Eliminating Health Disparities: Measurement and Data Needs (2004)

Patricia O’Campo and Jessica Burke*

The purpose of this paper is to explore what socioeconomic position measures can and should be collected along with racial and ethnic data to measure and better understand disparities in health.

Substantial research documents racial and ethnic disparities in health status and health care access. Racial disparities in disease incidence have been demonstrated for many health outcomes including cardiovascular disease, HIV/AIDS, diabetes, and infant morality (U.S. Department of Health and Human Services, 2002). For example, the national infant mortality rate (IMR) for African Americans is 2.5 times that of whites, with an IMR as high as 13.8 deaths per 1,000 live births for African Americans (Hoyert et al., 2000). The mortality rates for 8 of the 10 leading causes of death are much higher among African American/black populations as compared to whites (National Center for Health Statistics, 2000). The Institute of Medicine’s (IOM) recent report Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care is one of several documents that chronicle research highlighting racial inequalities in health (Institute of Medicine, 2003). A comprehensive discussion of racial inequalities in health is beyond the scope of this paper, but additional information can be found

in the IOM report and other sources (Williams, 1999; Collins, Hall, and Neuhaus, 1999; LaVeist, 2002; Krieger et al., 1993).

Evidence also supports a strong relationship between socioeconomic position (SEP) and both health status and health care access (Lynch and Kaplan, 2000; Krieger, Williams, and Moss, 1997). Explanations for the relationship between SEP and health tend to focus on how differences in material circumstances and in health-related knowledge and behavior are associated with differences in health outcomes (Townsend and Davidson, 1992; Lynch and Kaplan, 2000). Literature examines the relationships between pathways of SEP and health outcomes by suggesting that single indicators of SEP (i.e., education, occupation, or income and wealth) are interconnected (Krieger, Williams, and Moss, 1997). Specifically, education is an important determinant of future employment opportunities, and occupational status is related to earning potential, income, and future wealth. Income, in turn, can directly influence health by enabling the procurement of material circumstances associated with positive health outcomes. For example, individuals with higher incomes are more likely than their poorer counterparts to have the financial means to afford to live in a clean and safe environment and to afford health care insurance.

But the amount of money associated with an individual is only part of the picture. Knowledge of health-related issues and of where and how to seek health care, both typically associated with increased educational status, are also important factors contributing to health outcomes.

While much discussion has taken place regarding the endogeneity of health and socioeconomic position, evidence supporting a link between poor health and decreased SEP is limited and inconsistent across life stages. Instead, research has found stronger support for the pathway of lower socioeconomic position leading to worse health outcomes (Manor, Matthews, and Power, in press; McDonough and Amick, 2001).

It is also well recognized that race and socioeconomic position are highly interrelated and that both of these factors are strongly associated with health status (Adler et al., 1994; Krieger, Williams, and Moss, 1997; Williams, 1999). However, epidemiologists and other public health researchers in general fail to adequately account for the role of SEP factors in racial inequalities research. Socioeconomic position is often not taken into account at all. National Vital Statistics Reports produced by the Centers for Disease Control and Prevention are just one example of multiple official health reports that present data only on the racial distribution of health outcomes and fail to simultaneously address SEP issues (e.g., Anderson, 2002). While other epidemiologic studies find that racial disparities are attenuated once adjustments have been made to account for socioeconomic position, often such approaches are inadequate. Specifically, use of single indicators of socioeconomic position such as education or category of in-

come does not fully account for economic differences between the groups (Kaufman, Cooper, and McGee, 1997). For example, Schoendorf and colleagues (1992), in a study among college-educated parents, showed that racial disparities in infant mortality rates persisted despite accounting for the SEP indicator of education. Findings from this study show an IMR of 10.2 per 1,000 live births for black infants and a rate of 5.4 per 1,000 live births for white infants. The likelihood of death for a black infant was thus 1.82 times that of a white infant even after controlling for age and parity (Schoendorf et al., 1992). In the article’s conclusion, the authors acknowledge that the persistent gap in IMR may be attributable to economic and social differences between the black and white samples that were not addressed.

There are several problems inherent in the use of single indicators or crude measures to control or adjust for the impact of socioeconomic position that make it less than the ideal approach when studying racial inequalities in health (Kaufman, Cooper, and McGee, 1997). The concept of SEP is complex, and one-dimensional measures (e.g., education) do not fully capture it. In addition, the experiences associated with standard SEP indicators (income, education, and occupation) are not the same among different racial groups. For example, whites have approximately twice the income, four times the net financial assets, and a staggering nine times the net worth of blacks (Oliver and Shapiro, 1997). In addition, black men are more likely than white men to be employed in jobs that expose them to hazards and carcinogens (Robinson, 1984). In the few studies that have adjusted for a fuller range of socioeconomic position indicators, a gap in health between racial and ethnic groups often remains. For example, data from the National Longitudinal Mortality Survey (Muntaner, Sorlie, and O’Campo, 2001) showed unadjusted odds ratios for cardiovascular disease mortality for black men and women compared to whites to be 1.5 for men and 2.0, for women. Adjustment for several socioeconomic indicators, such as education, income, and occupational status, resulted in a reduction of those odds ratios to 1.3 and 1.8. Still, an unexplained gap in cardiovascular disease mortality for blacks compared to whites remained after accounting for numerous indicators of social class.

Adjustment using a single or crude indicator of socioeconomic position results in problems of residual confounding for economic position when comparing health status or health care utilization between racial or ethnic groups. Use of multiple indicators will minimize the degree to which residual confounding is a problem. However, as noted earlier, some reports of health status or health care utilization do not adjust for any SEP indicators (e.g., NCHS reports of U.S. live births or infant deaths). The reporting of health status by only racial or ethnic group gives the erroneous impression that the within-group heterogeneity may be less than the between-

group heterogeneity with respect to multiple factors including socioeconomic position. Under such circumstances, meaningful inferences about racial inequalities in health cannot be made. Therefore, when possible, adjustment with any SEP indicator is preferable to no adjustment at all.

It is apparent that greater efforts need to be made to identify and utilize sound measures of socioeconomic position in racial disparities research. While several solid reviews of SEP indicators have been published in recent years (Liberatos, Link, and Kelsey, 1988; Krieger, Williams, and Moss, 1997; Lynch and Kaplan, 2000; Duncan et al., 2002; Oaks and Rossi, 2003; Berkman and Macintyre, 1997), they offer little insight into which measures are best to use when studying racial disparities in health. For example, the recent review by Oaks and Rossi (2003) provides a history of SEP research, highlights the paucity of research on the measurement of socioeconomic position, and suggests that composite measures may offer the most insight into the complex relationship between SEP and health. However, they do not ultimately offer recommendations about which measures to use in general, or specifically in health disparities research. Although Nazroo’s (2003) discussion of social and economic inequalities as the fundamental causes of racial disparities in health draws attention to our limited understanding of these causal relationships and points out the limited availability of SEP data in large routine administrative data sets, he too fails to offer insight into which existing measures of SEP might best be incorporated in studies of racial health disparities. General SEP research has found that economic measures (i.e., income and wealth) outperform measures of education and occupation (Lynch and Kaplan, 2000; Duncan et al., 2002). Yet, given the racial differences in wealth described above, additional work is necessary before concluding that these findings are also applicable to health disparities research.

In order to adequately address how socioeconomic position is related to racial inequalities in health status, researchers must take significant strides toward identifying and incorporating appropriate measures in their investigations (Kaufman, Cooper, and McGee, 1997). In the following sections, we first summarize our findings from a review of SEP measures included in nationally available administrative and survey sources. We then provide an overview of various SEP indicators. And finally we offer recommendations about which measures to utilize in health disparities research.

Before proceeding, a brief discussion of the term race as used in this paper is necessary. Within public health there is much disagreement about the term. Often, public health researchers mistakenly base their notions of race on the idea that the human species can be separated into distinct human races identifiable through differences in physical characteristics. Setting the standards for the nation, the U.S. Office of Management and Budget (OMB) recommended using the categories of American Indian or

Alaska Native, Asian, Black or African American, Native Hawaiian or Other Pacific Islander, and White. The OMB recognizes that these categories represent a “sociopolitical construct” and “are not anthropologically or scientifically based.” We agree. In this paper, our interpretation and use of the term race refers to socially determined, not biological, characteristics.

We reviewed common health-related administrative and survey sources to determine the availability of socioeconomic position indicators. For each source, we noted the methods of identifying information, the inclusion or absence of geographic information, the frequency of data collection, the SEP indicators included, and the coding of each SEP measure. A summary of our findings can be found in Table C-1. All of the reviewed health-related administrative and survey sources, with the exception of the Medicare Program Data, include a measure of at least one SEP indicator. Consistent with prior reviews (Lynch and Kaplan, 2000; Duncan et al., 2002; Oaks and Rossi, 2003), we found that educational status was the most frequently cited measure of socioeconomic position. Multiple surveillance systems and surveys also include measures of both income and education. For example, the Medicare Current Beneficiary Survey includes indicators of income and education. While not a direct measure of income, source of payment (e.g., Medicare, Blue Cross/Blue Shield, self-pay, etc.) is also collected by several national health surveys such as the National Hospital Discharge Survey. This source-of-payment variable can sometimes serve as a proxy for income status. Only three of the sources reviewed address issues of wealth, and none utilize individual-level indices of socioeconomic position or indicators of social class.

Among the administrative and survey sources presented in Table C-1, the use of multiple SEP indicators varies dramatically. Administrative sources, in general, tend to rely on single or limited indicators of socioeconomic position. For example, the National Vital Statistics System and the Consumer Assessment of Health Plans both rely on education as the sole indicator of socioeconomic position. National surveys, on the other hand, tend to include more than one measure of socioeconomic position. For example, the National Survey of Early Childhood Health includes questions regarding maternal education and employment, and family income. Two prime examples of more comprehensive approaches to determining socioeconomic position can be seen in the Medical Expenditure Panel Survey (MEPS) and the National Health and Nutrition Examination Survey (NHANES). In addition to detailed questions about employment status (e.g., name of employer, type of employer, job title, job duties), the MEPS

also asks about sources of income and the value of assets that the respondent may hold. NHANES respondents are questioned about their education, employment (including type of business/industry and type of work), income, type of health insurance, and homeownership.

For many of the administrative and survey sources it is possible to link individual-level records to area-level SEP indicators (Krieger, Williams, and Moss, 1997). For example, the National Vital Statistics System includes mother’s residential address with the birth record file. This address information can be mapped to specific residential area designations (e.g., census tract) and then assigned area-level indicators of socioeconomic position

using available administrative data such as those from the census. While a few of the databases included in this review do not contain respondent street address information (e.g., National Hospital Discharge Survey), a residential Zip Code is provided for each respondent and can be used in a similar fashion to tie individual records to area-level information.

We describe here the specific indicators of socioeconomic position and how they might be used to increase our understanding of racial inequalities in health. Strengths, weaknesses, and comparisons of the indicators of socioeconomic position are described below. Details about measuring SEP indicators are contained in Table C-1.

For reasons noted earlier, wealth information is probably the most critical data to collect if one is attempting to account for economic differences between racial and ethnic groups. Numerous reports and studies have documented that the greatest differences between racial and ethnic groups are wealth differences and not educational or income differences. Blacks and Hispanics have one-tenth the household net worth of whites (U.S. Bureau of the Census, 1996). Educational attainment does not account for this difference as even among college-educated persons, blacks have 23 percent the household net worth of whites (Oliver and Shapiro, 1995).

Questions about wealth capture information, at a single point in time, about household financial assets (e.g., housing ownership, ownership of rental properties and businesses, investment income, vehicles, etc.). Net financial assets account for the household debt on these assets. Duncan and Petersen (2001, Appendix C) outline recommendations for short and long data collection instruments for wealth and discuss important aspects of ensuring high-quality complete data on wealth and income that should be considered by those collecting this information on surveys. For example, if a respondent answers “don’t know” to a question about housing value, this answer can be followed by other questions to attempt to get the respondent to give an approximation (e.g., “Would the house sell for more or less than $50,000 if it were sold today?”).

Information about wealth is not commonly requested in studies nor collected in administrative data sources because of perceptions and experiences indicating that it is both difficult to collect and subject to non-reporting. However, given that most U.S. residents have cars and homes as their sole assets, a short set of questions might ask only about these two items. (Many U.S. households also have retirement income, but for those

under 65 that income is not available to use.) Duncan and Petersen recommend collecting information on housing in a short set of questions. The utility of this short set will depend on the population being studied. For example, a great majority of higher-income individuals and families are likely to own homes or cars and such information may not differentiate disadvantaged higher-income groups as well as it might in lower-income populations. More detailed questions about wealth tap information on several factors as noted in Table C-1. This level of detail is possible only in surveys and not normally available from administrative sources. However, among those administrative sources that do collect information on income, another question on housing ownership and value might be considered. When possible, to promote a better understanding of the interaction of class and racial inequalities in health, information on wealth should be collected in future surveys and in administrative sources of data. The collection of wealth data, even through a short set of wealth questions, would greatly enhance our ability to understand racial and ethnic inequalities and health.

Income usually refers to wages and salary received either by a person (i.e., individual income) or by all members of a family or household (i.e., household income). Less frequently, a more comprehensive assessment of income from all sources (e.g., salary, government programs, business income, income from rental property, etc.) is collected for individuals or households (Krieger, Williams, and Moss, 1997; Duncan and Petersen, 2001).

While many researchers report that income data are difficult to obtain and suffer from a high proportion of missing information, Duncan and Petersen (2001) discuss ways to promote the reporting of this information (see above discussion regarding wealth). For example, for surveys, Duncan and Petersen (2001) do not recommend having the respondent choose from categories of salary and wages. Rather, a direct question on the dollar amount is recommended. Then, “unfolding scales” and prompts that follow answers of “don’t know” can be used to reduce missing information (Duncan and Petersen, 2001). These options are not always possible to use with administrative sources of data.

A simple set of indicators can reflect total income for either an individual or a household. Household is probably preferable to individual income as it is a reflection of the current economic resources available to the household or family. Information on individual income, on the other hand, because it varies widely within families, may not accurately reflect the

current resources available to that individual (e.g., the income of a respondent who works part-time might appear artificially low if other adults in the household also work. Adjusting household income for the number of members being supported also increases the accuracy of the measure of current economic resources available to that individual. The economic resources available to each family member in a family of two that earns $40,000 annually differ from those available to members of a family of four that earns the same amount.

Poverty is an alternative indicator to income. Poverty can be determined using information on salary and wages and household size. The advantage of using poverty indicators is that, when expressed in relation to the official poverty line, they provide a measure of whether the household has the minimum resources for basic necessities. Although the official poverty measure underestimates the amount a family needs to meet basic necessities, it is standardized and can be compared across studies. While not commonly used, income can alternatively be expressed in terms of other economic metrics such as a living wage, which may better reflect how close or far families are from being able to afford basic necessities. It should also be noted that indicators such as income and poverty show greater fluctuations from year to year than an indicator such as wealth (Stevens, 1999). If a stable indicator of economic standing is desired, wealth may be preferable to income or poverty status.

Education is the most commonly measured indicator of socioeconomic position. Some have argued that it is the best indicator because of its stability over the lifecourse as well as the ease and accuracy for which these data can be collected (Winkleby et al., 1992). However, the fact that education is usually stable once a person has passed early adulthood may be one of its major weaknesses in that it does not capture the volatility that is present in the lives of many, especially the poor (e.g., Stevens, 1999). Moreover, it does not directly reflect the economic resources available to a person or household as income and wealth indicators do. Finally, use of education as the sole indicator of socioeconomic position is likely to yield significant residual confounding given the variability in income and wealth seen between racial and ethnic groups for the same levels of educational attainment (e.g., Oliver and Shapiro, 1995).

Despite these limitations, education is often the only SEP indicator available (e.g., in administrative data sources) and collection of education data is relatively easy and subject to little misclassification. These advantages should be kept in mind when options are very limited in terms of survey or interview length.

Education data can be collected as a continuous variable (i.e., number of years of schooling completed), which can be categorized (see Table C-1) or, in addition, in terms of credentials obtained (i.e., high school diploma, college degree, master’s degree, etc.). The latter is important because credentials are often the means to advancement in the workplace. Yet for purposes of adjustment between racial and ethnic groups, the availability of information on both continuous education and credentials would be ideal.

Stratification indicators, the most commonly used measures of socioeconomic position in the United Stated, are based on Weberian notions of how power, privilege, resources, and prestige give persons differential access to life opportunities (Krieger, Williams, Moss, 1997; Muntaner et al., 2000; Lynch and Kaplan, 2000). While indicators of social class are rare in the U.S. literature, they yield important information about how health inequalities are created by the opposing interests of those in different classes (Krieger, Williams, and Moss, 1997; Muntaner, Eaton, and Diala, 2000; Lynch and Kaplan, 2000). Class struggle—“the struggle between such collectively organized actors over class interests”—is critical to understanding how inequalities between classes are generated (Wright, 1997). There is ample evidence of class struggle all around us in policies like the scaling back of social programs such as welfare, health care coverage, or unemployment benefits, or an increase in jobs without benefits, or a low, poverty-level, minimum wage for the nation. Because of this explicit link to mechanisms by which class processes can create inequalities, these social class indicators yield information that is different from the usual stratification measures (Muntaner, Eaton, and Diala, 2000).

The set of questions developed by Erik Olin Wright (1997) has been the most commonly implemented set of social class measures in the health literature. If our goal is to better understand the reasons for racial and ethnic inequalities in health we may want to consider including and using indicators of social class more frequently in future studies and analytic efforts. Questions on social class indicators could easily be included in surveys, but it is unlikely that this full set of data would be feasibly obtained from administrative sources.

Researchers have combined indicators on individual education, occupation, and income to create indices of socioeconomic position. While these indicators were developed and have been used primarily in sociology, some public health researchers have used them as well. Examples of the latter

include the Hollingshead Index of Social Position (Hollingshead, 1975), Duncan’s Socioeconomic Index (Duncan, 1961), and, for occupational specific indicators, the Nam-Powers Occupational Status Score (Nam and Powers, 1983). Because these indices, especially those that are occupationally based, were created decades ago and combine information from different indicators of socioeconomic position, their utility has been questioned (Liberatos, Link, and Kelsey, 1988; Krieger, Williams, and Moss, 1997). Indices such as those used in Nam-Powers are based on occupational prestige ratings that were devised several decades ago and that therefore may no longer be relevant to the current job market.

The collection of these data is more burdensome than the collection of the separate components alone (e.g., income, education). This should be kept in mind when time or interview length are limited.

Combining different indicators of socioeconomic position both precludes the identification of which factors may matter most and limits the interpretation of the overall index in terms of inequalities between racial and ethnic groups. Therefore, the use of multiple, separate indicators of SEP is recommended over the use of these indices.

Government program participation information, typically collected in administrative sources, is useful for the study of lower-income populations as, for program eligibility, there are often income cutoffs (e.g., Temporary Assistance for Needy Families, Food Stamps, Medicaid), not though for all assistance programs (e.g., Medicare, unemployment benefits). Thus, in the absence of specific information on economic resources, this information might be useful in determining which individuals or families are of low income. The information cannot, however, be used to directly estimate income as eligibility requirements differ by program and often across states.

Health insurance and payment information, like government program participation, can be used to categorize individuals or households as lower or higher income. Some specific types of insurance, primarily Medicaid, are based on low-income eligibility. Therefore, this information might serve as a proxy to indicate income level, with private insurance indicating higher incomes. However, given the ever-changing insurance situation in the United States, these categories do not always clearly correlate with income levels; for example, middle- or high-income individuals or families may be uninsured or self-pay for medical care. But in the absence of other direct income

information, these indicators can be used to suggest relative income levels of a sample.

Recently there has been increasing interest in using area-level indicators of social position (Krieger, Williams, and Moss, 1997; Lynch and Kaplan, 2000). Area-level indicators might be used in place of or in addition to individual data on social position. The advantages and disadvantages of this method have been discussed in depth in the literature (Kaufman, Cooper, and McGee, 1997; Krieger, Williams, and Moss, 1997; Geronimus, Bound, and Neidert, 1996; Mustard et al., 1999). There are several reasons for briefly mentioning this methodology here. First and foremost, given that many administrative sources lack sufficient characterization of social position (e.g., they may have only the educational status of the individual) but may have information on the home address that can be used to link to area-level data, this method may be a means by which individual social position data may be augmented with area-level social position information. Such a link may be particularly important when comparing racial or ethnic differences, as ample data have shown vast differences in the socioeconomic conditions of neighborhoods of different racial and ethnic groups (Jargowsky, 1997; Massey and Denton, 1993). Another reason is that if the goal for one type of comparison between racial and ethnic groups is to adjust as fully as possible for socioeconomic position, then more than one indicator, and indicators at more than one level, may be desirable for this purpose. Kaufman, Cooper, and McGee (1997) warn of the need to use these measures as continuous (e.g., continuous levels of income rather than as a fixed category of, say, greater than $20,000) in order to retain as much information as possible for this very purpose. Table C-1 thus includes several economic-, occupation-, and education-based indicators that can be combined to represent neighborhood socioeconomic position. To create these indicators, the person in the data survey or administrative base must simply provide information on residential address, which can then be used to link to census (or other administrative) sources of information to estimate the indicators.

To fully understand racial inequalities in health, we must begin to do a better job of untangling SEP from racial group membership. Thus far, few studies or reports have adequately accounted for SEP in a way that makes

racial or ethnic groups comparable. Currently, there is a wide range of practices that account for SEP when comparing the health of racial groups. Given that the indicators described in the previous section and in Table C-1 cannot act as proxies for one another, careful thought must go into selection of the single or multiple indicators to be used. For example, one cannot collect information on income and assume that it will serve as a good proxy for wealth or that education can serve as a proxy for income. The correlations of these measures have been examined and found to be moderate at best (Parker, Schoendorf, and Kiely, 1994; Braveman et al., 2001). Moreover, the correlation of socioeconomic position variables differs by race or ethnicity (Braveman et al., 2001). One must carefully identify the reason for the collection and measurement of particular SEP indicators, as described earlier in the paper, and collect information, when possible, on all indicators needed. Here we offer brief guidance as to which indicators might be given priority for collection in research surveys or for administrative data sets. We base these recommendations on the review of administrative data sources and the descriptions of SEP indicators and their advantages and disadvantages described in the previous sections.

We begin by discussing the socioeconomic indicators for existing administrative data sources. It was encouraging to see that most of the administrative data sources we reviewed contain indicators of SEP that can be used when examining racial and ethnic inequalities in health. While education is the most common SEP indicator collected in these administrative data sets, some sources also contained information on critical economic indicators such as income and even wealth. For the data sets that contain only education, additional collection of a more direct indicator of economic standing (e.g., income or wealth) would facilitate an even greater understanding of the nature of the racial inequalities we now often observe.

Administrative data sets with no information on SEP are the real challenge. It is critical that efforts be undertaken to collect SEP information. Where possible, the collection of more than one indicator should be considered. If the ability to do so is limited, then simple questions on education, participation in government programs, or the short set of questions on income might be considered. Another option is to use participant residential address to link census-based measures of SEP for purposes of analysis and reporting; this approach avoids the need to collect additional information in instances where it is just not possible.

In enrollment or service delivery encounters where there are often constraints on collecting more than one indicator of SEP, the collection of any SEP indicator is preferable to no such information. While indicators of economic standing (e.g., wealth and income) are desirable if time for data collection permits, information on client education or government program participation may be easier to collect, is less sensitive, and may require less

time. If address information is available, it can be linked to residential area socioeconomic data.

Given that research surveys often have the most flexibility for including questions on SEP we suggest that, when possible, multiple indicators be collected, with data on wealth receiving the highest priority. This approach of collecting multiple indicators allows for residual confounding to be addressed in the analyses of survey data on racial and ethnic differences in health. Moreover, if analyses are conducted on single groups, the availability of data on multiple indicators facilitates a more complete understanding of socioeconomic variation within a particular racial or ethnic group that may contribute to adverse health outcomes. Because surveys that include multiple indicators of socioeconomic standing will contribute the greatest insights into the magnitude and reasons for racial inequalities in health, placing greater emphasis on appropriate measurement in surveys of SEP is critical to future research efforts on racial and ethnic inequalities in health.

Finally, while the indicators discussed in this paper emphasize measures of SEP at a single point in time (i.e., at the time the survey is completed) it is recommended that, when possible, SEP information be collected for multiple points in time (Lynch and Kaplan, 2000). Childhood SEP—for example, the experience of long-term or multiple spells of poverty during childhood—can have consequences for later health status (Marmot and Wilkinson, 1999). Also, SEP during early to middle adulthood can have effects on health status for the elderly independent of their SEP after retirement. Therefore, where possible, most likely for research surveys, information on SEP for multiple time points across the lifespan should be collected.

The authors would like to acknowledge Adam Allston’s contribution to this work in compiling the table of socioeconomic status indicators available in national surveillance systems and health surveys contained in Table C-1.

Anderson, R.N. 2002 Deaths: Leading causes for 2000. National Vital Statistics Reports 50(16).

Berkman L., and S. Macintyre 1997 The measurement of social class in health studies: Old measures and new formulations. Pp. 51-64 in M. Kogevinas, N. Pearce, and P. Boffetta, eds. Social Inequalities and Cancer. Lyon, France: International Agency for Research on Cancer.

Braveman P., C. Cubbin, K. Marchi S., Egerter, and G. Chavez 2001 Measuring socioeconomic status/position in studies of racial/ethnic disparities: Maternal and infant health. Public Health Reports 116(5):449-463.

Collins, K.S., A. Hall, and T. Neuhaus 1999 U.S. Minority Health: A Chartbook. New York: The Commonwealth Fund.

Duncan G., and E. Petersen 2001 The long and short of asking questions about income, wealth and labor supply. Social Science Research 30:243-263

Duncan, G.J., M.C. Daly, P. McDonough, and D. Williams 2002 Optimal indicators of socioeconomic status for health research. American Journal of Public Health 92(6):1151-1157.

Duncan, O. 1961 A socioeconomic index for all occupations. In W. Reiss, ed., Occupations and Social Status. New York: Free Press.

Geronimus, A., J. Bound, and L. Neidert 1996 On the validity of using census geocode characteristics to proxy individual socioeconomic characteristics. Journal of the American Statistical Association 91:517529.

Hollingshead, A. 1975 Four Factor Index of Social Status. New Haven, CT: Department of Sociology, Yale University.

Hoyert, D.L., M.A. Freedman, D.M. Strobino, and B. Guyer 2000 Annual summary of vital statistics. Pediatrics 108(6):1241-55

Institute of Medicine 2003 UnequalTreatment: Confronting Racial and Ethnic Disparities in Health Care. B.D. Smedley, A.Y. Stith, and A.R. Nelson, eds. Committee onUnderstanding and Eliminating Racial and Ethnic Disparities in Health Care. Boardon Health Sciences Policy. Washington, DC: The National Academies Press.

Jargowsky, P.A. 1997 Poverty and Place: Ghettos, Barrios and the American City. New York: Russell Sage Foundation .

Kaufman, J.S., R.S. Cooper, and D.L. McGee 1997 Socioeconomic status and health in Blacks and Whites: The problem of residual confounding and the resiliency of race. Epidemiology 8:621-628.

Krieger, N. 1999 Embodying inequality: A review of concepts, measures, and methods for studying health consequences of discrimination. International Journal of Health Services 29(2):295-352.

Krieger, N., D.L. Rowley, A.A. Herman, B. Avery, and M.T. Phillips 1993 Racism, sexism, and social class: Implications for studies of health, disease, and well-being. American Journal of Preventive Medicine 9:82-122.

Krieger, N., D.R. Williams, and E. Moss 1997 Measuring social class in U.S. public health research: Concepts, methodologies and guidelines. Annual Review of Public Health 18:341-378.

LaVeist, T., ed. 2002 Race, Ethnicity and Health: A Public Health Reader. San Francisco, CA: Jossey-Bass.

Liberatos, P., B. Link, and J. Kelsey 1988 The measurement of social class in epidemiology. Epidemiologic Reviews 10:87-121.

Lynch, J., and G. Kaplan 2000 Socioeconomic position. In L. Berkman and I. Kawachi, eds., Social Epidemiology. New York: Oxford University Press.

Manor, O., S. Matthews, and C. Power in Health selection: The role of inter- and intra-generational mobility on social press inequalities in health. Social Science and Medicine.

Marmot, M., and R. Wilkinson, eds. 1999 Social Determinants of Health. New York: Oxford University Press.

Massey, D., and N. Denton 1993 American Apartheid: Segregation and the Making of the Underclass. Cambridge, MA: Harvard University Press.

McDonough, P., and B.C. Amick III 2001 The social context of health selection: A longitudinal study of health and employment. Social Science and Medicine 53:135-145.

Muntaner, C., W. Eaton, and C. Diala 2000 Social inequalities in mental health: A review of concepts and underlying assumptions. Health 4:89-113.

Muntaner C., P. Sorlie, and P. O’Campo 2001 Finance, government and production work and cardiovascular mortality among men and women: Findings from the National Longitudinal Mortality Study. Annals of Epidemiology 11(3):194-201.

Mustard, C., S. Derksen, J. Berthelot, and M. Wolfson 1999 Assessing ecologic proxies for household income: A comparison of household and neighborhood level income measures in the study of population and health status. Health and Place 5:157-171.

Nam, C., and M. Powers 1983 The Socioeconomic Approach to Status Measurement with a Guide to Occupational and Socioeconomic Status Scores. Houston: Cap and Gown Press.

National Center for Health Statistics 2000 Health, United States, 1999, with Socioeconomic Status and Health Chartbook. Hyattsville, MD: National Center for Health Statistics.

Nazroo, J.Y. 2003 The structuring of ethnic inequalities in health: Economic position, racial discrimination and racism. American Journal of Public Health 93(2):277-284.

Oaks, J.M., and P.H. Rossi 2003 The measurement of SES in health research: Current practice and steps toward a new approach. Social Science and Medicine 56(4):769-784.

Oliver, M., and T. Shapiro 1995 Black Wealth / White Wealth: A New Perspective in Racial Inequality. New York: Routledge.

Parker J., K. Schoendorf, and J. Kiely 1994 Associations between measures of socioeconomic status and low birth weight, small for gestational age, and premature delivery. United States Annals of Epidemiology 4:271-278

Robinson, J. 1984 Racial inequality and the probability of occupation-related injury or illness. Milbank Quarterly 62:567-590.

Schoendorf, K.C., C.J.R. Hogue, J.C. Kleinman, and D. Rowley 1992 Mortality among infants of black as compared with white college-educated parents. New England Journal of Medicine 326:1522-1526.

Stevens A. 1999 Climbing out of poverty, falling back in: Measuring the persistence of poverty over multiple spells. Journal of Human Resources 34(3):557-588.

1995 Asset Ownership of Households, 1993. Washington, DC: U.S. Government Printing Office.

Townsend, P., and Davidson, N. 1992 The Black report. In P. Townsend, M. Whitehead, and N. Davidson, eds., Inequality in Health: The Black Report and Health Divide. London, England: Penguin.

U.S. Bureau of the Census 1996 Money Income in the United States: 1995. Washington, DC: U.S. Bureau of the Census.

U.S. Department of Health and Human Services 2002 Healthy People 2010. Available: http://web.health.gov.

Williams, D.R. 1999 Race, socioeconomic status, and health: The added effect of racism and discrimination. Annals of New York Academy of Sciences 896:173-188.

Wilson, W.J. 1996 When Work Disappears: The World of the New Urban Poor. New York: Alfred Knopf.

Winkleby, M., D. Jatulis, E. Frank, and S. Fortmann 1992 Socioeconomic status and health: How education, income, and occupation contribute to risk factors for cardiovascular disease. American Journal of Public Health 82(6):816-820.

Wright E. 1997 Class Counts. Cambridge, England: Cambridge University Press.