Appendix F
Collection of Data on Race and Ethnicity by Private-Sector Organizations: Hospitals, Health Plans, and Medical Groups

David Nerenz and Connie Currier*

BACKGROUND

The published literature on racial and ethnic disparities in health, access to health care, and quality of care paints a distressing picture of inequities. Members of minority groups are generally less likely than their non-Hispanic white counterparts to have health insurance coverage, a regular physician, good control of chronic illnesses, access to invasive diagnostic or surgical procedures, adequate pain medications, or appropriate therapy for mental health problems (IOM, 2003). The higher prevalence of chronic conditions like hypertension and diabetes in some minority groups, coupled with problems in access to care for conditions like cancer, heart disease, or stroke, leads to significant disparities in measures of mortality and life expectancy (Satcher, 2001).

Although some of the disparities in broad measures of population health—e.g., life expectancy—may be related to socioeconomic factors such as income, education, and access to health insurance, there is ample evidence of disparities in care provided to individuals with similar jobs, similar health insurance coverage, and similar incomes (Fiscella et al., 2000). There is evidence, for example, of racial and ethnic disparities in the stage of breast cancer at diagnosis (presumed to at least partially reflect screening

*  

David R. Nerenz, Ph.D., and Connie Currier, Dr PH., are project director and assistant professor, respectively, at the Michigan State University Institute for Health Care Studies.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs Appendix F Collection of Data on Race and Ethnicity by Private-Sector Organizations: Hospitals, Health Plans, and Medical Groups David Nerenz and Connie Currier* BACKGROUND The published literature on racial and ethnic disparities in health, access to health care, and quality of care paints a distressing picture of inequities. Members of minority groups are generally less likely than their non-Hispanic white counterparts to have health insurance coverage, a regular physician, good control of chronic illnesses, access to invasive diagnostic or surgical procedures, adequate pain medications, or appropriate therapy for mental health problems (IOM, 2003). The higher prevalence of chronic conditions like hypertension and diabetes in some minority groups, coupled with problems in access to care for conditions like cancer, heart disease, or stroke, leads to significant disparities in measures of mortality and life expectancy (Satcher, 2001). Although some of the disparities in broad measures of population health—e.g., life expectancy—may be related to socioeconomic factors such as income, education, and access to health insurance, there is ample evidence of disparities in care provided to individuals with similar jobs, similar health insurance coverage, and similar incomes (Fiscella et al., 2000). There is evidence, for example, of racial and ethnic disparities in the stage of breast cancer at diagnosis (presumed to at least partially reflect screening *   David R. Nerenz, Ph.D., and Connie Currier, Dr PH., are project director and assistant professor, respectively, at the Michigan State University Institute for Health Care Studies.

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs services) and survival among members of the same managed care plan (Yood, Johnson, and Blount, 1999). When individuals who live in the same general area and share many socioeconomic characteristics other than race or ethnicity receive different patterns of care, explanations for these differences are likely to be found in the dynamics of physician decision making, doctor-patient interaction, individual attitudes and beliefs about illness and treatment, or organizational characteristics of health care systems (Grantmakers in Health, 2000). There have been some studies documenting differences in physician recommendations or referral patterns for cardiac surgery as a function of patient race or gender, even when other characteristics were controlled (Grantmakers in Health, 2002). Studies of disparities in dose of adjuvant chemotherapy for women with breast cancer have also identified individual physician decision making as a key factor (LaVeist, Morgan, and Arthur, 2002). When the causal factors related to disparities cannot be identified so closely with specific individuals, though, there are a number of characteristics of organizations or health care systems that have been linked to either the current existence of disparities or their potential reduction or elimination. These factors include institutional racism (Williams and Rucker, 2000) cultural competence (Cohen and Goode, 1999), or “patient-centeredness (Picker Institute).” As research and health policy attention shifts from the documentation of disparities to the testing of initiatives to reduce disparities, the role of organizations such as hospitals, health plans, and medical groups will be crucial. Each currently plays an important role in quality measurement, quality improvement, and establishing the norms, values, and systems for accountability of medical care. If we follow the recommendation of a recent IOM panel and view disparities as an important quality of care problem (IOM, 2003), then it is important to understand how hospitals, health plans, and medical groups can play a role in reducing disparities. In the domains of quality measurement and quality improvement, these organizations already play important roles, as indicated below. Hospitals: Responsible for assessing and ensuring quality for inpatient care Able to establish and enforce clinical policies and guidelines Able to produce clinician profiles for quality and cost of care Able to grant or rescind privileges Subject to external review for both accreditation and licensing purposes Health plans: Accountable to public and private purchasers for achieving stan-

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs dards of quality in terms of both processes and outcomes of care Subject to formal accreditation processes that include use of standard quality measures Health Plan Employer Data and Information Set (HEDIS) and member survey data Consumer Assessment of Health Plans (CAHPS) Responsible for implementing clinical guidelines and designing disease management programs Able to use a variety of financial and other incentive systems to alter clinician behavior Medical Groups: Able to enforce clinical guidelines and policies Able to select group members on the basis of compatible practice styles Able to maintain medical records systems, registries, and other sorts of data systems used to measure quality of care Able to design and implement compensation systems that reward quality From the perspective of the purchaser, responsibility for assessing and improving quality of care has largely been delegated to health plans and hospitals. For health plans, requirements for National Committee for Quality Assurance accreditation and comparison using HEDIS measures and CAHPS results represent methods by which purchasers attempt to improve quality of care. For hospitals, more recent initiatives by the Leapfrog Group and other purchasers have focused on technologies such as electronic order entry for pharmacies or the “volume-outcome relationship” to improve quality in inpatient settings. Joint Commission on Accreditation of Healthcare Organizations (JCAHO) accreditation requires that a “cultural assessment” be conducted in the context of patient and family education to ensure that social, ethnic, cultural, and emotional factors are considered in providing patient care (JCAHO, 1999). Medical groups have not been as directly engaged as agents for quality improvement, except perhaps in examples like the Buyers Health Care Action Group in the Minneapolis/ St. Paul area, where a “direct contracting” model has been used to try to manage both cost and quality (Buyers Health Care Action Group, 2002). When these types of organizations engage in quality improvement activities, it is generally presumed that they have the necessary clinical, administrative, and patient demographic information to create effective programs and monitor outcomes. Health plans working to improve breast cancer screening rates, for example, are presumed to have information that will enable them to identify women over the age of 50 who either have or have not had a mammogram during a certain time period. Hospitals working to improve use of beta-blockers after acute myocardial infarction are

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs expected to be able to identify patients with a diagnosis of acute MI and to know whether or not beta-blockers were prescribed at discharge. If health plans, hospitals, or medical groups are going to be engaged as active partners in reducing racial and ethnic disparities in care, it is important to know whether they have the essential data on the race or ethnicity of members or patients in order to: (a) assess disparities, (b) identify individuals in target populations for intervention, (c) evaluate the effects of those interventions, and (d) be accountable for reducing or eliminating disparities. The availability of data on race, ethnicity, and primary language is crucial to these organizations’ ability to focus their efforts on quality improvement and reduction or elimination of disparities. Initiatives that seek to expand translation services, community outreach programs, access to care, and quality improvement will be much more beneficial to members of racial, ethnic, or linguistic minority groups if they are guided by information on the groups most affected by disparities and the extent to which specific structural or process variables are responsible for those disparities. From the larger perspective, standardized data collection and reporting on patient race and ethnicity are essential to identify discriminatory practices in the process, structure, and outcomes of care, to demonstrate compliance with civil rights legislation, and to measure achievement and monitor progress toward DHHS Healthy People 2010 goals (IOM, 2003). In this paper, we will attempt to summarize what is known about the extent to which hospitals, health plans, and medical groups collect data on race/ethnicity and how they use that information to either assess or attempt to eliminate disparities in care. When organizations do collect and/or use information, we will attempt to describe the accuracy and completeness of that information and the methods used to obtain it. HOSPITALS In February 2003, a survey was sent to a nationally representative sample of 1,000 hospitals to determine whether they collected data on race and ethnicity, and the factors that went into the decision to collect the data. A total of 262 completed surveys were returned. Seventy-nine percent of respondents reported that they do collect racial and ethnic data about patients. The patient was identified as the primary source of information about race or ethnicity, although a large percentage of respondents reported that the admitting clerk obtained the information by observation. Information was most often collected upon admission. A complete summary of the results of the survey can be found in the survey summary at the end of this appendix.

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs Hospitals have been required to provide care without regard to the race or ethnicity of patients since 1966 with the implementation of regulations based on the 1964 Civil Rights Act (Smith, 1999). Title VI of the act required hospitals to be able to document absence of discriminatory treatment through the submission of an annual survey form (Assurance Form 441) and possible site visits. An Office of Equal Health Opportunity (OEHO) was established in the surgeon general’s office to monitor compliance. The requirement was imposed because there were patterns of blatant segregation in both northern and southern hospitals until then, evidenced in either fully segregated hospitals or segregated units and programs within hospitals. The passage of the Civil Rights Act and subsequent enforcement actions by the OEHO (or credible threats of such actions) served to eliminate the most overt forms of segregation and discrimination in relatively few years, primarily because the subsequent passage of Medicare and Medicaid legislation gave federal officials a powerful financial tool to use to force compliance with Title VI requirements. Hospitals were threatened with ineligibility for Medicare and Medicaid reimbursement if they failed to comply, and even recalcitrant hospitals in the Deep South eventually gave in (Smith, 1999). The requirement to document nondiscriminatory practices did not apparently include a specific requirement to collect data on the race or ethnicity of individual patients. Gross violations of Title VI in the mid-60, such as the presence of racially segregated units or separate dining rooms, did not require the presence of data on individual patients in order to be apparent. In the late 1970s, the Office for Civil Rights and Health Care Financing Administration were still working on the development of hospital surveys that would include information on the race and ethnicity of patients in order to monitor compliance with Hill-Burton Act requirements.1 In the mid-1990s, the uniform hospital claim forms (UB-92 and HCFA-1500) did not include a data field for race or ethnicity in spite of strong efforts by advocacy groups to have this information included. HCFA argued at the time that information on race or ethnicity could be obtained from Social Security enrollment files (ibid). It appears, then, that although many hospitals developed mechanisms for collecting data on the race or ethnicity of patients as a way to document compliance with Title VI, such data collection has not been universally implemented or enforced. 1   The Hill Burton Act, or Hospital Survey and Construction Act, passed in 1946, originally allowed the construction of racially separated hospitals if the planned facility, equitably provided, on the basis of need, facilities, and services of like equality. The provision was later challenged in Simkins v. Moses Cone Memorial Hospital and overridden by Title VII of the Civil Rights Act.

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs There appears to be considerable variation from state to state and region to region in interpretation of current requirements and practices. An attorney at a large hospital in Michigan who had previously worked in the DHHS Office for Civil Rights (OCR) stated that hospitals were technically required to collect data, but that there was little or no incentive for doing so and that there was no active enforcement of the requirement.2 An official at another large hospital in Michigan stated that the hospital did not routinely collect data on race or ethnicity and that he was not aware of a legal requirement. Even if enforcement of a legal requirement is minimal, it appears that large numbers of hospitals do collect data on race or ethnicity and incorporate that information into patient registration systems and discharge abstracts. According to an official at the Michigan Health and Hospital Association, the Michigan Inpatient Database (a collection of discharge abstracts from essentially all general medical/surgical hospitals in the state) includes racial and ethnic data for approximately 75 percent of patient records (Robert Zorn, 2002, personal communication, Sparrow Regional Health System). A study conducted in 1993 in New York found data on race or ethnicity assigned to virtually all patients whose records were reviewed in a study on care for acute myocardial infarction (Blustein, 1994). The accuracy of the data was called into question, though, when records were compared for patients who had two separate admissions in the same study period. Although there was agreement in classification for over 93 percent of the patients with two admissions, kappa statistics for agreement were relatively low for groups other than black or white. The authors of the study observed that the designation of race or ethnicity was generally assigned by an admitting clerk or some other administrative staff person based on his or her own judgment rather than on the basis of patient self-report, and cited a prior study from California making the same observation (California Department of Health, 1990). The Hospital Cost and Utilization Project (HCUP) databases, which have been used for several studies of racial and ethnic disparities in use of procedures (Harris, Andrews, and Elixhauser, 1997), are created from discharge abstract data from hospitals in approximately 30 states. Patient’s race is one of the fields in the HCUP National Inpatient Sample (NIS) database, suggesting strongly that race is an element of the discharge abstract or underlying medical record in most hospitals (at least those hospitals in states that maintain state discharge abstract systems and were included in HCUP-NIS). Race is also an element included by 12 of 18 states 2   Personal communication, Peter Jacobson, J.D, M.P.H., November 15, 2002.

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs included in the State Inpatient Database (SID) in HCUP. Studies of hospital care using HCUP can identify racial and ethnic disparities, but some states cannot be included in the analyses. In one study of disparities in care for patients with AIDS using HCUP, for example, the investigators had to delete Illinois from the analysis because no data on the race or ethnicity of patients were available from Illinois (Hellinger and Fleishman, 2001). Analyses of Medicare claims files have been conducted frequently to identify racial and ethnic disparities in care (McBean and Gornick, 1994). However, the racial and ethnic data in these files should be used with caution. Data on race and ethnicity for these Medicare Provider and Review (MEDPAR) files come from the Medicare Enrollment Database (EDB), populated from the Social Security Administration (SSA) Master Beneficiary Record (MBR), which carries only four racial codes: white, black, other, and unknown. The Medicare EDB is updated using information from another SSA file called the NUMIDENT file that carries seven codes: white, black, other, unknown, Asian, Hispanic, Northern American Indian or Alaska Native. The racial and ethnic categories of the Medicare EDB are updated periodically using data from the SSA NUMIDENT file, and also by mailings to certain ethnic-sounding names once a year, though not on a set schedule.3 Procedures for updating the EDB do not guarantee that all records get updated on a regular basis. Many minority beneficiaries remain misclassified in the Medicare EDB (Arday et al., 2000). Even if some hospitals do not collect routinely data on race or ethnicity for all patients, virtually all hospitals collect such data for special purposes such as birth and death certificates, tumor registries, and reportable diseases like HIV/AIDS. A study of the completeness and validity of information on AIDS cases provided by hospitals to the CDC, for example, found 83 percent concordance between information on race or ethnicity in the original case report and information found subsequently in a medical record review (Klevens, Fleming, and Lee, 2001). An analysis of the accuracy of racial or ethnic categorizations in California hospital birth certificate data found generally complete and valid information when compared against subsequent self-report in the context of a postpartum follow-up survey of mothers, but also noted that birth clerks used a mix of mothers’ self-identification and their own observations as methods for assigning race and ethnicity (Baumeister, Marchi, and Pearl, 2000). Some of these studies are now 10-15 years old, and procedures for the collection of data on race and ethnicity may have changed in the interim because of changes in categories used in the U.S. Census, attention to the 3   Personal communication, Medicare information from Medicarestats@cms.hhs.gov, December 18, 2002.

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs issue of race and ethnicity in the 1998 DHHS initiative on disparities,4 or attention to the collection and usage of personal information in the context of HIPAA (IOM, 2003). HEALTH PLANS In March 2003, the survey was sent to a sample of 158 health plans to better understand the current state of data collection related to race and ethnicity; a total of 38 completed surveys were returned. Although we are reluctant to draw conclusions from such a relatively small sample, it appears that health plans are less likely than hospitals to collect data on enrollees’ race and ethnicity. Those that do, use the information primarily to translate materials, for quality improvement purposes, and to inform disease management programs. Health plans that do not collect data expressed concern about enrollees’ perceptions regarding the need for and use of such data. Both health plans that do and do not collect data are concerned about the lack of a reliable system for data collection.5 A complete summary of the survey results can be found in at the end of this appendix. None of the various types of health plans—health maintenance organizations, preferred provider organizations, indemnity plans, or point-of-service plans—have been required to collect data on the race or ethnicity of their members as part of any federal legislation. In fact, most of the legal issues related to insurance companies’ collection and use of data on race and ethnicity have been based on concerns about how such information might be used to adversely affect services offered to members of minority groups. Concerns about insurance “redlining” and charging higher premiums to members of minority groups have led several states to either pass laws or develop language in state insurance regulations limiting or prohibiting companies’ ability to ask questions about race or ethnicity. These concerns, which typically arose in insurance areas other than health insurance, led many health plans surveyed in the context of a 1999 DHHS conference on health plans’ use of data on race and ethnicity to express the belief that it was illegal for them to collect such data (Bierman et al., 2002). Subsequent reviews of both federal law (Perot and Youdelman, 2001) and applicable state laws (Berry et al., 2001) have shown that there are no federal laws or regulations barring health plans from collecting data on the race or ethnicity of their members, and that only four states have 4   U.S. Department of Health and Human Services. The Initiative to Eliminate Racial and Ethnic Disparities in Health. http://raceandhealth.hhs.gov. Accessed October 15, 2002. 5   Personal communication, Deborah Wheeler, deputy director, Quality Initiatives and Industry Standards and Teresa Chovan, director, Policy Research, American Association of Health Plans, 04/23/03.

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs laws that can be interpreted as prohibiting such data collection. In those four states, the specific language of the laws typically addresses the collection of data on race or ethnicity at the time of application for insurance, and does not address health plans’ ability to collect data at a later time (for example, in the context of a health risk appraisal for new members). An informal survey of health plan representatives invited to attend the 1999 DHHS meeting, and a concurrent telephone and e-mail survey of large not-for-profit health plans involved in the HMO Research Network, indicated that essentially no plans routinely collected data on the race or ethnicity of their members at that time. Some plans had done special research projects or Quality Improvement initiatives that had involved collection of that information, but no plans reported the routine collection of racial or ethnic information as part of the regular membership process. In addition to the perceived legal barriers, plans cited concerns about potential adverse reactions in minority communities, or the risk of adverse publicity associated with finding disparities in care, as reasons for not collecting the information. Virtually all of these plans were organized as not-for-profit HMOs, but there is no reason to believe that the policies and practices of for-profit HMOs, PPOs, or other forms of managed care are fundamentally different. Subsequent discussions with the National Committee for Quality Assurance (NCQA) indicated that, because of the lack of consistent data on race and ethnicity available to managed-care plans, NCQA did not require collection of these data as part of its accreditation process. Later communications with Dr. Greg Pawlson of NCQA indicated that the committee is exploring a variety of approaches to increasing the reporting of data on race and ethnicity in the context of quality assessment. A recent demonstration project conducted with 15 managed-care plans from different parts of the country showed that the health plans were able to obtain data on the race or ethnicity of their members through a variety of mechanisms, although none had data routinely available at the start of the project. Plans were able to obtain these data from state Medicaid program enrollment files, from medical records and patient registration systems of contracting medical groups, from self-report items in member surveys, and through the use of surname-recognition software that could distinguish Hispanic from non-Hispanic plan members (Nerenz, Gunter, and Garcia, 2002). Plans were then able to link the information to procedures for producing HEDIS and CAHPS reports in order to stratify those reports by race and ethnicity. MEDICAL GROUPS In early March 2003, a survey was sent to 250 medical groups to gather information about current practices and policies regarding the collection of

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs data on race and ethnicity; a total of 83 valid surveys were returned. Like health plans, medical groups are less likely than hospitals to collect data on patients’ race and ethnicity. Those groups that do collect data do so primarily for internal quality improvement or disease management purposes rather than because of an external reporting requirement. Groups that do not collect data believe that it is either unnecessary or potentially troubling to patients, or that it cannot be done reliably. A complete summary of survey results can be found at the end of this appendix. In the 1964 Civil Rights Act and subsequent implementing regulations (Smith, 1999), physician offices and group practices were exempted from any requirements to collect data on patient race or ethnicity or to document nondiscriminatory practices. Therefore very little is known about the current policies and practices of medical groups. Some multispecialty group practices that have been closely affiliated with hospitals have developed procedures for collecting data on race and ethnicity as part of the patient registration process. Those data have been used for studies of racial and ethnic disparities in patterns of care and disease outcomes. For example, the Henry Ford Medical Group in Detroit has been able to use data on race and ethnicity in its administrative databases to support studies of breast cancer survival rates and patterns and outcomes of care for patients with diabetes. CONCLUSIONS Collection of data on race and ethnicity seems to be far more common, routine, and complete in hospitals than in health plans or medical groups. We presume that the main reason for this difference is the legal requirements imposed on hospitals by the 1964 Civil Rights Act, even if the enforcement of those requirements is not very stringent today and even if the original requirements did not specifically include collection of data on the race or ethnicity of individual patients. State-level requirements for collection and reporting of data seem to be an important driving force for those hospitals that do collect racial and ethnic data. The regulations associated with the Health Insurance Portability and Accountability Act (HIPAA) will have an effect on the collection and use of demographic information, including data on race and ethnicity. It is not clear, though, whether HIPAA will serve to promote or inhibit the collection and use of such data overall. On one hand, HIPAA has made health care organizations extremely cautious about their policies for collection, communication, and use of all forms of personally identifiable health information. This atmosphere of caution will probably inhibit organizations that might otherwise move toward collection and use of data to reduce

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs racial, ethnic, or SES disparities in care. On the other hand, the protections built into HIPAA may have the effect of reassuring individuals and community groups that information provided will be used for legitimate health care and quality improvement purposes and will not be used inappropriately. Until some years of experience with HIPAA regulations have accumulated, though, it will be difficult to say with any certainty how these competing forces will balance out. Key questions remain about the procedures used to collect and categorize information on race and ethnicity, and about the extent to which practices in hospitals or other organizations conform to current U.S. Census Bureau standards and the recommendations of expert panels and advisory groups. Specifically, it will be important to determine: the extent to which categories and procedures used to define race and ethnicity conform to categories and procedures used in the 2000 U.S. census or to the recommendations made in OMB Directive #15 (Ford et al., 2002; Friedman et al., 2000); the extent to which local variations from those categories and procedures (e.g., “fine-grained” categorization of Hispanic or Asian groups) are either used at all or used in ways that would allow “roll-up” to broader U.S. census categories; the extent to which racial and ethnic information is obtained on the basis of patient or health plan member self-report (preferred) versus on some other basis such as visual observation on the part of a registration clerk; the extent to which information on race and ethnicity is complete and accurate as assessed by agreement with other data sources; and the extent to which information on race and ethnicity is used in the context of quality improvement and/or disparity reduction initiatives. These issues of “process” for the collection of data on race and ethnicity (and other demographic factors such as SES) are extremely important if health care organizations are going to work together in any sort of collaborative fashion on disparities, if research is going to be conducted on disparity reduction initiatives, and if governmental agencies and accrediting bodies are going to effectively monitor and encourage those initiatives. A single organization could conceivably use any method it wants to collect, categorize, and use demographic information, as long as it does so in a way that fits local circumstances and is consistent over time. Comparison of data across organizations, though, requires broad adherence to standards for methods of collecting and categorizing data. It is possible to have standardization and flexibility to adapt to local circumstances (e.g., multiple sub-

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs nonexistent. Health plans and medical groups are not legally required to collect racial or ethnic data, some states prohibit data collection, and most have no formal process for doing so. Data collected for special purposes, though (e.g., CAHPS data for health plans; disease registries for medical groups) can be used effectively for initiatives aimed at reducing disparities. ACKNOWLEDGMENTS Several surveys were conducted in preparation for this paper. These surveys obtained information on data collection practices of hospitals, health insurance plans, and medical group practices. Romana Hasnain-Wynia, Ph.D., of the Health Research Education Trust coauthored the survey of American Hospital Association members. Julia Sanderson-Autin, R.N., of the American Medical Group Association coauthored the survey of members of that organization. REFERENCES Arday, S.L., D.R. Arday, S. Monroe, and J. Zhang 2000 HCFA’s racial and ethnic data: Current accuracy and recent improvements. Health Care Financing Review 21:107-116. Baumeister, L., K. Marchi, and M. Pearl 2000 The validity of information on “race” and “Hispanic ethnicity” in California birth certificate data. Health Services Research 35:869-883. Berry, E., S. Hitove, J. Perkins, D. Wong, and V. Woo 2001 Assessment of State Laws, Regulations, and Practices Affecting the Collection and Reporting of Racial and Ethnic Data by Health Insurers and Managed Care Plans. Presented at the Annual Meeting of the American Association of Health Plans, Washington, DC. Bierman, A.S., N. Lurie, K. Scott Collins, and J.M. Eisenberg 2002 Addressing racial and ethnic barriers to effective health care: The need for better data. Health Affairs May/June(3):91-102 Blustein, J. 1994 The reliability of racial classifications in hospital discharge abstract data. American Journal of Public Health 84:1018-1021. Buyers Health Care Action Group 2002 A health care trailblazer says Harvard Business Review. Buyers Health Care Action Group Newsletter 1(2):September. California Department of Health 1990 Report of Results of the OSHPD Reabstracting Project. Sacramento, CA: California Department of Health, Office of Statewide Planning and Development. Cohen, E., and T.D. Goode 1999 Policy Brief 1: Rationale for Cultural Competence in Primary Health Care. Washington, DC: National Center for Cultural Competence, Georgetown University Child Development Center. Fiscella, K., P. Franks, M.R. Gold, and C.M. Clancy 2000 Inequality in quality: Addressing socioeconomic, racial, and ethnic disparities in health care. Journal of the American Medical Association 283:2579-2584.

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs Ford, M.E., D.D. Hill, D. R. Nerenz, M. Hornbrook, J. Zapka, R. Meenan, S. Greene, and C.C. Johnson 2002 Categorizing race and ethnicity in the HMO Cancer Research Network. Ethnicity and Disease 12:135-140. Friedman, D.J., B.B. Cohen, A.R. Averbach, and J.M. Norton 2000 Race/ethnicity and OMB Directive 15: Implications for state public health practice. American Journal of Public Health 90:1714-1719. Grantmakers in Health 2002 Racia/lEthnic Differences in Cardiac Care: The Weight of the Evidence. Washington, DC: Henry J. Kaiser Family Foundation. 2000 Strategies for Reducing Racial and Ethnic Disparities in Health. Washington, DC: Grantmakers in Health. Harris, D.R., R. Andrews, and A. Elixhauser 1997 Racial and gender differences in use of procedures for black and white hospitalized adults. Ethnicity & Disease 7:91-105. Hellinger, F.J., and J.A. Fleishman 2001 Location, race, and hospital care for AIDS patients: An analysis of 10 states. Inquiry 38:319-330. Institute of Medicine 2003 Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care. B.D. Smedley, A.Y. Stith, and A.R. Nelson, editors. Committee on Understanding and Eliminating Racial and Ethnic Disparities in Health Care. Board on Health Sciences Policy. Washington, DC: The National Academies Press. Joint Commission on Accreditation of Healthcare Organizations 1999 Using Performance Measurement to Improve Outcomes in Behavioral Health Care. Oakbrook Terrace, IL: Joint Commission on Accreditation of Healthcare Organizations. Klevens, R.M., P.L. Fleming, and J. Li 2001 The completeness, validity, and timeliness of AIDS surveillance data. Annals of Epidemiology 11:443-449. LaVeist, T.A., A. Morgan, and M. Arthur 2002 Physician referral patterns and race differences in receipt of coronary angiography. Health Services Research 37:949-962. McBean, A.M., and M. Gornick 1994 Differences by race in the rates of procedures performed in hospitals for Medicare beneficiaries. Health Care Financing Review 15:77-85. Nerenz, D.R., M. Gunter, and M. Garcia 2002 Quality of Care for Underserved Populations Developing a Health Plan Report Card on Quality of Care for Minority Populations. New York: The Commonwealth Fund. Perot, R.T., and M. Youdelman 2001 Racial, Ethnic, and Primary Language Data Collection in the Health Care System: An Assessment of Federal Policies and Practices. New York: The Commonwealth Fund. Satcher, D. 2001 Our commitment to eliminate racial and ethnic health disparities. Yale Journal of Health Policy, Law, and Ethics (1):1-14. Smith, D.B. 1999 Health Care Divided: Race and Healing a Nation. Ann Arbor, MI: University of Michigan Press.

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs Williams, D.R., and T.D. Rucker 2000 Understanding and addressing racial disparities in health care. Health Care Financing Review 21:75-90 Yood, M.U., C.C. Johnson, and A. Blount 1999 Race and differences in breast cancer survival in a managed care population. Journal of the National Cancer Institute 91:1487-1491. SUMMARY OF SURVEY OF AMERICAN HOSPITAL ASSOCIATION HOSPITALS In 2003, the Institute of Medicine released the report Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care, which highlighted the importance of collecting patient data by race, ethnicity, and primary language. The report noted that if such data were collected, researchers, policy makers, and clinicians could disentangle factors associated with health care disparities, facilitate monitoring of performance, ensure accountability to health plan members and health care purchasers, improve patient choice, and identify discriminatory practices. The growing body of evidence documenting disparities in health and health care underscores the importance of such data collection. Fortunately, a number of public and private organizations, including the American Association of Heath Plans, are beginning to consider how to incorporate minority data collection into overall quality measurement initiatives. Currently, information about racial and ethnic data collection in hospitals is very limited. We know that overall the reasons for disparities are poorly understood and that there are significant gaps in our understanding. One critical problem in understanding these reasons is the lack of uniform data by race, ethnicity, and primary language at the hospital and health system level. Currently, the methods for data collection by race, ethnicity, and primary language are inconsistent and incompatible across most hospitals and health systems. The Health Research and Educational Trust (HRET), an affiliate of the American Hospital Association, has a track record of working with hospitals in diverse communities across the country. HRET is currently leading a research project with support from the Commonwealth Fund, to develop a Uniform Framework for Collecting Race, Ethnicity, and Primary Language Data in Hospitals. As part of this project, HRET is working closely with a consortium of six hospitals and health systems. In an effort to better understand the current state of data collection in hospitals and to inform the National Academies Panel on DHHS Collection of Race and Ethnicity Data, in February 2003, a survey on the collection of such data was developed collaboratively by the Michigan State University Institute for Health Care Studies and the Health Research and Educational Trust. The survey was sent to a nationally representative sample of 1,000

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs hospitals and was designed to determine whether hospitals collected data on race and ethnicity and what factors go into the decision to collect the data. A total of 262 completed surveys have been returned to date, and we provide a preliminary analysis based on these responses. Survey Content Respondents were asked to indicate who provides information about patients’ race and ethnicity and when it is collected for up to three units or clinics that they identified within the organization. After reviewing the units listed by respondents, four categories of units were created: (1) admitting/ registration, (2) emergency department, (3) outpatient/specialty clinics, and (4) hospital (general). Respondents were asked to “check all that apply” for the question, “Who provides information about the patient’s race or ethnicity?” using the following categories: Patient self-identifies Caretaker/guardian provides information Admitting clerk obtains information from patient Admitting clerk provides information based on observation Health care provider obtains information from patient Health care provider provides information based on observation Don’t know Respondents were asked, “When is the information collected?” and again to “check all that apply” using the following categories: Upon admission At discharge At first visit/new patient registration Included in health care provider’s discharge notes/medical record Don’t know Findings The majority of hospitals (79 percent) reported that they collect racial and ethnic data about patients. Fifty-seven percent of those respondents indicated that more than one unit or clinic within the hospital collected data. The patient was identified as the primary source of information on race or ethnicity for all categories of units. Either an admitting clerk obtains the information from the patient and completes a form or inputs the informa-

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs tion into a computer (67 percent of responses among hospitals collecting data at all, n = 206), or the patient self-identifies by completing a form him or herself (65 percent). In the emergency department and admitting/registration, the information was more often obtained by an admitting clerk. In outpatient/specialty clinics and hospitals (general), the patient was more likely to self-identify. Across all units, 53 percent of respondents reported that information was provided by a caretaker or guardian. Information was more likely to be obtained from a caretaker or guardian in a hospital (general) or emergency department than in other units. The admitting clerk frequently obtained information by observation of the patient’s race or ethnicity (51 percent), though this occurred most often in the emergency department. Information was most often collected upon admission for all units (checked 85 percent of the time), or at first visit/new registration (59 percent). It was collected less often through health care provider notes or the medical record (11 percent), and rarely at discharge (2 percent). These percentages were fairly consistent across all units. Racial and ethnic information is also collected during the hospital stay, upon subscribing to an HMO, during preadmission screening, at an initial intake assessment, when obtaining a birth certificate, and when making an initial appointment. A referring facility may also provide data on a patient’s race and ethnicity. Hospitals were asked why they established policies/practices to collect data on patient race and ethnicity and were given the option to check all responses that applied. The largest single set of respondents reported that their hospital established policies to collect data on patient race and ethnicity because it was required by law or regulation (42 percent). Twenty-three percent reported collecting the data for quality improvement, 19 percent felt it was important for community relations, and 12 percent reported collecting it because it helped target marketing efforts. Seventeen percent indicated other reasons for collecting these data such as: it is required by the state; it is included as basic demographic information for medical records or a hospital database or cancer registry; it is required by the state hospital association; and it is used for teaching and for conducting research on best practices, trends, and preventive care. Eighty-six percent of respondents indicated that they provide specific categories for patients or guardians to check off when data on race and ethnicity are collected, and 14 percent of the respondents reported collecting the information using a “fill in the blank” open question. For hospitals that reported using specific categories, respondents were asked which categories they used from the minimum racial and ethnic classifications used by the Census Bureau (American Indian and Alaska Native, Asian, Black or African American, Native Hawaiian and Other Pacific Islander, and White), and other racial and ethnic categories that were reported in the profile of

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs general U.S. demographic characteristics from the 2000 census. Again, respondents were asked to “check all that apply.” Respondents were also given the option to indicate any additional categories they used that were not included among those used on the U.S. census. Respondents added Cambodian, Czech, Hindu, Hmong, Laotian, Middle Eastern, Persian, Polish, Portuguese, Russian, Thai, and Ukrainian. None of the broader categories used to specify race or ethnicity—i.e., white, black, Hispanic, Asian, Native American and Alaska Native—were used by more than 95 percent of the respondents, and a number of the smaller, more “fine-grained” categories were used by 4 percent or more of the respondents. These two observations suggest that hospitals do some significant tailoring of the standard U.S. census categories to adjust to local circumstances. Specific racial and ethnic categories and the percentage of hospitals that used them are listed below. Caucasian/White 95% African American/Black 94% Spanish/Hispanic/Latino 81% American Indian 77% Asian 77% Alaska Native 26% Other Pacific Islander 24% Mexican, Mexican American, Chicano 11% Native Hawaiian 9% Asian Indian 8% Chinese 8% Filipino 7% Japanese 7% Puerto Rican 5% Cuban 4% Vietnamese 4% Samoan 4% Korean 4% Guamanian or Chamorro 1% Hospitals were asked in what percentage of cases data on race and ethnicity were missing or unavailable. Respondents gave the widest possible range of responses, from 0 to 100 percent. The 100 percent figure presumably represents data that are not collected in this area at all, thus indicating 100 percent missing; the “0 percent” figure seems too good to be true, if in fact this indicates that all data fields for race and ethnicity are always filled. Another more likely interpretation could be that in that hospital there is at

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs least some entry in the data field for race or ethnicity for every patient, regardless of whether the entry is “unknown” or “other.” Eighty-nine percent of the respondents reported that race and ethnicity data are stored in an electronic database in their hospital. Hospitals were asked who may obtain access to the data and how the hospital uses the data that are collected. Again, a number of possible answers were provided, and respondents were asked to check all that applied. Seventy-nine percent of respondents reported that hospital employees had access to the data, followed by health care providers (41 percent), researchers (15 percent), grantees/contractors (8 percent), and the public 3 percent. Five percent of respondents said they did not know who had access to the data. Hospitals were asked how data on race and ethnicity are used. Responses indicated that they used for a variety of internal purposes, including ensuring the availability of interpreter services (36 percent), quality improvement or disease management programs (36 percent), program/benefit design (17 percent), marketing (13 percent), actuarial purposes (2 percent), and underwriting (1 percent). Data are also shared with federal, state, and local governmental agencies including state health departments (36 percent), Medicare (27 percent), Medicaid (26 percent), local health departments (16 percent), and the Veterans Administration (10 percent). Nongovernmental agencies/organizations are also given access to the data, including accrediting bodies (21 percent), community groups (5 percent), and purchasers (1 percent). A small number of respondents indicated that the data are collected but are not used. The majority of respondents, (73 percent) reported that policies regarding collection of data on race and ethnicity were not currently undergoing revision. Six percent said the policies were being revised and 16 percent reported that they did not know. The main revision to policies related to increasing the number of categories patients had to choose from when self-identifying race and ethnicity. Seventy-two percent of the respondents that collect racial or ethnic data did not see any drawbacks to collecting such data. Drawbacks that were reported most often included discomfort on the part of the registrar or admitting clerk asking the patient for the information; problems associated with the accuracy of the data collected; a sense that patients might be insulted or offended, or resist answering questions about their race and ethnicity; patients often don’t “fit” the categories that are given; and a fear that data may not be kept confidential. Also mentioned were the possibility that collecting data on race and ethnicity might be used to profile patients and discriminate in the provision of care. In the case of primary language, we simply asked whether hospitals collected data on a patient’s primary language, but did not include detailed follow-up questions in order to keep the length of the survey reasonable.

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs Forty percent of hospitals collect data on patients’ primary language, 52 percent do not, 3 percent of respondents reported that they did not know, and 5 percent did not respond to the question. Hospitals that did not collect data on race and ethnicity (n = 54) were asked to give reasons why. Sixty-one percent stated that it was unnecessary to collect data on patients’ race and ethnicity, 18 percent felt there was no reliable system for data collection, 17 percent said there was no good classification system for race or ethnicity, 7 percent said the data were too costly to maintain, 7 percent said the data would be unreliable, 5 percent said it was legally allowed but not authorized by the hospital, and 5 percent stated it was prohibited by law or external regulation. Only one of the 54 hospitals not currently collecting racial or ethnic data indicated it was engaged in efforts to change its policies related to the collection of these data for the purposes of determining patients’ primary language. Respondents from hospitals that were not changing their existing policies stated there was no need to collect data on race and ethnicity because it was unnecessary, it was not a priority or a concern, the patient population was over 95 percent Caucasian, and one stated it was “against the law.” Hospitals that did not collect data on race or ethnicity were also asked whether they saw any drawbacks to collecting these data; of these hospitals, 44 percent answered “no” and 56 percent answered “yes.” Respondents stated that the time and resources involved in collecting and managing the data would be a barrier to its collection. One thought it was an invasion of privacy. It appears there is concern that providers will use the information to discriminate in the provision of care. Whether this occurs or not, there is concern that patients will perceive that care will be different based on their race or ethnicity if the information is provided. One respondent voiced concern that knowledge of a patient’s race and ethnicity would lead to “segmenting service delivery, discrimination, and multiple standards of care.” Another wrote, “some people feel these questions signify that they will be treated differently from other patients.” Others took an almost defensive stance, questioning the need for such data and stating that “all of their patients are treated the same” and asking “does it make a clinical difference?” It is interesting to note the differences and similarities in the drawbacks identified by hospitals that do and do not collect data on race and ethnicity. Hospitals that did not collect data indicated that the time and resources needed to collect and manage the data were a barrier to collection, whereas hospitals that collect the data did not. Both expressed concern about the possibility that discrimination would result in the provision of care, although hospitals that do not collect the data mentioned it considerably more often than hospitals that did. The drawbacks mentioned most fre-

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs quently by hospitals that do collect data were related to staff and patients feeling uncomfortable or offended by questions about race and ethnicity, and concerns about the accuracy of the data, particularly for individuals who do not “fit” the categories provided. SUMMARY OF SURVEY OF AAHP MEMBER HEALTH PLANS In February 2003, a survey on the collection of racial and ethnic data was developed collaboratively by the MSU Institute for Health Care Studies and the American Association of Health Plans. In March 2003, the survey was sent to a sample of 158 health plans based on the sample methodology described below. A total of 38 completed surveys were returned. Sample Methodology The sample frame was a database from The InterStudy Competitive Edge, Part I: HMO Directory, Version 12.1 (using data as of July 1, 2001), consisting of 500 HMOs. This source was chosen because of its reliability in reporting both enrollment information and types of products offered. From the HMOs listed in the database, health plans were selected that were AAHP members as of February 1, 2003. From this sublist, AAHP member health plans that had recently been included in a sample for a similar project were removed. The final sample consisted of 170 health plans. In all, 12 plans were deemed ineligible. The remaining 158 health plans represented 59 million enrollees in HMO, POS, and PPO plans. The final sample of health plans responding to the survey represented approximately 16.3 million enrollees from the various health plan products and provided information about 69 commercial, Medicaid, and Medicare health plan products. Findings Of the 38 health plans that responded to the survey, 13, or 34 percent voluntarily collected data on the race and ethnicity of their health plan members and 23, or 61 percent did not at this time. Two other respondents either did not know or did not respond to the question. Of the 13 health plans that reported voluntarily collecting data, the following reasons were provided for establishing policies or practices to collect such data: six plans reported that they collected data in order to design translation materials or to improve quality, another five plans indicated that they recognized the benefits of data collection or that they use the data for disease management. Other reasons given were that data collection is required by law or regulation (3), the data were used to screen high-risk

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs populations (3), the data were used to help marketing efforts (3), and data collection was permitted by statute (1). Data collection primarily occurred upon health plan enrollment. Fourteen responses indicated that data on race and ethnicity were voluntarily collected from the state Medicaid or federal Medicare enrollment files (9), directly from the enrollee (4), or by the caretaker or guardian (1). Among the responses received from health plans that did not voluntarily collect data, eight stated that data collection was unnecessary, seven stated that it was legally allowed but not authorized by the health plan, six stated that either they had concerns about the enrollee’s reactions, or there was no good or reliable system for collection, two stated that either it was prohibited by law or external regulation or they did not know, and one response indicated that either the data would be unreliable, too costly to collect and maintain, or there was no good or reliable classification system for race and ethnicity. Fifteen health plans stated that there were drawbacks to collecting these data. Specific issues identified were concerns about enrollees’ perceptions, there was no good or reliable system for data collection, the data collection was legally allowed but not authorized by the plan, data would be unreliable, too costly to collect and maintain, or there was no good or reliable system to classify race and ethnicity. SUMMARY OF SURVEY OF AMGA-MEMBER MEDICAL GROUPS In early March 2003, an e-mail survey was sent to a designated contact person in each of the 250 member organizations of the American Medical Group Association. The e-mail was sent from the AMGA offices in Alexandria, Virginia, and the cover note described the purpose of the survey as gathering information about current practices and policies with regard to collection of data on race and ethnicity. An e-mail response option was provided; all responses were returned to the AMGA office and automatically entered into a database. A total of 83 valid surveys were returned, and another seven respondents replied by e-mail that they did not collect the information. Findings In response to the opening question about whether the group collected data on the race or ethnicity of patients, 21 of the 83 respondents (25 percent) answered “yes.” All but one of the others responded “no” and one was “not sure.” The groups that did not collect data gave a variety of reasons for not doing so. The most frequently chosen reason for not collecting data was “concerns about patient reactions” (26/62 or 42 percent of those not col-

OCR for page 249
Eliminating Health Disparities: Measurement and Data Needs lecting data). Other reasons included “unnecessary” (37 percent), “no good/ reliable system for collection” (25 percent), and “data would be unreliable” (18 percent). Only two respondents felt that data collection was prohibited by law or external regulation. (Percentages add to more than 100 percent because respondents could choose more than one option.) Those groups that did report collecting data on race or ethnicity did so mainly for internal purposes rather than for external reporting. Fifteen of the 21 groups collecting data on race or ethnicity said that they used the data for internal purposes. Ten of the 15 mentioned using the data for quality improvement or disease management purposes. Six of the 21 groups said that they shared data with federal, state, or local government agencies—primarily state or local health departments. In most instances, the patient was the source of data on race or ethnicity. Seventeen of the 21 medical groups chose “patient self-identifies” in response to a question about who provides data. Caregivers or health care providers were much less frequently mentioned as the source of data. Data were collected most often at the first visit or new patient registration (15 out of 21 groups), but in eight groups the provider’s note in the medical record was identified as the point at which data were collected. (Again, groups had the option of choosing more than one answer.) Summary Like health plans, medical groups are less likely than hospitals to collect data on patients’ race and ethnicity. Those groups that do collect data do so primarily for internal quality improvement or disease management purposes rather than because of an external reporting requirement. Groups that do not collect data feel that it is either unnecessary or potentially troubling to patients, or that it cannot be done reliably.