This Article Abstract Full Text (PDF) Services Download to citation manager Citing Articles Citing Articles via HighWire PubMed PubMed Citation

Neighborhood Effects on the Self-Rated Health of Elders: Uncovering the Relative Importance of Structural and Service-Related Neighborhood Environments

Department of Society, Human Development, and Health, Harvard School of Public Health, Boston, Massachusetts.

Address correspondence to S. V. Subramanian, Department of Society, Human Development, and Health, Harvard School of Public Health, 677 Huntington Ave., KRESGE 7^th Floor, Boston, MA 02115. E-mail: svsubram{at}hsph.harvard.edu

	Abstract

TOP Abstract Typology of Neighborhood Effects Study Objectives Methods Results Discussion References

 
Objectives. The purpose of this study was to investigate the independent relationship between neighborhood context (characterized through age structure, economic conditions, service provision, and residential stability) and self-reported health among elders in one U.S. city.

Methods. By using multilevel statistical models, we examined the cross-sectional relationships between markers of neighborhood environment (derived from the 1980 U.S. Census and the Yellow Pages of the 1985 New Haven, Connecticut, telephone book) and self-rated health among elders. We used survey data from the 1985 New Haven Established Populations for Epidemiologic Studies of the Elderly, which comprised 1,926 elders nested within 28 census tracts.

Results. When controlled for individual age, gender, race, marital status, education, and income, neighborhood measures of percent poverty were positively associated with poor self-rated health (odds ratio [OR] = 1.09; 95% confidence interval [CI] = 1.02–1.17), whereas residential stability (OR = 0.90; 95% CI = 0.84–0.96) and concentration of elders (OR = 0.82; 95% CI = 0.72–0.94) were inversely associated with poor self-rated health. Neighborhood service density was not associated with self-rated health.

Discussion. We found support for the role of neighborhood structural context (reflected through measures of poverty, residential stability, and age-based demographic concentration) in predicting the health of elders. Density of neighborhood services did not appear to have an independent effect on the self-rated health of elders.

RESEARCH on understanding the influences of residential environments on individual health outcomes is becoming increasingly common (Berkman & Kawachi, 2000; Kawachi & Berkman, 2003; Marmot & Wilkinson, 1999). Indeed, during the past 10 years or so, empirical research has shown—with some degree of consistency—an association between neighborhood factors and individual health (for a review of these works, see Diez Roux, 2001; Ellen, Mijanovich, & Dillman, 2001; O'Campo, 2003; Pickett & Pearl, 2001; Sampson & Morenoff, 2002). This study investigated whether neighborhood environment was predictive of self-rated health at older life stages, with the neighborhood environment being defined by both census data and by data characterizing the service environments for neighborhoods in the city of New Haven, Connecticut.

Why should neighborhood environments matter for the health of elders? With populations, on average, living longer, it is becoming increasingly important to understand determinants of health at later stages of the life course (Rubenstein, Moss, & Klebow, 1999). Gerontologists have shown considerable interest in how physical environment affects older adults; the focus has not been so much on health per se (Carp, 1994), although there are a few notable exceptions (Krause, 1996). Importantly, researchers have seldom successfully integrated this work with the epidemiology of aging in order to identify how living under various environmental conditions affects the health of elders. Even when studies have investigated the effects of related factors like residential stability on well-being in late life, researchers have tended to conceptualize such effects at the individual level. Such research cannot address questions related to how places influence individuals. One can argue that health in late life is particularly sensitive to residential environments (Friedrich, 1991; Lawton, 1990; Rubenstein et al., 1999). Dilapidated neighborhood environments typically go hand in hand with higher levels of environmental pollutants, over-crowding, violence, less social cohesion, and low availability of services. Several studies have also shown that residential instability is especially stressful for older persons (Armer, 1993; Colsher & Wallace, 1990; Schulz & Brenner, 1977). Although it is often measured at the level of the individual, stability of residence is likely to be a function of neighborhood factors because dilapidated environments often discourage residential stability (Sampson, 2003; Sampson & Morenoff, 2002). Older adults who reside in disadvantaged neighborhoods experience more physical health problems than those who dwell in more favorable environments (Chapman & Beaudet, 1983; Krause). Investigators have also shown that neighborhood characteristics such as low crime rates and lack of noise promote active and satisfying lives for older people (Krause; Lawton, Nahemow, & Tsong, 1980). Almost all of these studies tend to treat neighborhood characteristics as individual characteristics, an aspect from which the present study makes a distinct departure. Testing the proposition that neighborhood contexts contribute to health outcomes independently of individual characteristics requires the use of a multilevel study design and analytic strategies (Subramanian, 2004a, 2004b; Subramanian, Jones, & Duncan, 2003).

Research on neighborhoods and health has typically characterized neighborhoods by using data routinely available from administrative sources such as the census—data that may not adequately capture the multidimensional nature of neighborhood contexts (Macintyre, Ellaway, & Cummins, 2002). Furthermore, although extremely valuable in defining the structural environment of a neighborhood, census variables are by definition aggregates of individual data, expressed commonly as percentage of people in a given area who live below the poverty line, or percentage of people in a given area who are unemployed, or median income of a given area, and so on. Thus, data restricted to capturing the characteristics of individuals in an area are likely to reflect only one aspect of the neighborhood residential environment. At the same time, such aggregate characteristics of individuals tend, on average, to correlate rather strongly with other aspects of the neighborhood environment, including social cohesion, forms of control, stress, inadequate local resources (such as public transportation services, recreational facilities, health centers, and other services; Sampson, 1998, 2003; Sampson & Morenoff, 2002). The literature needs studies that uncover the mechanisms by which neighborhood disadvantage (as defined using census measures) leads to adverse health outcomes. That is, researchers need to understand what it is about neighborhood deprivation that produces differential patterns of risk and protection. This task requires going beyond the use of census-derived indicators of disadvantage and moving toward defining, operationalizing, and measuring specific neighborhood characteristics (e.g., local access to services and amenities), exposures (e.g., crime and vandalism), and social processes (e.g., behavioral contagion, social cohesion; Kawachi & Berkman, 2003).

One aspect of neighborhoods that investigators have hypothesized to be relevant for health is the service environment—the availability of services that may either enhance an individual's ability for self-care or increase ease of social integration. There has been some movement in the empirical literature toward considering non-census measures in order to characterize areas (O'Campo, Xue, Wang, & Caughy, 1997; Sooman & Macintyre, 1995). However, we are not aware of any study that has examined the relationship between non-census measures of an area designed to capture the service environment of a neighborhood and self-rated health at later life, over and above the effect of census measures and individual factors.

	Typology of Neighborhood Effects

TOP Abstract Typology of Neighborhood Effects Study Objectives Methods Results Discussion References

 
In this study, we conceptualized and operationalized the neighborhood effects along three domains.

Compositional Effects
The first type of effect was based on characteristics specific to the older individuals who lived in the particular neighborhoods (i.e., each individual's age, gender, race, marital status, education, and income). We identified elders' individual characteristics as an intrinsic component of the neighborhood because we did not expect these characteristics to be distributed randomly across neighborhoods. Consequently, even though these measures pertained to individuals, they could not automatically be equated with individual effects. Explanation of neighborhood differences as a function of the concentration of the individual characteristics of the people who live in these places is often called a compositional explanation to neighborhood variation in health (Macintyre, 1997). We should note that compositional explanations are not necessarily the same as individual explanations (even though a majority of the literature on neighborhoods and health tends to use the terms interchangeably), because differential composition between areas may well be due to processes that are extraindividual and need not always be a reflection of individual choice.

Collective Effects
The second type of neighborhood effect relates to what researchers refer to as the collective (Macintyre, 1997; Macintyre & Ellaway, 2000). Investigators often interpret such effects based on measures drawn from census data that are aggregates of characteristics associated with all individuals in an area. In this sense, it is distinct from characterizing a neighborhood based only on its elders; this conceptual distinction is especially critical when examining specific population groups. The rationale for using such collective measures is that they are greater than the sum of their parts. In other words, collective properties of an area are likely to reflect a "structural" component of the neighborhood that both serves as a backdrop for other more dynamic aspects of the neighborhood environment and also exerts some independent influence. For instance, having a greater percentage of people living below the poverty line (an aggregate census-based measure) is likely to affect other aspects of neighborhood environment, such as the quantity and quality of service provision. In addition, having a greater percentage of people living below the poverty line may also exert social or cultural effects that occur as a function of living in the midst of poor people. Such effects are likely to influence all individuals, not simply the poor in that area.

Contextual Effects
In addition to using compositional and collective domains to distinguish neighborhood effects, we also conceptualized neighborhood in terms of its service environment. For instance, age-related losses in ability for self-care mean that older persons rely more on municipal services. Research has shown that older persons are more affected than younger persons by the availability of services (Davies & Fleishman, 1981), including services such as transportation (Rittner & Kirk, 1995). Other work has indicated that attendance at senior centers becomes more frequent with advancing age (Minor, 1991), suggesting that the events of older age (including widowhood, deaths of friends, and children moving away) may result in older persons relying more heavily on community resources. By using data abstracted from the Yellow Pages of the 1985 New Haven telephone book, we sought to develop a set of indicators in order to characterize the neighborhood service environment. We developed neighborhood-level density measures of four types of services: (a) services that promoted social organization, (b) services that promoted social interaction, (c) services that were directly health related, and (d) services that may have adversely affected the reputation of neighborhoods or promoted deleterious health behaviors; we define each of these service types in the Methods section. Such measures are likely to reflect the contextual pathways through which neighborhood influences health, including the pathways that may be amenable to policy interventions. For instance, planners often base decisions related to location and allocation of key services on characteristics of individuals in a neighborhood. But if it becomes evident that particular aspects of the service environment can either help or harm neighborhood residents, location and allocation of key services could be regulated.

	Study Objectives

TOP Abstract Typology of Neighborhood Effects Study Objectives Methods Results Discussion References

 
With this conceptual typology of neighborhood effects in mind, this study presents a cross-sectional multilevel analysis of neighborhood effects on self-rated health in late life by using a combination of individual, census, and area-based data. Specifically, we examined: (a) the extent to which neighborhood variation in self-rated health among elders can be explained by compositional effects, as evidenced by the concentration of individual characteristics of elders in that neighborhood; (b) the association between census-based neighborhood characteristics and self-reported health of elders, after controlling for their individual characteristics; and (c) the association between the neighborhood service environment and self-rated health of elders and whether neighborhood service environment can account for any of the relationship between census-based neighborhood socioeconomic characteristics and self-rated health of elders.

In aggregate studies, investigators have reported that percentage of persons living in poverty is reliably and consistently associated with a number of health outcomes, including mortality (Krieger, Chen, Waterman, Rehkopf, & Subramanian, 2005). We hypothesized that economic conditions at the census tract level would be associated with worse self-rated health among elders, even after controlling for individual compositional characteristics. Residential stability, meanwhile, is a key neighborhood characteristic that determines residents' ability to maintain shared values and realize common goals (Sampson, 2003), and instability has been linked to low social cohesion (including the inability to exercise social control) and undesirable outcomes such as higher crime rates (Sampson & Morenoff, 2002; Sampson & Raudenbush, 2004). We hypothesized that residential stability would be associated with lower risk of poor self-rated health among elders. Similarly, demographic concentration—in particular higher concentrations of older adults at the neighborhood level—has been hypothesized to be a marker either for the presence of services catering to elders or for the availability of dense social support networks (Glass & Balfour, 2003). For example, one examination of the patterns of service use and attitudes toward aging in 1,100 seniors in neighborhoods of high, moderate, and lower concentrations of elders found that living in areas of higher concentrations of elders was associated with increased knowledge of, and access to, services catering to elders (Sherman, Ward, & Lagory, 1985). Following these findings, we hypothesized that higher neighborhood concentration of older adults would be associated with better self-rated health among older residents. Finally, after accounting for individual-level demographic and socioeconomic factors, we explored the extent to which neighborhood structural characteristics and neighborhood service density independently predicted the self-rated health of elders.

	METHODS

TOP Abstract Typology of Neighborhood Effects Study Objectives Methods Results Discussion References

 
Delineating Neighborhood
We operationalized the concept of neighborhood by using census tracts that, on average, contained 4,000 persons (U.S. Census Bureau, 2000). Although by no means ideal, the appeal of census tracts as a proxy for people's residential environment is that they were originally "designed to be homogeneous with respect to population characteristics, economic status, and living conditions" (U.S. Census Bureau web site). Even if their creation was largely administrative (including the collection and dissemination of information on the socioeconomic characteristics of their populations), such partitioning of space is likely to provide a setting for the unfolding of important processes that may potentially influence residents' health. Importantly, this may still be the case even if such partitioning does not map neatly onto residents' perception of what constitutes their neighborhood boundary. For instance, federal, state, and local governments (including public health departments) routinely use census tracts in order to characterize jurisdictions, to determine eligibility for various programs, and to allocate resources (Boston Public Health Commission, 2002; Bureau of Health Professions, 2002) that may have substantive implications for the functioning of their residents (Monmonier, 1997).

Sources of Data
We used data from the 1985 New Haven Established Populations for Epidemiologic Studies of the Elderly (EPESE), the details of which are discussed elsewhere (Berkman, et al., 1986). We derived the data used for this analysis from a probability sample of 2,487 noninstitutionalized men and women aged 65 years and older who lived in the city of New Haven, Connecticut, in 1985. We geo-coded individual records from 1985 to the 1980 census tracts. We derived census tract characteristics from the 1980 Census and determined services by using the Yellow Pages of the 1985 New Haven telephone book.

Outcome
We used the response to the following question related to self-rated health (available in the 1985 EPESE) as our outcome of interest: "How would you rate your health at the present time? Excellent, good, fair, poor, or bad." More than two dozen prospective studies in the United States and elsewhere have established that self-reported health is highly predictive of subsequent mortality independent of other medical, behavioral, and/or psychosocial factors (Idler & Benyamini, 1997). We dichotomized the 5-scale item response into 1 (fair, poor, or bad) or 0 (excellent or good). In the sample data used for analysis, approximately 38% of the respondents reported being in fair, poor, or bad health (Table 1).

Neighborhood Predictors: Yellow-Page–Derived Measures of Service Density
Following a prior study (Yen & Kaplan, 1999), we developed measures of service density at the census tract level that we based on data abstracted from the Yellow Pages of the New Haven telephone book. We distinguished four types of services: (a) services that promoted social organization (e.g., churches, synagogues); (b) services that promoted social interaction (e.g., beauty parlors, cafés, libraries); (c) services that were directly health related (e.g., hospitals, audiologists, pharmacies); and (d) services that may have adversely affected the reputation of neighborhoods and/or promoted deleterious health behaviors (e.g., liquor outlets, pawnbrokers, tattoo parlors, fast food outlets). Based on a page-by-page analysis, two independent investigators abstracted a list of these four types of neighborhood services and amenities from the Yellow Pages of the 1985 New Haven telephone book. The two independent listings had an interrater agreement of more than 85%. By using the street addresses, we subsequently geo-coded every service establishment to the census tract in which it was located. We then used the counts of each of the four types of services in order to create the service density measure, based on the total census tract population, for each type of service (Table 1). We then dichotomized these population-based service density measures (the number of services in a census tract ÷ total population) into those above and below the median.

Individual Predictors
Individual-level predictors included age (specified as a continuous variable), gender, race, marital status, educational attainment, and income as obtained from the EPESE survey (Table 1).

Statistical and Modeling Strategy
After excluding observations that were missing on the variables considered, we used the multilevel statistical approach (Goldstein, 2003) on a final analytic sample of 1,926 individuals (Level 1) nested within 28 neighborhoods (Level 2). Investigators have described the principles underlying multilevel modeling procedures elsewhere (Subramanian, 2004a, 2004b; Subramanian, et al., 2003). In the context of the present analysis, multilevel statistical models enabled estimation of (a) the conditional relationship between poor self-rated health and individual predictors (fixed parameters); (b) the variation between census tracts that could not be accounted for by individual predictors (random parameters); and (c) the main effect of neighborhood predictors on poor self-rated health (fixed parameters), conditional upon the individual-level relationship between poor self-rated health and individual demographic and socioeconomic markers. We adopted a two-level logistic modeling approach (with binomial assumption at Level 1 and normality assumption at Level 2) with predictive or penalized quasi-likelihood second approximation linearization procedures (Goldstein & Rasbash, 1996). We calibrated models by using restricted maximum likelihood estimation (Goldstein, 2003; Rasbash, Steele, Browne, & Prosser, 2004). Because the survey data oversampled elders from certain housing units (stratified by age and gender; Berkman et al., 1986) we weighted model estimates to a comprehensive sampling weight.

We calibrated three types of models. The first modeled the fixed relationship between individual predictors and poor self-rated health conditional on a random effect for census tracts but unadjusted for neighborhood predictors (Aim 1). The second examined the unadjusted fixed effect of each of the neighborhood predictors (census and non-census measures) on individual self-rated health before and after adjusting for individual predictors (Aim 2). The final model examined the contribution of neighborhood-level service predictors to predicting individual self-rated poor health conditional on individual predictors and census-based neighborhood predictors, one at a time (Aim 3).

	RESULTS

TOP Abstract Typology of Neighborhood Effects Study Objectives Methods Results Discussion References

 
Table 2 reports the conditional relationship between various individual demographic and socioeconomic markers and poor self-rated health. The odds ratios (ORs) we report here were all statistically significant at p =.05. The reference category was a 67-year-old non-Hispanic White married man with a college education or higher and with an annual income greater than $15,000; for this "most advantaged" group, the probability of reporting poor health was 16%. Age was positively associated with the probability of reporting poor health, with each 5-year increment in age increasing the odds of reporting poor health by 18%. Gender and race were not statistically associated with self-rated health. Widowed individuals (OR = 0.60) and persons in the residual marital category of "other" (OR = 0.65) were less likely than married individuals to report poor health. Individuals with less than a high school education were more likely to report poor health (OR = 1.87) compared with the reference group of some college or more. We did not find a statistically significant gradient in self-rated health across educational categories. Compared with individuals in the highest income category (greater than $15,000), persons in the lowest income category (less than $5,000) were more than twice as likely to report poor health, followed by individuals who earned $5,000–$10,000 (OR = 1.61). Models testing for an association between neighborhood and poor self-rated health showed evidence of significant association after we had adjusted for the individual-level characteristics. The between-neighborhood variation (associated with the random effects for neighborhoods) decreased (p =.04 vs p =.07, respectively).

Table 4 presents the results from 30 different models (each of which tested the independent predictive power of the four types of service domains) after adjusting for individual characteristics and one census-based structural neighborhood marker at a time. Structural measures related to neighborhood poverty, residential stability, and elder concentration continued to be associated with individual poor self-rated health even after controlling for individual socioeconomic markers as well as the different types of service (Table 4). Individuals who resided in neighborhoods with fewer services that promoted social organization were less likely to report poor health (OR = 0.75) in a model that also included the census measure of neighborhood affluence. We observed similar effects for this service component when we included residential stability in the model.

	DISCUSSION

TOP Abstract Typology of Neighborhood Effects Study Objectives Methods Results Discussion References

Because census-based neighborhood measures are formed by aggregating individuals, some researchers have argued that such measures may reflect only the compositional characteristics of a neighborhood (Yen & Kaplan, 1999) rather than its collective properties (Macintyre et al., 2002), which in turn exert some sort of contextual effect. However, as we emphasized earlier, in the context of empirical analysis focused on a subset of the population (e.g., elders), it is important to distinguish composition of the neighborhood based on only elders from collective composition of the neighborhood based on all the individuals living in that neighborhood. The statistically significant effect of neighborhood poverty and elderly concentration on elders' self-rated health is quite critical from an etiologic perspective. Investigators have long hypothesized that the demographic structure of neighborhoods affects the availability of services (e.g., schools, clinics, parks, and playgrounds) to residents (Glass & Balfour, 2003). However, the empirical evidence has been mixed. For example, Sherman and colleagues (1985) found that seniors who lived in neighborhoods with higher concentrations of older adults had better knowledge of, as well as access to, various kinds of support services compared with seniors who lived in areas with low concentrations of older adults. By contrast, Usui and Keil (1987) found that a high concentration of older adults was associated with reduced well-being and life satisfaction for residents, possibly because of the burdens of providing support to aging neighbors. In the present study, we sought to test the hypothesis that higher concentrations of older adults in New Haven neighborhoods are associated with better self-rated health for older residents and that this association is mediated by higher density of various kinds of services. Although being poor and being older is associated with a higher likelihood of reporting poor health at the individual level, there seems to be an additional collective effect of neighborhood poverty and elderly concentration (a multilevel contextual relationship). Notably, living in neighborhood with a greater proportion of elders seemed to have a protective effect on the likelihood of reporting poor health. It is possible that older age concentration in a neighborhood is a marker for better service provision targeted towards elders. Alternatively, it may reflect the positive influence of relative comparisons (vis à vis expected health status) among older residents who live in close proximity to one another.

With regard to service density within a neighborhood, a previous analysis from the Alameda County Study (Yen & Kaplan, 1999) reported that a higher number of commercial stores (grocery stores, supermarkets, pharmacies, beauty parlors or barber shops, and laundromats or dry cleaners) in a census tract was associated with increased all-cause mortality (age, gender-adjusted OR = 1.40; 95% confidence interval = 1.02–1.93) and, importantly, that "neighborhood socioeconomic status" and number of commercial stores were not correlated (r =.04, p =.49). Although the present study disaggregated the characterization of neighborhood service environment to four different types, we too found worse health status among elders in neighborhoods with more services (particularly those services that promoted social interactions). This seems to suggest the presence of other mechanisms that may explain the structural relationship between neighborhood poverty, residential stability, demographic concentration, and poor self-rated health among elders.

Readers should recognize the following caveats specific to the present study. First, our operationalization of the service environment in terms of its per capita population density is only one possible approximation of the neighborhood service environment. As such, although this measure may capture the crude availability of services, it does not capture elements of accessibility (which may be more influenced by availability of transportation services than by proximity to one's residence) and quality. Unfortunately, we did not have data on transportation services or quality of the services, and, therefore, we were unable to look at these issues. Second, we deliberately focused our conceptualization of the service environment on the overall neighborhood service environment rather than restricted it to services that were likely to be relevant primarily for elders. These two limitations may also partially account for the lack of apparent correspondence between the availability of services and self-reported health. It is necessary to undertake further research on measuring service environments in order to better test the links between neighborhood service and built environments and health. Third, the data we used in the present study were not the most current, and society's continual transformation has implications for the temporal appropriateness of our results. Indeed, the lack of association between the neighborhood service environment and health leads to the question of whether the service environment may have become more or less important with time. Given our substantive interest in the question of self-rated health (as opposed to only mortality), it seemed more appropriate to consider the baseline survey data for statistical efficiency. However, before ruling out any statistical association between service environment and health in later life, we must examine more closely the temporal stability of this postulated relationship.

The value of the present study is in its empirical demonstration of the relationship between structural aspects of neighborhood environments (as measured by area poverty, residential stability, and concentration of older adults) and the health of elders. We did not find any statistical support for a relationship between the density of neighborhood services and health, suggesting that the provision of such services may be of less import than has been previously assumed either in terms of a direct impact on health or as a pathway through which neighborhood structural disadvantage may influence health. In what identifiable and mutable ways residential context might influence health in late life is an important component of any research and policy agenda on successful aging. Issues related to measuring neighborhood characteristics are central to this research, and this study is a first empirical step in that direction.

	Acknowledgments

 
This study was supported by the National Institute of Aging (Grant no. AG018369; principal investigator: Lisa Berkman). S. V. Subramanian is supported by the National Institute of Health, Lung, and Blood Institute Career Development Award (K25 HL081275-01).

	Footnotes

 
Decision Editor: Charles F. Longino, Jr., PhD

Received for publication August 9, 2004. Accepted for publication July 4, 2005.

	References

TOP Abstract Typology of Neighborhood Effects Study Objectives Methods Results Discussion References

 

• Armer, J. M. (1993). Elderly relocation to a congregate setting. Issues in Mental Health Nursing, 14,157-172.[Medline]
• Berkman, L. F., Berkman, C. S., Kasl, S., Freeman, D. H., Jr., Leo, L., Ostfeld, A. M., et al. (1986). Depressive symptoms in relation to physical health and functioning in the elderly. American Journal of Epidemiology, 124,372-388.[Abstract/Free Full Text]
• Berkman, L. F., Kawachi, I., (Eds.). (2000). Social epidemiology. New York: Oxford Press.
• Boston Public Health Commission. (2002). Race and the elimination of health disparities in the city of Boston: Promoting the health of the underserved. Boston: Author.
• Bureau of Health Professions. (2002). Health professional shortage areas. Retrieved February 2, 2006, from http://bhpr.hrsa.gov/shortage/.
• Carp, F. M. (1994). Assessing environments. In M. P. Lawton & J. A. Teresi (Eds.), Annual review of gerontology and geriatrics (Vol. 14, pp. 302–323). New York: Springer.
• Chapman, N. J., Beaudet, M. (1983). Environmental predictors of well-being for at-risk older adults in a mid-sized city. Journal of Gerontology, 38,237-244.[Abstract/Free Full Text]
• Colsher, P. L., Wallace, R. B. (1990). Health and social antecedents of relocation in rural elderly persons. Journal of Gerontology: Social Sciences, 45,S32-S38.
• Davies, A. M., Fleishman, R. (1981). Health status and use of health services as reported by the older residents of the Baka neighborhood, Jerusalem. Israel Journal of Medical Sciences, 17,(2/3), 138-144.[Medline]
• Diez Roux, A. V. (2001). Investigating neighborhood and area effects on health. American Journal of Public Health, 91,1783-1789.[Abstract/Free Full Text]
• Ellen, I. G., Mijanovich, T., Dillman, K.-N. (2001). Neighborhood effects on health: Exploring the links and assessing the evidence. Journal of Urban Affairs, 23,391-408.
• Friedrich, K. (1991). Spatial pattern and consequences of demographic aging processes for regional planning. Zeitschrift für Gerontologie, 24,257-265.[Medline]
• Glass, T., Balfour, J. L. (2003). Neighborhoods, aging and functional limitation. In I. Kawachi & L. F. Berkman (Eds.), Neighborhoods and health (pp. 303–334). New York: Oxford University Press.
• Goldstein, H. (2003). Multilevel statistical models. London: Arnold.
• Goldstein, H., Rasbash, J. (1996). Improved approximations for multilevel models with binary responses. Journal of the Royal Statistical Society A, 159,505-513.
• Idler, E. L., Benyamini, Y. (1997). Self-rated health and mortality: A review of twenty-seven community studies. Journal of Health and Social Behavior, 38,21-37.[Medline]
• Kawachi, I., Berkman, L. F., (Eds.). (2003). Neighborhoods and health. New York: Oxford University Press.
• Krause, N. (1996). Neighborhood deterioration and self-rated health in later life. Psychology & Aging, 11,342-352.
• Krieger, N., Chen, J. T., Waterman, P. D., Rehkopf, D. H., Subramanian, S. V. (2005). Painting a truer picture of U.S. socioeconomic and racial/ethnic health inequalities: The Public Health Disparities Geocoding Project. American Journal of Public Health, 95,312-323.[Abstract/Free Full Text]
• Lawton, M. P. (1990). Residential environment and self-directedness among older people. American Psychologist, 45,638-640.[Medline]
• Lawton, M. P., Nahemow, L., Tsong, M. Y. (1980). Neighborhood environment and the wellbeing of older tenants in planned housing. International Journal of Aging and Human Development, 11,211-227.[Medline]
• Macintyre, S. (1997). What are the spatial effects and how can we measure them? In A. Dale (Ed.), Exploiting national surveys and census data: The role of locality and spatial effects (pp. 1–28). Manchester, UK: Center for Census and Survey Research, University of Manchester.
• Macintyre, S., Ellaway, A. (2000). Ecological approaches: Rediscovering the role of physical and social environment. In L. F. Berkman & I. Kawachi (Eds.), Social epidemiology (pp. 332–348). New York: Oxford Press.
• Macintyre, S., Ellaway, A., Cummins, S. (2002). Place effects on health: How can we conceptualise, operationalise and measure them? Social Science and Medicine, 55,125-139.
• Marmot, M., Wilkinson, R. G. (1999). The social determinants of health. Oxford, England: Oxford University Press.
• Minor, M. (1991). Physical activity and management of arthritis. Annals of Behavioral Medicine, 13,117-124.
• Monmonier, M. (1997). Cartographies of danger: Mapping hazards in America. Chicago: University of Chicago Press.
• O'Campo, P. (2003). Advancing theory and methods for multilevel models of residential neighborhoods and health. American Journal of Epidemiology, 157,9-13.[Free Full Text]
• O'Campo, P., Xue, X., Wang, M. C., Caughy, M. (1997). Neighborhood risk factors for low birthweight in Baltimore: A multilivel analysis. American Journal of Public Health, 87,1113-1118.[Abstract/Free Full Text]
• Pickett, K. E., Pearl, M. (2001). Multilevel analyses of neighborhood socioeconomic context and health outcomes: A critical review. Journal of Epidemiology and Community Health, 55,111-122.[Abstract/Free Full Text]
• Rasbash, J., Steele, F., Browne, W., Prosser, B. (2004). A user's guide to MLwiN version 2.0. London: Centre for Multilevel Modelling, Institute of Education.
• Rittner, B., Kirk, A. B. (1995). Health care and public transportation use by poor and frail elderly people. Social Work, 40,365-373.[Medline]
• Rubenstein, R., Moss, M., Klebow, M., (Eds.). (1999). The many dimensions of aging. New York: Springer.
• Sampson, R. J. (1998). What "community" supplies. In R. F. Ferguson & W. T. Dickens (Eds.), The future of community development: A social science synthesis (pp. 241–292). Washington, DC: Brookings Institution Press.
• Sampson, R. J. (2003). Neighborhood-level context and health: Lessons from sociology. In I. Kawachi & L. F. Berkman (Eds.), Neighborhoods and health (pp. 132–146). New York: Oxford University Press.
• Sampson, R. J., Morenoff, J. D. (2002). Assessing neighborhood effects: Social processes and new directions in research. Annual Review of Sociology, 28,443-478.
• Sampson, R. J., Raudenbush, S. W. (2004). Seeing disorder: Neighborhood stigma and the social construction of broken windows. Social Psychology Quarterly, 67,319-342.
• Schulz, R., Brenner, G. (1977). Relocation of the aged: A review and theoretical analysis. Journal of Gerontology, 32,323-333.[Abstract/Free Full Text]
• Sherman, S. R., Ward, R. A., LaGory, M. (1985). Socialization and aging group consciousness: The effect of neighborhood age concentration. Journal of Gerontology, 40,102-109.[Abstract/Free Full Text]
• Sooman, A., Macintyre, S. (1995). Health and perceptions of the local environment in socially contrasting neighborhoods in Glasgow. Health and Place, 1,27-40.
• Subramanian, S. V. (2004a). Multilevel methods, theory and analysis. In N. Anderson (Ed.), Encyclopedia on health and behavior (pp. 602–608). Thousand Oaks, CA: Sage.
• Subramanian, S. V. (2004b). The relevance of multilevel statistical models for identifying causal neighborhood effects. Social Science and Medicine, 58,1961-1967.
• Subramanian, S. V., Jones, K., Duncan, C. (2003). Multilevel methods for public health research. In I. Kawachi & L. F. Berkman (Eds.), Neighborhoods and health (pp. 65–111). New York: Oxford University Press.
• U.S. Census Bureau. (2000). Census 2000 geographic terms and concepts. Washington, DC: U.S. Department of Commerce. Retrieved February 9, 2006, from http://www.census.gov/geo/www/cen_tract.html.
• Usui, W. M., Keil, T. J. (1987). Life satisfaction and age concentration of the local area. Psychology and Aging, 2,30-35.[Medline]
• Yen, I. H., Kaplan, G. A. (1999). Neighborhood social environment and risk of death: Multilevel evidence from the Alameda County Study. American Journal of Epidemiology, 149,898-907.[Abstract/Free Full Text]

This article has been cited by other articles:

				G. C. Smith and G. M. Sylvestre Effects of Neighborhood and Individual Change on the Personal Outcomes of Recent Movers to Low-Income Senior Housing Research on Aging, September 1, 2008; 30(5): 592 - 617. [Abstract] [PDF]

				S. V. Subramanian, J. T. Chen, D. H. Rehkopf, P. D. Waterman, and N. Krieger Comparing Individual- and Area-based Socioeconomic Measures for the Surveillance of Health Disparities: A Multilevel Analysis of Massachusetts Births, 1989-1991 Am. J. Epidemiol., November 1, 2006; 164(9): 823 - 834. [Abstract] [Full Text] [PDF]

				S. V. Subramanian, J. T. Chen, D. H. Rehkopf, P. D. Waterman, and N. Krieger Subramanian et al. Respond to "Think Conceptually, Act Cautiously" Am. J. Epidemiol., November 1, 2006; 164(9): 841 - 844. [Full Text] [PDF]

This Article Abstract Full Text (PDF) Services Download to citation manager Citing Articles Citing Articles via HighWire PubMed PubMed Citation