.034 *** 0.000 0.024 * 0.000 0.021 *** 0.000 Model X 2 (df) .493 (4) 28.243 (30) 49.701 (42) CFI 1.000 1.000 0.984 13685.53 (AIC) TLI 1.036 1.008 0.968 14013.82 (BIC) 81 Table 4.5, Continued RMSEA 0.000 0.000 0.009 SRMR .005/.014 .010/.005 .011/.016 ***P<.001; **P<.01; *P<.05. 82 Cognitive Function Table 4.6 presents model fit indices and parameter estimates for cognitive function. Model 1 shows that the overall mean of cognitive function is 12.083, and it declines .476 by each wave of the survey. The significant variance components on constant and non-significant variance components on change suggest that individuals and communities differed significantly in their initial cognitive function, but not in its subsequent development. Individual-level factors were added into Model 2 to test the effects of individual- level factors on cognitive function. In the initial testing of the model, the age square term was not significant; therefore, it was dropped out the model. The final model implies that the distribution of cognitive function assumes a linear form, suggesting that cognitive function declines as age increases. Being older or in lower occupational class are related to worse initial cognitive function, whereas being male, having higher education, or having health insurance are associated with better cognitive function. Surprisingly, male experienced a steeper decline in cognitive function compared to female. Those in lower occupational class (class 4) or those who had health insurance had a relatively gradual cognitive function decline over time. The effects of community-level factors on cognitive function are presented in Model 3. Annual median household income is significantly related to better initial cognitive function after controlling for individual-level factors. The cross-level Interaction term between education and community median household income was tested, 83 and the results show that the interaction term is not significant in affecting cognitive function of older adults. 84 Table 4.6: Multi-level growth curve model on cognitive function Variable Model 1 Model 2 Model 3 Model 4 Constant Change Constant Change Constant Change Constant Change Intercept 12.083 *** -0.476 *** 13.907 *** -0.609 *** 13.369 *** -0.476 *** 13.241 *** -0.476 *** Individual-level factors Age -2.226 *** -0.103 -2.253 *** -0.084 -2.227 *** -0.074 Age 2 0.129 0.036 0.09 0.015 0.075 0.011 Male 1.041 *** -0.083 * 1.094 *** -0.061 1.121 *** -0.057 Married 0.243 0.054 0.315 0.037 0.318 0.036 Education 0.139 *** 0.007 ** 0.140 *** 0.005 * Class5 -0.957 * 0.151 -0.640 0.068 -0.532 0.06 Class4 -0.356 0.252 ** -0.248 0.106 -0.147 0.091 Class3 -0.815 0.098 -0.527 0.046 -0.439 0.041 Class2 -0.373 0.120 -0.347 0.026 -0.301 0.014 Income 1 (lower 20%) -0.241 -0.017 -0.090 -0.020 -0.120 -0.016 Income 2 (top 20%) 0.033 -0.020 -0.049 -0.009 -0.036 -0.007 Health insurance 1.002 *** 0.121 * 0.835 ** 0.099 * 0.836 ** 0.097 * 85 Table 4.6, Continued Chronic illness -0.162 -0.075 * -0.239 -0.087 * -0.237 -0.081 Community-level factors Median household income 4.304 *** 0.069 4.181 *** 0.123 Communication 0.106 -0.045 0.115 -0.049 Education investment -0.303 0.011 -0.334 0.011 Time spent for care -0.008 -0.002 -0.01 -0.001 Cost for care -0.005 0.001 -0.05 0.001 Cross-level interaction Education X Median household income -0.009 0.000 Variance (within) 8.054 *** 0.040 2.991 *** 0.022 3.350 *** 0.009 Variance (between) 3.957 *** 0.000 1.059 ** 0.000 0.610 0.002 Model X 2 (df) 37.709 (10) 59.178 (36) 46.539 (40) 28616.81 (AIC) CFI 0.938 0.982 0.995 28926.82 (BIC) TLI 0.925 0.967 0.989 RMSEA 0.037 0.018 0.009 SRMR .031/.077 .015/.027 .010/.024 ***P<.001; **P<.01; *P<.05. 86 Discussion Despite the rapid economic and social development in China over the past three decades, there has been little empirical evidence to show how the development has affected individual health over time. This study addresses this gap by investigating the role of social environment in explaining health trajectories among older Chinese adults. The results of the study provide some support for the hypothesis that social environment has direct effects on individual health changes. Specifically, higher community median household income was associated with better self-rated health and cognitive function of older adults. Community communication development was positively associated with functional health, however, it benefited people living in less developed community but have higher education level. The longer time needed for obtaining healthcare predicted poorer self-rated health, particularly for those at top income category. Those results indicate that community economic and social development exerts independent effects on health over the effects of individual-level factors. In addition, the results also demonstrated that the effects of communication and differ by individual SES. Another gap in the literature which this study addressed is that social environment is often assumed to exert lag effects on individual health without taking into account the development and changes in social environment. In this study, the composite scores representing the five aspects of social environment were calculated based on the average of four waves of data, which underlines the overall development of each community over time. 87 The results of this study are generally consistent with previous research, which has found an independent but weak relationship between community-level factors and individual-level health (Pickett & Pearl, 2001; Riva, Gauvin, and Barnett, 2007; Robert, 1998). However, these results differ from previous research in a number of ways. First, previous work examining the ecological effects on health has rarely considered the effects on cognitive function. There is also very little evidence of the relationship between community-level factors and cognitive function in the literature. The finding in this study echo with a previous study which shows that urban residence in early life was associated with lower odds of cognitive impairment (Zhang et al., 2008). Although the data did not allow further examination of the mechanisms behind the protective effects of community economic development, it is reasonable to suspect that, in communities where median household income was high, residents are less likely to experience malnutrition, more likely to access healthcare, which in turn could affect cognitive function. This study also differs from most previous studies in examining the effects of healthcare accessibility and affordability on health outcomes. Previous studies on the relationships between healthcare services and health outcomes remain equivocal. Some have found significant effects of number of physician and nurses per capita on mortality and life expectancy (Flegg, 1982; Shi & Starfield, 2000; Shi, Starfield, Kennedy, and Kawachi, 1999). Others, however, did not detect any effects of healthcare services (Judge, Mulligan, and Benzeval, 1998; van Doorslaer , Wagstaff , Bleichrodt , Calonge , Gerdtham , Gerfin et al., 1997). One possible explanation given is that the global measures of healthcare may be stronger predictors of overall population health status in developing countries than in wealthier ones (Macinko et al., 2003). In this study, more 88 specific measures of healthcare were used. A significant relationship was found between time needed for obtaining healthcare and self-rated health. One clear limitation of this study is that the data provided less ideal information for community-level measures. The choice of some community-level factors under studied was a compromise between theoretically-driven approach and data limitations. Measures of community economic development could be those of direct measures such as per capita GDP, cost of living, or economic vitality instead of aggregated data from household-level data. The use of derived individual or household data has led to the question of whether health outcome differences between different communities are due to the composition of individuals residing in those communities or “true” community effects (Macintyre et al., 2002). Collecting theoretically-driven community data in future studies should address this limitation better. Another limitation of this study is sample attrition, which may lead to an underestimation of community-level effects on health. Attrition is a commonly-observed phenomenon in longitudinal studies, particularly in studies among older adults. Attrition induces a mortality selection bias, which means that ignoring those who drop from the study may lead to underestimating the effects of community-level factors on health. It is because respondents from communities with a lower developmental level tend to drop out as they may have worse health status in the first place. Future research should explore methods to address the issue of attrition and adjust for it in the regression model for more accurate estimation. 89 In addition, the effects of community-level factors on health may have been underestimated in this study because only the direct effects of community-level factors on health was considered. It is possible that communities affect health indirectly, through its mediation path of individual-level factors. As we know, an individual’s SES could be largely shaped by his/her social environment. For example, the education level of an individual depends on the availability and quality of educational resources in his community. In this study, controlling for individual education may suppress possible indirect effects of community-level education investment. That may explain the reason that no significant effects of education investment of a community on health outcomes have been found. However, to disentangle the relationships between individual education and community education investment and to test their effects on health outcomes can be statistically challenging due to the availability of data and the complexity of model building. In this study, the changes in social environment over time, referring to the evolution of the same communities over time has explored. However, this perspective is limited in that it studies only people who stay in the same community without considering those who move from one community to another. The reality is that people do live in multiple social contexts either sequentially or simultaneously. For example, one could spend his/her childhood in a rural area and then move to an urban area for employment as an adult (sequentially). It could also be that one lives in one city and drives to work in another city on the daily basis (simultaneously). Only by taking into consideration the influence of multiple social contexts on health can the holistic and true 90 contextual effects on health be found. Future research could move toward this direction of conceptualizing ecological effects on health in multiple contexts. Furthermore, this study has demonstrated that various community-level factors are associated with health outcomes; however, the mechanisms underlying those associations remain largely unexplained. Important factors which lie along the causation chain are not included in the analysis. For example, stress or anxiety induced by living in a community where the cost of living is rising fast while the income level remains low may help explain more fully the relationship between median household income level and cognitive function. Developing theoretically-grounded and culturally-appropriate approaches to measuring the true aspects of social environments is necessary and important for future research. In conclusion, this study has shed light on the necessity of studying the social environment and its effects on health. Only by adding macro-level health determinants into the picture can our thinking on how to address health disparities be extended to areas to where promising methods and approaches are pointed. 91 Chapter 5: Studying the Effects of Social Environment on Health among Older Adults in China Chapter 5 Abstract Since the economic reform in the late 1970s, there have been increasing health disparities across regions and populations in China. The increasing health disparities have generated a lot of debate and discussion. A series of societal-, environmental-, and policy-level factors have been identified as the predictors of the inequalities. However, health research has traditionally focused on individual- and family-level determinants; few studies have studied the effects of social environment on health and little empirical evidence is available to support the claimed effects. In this paper, the aspects of social environment to be studied in relation to health disparities in China are proposed with reference to the western literature. In addition, the conceptual and methodological challenges in the current literature are critiqued and an integrated life course and ecological perspective are suggested to address some of those challenges identified. 92 Introduction China has undergone tremendous socioeconomic changes since the economic reform in the late 1970s. One of the consequences of these socioeconomic changes is an increase in health disparities. A large gap of life expectancies, infant mortality, and morbidity are observed between urban and rural populations (Liu et al., 1999; Zhao, 2006). Rising health disparities among the older population have also been documented in terms of high mortality rates (Liang et al., 2000; Zimmer et al., 2007), incidence of functional disabilities (Beydoun & Popkin, 2005; Liang et al., 2001; Yi & Vaupel, 2002), incidence of chronic illness (Zimmer & Kwong, 2004), and poor self-rated health (Liu & Zhang, 2004; Yip et al., 2007; Zeng et al., 2002). The explanatory variables for the increasing health disparities are mainly focused on individual-level factors, including individual socioeconomic status (SES) (Anson & Sun, 2004; Beydoun & Popkin, 2005; Li, Aranda, and Chi, 2007; Liu & Zhang, 2004; Zeng et al., 2002; Zhang et al., 2008; Zimmer & Kwong, 2004); urban and rural residency (Lawson & Lin, 1994; Raudenbush, 2000); health insurance coverage (Liu, Zhao, Cai, Yamada, and Yamada, 2002; Yu, 2005); gender (Yi, Liu, and George, 2003; Yu & Sarri, 1997); social support (Cong & Silverstein, 2008; Liu, Liang, and Gu, 1995), and health knowledge and behaviors (Feng et al., 2005; Hesketh et al., 2003). Such an individually focused approach does not address the causes of health disparities satisfactorily by leaving out the macro-environment, which shapes the individuals’ SES, healthcare access, or exposure to environmental hazards. In response, increasing attention has been turned to examine macro-level factors in recent years. A 93 few studies have looked at urbanization (Liu, Wu, Peng, and Fu, 2003; Monda, Gordon- Larsen, Stevens, and Popkin, 2007), income inequality (Chen & Meltzer, 2008; Li & Zhu, 2006; Liu, Dow, Fu, Akin, and Lance, 2008; Pei & Rodriguez, 2006; Zhao, 2006), social capital (Yip et al., 2007), community environment and amenities (Takano , Fu , Nakamura , Uji , Fukuda , Watanabe et al., 2002; Zimmer et al., 2007), and air pollution (Sun & Gu, 2008). These studies focus on a wide range of macro-level factors, and use different measures to estimate the environmental effects on health. The existence of different measures is problematic because researchers cannot compare and confirm the effects of social environment on health easily. Moreover, in these studies, the choice of measurement is frequently chosen on the basis of data availability rather than theory- driven concepts. In other words, the association relationship in these studies cannot provide the explanatory power for how the social environmental factors affect health. In addition to the methodological differences among the studies, the number of studies, as it stands currently, is also too small to conduct meta-analyses. In particular, with regard to the social-environmental effects on health among the Chinese population, there is no consensus on how to conceive and measure social environments, how to conceptualize the complicated relationships between individual health outcomes and environmental factors, and how to study the relationship between changes in social environment and individual health. The purpose of this chapter is to start such a conversation on these basic but important questions. In this chapter, the importance of research on the effects of social environment on older population is illustrated. This is followed by a review of the 94 relevant literature on this topic. Then the theoretical framework for understanding the social environment in China is discussed, and the aspects of social environment relevant to the study of health among older adults are defined. Finally, the challenges in current research are illustrated, and an integrated life course and ecological perspectives is proposed to address some of those challenges. Social Environment and Older Population The social environment may be particularly pertinent to the lives of older adults because of their long exposure to the social environment and their vulnerability in making appropriate adjustment to environmental changes. First, older adults tend to be far less geographically mobile than younger adults, so they tend to have longer-term exposure to their communities and more frequent contact with community resources and services (Balfour & Kaplan, 2002). Therefore, their social environments are more likely to exert an influence on them (Roux, 2002). Second, older adults may be compromised in their abilities to make adaption to their social environments. Lawton (1974) suggests that the compromised cognitive, psychological, or physical competence among older adults make it more difficult for them to adjust to demanding social environments. The same level of demand or hardship in the social environments may affect the health outcomes of older adults more than that of younger adults. Theoretical Framework for Understanding Social Environment In the study of health, the political economy perspective addresses the economic and political determinants of health and the distribution of disease within and across 95 societies (Doyal & Pennell, 1979). Its core assumption is that the fundamental causes of social inequalities in health lie in the economic and political institutions that shape the decisions to create, enforce, and perpetuate economic and social privilege and economic and social inequalities (Krieger, 2001a). The political economy of health, pioneered by Estes (1979; 1984; 2001) has been extensively developed to analyze the broad implications of economics, politics, and socio-cultural factors on the health of the aged population. Central to this theoretical framework are questions related to the social determinants of health and illness; the social creation of dependency and the management of that dependency status through public policies and health services; medical care as an ideology and as an industry in the control and management of aging; the consequences of public policies for the elderly as a group and as individuals; the role and functions of the state in relation to aging and health; and the social construction of the realities of both old age and health (Estes, 1984). According to Estes (1984), a capitalist state plays a critical role in shaping public policies and resource allocations through its two primary functions. One is to assure the survival of economic systems. The other is to maintain the legitimacy and operation of the social order to avoid destructive social unrest. As a consequence of these two functions, there exists a tension between the two competing demands for social spending: favorable treatment for business on the one hand and the provision of social and welfare programs on the other. Specifically, social spending has to ensure smooth and continuous economic growth by providing a supply of productive labor and by creating effective demands for goods and services. Social spending is used to assist businesses and industries in their accumulation of wealth through favorable tax treatment and 96 government subsidies. At the same time, social spending needs to ameliorate conditions that might be politically threatening by promoting the legitimacy of the state and the economic system and by strengthening the control of social behaviors. Therefore, social spending is also used to create social and welfare programs of education, job training, nutrition, shelter, healthcare, and unemployment compensation to ameliorate the displacement costs resulting from the operation of the economic system. The political economy perspective highlights three important points. First, health inequalities are socially constructed. Second, public policies and resource allocations are major factors in creating and perpetuating health inequalities. Third, public policies and resource allocations are the results of conflicts and power struggles, which arise from the state’s two competing functions and its social spending. Borrowing from these ideas, social environments, in which health inequalities occur, are considered to be shaped by two driving forces, economic development and social development. Social spending allocated to serve those two purposes leads to the creation, enforcement, and perpetuation of health inequalities. Concept of Social Environment There lacks an agreed-upon definition of social environment despite its pervasive use in ecological studies. Social environments are often conceived differently by various researchers. Yen and Syme (1999) classified the aspects of social environments into three categories: community socioeconomic status, social structure, and the quality of the environment. Others have conceived social environments as socioeconomic position and income inequality, racial discrimination, social cohesion and social capital, neighborhood 97 factors, and a five- dimension construct inclusive of social support and social networks (McNeill et al., 2006). In this paper, the political perspective is adopted and social environments are considered from the economic development and social development aspects, which are relevant and important to the social context in China. Economic development is defined along two dimensions, the level of economic development and the distribution of economic well-being. Social development describes the improvement of human well-being by investing on human capital and developing individual and community assets. Figure 5.1 Conceptualization of Social Environment Economic development describes changes involving qualitative as well as quantitative improvements in a country's economy (Economic development, 2008). In a broader sense, economic development focuses not merely on the increasing production of goods and services but also on improving other aspects of social life such as employment opportunities, working conditions, equality in income distribution and social treatment as well as life satisfaction (United Nations Development Program, 2004). However, many 98 countries have adopted a narrow definition of economic development that focuses solely on growth, profits, and the accumulation of goods and services. Sometimes it is equated with economic growth. Such a narrow definition introduces tensions in policy-making, which tend to favor economic development over social development at the cost of social well-being. It may also lead to greater economic and social inequalities due to the lack of consideration of proper wealth redistribution mechanisms. Likewise, social development is not only about reducing poverty but empowering people by creating more inclusive, cohesive, and accountable societies (World Bank., 2008). Midgely (1999) offered an alternative perspective on social development which is productivity and investment-oriented. It is concerned with “increasing labor market participation; promoting the formation of human capital; accumulating assets; mobilizing social capital in poor communities; developing microenterprises; and enhancing the efficiency of social programs to minimize waste and assure cost effectiveness”. He further suggested that, at a broader level, social development seeks to remove impediments to economic participation, such as racial and gender discrimination, and to create a climate conducive to economic development. Midgely (1999) recognized the powerful dynamic of economic development for progress but argued that if left alone, economic growth could distort development by generating contrasts between wealth and poverty and excluding substantial numbers of people from participating in the productive economy. His ideas on the relationships between economic development and social development are different from another school of thoughts which described social development as a product of economic 99 development. Based on the Rostovian model (1960), this school argued that only by assuring sustained economic growth can social development be achieved. It treats spending on social programs as transferring resources from the productive economy into unproductive social expenditures that could impede economic development. In order to test the causal relation between social development and economic growth empirically, Mazumdar (1996) analyzed data collected from 93 countries and found that the causal relationship is not universal across all countries but differs in different contexts for different income groups. For example, for the high income countries, social development is independent of economic growth, but for the middle income and low income countries, social development actually propels economic growth. Social Environment and Health Outcomes Though the relationships between social environment and health have been observed and documented since the 20th century, a surge in enthusiastically identifying, quantifying, and testing the effects of social environment on health has not been seen until the early 1990s (Entwisle, 2007). Because of the limited number of studies focusing specifically on older adults, the following discussion on the relationship between economic and social development and health made reference to extant research studies conducted among adult population. Economic Development and Health One of the most obvious impacts brought by economic development is a reduced poverty level. Reduced poverty has consistently shown to be related to positive health outcomes at both societal and individual levels (World Health Organ, 1999). At the 100 individual level, income and wealth, as measured by indicators such as per capita income, durable assets, and savings have demonstrated as being related to better health outcomes (Adler & Ostrove, 1999; Zhao, 2006; Zimmer & Kwong, 2004). At the societal level, increasing empirical findings have pointed to the fact that besides community SES, equality in the distribution of resources, mostly measured by the Gini coefficient, also matters for improving average population health, especially for reducing inequalities in health (Babones, 2008; Diez Roux , Merkin , Arnett , Chambless , Massing , Nieto et al., 2001; Freedman, Grafova, Schoeni, and Rogowski, 2008; Haan, Kaplan, and Camacho, 1987; Kawachi & Kennedy, 1997; Subramanian, Belli, and Kawachi, 2002). The relationship between economic development and individual health among the Chinese population is less clear. Zimmer (2007) used the average wage within a community as the community-level SES measure and found it important in predicting mortality. Pei and Rodriguez (2006) found that one unit increase in the provincial-level Gini coefficient is related to an increased risk of about 10-15% on average for fair or poor health for people. Li and Zhu (2006) found that after controlling for per capita income, there is an inverted-U association between self-reported health status and income inequalities, suggesting that high inequality in a community poses a threat to health. Despite the fact that all of the above three studies analyzed the same China Health and Nutrition Survey data, it is still hard to conclude at which level income inequalities affect health. Moreover, findings have been inconsistent regarding the relationship between community SES and health outcomes. The inconsistencies may be attributed to variations in measures. Contextual studies often measure community characteristics using 101 derived data from aggregated individual-level variables, such as median household income; percentage of community members living below poverty level; or percentage of residences with more than one person per room. The use of derived individual-level data has led to the question of whether health outcome differences between different communities are due to the composition of the individuals residing in those communities or “true” community effects (Macintyre et al., 2002). Such a cross-level dependence on measures has resulted in analyses in which community-level SES variables lose significance in predicting health outcomes once individual-level SES is controlled for (Sloggett & Joshi, 1998; Waitzman & Smith, 1998). In other words, the contextual indicator of SES may simply be a manifestation of individual-level components. Social Development and Health Different from the comparatively straight-forward relationships between economic development and health, the relationships between social development and health have remained equivocal, largely because social development is conceived differently and encompasses different contents across various studies. Considering their relevance to the Chinese social context, the aspects of social environment which have been studied or mentioned are categorized into the following groups. Infrastructure Studies of the quality of the infrastructure have examined the physical features of the community such as ambient pollution, housing tenure, and car access (Caughy, O'Campo, and Brodsky, 1999; Macintyre, Ellaway, Der, Ford, and Hunt, 1998), residential noise levels and overcrowding (Krause , Lynch , Kaplan , Cohen , Goldberg & 102 Salonen, 1997) and the availability of services (Caughy et al., 1999; Takano et al., 2002). Takano (2002) found that the aspects of residential environments such as the floor space of dwelling per person, proportion of parks, gardens, and green areas to total land area, number of health professionals per person, and the number of retail employees per person are inversely related to age-adjusted mortality in Shanghai. Monda et al. (2007) used a detailed measure of urbanicity comprised of 10 dimensions of urban services and infrastructure to examine its effects on the patterns of occupational physical activity among Chinese adults. Their findings showed that the intensity of occupational activity reduces as the level of urbanicity increases. Education Investment Community education level, usually measured by the percentage of adults without high school completion or median education level, has been found to be related to the incidence of coronary heart disease (Diez-Roux et al., 2001). Most often, community education level is measured together with other general community SES indicators instead of being regarded as an independent measure on its own. Therefore, it has been suggested that other aspects of education deserve a focused assessment such as the levels of funding, the characteristics of school systems, curricula, and learning-related aspects of community life such as the prevalence of television viewing and the numbers of library books per capita. These aspects indicate the emphasis given to education and the investment committed by an area (Hillemeier et al., 2003). 103 Healthcare Services Relatively little attention has been paid to healthcare system characteristics, despite evidence suggesting the importance of the role of health and social service availability in determining health outcomes (Macinko et al., 2003). The distribution of health resources, measured by number of physicians and nurses per capita, has been found to be predictive of health outcomes after controlling for income inequalities (Flegg, 1982). Shi and Starfield (2000) found that individuals living in the U.S. with a higher ratio of primary care physicians relative to the population are more likely to report good health than those living in states with a lower ratio. They emphasized the role of primary care and explained that the observed relationship between primary care and health outcomes could be due to the following reasons: 1) having access to a regular source of primary care may facilitate health promotion activities; 2) detecting diseases at a earlier stage can lead to appropriate and more efficient secondary and tertiary care; 3) developing social ties with primary care providers may compensate for the negative effects of the relatively low income on health. Challenges in Current Research Despite the fact that many studies have investigated the effects of social environment on health, there are several conceptual and methodological challenges in current research. For conceptual challenges, there are at least three issues which have not been satisfactorily addressed. The first challenge is that changes in the social environment are not conceptually incorporated, and the impact of these changes on health outcome is not estimated. Current research often assumes the lag effects of social 104 environment on health, or regards the magnitude of changes in the social environment as negligible. However, it is possible that individuals living in a fast-changing social environment may experience life changes more dramatically than others who live in a comparatively stable environment. It is also possible that individuals living in areas where the economy experiences recession may have more stress financially than individuals from areas where economy grows fast. The changes in larger social environments, which may induce individual-level life changes or stress, may have direct or indirect effects (through individual-level changes) on individual health. The second conceptual challenge is that the changes in social contexts are rarely considered. Current research dominantly considers only one social context for each individual, usually using the home address to locate individual’s social context. However, the reality is that individuals live in multiple social contexts either sequentially or simultaneously. For example, one could spend his/her childhood in a rural area and then move to an urban area for employment as an adult (sequentially). It could also be that one lives in one city but works in another city on the daily basis (simultaneously). We need to take the multiple social contexts into consideration conceptually, then our estimation on the influences of social contexts on health can the holistic and complete. Maybe the “true” contextual effects on health can be found. The third conceptual challenge is that the changing relationship between social environment and health is not well addressed and rarely examined. One exception is a study by Robert and Li (2001) in which they found that the effects of community SES on health change over time: it is weak during younger adulthood, stronger through middle 105 ages, strongest at ages 60 to 69, and weak again at ages 70 and older. The observed pattern supports the argument that the impact of environmental risk on health may vary according to the individual changes in biological and potential psychological vulnerability (House et al., 2005). It is important to realize that the relationship between social environment and health are contingent on age, time, or other individual factors that change over time. In addition to above mentioned conceptual challenges, there are two methodological issues existing in current ecological research. The first challenge has to do with the measurement of social environment. The second challenge has to do with the measurement of health outcomes. With regard to social environment, there is no agreed- upon definition of social environment. There are large variations in the conceptualizations of social environment, ranging from as many as 12 overarching dimensions of contextual characteristics (Hillemeier et al., 2003) to five types of features or characteristics of local areas (Macintyre et al., 2002; Starfield, 2004). Moreover, authors of these studies rarely suggest how the various aspects or characteristics of social environments should be measured. And standardized measures for social environment are not yet available, making it difficult to compare and confirm the results across studies. In addition, quality control in collecting environmental-level data warrants particular attention. In large surveys in which community-level data are collected, protocols for data collection are not as well developed as those used at the individual or household level. The consequence is poor data quality and compromised study. 106 Though community-level data collected using census data or other administration data is less of problem in terms of reliability because of the higher credibility of census data, the census data has its own problem as well. Census data is basically aggregated individual-level data, such as median household income, percentage of community members living below poverty level, or percentage of residences with more than one person per room. It is questionable whether the characteristics of communities measured by aggregated data are the true representation of the communities, or instead, only the composition of the individuals residing in those communities. Another issue with the measurement of social environment has to do with the choice of methodology. Current studies have predominantly used quantitative methods to quantify social environments (Pickett & Pearl, 2001) . Rarely has qualitative methodology been adopted. For example, the quantity of doctor-patient ratio tells us the density of services in an area. It cannot, however, differentiate two areas with similar ratios but varied health outcomes. Acceptability, affordability, or accessibility of health service from the perspective of service users may explain the differences better than a crude measure of service density. Recent studies have paid attention to this issue and began to incorporate subjective measures into objective measures of communities (Weden, Carpiano, and Robert, 2008; Yip et al., 2007). The second methodological challenge has to do with the measurement of health outcomes. The measurement of individual health outcomes is usually based on a biomedical model rather than a holistic approach. Specific diseases or dysfunctions are often the health outcomes measured, instead, the measure of health should be broadened 107 to include the impact of co-morbidity and the phenomena of vulnerability and resilience (Starfield, 2004). The Integrated Life Course and Ecological Perspective In order to address some of the conceptual challenges facing the ecological studies, an integrated life course perspective and ecological approach is proposed as a guiding framework to conceptualize changes in social environments in relation to individual health. This model is built upon previous work (Macintyre et al., 2002; Starfield, 2004; Wheaton & Clarke, 2003) but is designed with an emphasis on the contemporary social environment of China in mind. As shown in Figure 5.2, the time dimension, as emphasized in the life course perspective is represented by the vertical line, referring to the changes in health status, as well as the changes in social context over time. The contextual dimension, as stressed in the ecological perspective is depicted as the horizontal line on the right side. It implies that individuals are nested within the social contexts and influenced by surrounding social environment. Social environment consists of two major aspects, economic development and social development, which are interrelated to each other. Social environment evolves and changes over time. There is a continuity in the changes, as indicated by the dotted line between earlier social environment and current social environment. Social environment has lag effects as well as concurrent effects on health outcomes. Such ecological effects are depicted using dotted lines in the figure. Earlier social environment influences earlier 108 health status, at the same time, its lag effects can also be observed on current health status. Concurrent effects of social environment are also seen in its influence on current health status. Health outcomes are not only affected by discrete social environments at specific time point, the changes in social environments, including the rate of change, the direction of change, and the magnitude of change, also affect health outcomes. Of the many determinants of individual health, only individual SES in the figure is used as an illustration of the relationships between individual-level health predictors, individual health outcomes, and social environment. Individual SES affects both earlier and current health status, at the same time, individual SES is affected by his social environment. Individual SES mediates the effects of social environment on health. In addition, there are also possible cross-level effects, referring to the moderation effects of individual SES on the relationship between social environment and health. The individual- level effects on health outcomes are depicted using solid lines in the figure. This integrated life course and ecological perspective incorporates changes in social environment into its understanding of the ecological effects on health. This perspective views individual and social environment as dynamic entities and models the dynamic relationship between the two. It also addresses the mediation and moderation relationship of individual-level factors on the relationship between social environment and individual health. 109 Figure 5.2 The Integrated Life Course and Ecological Perspective Conclusion In this conceptual paper, the political economy perspective is used to conceptualize the social environment in China. Making reference to the existing western literature, we conceive aspects of social environment to be studied in relation to health. Challenges in current ecological research are identified, and an integrated life course and ecological perspective is proposed to address some of those challenges. There are growing interest and efforts to understand the macro-level determinants of health disparities in China over the past years. Efforts to understand the ecological effects of social environment on health in China can be made more effective and efficient when informed by well-conceptualized and well-conducted investigation. It is hoped that this paper can provide an impetus to discuss the conceptual and methodological challenges facing the current research. 110 Chapter 6: Discussion and Conclusion Summary of Findings Each of the three papers in this dissertation has made a meaningful contribution to our knowledge about the social determinants of health among older Chinese adults from the dimension of time and space. The two independent studies each addressed one specific question under the overarching theme of how socioeconomic changes and social stratification affect the health outcomes of older adults in China. In the first paper, the results from latent growth curve models confirmed the hypothesis of cumulative advantage on the relationships between education and self-rated health, education and cognitive function, as well as income and functional health. It suggests that education has a lasting and beneficial effect on the development of cognitive function and self-rated health. The effects of income on health seem less clear cut: it benefits functional health over time, but not on self-rated health. Less clear are the effects of occupational class and cognitive function: higher occupational class is associated better cognitive function at baseline, but those in lower class (class 4) managed to experience a less steep decline later before cognitive function deteriorates fast in old age. This research suggests that the effects of SES on health continue over time, but the effects could not all be explained by the cumulative advantage hypotheses. In the second paper, analyses based on multi-level growth models show that the community economic development, community communication, and community healthcare access had significant effects on health status over and above individual factors. Particularly, living in communities where household median income is high is associated with better self-rated health as well as better cognitive function. Community 111 communication development benefits older adults on their functional health, however, it seems that those with higher education but living in a less developed community can gain more from the communication development. This study also found that high income group who lived in a community where a longer time is needed for accessing healthcare reported worse health. The research suggests that some aspects of social environment are associated with the health status of older Chinese adults but not the rate of change in health status. In the third paper, the conceptual and methodological challenges in studying the effects of social environment on health are acknowledged. Using the political economy perspective as the guiding framework, it is proposed that social environment should be conceptualized as consisting of economic and social development, which reflects the contemporary situation in China. An integrated life course and ecological perspective is suggested to integrate the changes in social environment into the modeling of ecological effects on health. In summary, this dissertation demonstrates that changes in individual socioeconomic status as well as the development in macro-socioeconomic environment have substantial influence on the health status and health development among older Chinese adults. Resounding Themes This dissertation emphasizes the importance of two perspectives. They are the life course perspective and the integrated life course perspective and ecological effects. Guided by those two perspectives, time-related and place-related effects on the health status of older adults were examined. 112 The life course perspective This dissertation demonstrates the cumulative advantage of education and income on health over time. It confirms the cumulative advantage hypothesis generated from the life course perspective, which suggests that various factors, both biological and social, influence health and disease independently, cumulatively, and interactively (Kuh et al., 2003). Although some components of SES did not lead to increased health disparities over a nine-year period, they were associated with health disparities observed at baseline (age 55), suggesting the continuing effects of SES on health status in old age. This study suggests that SES-based health disparities exist among older Chinese adults throughout the life course, but the effects may not all be cumulative. Integrated life course perspective and ecological approach The study of ecological effects should be integrated with the life course perspective in order to incorporate changes in social environment into the investigation of individual health trajectories. The second paper of this dissertation shows that individual health trajectories are affected by a combination of individual and contextual factors. Individual factors outweigh contextual factors for their effects on health development. Economic development, communication development, and time cost for accessing healthcare had significant effects on health status at baseline but not the health status changes afterwards. Limitations There are several limitations of this dissertation. First, this dissertation only examines health trajectories over a nine-year period, and the patterns of health convergence may not be observed within such a time frame. This is particularly true 113 because the respondents included in our study were younger old adults with a mean age of 65.9. Second, the measurement of SES, particularly the measure of income may not capture some important income sources for the older population in this survey. The effect of income on health may be strengthened with a more accurate measure of income. Third, the measure of social environment is less than ideal due to the inconsistency in community questionnaires from wave to wave, as well as the lack of more direct measures of social environment. Theoretically-driven measures of social environment should be developed in the future. Fourth, this dissertation did not consider the possibility of multi-contexts in relation to health. Although this is less a problem in this dissertation because respondents were older adults, and they tend to live in one community longer instead of moving around, it is interesting theoretically and realistically to take multiple social environments into consideration when conceptualizing the ecological effects on health among younger populations. Fifth, income inequality has been demonstrated as an important factor in predicting health disparities (Chen & Meltzer, 2008; Pei & Rodriguez, 2006). This study did not look into this factor due to a concern for availability and reliability of income inequality data at community level. Finally, this dissertation did not explore the possible mediational paths between social environment and individual health. Testing of these effects calls for a more complex model speciation and a larger ecological effect size, for which this dataset does not allow. Future Studies In the future, studies on the health determinants of older Chinese adults should take a life course perspective, and incorporate changes in communities into the examination of ecological effects on health. Currently, many ecological studies are 114 constrained by the availability of community-level data, especially data which contain information on community changes. Thus, the investigation of ecological effects assumes the unrealistic assumption that the community does not change much or the effects of community characteristics on health are either simultaneous or lagged. This greatly undermines a more accurate estimation of ecological effects on health, and impedes theoretical development. To better theorize the ecological effects on health, a better conceptualization and quantification of social environment is needed. Also needed is a clear map for understanding the mediation or moderation relationships between social environment and health. In addition, very few studies have examined the ecological effects on cognitive function. In this dissertation, a significant direct effect of social environment on cognitive function was found. It is interesting to explore possible mechanisms through which social environment influences health. A few potential aspects which future research should explore further are community-related stresses or environmental hazards. Conclusions This dissertation studies health determinants at the individual and community level, and examines their dynamic relationships to health development among older Chinese adults. By taking the life course perspective and the integrated life course and ecological perspective, this dissertation addresses several research questions that have not been addressed before in the literature. This dissertation studies a period of time when socioeconomic development in China underwent dramatic changes and individuals living in that social environment also experienced the effects of those changes on their health 115 status. The results of this dissertation will have theoretical as well as policy implications in a time when global aging has brought tremendous changes to healthcare. 116 References Adler, N. E., & Ostrove, J. M. (1999). Socioeconomic status and health: What we know and what we don't. New York Academy of Sciences, 896, 3-15. Amaducci, L., Maggi, S., Langlois, J., Minicuci, N., Baldereschi, M., Di Carlo, A., et al. (1998). Education and the risk of physical disability and mortality among men and women aged 65 to 84: The Italian longitudinal study on aging. Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 53(6), M484- M490. Anson, O., & Sun, S. F. (2004). Health inequalities in rural China: Evidence from HeBei Province. Health & Place, 10(1), 75-84. Babones, S. J. (2008). Income inequality and population health: Correlation and causality. Social Science & Medicine, 66(7), 1614-1626. Balfour, J. L., & Kaplan, G. A. (2002). Neighborhood environment and loss of physical function in older adults: Evidence from the Alameda County Study. American Journal of Epidemiology, 155(6), 507-515. Beckett, M. (2000). Converging health inequalities in later life: An artifact of mortality selection. Journal of Health and Social Behavior, 41(1), 106-119. Bengtson, V., Elder, G. J., & Putney, N. (2005). The lifecourse perspective on aging: Linked lives, timing, and history. In Johnson, M. L., V. L. Bengtson, P. G. Coleman & T. B. L. Kirkwood (Eds.), The Cambridge handbook of age and aging (pp. 493-501). Cambridge, UK ; New York: Cambridge University Press. Bentler, P. M. (2003). EQS 6 for Windows program manual. Los Angeles, CA: Multivariate Software. Beydoun, M. A., & Popkin, B. M. (2005). The impact of socio-economic factors on functional status decline among community-dwelling older adults in China. Social Science & Medicine, 60(9), 2045-2057. 117 Breeze, E., Fletcher, A. E., Leon, D. A., Marmot, M. G., Clarke, R. J., & Shipley, M. J. (2001). Do socioeconomic disadvantages persist into old age? Self-reported morbidity in a 29-year follow-up of the Whitehall Study. American Journal of Public Health, 91(2), 277-283. Caughy, M. O., O'Campo, P., & Brodsky, A. E. (1999). Neighborhoods, families, and children: Implications for policy and practice. Journal of Community Psychology, 27(5), 615-633. Chen, Z., & Meltzer, D. (2008). Beefing up with the Chans: Evidence for the effects of relative income and income inequality on health from the China Health and Nutrition Survey. Social Science & Medicine, 66(11), 2206-2217. China Research Center on Aging. (2003). The initial analysis of data from the national survey of the aged population in urban/rural China. Beijing: China Biao Zhun Press. Cong, Z., & Silverstein, M. (2008). Intergenerational support and depression among elders in rural China: Do daughters-in-law matter? Journal of Marriage and the Family, 70(3), 599-612. Damian, J., Ruigomez, A., Pastor, V., & Martin-Moreno, J. M. (1999). Determinants of self assessed health among Spanish older people living at home. Journal of Epidemiology and Community Health, 53(7), 412-416. Diez-Roux, A. V., Kiefe, C. I., Jacobs, D. R., Jr., Haan, M., Jackson, S. A., Nieto, F. J., et al. (2001). Area characteristics and individual-level socioeconomic position indicators in three population-based epidemiologic studies. Annals of Epidemiology, 11(6), 395-405. Diez Roux, A. V. (2002). Invited commentary: Places, people, and health. American Journal of Epidemiology, 155(6), 516-519. Diez Roux, A. V., Merkin, S. S., Arnett, D., Chambless, L., Massing, M., Nieto, F. J., et al. (2001). Neighborhood of residence and incidence of coronary heart disease. New England Journal of Medicine 345(2), 99-106. 118 Doyal, L., & Pennell, I. (1979). The political economy of health. London: Pluto Press. Economic development. (2008). In Encyclopædia Britannica. (Publication. Retrieved May 29, 2008, from Encyclopædia Britannica Online: http://www.britannica.com/eb/article-9106199 Entwisle, B. (2007). Putting people into place. Demography, 44(4), 687-703. Estes, C. L. (1979). The aging enterprise (1st ed.). San Francisco: Jossey-Bass Publishers. Estes, C. L. (1984). Political economy, health, and aging. Boston: Little, Brown. Estes, C. L., and Associates. (2001). Social policy & aging: A critical perspective. Thousand Oaks, California: Sage Publications. Feng, W., Ren, P., Shaokang, Z., & Anan, S. (2005). Reproductive health status, knowledge, and access to health care among female migrants in Shanghai, China. Journal of Biosocial Science, 37(5), 603-622. Flegg, A. T. (1982). Inequality of income, illiteracy and medical-care as determinants of infant-mortality in underdeveloped-countries. Population Studies: A Journal of Demography, 36(3), 441-458. Folstein, M. F., Folstein, S. E., & Mchugh, P. R. (1975). Mini-Mental State: Practical method for grading cognitive state of patients for clinician. Journal of Psychiatric Research, 12(3), 189-198. Freedman, V. A., Grafova, I. B., Schoeni, R. F., & Rogowski, J. (2008). Neighborhoods and disability in later life. Social Science & Medicine. Gao, J., Qian, J. C., Tang, S. L., Eriksson, B., & Blas, E. (2002). Health equity in transition from planned to market economy in China. Health Policy and Planning, 17, 20-29. Haan, M., Kaplan, G. A., & Camacho, T. (1987). Poverty and Health: Prospective evidence from the Alameda county study. American Journal of Epidemiology, 125(6), 989-998. 119 Haas, S. (2008). Trajectories of functional health: The 'long arm' of childhood health and socioeconomic factors. Social Science & Medicine, 66(4), 849-861. Hesketh, T., Ding, Q. J., & Tomkins, A. M. (2003). Health and health care-seeking behavior of adolescents in urban and rural China. Journal of Adolescent Health, 33(4), 271-274. Hillemeier, M. M., Lynch, J., Harper, S., & Casper, M. (2003). Measuring contextual characteristics for community health. Health Services Research, 38(6), 1645-1717. Ho, S. C., Woo, J., Sham, A., Chan, S. G., & Yu, A. L. M. (2001). A 3-year follow-up study of social, lifestyle and health predictors of cognitive impairment in a Chinese older cohort. International Journal of Epidemiology, 30(6), 1389-1396. House, J. S., Lantz, P. M., & Herd, P. (2005). Continuity and change in the social stratification of aging and health over the life course: Evidence from a nationally representative longitudinal study from 1986 to 2001/2002 (Americans' Changing Lives Study). Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 60, 15-26. Judge, K., Mulligan, J. A., & Benzeval, M. (1998). The relationship between income inequality and population health. Social Science & Medicine 47(7), 983-985. Kawachi, I., & Kennedy, B. P. (1997). Socioeconomic determinants of health: Health and social cohesion: Why care about income inequality? British Medical Journal, 314(7086), 1037-1040. Kim, J. Y., & Durden, E. (2007). Socioeconomic status and age trajectories of health. Social Science & Medicine, 65(12), 2489-2502. Krause, N., Lynch, J., Kaplan, G. A., Cohen, R. D., Goldberg, D. E., & Salonen, J. T. (1997). Predictors of disability retirement. Scandinavian Journal of Work Environment & Health, 23(6), 403-413. Krieger, N. (2001a). A glossary for social epidemiology. Journal of Epidemiology and Community Health, 55(10), 693-700. 120 Krieger, N. (2001b). Theories for social epidemiology in the 21st century: An ecosocial perspective. International journal of epidemiology, 30(4), 668-677. Kuh, D., Ben-Shlomo, Y., Lynch, J., Hallqvist, J., & Power, C. (2003). Life course epidemiology. Journal of Epidemiology and Community Health, 57(10), 778-783. Lawson, J. S., & Lin, V. (1994). Health status differentials in the Peoples Republic of China. American Journal of Public Health, 84(5), 737-741. Lawton, L., Silverstein, M., & Bengtson, V. (1994). Affection, social contact, and geographic distance between adult children and their parents. Journal of Marriage and the Family, 56(1), 57-68. Lawton, M. P. (1974). Social ecology and the health of older people. American Journal of Public Health, 64(3), 257-260. Li, H., & Zhu, Y. (2006). Income, income inequality, and health: Evidence from China. Journal of Comparative Economics, 34(4), 668-693. Li, Y., Aranda, M. P., & Chi, I. (2007). Health and life satisfaction of ethnic minority older adults in mainland China: Effects of financial strain. International Journal of Aging & Human Development, 64(4), 361-379. Liang, J., Liu, X., & Gu, S. Z. (2001). Transitions in functional status among older people in Wuhan, China: Socioeconomic differentials. Journal of Clinical Epidemiology, 54(11), 1126-1138. Liang, J., McCarthy, J. F., Jain, A., Krause, N., Bennett, J. M., & Gu, S. Z. (2000). Socioeconomic gradient in old age mortality in Wuhan, China. Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 55(4), S222- S233. Liu, G. G., Dow, W. H., Fu, A. Z., Akin, J., & Lance, P. (2008). Income productivity in China: On the role of health. Journal of Health Economics, 27(1), 27-44. Liu, G. G., Wu, X. D., Peng, C. Y., & Fu, A. Z. (2003). Urbanization and health care in rural China. Contemporary Economic Policy, 21(1), 11-24. 121 Liu, G. G., Zhao, Z. Y., Cai, R. H., Yamada, T., & Yamada, T. (2002). Equity in health care access to: Assessing the urban health insurance reform in China. Social Science & Medicine, 55(10), 1779-1794. Liu, G. P., & Zhang, Z. (2004). Sociodemographic differentials of the self-rated health of the oldest-old Chinese. Population Research and Policy Review, 23(2), 117-133. Liu, X., Liang, J., & Gu, S. Z. (1995). Flows of social support and health-status among older persons in China. Social Science & Medicine, 41(8), 1175-1184. Liu, Y. L., Hsiao, W. C., & Eggleston, K. (1999). Equity in health and health care: The Chinese experience. Social Science & Medicine, 49(10), 1349-1356. Lu, X. (2002). Dang dai Zhongguo she hui jie ceng yan jiu bao gao (1st ed.). Beijing: She hui ke xue wen xian chu ban she : Jing xiao xin hua shu dian zong dian Beijing fa xing suo. Luo, Y., & Wen, M. (2002). Can we afford better health? A study of the health differentials in China. Health, 6(4), 471-500. Lynch, J., & Smith, G. D. (2005). A life course approach to chronic disease epidemiology. Annual Review of Public Health, 26, 1-35. Lynch, S. M. (2003). Cohort and life-course patterns in the relationship between education and health: A hierarchical approach. Demography, 40(2), 309-331. Macinko, J. A., Shi, L. Y., Starfield, B., & Wulu, J. T. (2003). Income inequality and health: A critical review of the literature. Medical Care Research and Review, 60(4), 407-452. Macintyre, S., Ellaway, A., & Cummins, S. (2002). Place effects on health: How can we conceptualise, operationalise and measure them? Social Science & Medicine, 55(1), 125-139. Macintyre, S., Ellaway, A., Der, G., Ford, G., & Hunt, K. (1998). Do housing tenure and car access predict health because they are simply markers of income or self esteem? A Scottish study. Journal of Epidemiology and Community Health, 52(10), 657-664. 122 Macintyre, S., Maciver, S., & Sooman, A. (1993). Area, class and health: Should we be focusing on places or people? Journal of Social Policy, 22, 213-234. Maddox, G. L., Clark, D. O., & Steinhauser, K. (1994). Dynamics of functional impairment in late adulthood. Social Science & Medicine, 38(7), 925-936. Marmot, M. G., & Shipley, M. J. (1996). Do socioeconomic differences in mortality persist after retirement? 25 year follow up of civil servants from the first Whitehall study. British Medical Journal, 313(7066), 1177-1180. Mazumdar, K. (1996). An analysis of causal flow between social development and economic growth: The social development index. American Journal of Economics and Sociology, 55(3), 361-383. McMunn, A., Breeze, E., Goodman, A., Nazroo, J., Oldfield, Z.,. (2006). Social determinants of health in older age. In Marmot, M. G. & R. G. Wilkinson (Eds.), Social determinants of health (2nd ed.). Oxford: Oxford University Press. McNeill, L. H., Kreuter, M. W., & Subramanian, S. V. (2006). Social environment and physical activity: A review of concepts and evidence. Social Science & Medicine, 63(4), 1011-1022. Midgley, J. (1999). Growth, redistribution, and welfare: Toward social investment. Social Service Review, 73(1), 3-21. Miech, R. A., & Hauser, R. M. (2001). Socioeconomic status and health at midlife: A comparison of educational attainment with occupation based indicators. Annals of Epidemiology, 11(2), 75-84. Mirowsky, J., & Ross, C. E. (2008). Education and self-rated health: Cumulative advantage and its rising importance. Research on Aging, 30(1), 93-122. Monda, K. L., Gordon-Larsen, P., Stevens, J., & Popkin, B. M. (2007). China's transition: The effect of rapid urbanization on adult occupational physical activity. Social Science & Medicine, 64(4), 858-870. National Bureau of Statistics of China. (2006). China statistical yearbook: 2006. Beijing: China Statistics Press. 123 Pei, X., & Rodriguez, E. (2006). Provincial income inequality and self-reported health status in China during 1991-7. Journal of Epidemiology and Community Health, 60(12), 1065-1069. Pickett, K. E., & Pearl, M. (2001). Multilevel analyses of neighborhood socioeconomic context and health outcomes: A critical review. Journal of Epidemiology and Community Health, 55(2), 111-122. Poortinga, W., Dunstan, F. D., & Fone, D. L. (2007). Perceptions of the neighbourhood environment and self rated health: a multilevel analysis of the Caerphilly Health and Social Needs Study. BMC Public Health, 7(1), 285. Raudenbush, S. W., Liu, X. (2000). Statistical power and optimal design for multisite randomized trials. Psychological Methods, 5(2), 199-213. Riva, M., Gauvin, L., & Barnett, T. A. (2007). Toward the next generation of research into small area effects on health: A synthesis of multilevel investigations published since July 1998. Journal of Epidemiology and Community Health, 61(10), 853-861. Robert, S. A. (1998). Community-level socioeconomic status effects on adult health. Journal of Health and Social Behavior, 39(1), 18-37. Robert, S. A., & Li, L. W. (2001). Age variation in the relationship between community socioeconomic status and adult health. Research on Aging, 23(2), 233-258. Ross, C. E., & Wu, C. L. (1996). Education, age, and the cumulative advantage in health. Journal of Health and Social Behavior, 37(1), 104-120. Roux, A. V. D. (2002). Invited commentary: Places, people, and health. American Journal of Epidemiology, 155(6), 516-519. Satariano, W. (2006). Epidemiology of aging: An ecological approach. Sudbury, Mass: Jones and Bartlett Publishers. Scherr, P. A., Albert, M. S., Funkenstein, H. H., Cook, N. R., Hennekens, C. H., Branch, L. G., et al. (1988). Correlates of cognitive function in an elderly community population. American Journal of Epidemiology, 128(5), 1084-1101. 124 Shi, L., & Starfield, B. (2000). Primary care, income inequality, and self-rated health in the United States: A mixed-level analysis. International Journal of Health Services 30(3), 541-555. Shi, L. Y., Starfield, B., Kennedy, B., & Kawachi, I. (1999). Income inequality, primary care, and health indicators. Journal of Family Practice, 48(4), 275-284. Sloggett, A., & Joshi, H. (1998). Deprivation indicators as predictors of life events 1981- 1992 based on the UK ONS longitudinal study. Journal of Epidemiology and Community Health, 52(4), 228-233. Starfield, B. (2004). Promoting equity in health through research and understanding. Developing World Bioeth, 4(1), 76-95. Starfield, B., & Shi, L. (1999). Determinants of health: Testing of a conceptual model. New York Academy of Sciences, 896, 344-346. Subramanian, S. V., Belli, P., & Kawachi, I. (2002). The macroeconomic determinants of health. Annual Review of Public Health, 23, 287-302. Sun, R. J., & Gu, D. N. (2008). Air pollution, economic development of communities, and health status among the elderly in urban China. American Journal of Epidemiology, 168(11), 1311-1318. Susser, M., & Susser, E. (1996). Choosing a future for epidemiology: II. From black box to Chinese boxes and eco-epidemiology. American journal of public health, 86(5), 674-677. Takano, T., Fu, J., Nakamura, K., Uji, K., Fukuda, Y., Watanabe, M., et al. (2002). Age- adjusted mortality and its association to variations in urban conditions in Shanghai. Health Policy, 61(3), 239-253. Turrell, G., Lynch, J. W., Kaplan, G. A., Everson, S. A., Helkala, E. L., Kauhanen, J., et al. (2002). Socioeconomic position across the lifecourse and cognitive function in late middle age. Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 57(1), S43-S51. 125 United Nations. (2005). World population prospects the 2004 revision: Population database. New York, N.Y.: Dept. of Economic and Social Affairs. Population Division, United Nations Population Division. United Nations Development Program. (2004). Human Development Report 2004: Cultural Liberty in Today's Diverse World. Washington D.C. van Doorslaer, E., Wagstaff, A., Bleichrodt, H., Calonge, S., Gerdtham, U. G., Gerfin, M., et al. (1997). Income-related inequalities in health: some international comparisons. Journal of Health Economics, 16(1), 93-112. Wadsworth, M. E. J. (1997). Health inequalities in the life course perspective. Social Science & Medicine, 44(6), 859-869. Waitzman, N. J., & Smith, K. R. (1998). Phantom of the area: Poverty-area residence and mortality in the United States. American Journal of Public Health, 88(6), 973-976. Weden, M. A., Carpiano, R. A., & Robert, S. A. (2008). Subjective and objective neighborhood characteristics and adult health. Social Science & Medicine, 66(6), 1256-1270. Wheaton, B., & Clarke, P. (2003). Space meets time: Integrating temporal and contextual influences on mental health in early adulthood. American Sociological Review, 68(5), 680-706. Willson, A. E., Shuey, K. M., & Elder, G. H. (2007). Cumulative advantage processes as mechanisms of inequality in life course health. American Journal of Sociology, 112(6), 1886-1924. World Bank. (2008). Social Development at the World Bank. Retrieved June 2, 2008, from http://web.worldbank.org/WBSITE/EXTERNAL/TOPICS/EXTSOCIALDEVEL OPMENT/0,,contentMDK:20782713~menuPK:199462~pagePK:148956~piPK:2 16618~theSitePK:244363,00.html World Health Organ. (1999). World Health Report: Making a Difference. Geneva: WHO. 126 Yen, I. H., & Syme, S. L. (1999). The social environment and health: A discussion of the epidemiologic literature. Annual Review of Public Health, 20, 287-308. Yi, Z., Liu, Y. Z., & George, L. K. (2003). Gender differentials of the oldest old in China. Research on Aging, 25(1), 65-80. Yi, Z., & Vaupel, J. W. (2002). Functional capacity and self-evaluation of health and life of oldest old in China. Journal of Social Issues, 58(4), 733-748. Yip, W., Subramanian, S. V., Mitchell, A. D., Lee, D. T. S., Wang, J., & Kawachi, I. (2007). Does social capital enhance health and well-being? Evidence from rural China. Social Science & Medicine, 64(1), 35-49. Yu, B. R. (2005). Influences of health insurance status on clinical treatments and outcomes for 4,714 patients after acute myocardial infarction in 14 Chinese general hospitals. Journal of Medical and Dental Sciences, 52(2), 143-151. Yu, M. Y., & Sarri, R. (1997). Women's health status and gender inequality in China. Social Science & Medicine, 45(12), 1885-1898. Zeng, Y., Vaupel, J. W., Xiao, Z. Y., Zhang, C. Y., & Liu, Y. Z. (2002). Sociodemographic and health profiles of the oldest old in China. Population and Development Review, 28(2), 251-273. Zhang, Z., Gu, D., & Hayward, M. D. (2008). Early life influences on cognitive impairment among oldest old Chinese. Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 63(1), S25-S33. Zhao, Z. W. (2006). Income inequality, unequal health care access, and mortality in China. Population and Development Review, 32(3), 461-483. Zimmer, Z., Kaneda, T., & Spess, L. (2007). An examination of urban versus rural mortality in china using community and individual data. Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 62(5), S349-S357. Zimmer, Z., & Kwong, J. (2004). Socioeconomic status and health among older adults in rural and urban China. Journal of Aging and Health, 16(1), 44-70.

Abstract (if available)

Abstract This dissertation consists of two independent studies and one conceptual paper, addressing one overarching question: how changes in socioeconomic status (SES) based social stratification and changes in social environment affect the health status of older adults in China across time and space. In the two empirical studies, analyses were performed using the four waves of data from the China Health and Nutrition Survey (CHNS), a longitudinal survey of nine provinces in China. Latent growth curve analyses from the first study (2,345 individuals aged 55 and above) show that the health of older Chinese adults continued to be shaped by individual education, occupational class, and income. Cumulative advantage hypotheses are supported by the findings that higher education predicts both better self-rated health and cognitive function at baseline and slower declines over time. Higher income is associated with better functional health at baseline as well as a more gradual decline over time. In addition, this study reveals that the relationship between occupational class and cognitive function as well as between income and self-rated health do not support the cumulative advantage hypothesis.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Family relationships and their influence on health outcomes over time among older adults in rural China

PDF

Life transitions, leisure activity engagement, and cognition among older adults

PDF

Intergenerational support between grandparents and grandchildren in rural China and its effect on the psychological well-being of older adults

PDF

Examining the longitudinal influence of the physical and social environments on social isolation and cognitive health: contextualizing the role of technology

PDF

Cognitive health and self-rated memory in later life: linkages to race/ethnicity, multimorbidity, and survival

PDF

Mechanisms of the effect of involuntary retirement on older adults' health and mental health

PDF

A study of depression disclosure among Latino older adults in Los Angeles

PDF

Association of carotid artery stiffness with measures of cognition in older adults

PDF

Self-perceptions of Aging in the Context of Neighborhood and Their Interplay in Late-life Cognitive Health

PDF

Subclinical carotid atherosclerosis, psychosocial measures, and cognitive function in middle- to older-aged adults

PDF

Children 's migration and the financial, social, and psychological well-being of older adults in rural China

PDF

Impacts of caregiving on wellbeing among older adults and their spousal caregivers in the United States

PDF

Integrative care strategies for older adults experiencing co-occurring substance use and mental health disorders (I-CARE)…

PDF

Personality and cognitive aging: are positive traits protective?

PDF

Investigating the effectiveness of a social work intervention on reducing hospital readmissions among older adults

PDF

Examination of the long-term psychosocial and functioning effects of a healthy living, life management behavior intervention for older adults

PDF

Fall related injuries among older adults in the Los Angeles region

PDF

Thinking generatively versus acting generatively: exploring the associations of generative self-concept and generative activity with cognitive function among older adults

PDF

Motivation to move: seeking to understand the motives and barriers in older adults to participate in physical activity classes

PDF

State variations in linguistic competency policies and the effects on immigrant access to health services

Asset Metadata

Creator Li, Yawen (author)

Core Title The effects of time and space on health status of older adults in China

School School of Social Work

Degree Doctor of Philosophy

Degree Program Social Work/ Gerontology

Publication Date 08/06/2009

Defense Date 05/11/2009

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag cognitive function,functional health,health disparities,OAI-PMH Harvest,older Chinese adults,self-rated health,social environment,socioeconomic status

Place Name China (countries)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Chi, Iris (committee chair), Palinkas, Lawrence A. (committee member), Silverstein, Merril (committee member)

Creator Email li.yawen@gmail.com,yawenli@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m2500

Unique identifier UC1220160

Identifier etd-Li-3015 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-180550 (legacy record id),usctheses-m2500 (legacy record id)

Legacy Identifier etd-Li-3015.pdf

Dmrecord 180550

Document Type Dissertation

Rights Li, Yawen

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu