This Article Extract Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when Comments are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via ISI Web of Science (15) Citing Articles via Google Scholar Google Scholar Articles by Nerenz, D. R. Articles by Gunter, M. Search for Related Content PubMed PubMed Citation Articles by Nerenz, D. R. Articles by Gunter, M. Related Collections Health Philanthropy Managed Care Minority Health Quality Of Care

GrantWatch

SPECIAL REPORT

Eliminating Racial/Ethnic Disparities In Health Care: Can Health Plans Generate Reports?

Health Plans, Quality Improvement, And Disparities

 
Most health plans are involved in some form of quality measurement and improvement. If a health plan improves its quality of care in general, the care of minority patients may improve also. However, a set of performance measures applied to all patients may mask or miss key disparities in quality-of-care for minorities. These considerations have led to calls for stratified analysis and reporting of quality-of-care information.⁴

Although plans could conceivably report many or all of the Health Plan Employer Data and Information Set (HEDIS) measures separately by race and ethnicity, essentially none do so. The National Committee for Quality Assurance (NCQA) does not require this kind of reporting, but the NCQA is going to recommend a change to its standards that would encourage health plans to work with their physicians to incorporate information on race/ethnicity that would allow analysis to look for disparities in health care.⁵

A key barrier to the creation and use of quality-of-care reports for racial or ethnic minority groups is the absence of data on race or ethnicity either in plan databases or in the claims and encounter databases maintained by providers in the plans’ delivery networks.

The Commonwealth Fund-HRSA Project

Top Health Plans, Quality... The Commonwealth Fund-HRSA... Conclusions NOTES

 
To address these concerns, the Commonwealth Fund and the Health Resources and Services Administration (HRSA) are supporting the development of a report card to assess quality at the health plan level for members of various racial/ethnic minority groups. The project’s goal is to determine whether health plans can obtain data on the race/ethnicity of their members through one or more possible methods and whether those data can be used to generate reports on quality of care stratified by race/ethnicity. The project is a collaborative effort by Michigan State University, the Henry Ford Health System, the Lovelace Clinic Foundation, the University of Texas School of Public Health, and thirteen health plans.

As part of the project, two groups of community leaders—one African American and one Hispanic—met in the fall of 1998 (and again in the fall of 1999) to make recommendations about the format and content of the report card and the underlying processes of data collection. A third group of community leaders representing Asian and Pacific Islander groups met in January 2001. Administrative staff of participating health plans were asked to identify leaders of community organizations or others who could represent the views of "lay users" of quality-of-care reports in specific racial/ethnic communities. Each group had approximately twelve members who were from a mix of backgrounds, including community activists, business owners, attorneys, and union officers. A technical expert panel made up of representatives from the NCQA, the Foundation for Accountability (FACCT), the Picker Institute, and academic research centers also advised on defining, collecting, analyzing, and reporting on the data.

Open-ended discussion was followed by more formal priority setting and ranking exercises to identify the most important quality-of-care concepts (such as member satisfaction, provision of preventive services, and cultural/linguistic competence); specific quality measures within those concepts; high-priority clinical conditions; and preferred ways to get data on race/ethnicity. All groups consistently recommended that plans should collect data on the race and ethnicity of their members, assuming that the data will be used for public reporting and internal quality improvement purposes and not for any illegal or unethical discriminatory purposes.

The community-leader groups and panel of technical experts did not agree on all specific questions regarding the design and content of a report card for minority populations. For example, the community-leader groups gave relatively high rankings to some conditions or issues (such as prostate cancer, stress, and violence) for which there are no quality measures derived from evidence-based guidelines. The groups did agree, though, on four basic report card components: (1) HEDIS effectiveness-of-care measures; (2) Consumer Assessment of Health Plans (CAHPS) survey of health plan members’ experiences with care; (3) surveys of patients with chronic disease (asthma or diabetes) or newly diagnosed prostate cancer as a means to develop measures of adequacy of information, patient involvement in decision making, respect for patients’ preferences and values, and other aspects of provider-patient communication; and (4) a survey of several dimensions of plans’ cultural and linguistic competence.

Identifying minority populations: Phase I pilot tests. Participating health plans examined possible sources of data on race/ethnicity and developed preliminary solutions to problems of lack of data at the plan level. Data on race/ethnicity were not available directly from the plan; alternatives had to be devised for use of data from other sources.

In Phase I of the project (1998–2000), three plans used three general methods to calculate HEDIS or other similar quality-of-care measures separately for different racial/ethnic groups: (1) A self-report item on race/ethnicity in surveys (CAHPS or a special asthma quality-of-care survey) was used to divide survey respondents into racial/ethnic groups, and analyses were performed comparing responses across groups. (2) Software using surname to estimate Hispanic versus non-Hispanic ethnicity was used to calculate HEDIS scores for those two groups of plan members. (3) Information from providers’ medical records and electronic encounter databases was used to assign persons to a racial/ethnic group. This information was then used to analyze disparities in processes of care.

We recognize that all sources of race/ethnicity data (including self-reports) have problems with completeness and accuracy. The advisory groups favored self-reports as the best method of assigning persons to race/ethnicity categories, but until health plans have self-reported data on race/ethnicity for all of their members, other methods, with inherent weaknesses, must be used as proxies.

Surname-recognition Collaboration with the Lovelace Clinic Foundation offered a unique opportunity to refine methods for assigning race or ethnicity categories to patients when such information was missing or inaccurate in databases. HEDIS data for the project were obtained from the Lovelace Health Plan.

Ethnicity/race was assigned using software that identifies Hispanic and non-Hispanic ethnicity/race based on surname. This software is a revision of the GUESS program (Generally Useful Ethnicity Search System) developed by the University of New Mexico. The original program was based on data derived from the U.S. Bureau of the Census, which include surname and self-declared race/ethnicity. Name characteristics were identified that correctly assigned Hispanic ethnicity for 90 percent of Hispanics.⁶ The revised Lovelace ethnicity identification program is approximately 95 percent accurate in identifying Hispanic ethnicity based on validation against ethnicity information contained in state tumor registry records.⁷ The software assigns persons to either "Hispanic" or "non-Hispanic" categories. It cannot be used to separately identify non-Hispanic whites or non-Hispanic African Americans or other groups; nor can it be used to identify members of Hispanic subgroups. The composition of both Hispanic and non-Hispanic groups varies considerably across regions of the country and perhaps across plans in a given region, so the system will generally be more useful for within-plan than for across-plan comparisons.

It was possible to calculate separate performance rates for Hispanic and non-Hispanic members on essentially all of the HEDIS "effectiveness-of-care" measures (for example, well-child visits in the first fifteen months, adolescent immunizations, and mammography). Very few measures showed noticeable differences in performance across the two groups. Only three of thirteen measures showed statistically significant differences between groups, and the magnitude was typically in the range of three to four percentage points (for example, breast cancer screening—74 percent for non-Hispanic women and 69.5 percent for Hispanic women).

Race/ethnicity data from self-report survey Analyses were conducted, using data from the Henry Ford Health System in Detroit, on survey measures being considered by the NCQA for inclusion in future iterations of HEDIS. The analyses focused on care of asthma in children and included concepts such as emergency department visit rates, hospital admission rates, and use of medications according to national guidelines. The sample allowed for separate analysis of privately insured children and children enrolled in Medicaid; analyses were designed to compare performance for African American versus non–African American children (nearly all of the non–African American sample was Caucasian).

The eligible population for the surveys was identified using the criteria for high-risk asthma taken from the draft HEDIS 2000 specifications: at least four prescriptions, or one emergency department visit, or one hospitalization, or at least four outpatient visits within a year. Data were obtained from administrative encounter and pharmacy claims data that did not include race/ethnicity. Surveys were sent to parents of 569 randomly selected patients who met these criteria; the returned surveys included data on race/ethnicity and also allowed respondents to confirm a diagnosis of asthma. Survey questions addressed issues of health care use (emergency department and physician office visits, and hospital admissions), routine use of medications (use of preventive versus "rescue" medications), knowledge of basic principles of asthma care, and satisfaction with care.

One of the key questions for the analysis was simply whether the information could be obtained and whether it would be possible to identify the relevant persons as African American. The answer to both questions is clearly "yes" using the combined encounter data and survey approach, and the results were substantively interesting. There was a trend on virtually all measures (emergency department visits, use of anti-inflammatory drugs, access to a specialist, and satisfaction with care) toward higher quality scores for African American children in the Medicaid group and the opposite trend for the privately insured group. Without additional analysis it is not clear whether these trends reflect performance of different parts of the provider network (that is, some physicians are more likely to see African American or Medicaid-insured patients than others are) or other factors.

We are now analyzing patterns of agreement between administrative data system assignment of race/ethnicity and self-reports in the survey, using data from both the Phase I survey and a similar survey of parents of children with asthma in Phase II.

Cultural and linguistic competence survey Community-leader groups were particularly interested in measures that could be used to assess the cultural and linguistic competence of health plans and provider networks. Structural measures like the number or proportion of minority primary care providers in a network, or the number of primary care physicians who spoke Spanish or another commonly spoken non-English language, were recommended as part of a report card.

Magda Garcia of the University of Texas School of Public Health developed a health plan survey and tested it in the three plans in Phase I. All plans completed the survey, which also included items on availability of translated materials for members; racial/ethnic composition of health plan staff, board, and provider network; and availability of training in cultural competence. Completing the survey, though, did not mean that plans necessarily had information readily available on all of the questions or dimensions of cultural and linguistic competence. Answers of "don’t know" or "data not available" were useful in that they reflected the state of data availability. Garcia found that the three plans were usually able to obtain information on patients’ reports of general satisfaction with communication (as measured in CAHPS), on the presence of educational materials in Spanish or on the availability of interpreter services, and whether these services are provided through AT&T or a department within the organization. Plans had difficulty obtaining information on numbers of Hispanic and African American members (except for Medicaid members and only through special inquiry to each state’s department of health) and on the providers’ language proficiency.

Report card project: Phase II. Phase II of the report card project involves working with thirteen health plans to demonstrate the feasibility of reporting HEDIS, CAHPS, and other quality-of-care measures separately by race/ethnicity. Plans involved in Phase II are using five methods for obtaining race/ethnicity data to use in preparing comparative quality reports: (1) self-reported data from surveys (CAHPS and chronic disease surveys); (2) data from state Medicaid enrollment files linked to plan membership files through a recipient’s identification number; (3) data from medical records obtained during the chart-review stage of the HEDIS hybrid method; (4) the GUESS software for estimating Hispanic ethnicity based on surname; and (5) geocoding, with race/ethnicity of selected persons being imputed on the basis of street address and census data.

At the time of this writing, plans had incorporated race/ethnicity data into their procedures for HEDIS and CAHPS in 2001 and had generated stratified reports. Preliminary data reviewed through November 2001 indicate that (at least in the mechanical sense of being able to obtain race/ethnicity data, link to HEDIS/CAHPS databases, and produce stratified quality reports) these methods can be used. All but one plan attempting to produce stratified HEDIS reports were able to do so, and all plans attempting to produce stratified CAHPS reports were able to do so.

Conclusions

Top Health Plans, Quality... The Commonwealth Fund-HRSA... Conclusions NOTES

 
Except in perhaps four states, there are no legal barriers to health plans’ use of data on race/ethnicity to improve quality.⁸ Initial experience in a demonstration project by thirteen volunteer health plans suggests that data on race/ethnicity can be obtained through one or more of several "work-around" methods until more direct methods of data collection are implemented, and that the data can be used to generate separate performance reports for major racial/ethnic groups.

Plans participating in Medicare+Choice will be required in 2003 to participate in a special project on either racial and ethnic disparities in care or cultural and linguistically appropriate services. These requirements are part of the regulations implementing the Medicare, Medicaid, and SCHIP Benefits Improvement and Protection Act (BIPA) of 2000, which are being carried out as part of the Quality Improvement System for Managed Care (QISMC).⁹ If state Medicaid programs, accrediting bodies such as the NCQA or Joint Commission on Accreditation of Healthcare Organizations (JCAHO), and private purchasers make similar requirements, plans will have a strong incentive to learn how to gather information on members’ race/ethnicity and use that information to reduce disparities and improve quality.

Editor's Notes

 
David Nerenz is director of the Institute for Health Care Studies and a professor in the College of Human Medicine at Michigan State University in East Lansing. Vence Bonham is associate professor in the university’s Department of Medicine, College of Human Medicine. Robbya Green-Weir is senior project coordinator in the Center for Health Services Research, Henry Ford Health System, in Detroit. Christine Joseph is an epidemiologist in the Department of Biostatistics and Research Epidemiology at Henry Ford Health Sciences Center. Margaret Gunter is executive director of the Lovelace Clinic Foundation in Albuquerque.

NOTES

Top Health Plans, Quality... The Commonwealth Fund-HRSA... Conclusions NOTES

 

1. See, for example, R. Coward et al., "Race Differences in the Health of Elders Who Live Alone," Journal of Aging and Health 9, no. 2 (1997): 147–170[Abstract/Free Full Text]; B. Jones et al., "Severe Obesity as an Explanatory Factor for the Black/White Difference in Stage at Diagnosis of Breast Cancer," American Journal of Epidemiology 146, no. 5 (1997): 394–404[Abstract/Free Full Text]; F. Chevarley and E. White, "Recent Trends in Breast Cancer Mortality among White and Black U.S. Women," American Journal of Public Health 87, no. 5 (1997): 775–781[Abstract/Free Full Text]; K.S. Collins, A. Hall, and C. Neuhaus, U.S. Minority Health: A Chartbook (New York: Commonwealth Fund, 1999); A. Lee et al., "Medicare Treatment Differences for Blacks and Whites," Medical Care 35, no. 12 (1997): 1173–1189[Medline]; and J. Mitchell and L. McCormack, "Time Trends in Late-Stage Diagnosis of Cervical Cancer: Differences by Race/Ethnicity and Income," Medical Care 35, no. 12 (1997): 1220–1224.[Medline]
2. E. Perez-Stable et al., "The Effects of Ethnicity and Language on Medical Outcomes of Patients with Hypertension or Diabetes," Medical Care 35, no. 12 (1997): 1212–1219[Medline]; and G. Xu et al., "The Relationship between the Race/Ethnicity of Generalist Physicians and Their Care for Underserved Populations," American Journal of Public Health 87, no. 5 (1997): 817–822.[Abstract/Free Full Text]
3. M.H. Chin et al., "Diabetes in the African-American Medicare Population," Diabetes Care 21, no. 7 (1998): 1090–1095[Abstract]; D.M. Carlisle et al., "Racial and Ethnic Disparities in the Use of Cardiovascular Procedures: Associations with Type of Health Insurance," American Journal of Public Health 87, no. 2 (1997): 263–267[Abstract/Free Full Text]; Collins et al., U.S. Minority Health; J. Blustein and L. Weiss, "Visits to Specialists under Medicare: Socioeconomic Advantage and Access to Care," Journal of Health Care for the Poor and Underserved (May 1998): 153–169; and V. Velanovich et al., "Racial Differences in the Presentation and Surgical Management of Breast Cancer," Surgery (April 1999): 375–379.
4. K. Fiscella et al., "Inequality in Quality: Addressing Socioeconomic, Racial, and Ethnic Disparities in Health Care," Journal of the American Medical Association (17 May 2000): 2579–2584; and B.D. Smedley, A.Y. Stith, and A.R. Nelson, eds., Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care (Washington: National Academy Press, March 2002).
5. Gregory Pawlson, NCQA, personal communication, 4 February 2002.
6. R.W. Buechley, "A Reproducible Method of Counting Persons of Spanish Surname," Journal of the American Statistical Association (March 1961): 88–97.
7. I. Rosenwaike, "The Status of Death Statistics from the Hispanic Population of the Southwest," Social Science Quarterly 69, no. 3 (1988): 722–736.
8. R.T. Perot and M. Youdelman, Racial, Ethnic, and Primary Language Data Collection in the Health Care System: An Assessment of Federal Policies and Practices (New York: Commonwealth Fund, 2001); and E. Berry et al., "Assessment of State Laws, Regulations, and Practices Affecting the Collection and Reporting of Racial and Ethnic Data by Health Insurers and Managed Care Plans" (Presentation at American Association of Health Plans Annual Meeting, Los Angeles, April 2001).
9. H.R. 5661, 106th Cong., 2d sess. (2000); see also www.hcfa.gov/medicare/bipaletter.htm (24 July 2001).

                      What's this?