Addressing Racial and Ethnic Disparities in the Context of
Medicaid Managed Care: A Six-State Demonstration Project

Executive Summary

Final Report of Contract # 250-02-0010
November 2004

Previous work has demonstrated that health plans can obtain data on race/ethnicity of enrollees and use the information to identify racial/ethnic disparities in quality of care. In the current project, we attempted to answer the question: Can managed care plans obtain data on race/ethnicity of enrollees from state Medicaid programs, and use that information to identify and reduce or eliminate disparities in quality of care? We also attempted to address questions of how state Medicaid agencies could act as coordinators of health plans’ efforts to address disparities and how disparities initiatives could be organized within the larger context of quality management programs and the Medicaid managed care program in general.

Purpose

The current project was a demonstration project with six closely-related objectives:

• to recruit several state Medicaid agencies to address racial/ethnic disparities in quality of care as an important quality of care issue;
• to recruit one or more managed care plans within each participating State, provide those plans with data on race/ethnicity of members in order to allow analysis of quality of care measures by race/ethnicity, and provide technical assistance to plans as they analyze data and organize quality improvement projects aimed at reducing or eliminating disparities;
• to obtain preliminary estimates of across-plan and within-plan variation in quality of care by race/ethnicity;
• to show that participating Medicaid agencies can coordinate, and health plans can organize, quality improvement initiatives designed to reduce or eliminate racial and ethnic disparities in quality of care;
• to document the results and impact of quality improvement initiatives developed by participating Medicaid managed care plans; and
• to disseminate findings from the project, in terms of initial experience with data analyses to identify disparities, organization of Quality Improvement (QI) projects, and assessment of impact of those projects on disparities in quality.

The project, which was sponsored by the Health Systems Organization and Financing Group in the Health Resources and Services Administration (HRSA), began in September of 2002 and ended June 30, 2004.

Project Design Summary

This project was structured as a demonstration project involving state Medicaid programs and managed care plans that volunteered to participate. There was no formal sampling of either state Medicaid programs or managed care plans within participating states. All state Medicaid programs were invited to participate in the project. Medicaid programs in six States (Washington, Oregon, Montana, Texas, Michigan, and Virginia) accepted the invitation to participate, and they in turn invited managed care plans in each State to participate. The Medicaid programs were given a list of criteria to use in selecting health plans for participation in the project, but States were allowed to use additional criteria for choosing plans if they wished. The list of plans participating in each state is presented in Table I. A total of 12 health plans (plus the Primary Care Case Management program in Montana) were involved. No state Medicaid programs or health plans dropped out of the project after agreeing to participate, and no state Medicaid programs or health plans were added to the project after it began.

Table I. States and Managed Care Plans Participating in the Demonstration Project

Methods

Medicaid program staff in each State provided data on race/ethnicity from eligibility/enrollment files to participating managed care plans in January of 2003. Health plans merged these data with membership files or The Health Plan Employer Data and Information Set (HEDIS) data files to produce quality of care reports stratified by race/ethnicity. In some plans, this process took longer than in others, and one or two of the participating plans had actually done a similar analysis in 2001 and/or 2002, so they had a base of experience from which to work.

All of the plans were able to obtain data on race/ethnicity from their state Medicaid programs and link that information to either membership files or HEDIS data analysis files. This linkage allowed all of the plans to generate quality of care reports stratified by race/ethnicity. The plans varied somewhat in the data available for analysis at baseline. As indicated in Table II below, most of the plans waited until calendar year 2002 (reporting year 2003) data were available and used those data for baseline analyses. Three plans did not analyze data according to standard HEDIS definitions, but were able to use the race/ethnicity data to generate stratified quality of care reports in “HEDIS-like” format. Conceptually, the measures in these analyses were very similar to HEDIS measures, focusing on measures of breast and cervical cancer screening rates, prenatal care visit rates, and laboratory testing rates for patients with diabetes.

Table II. Baseline Data Sources

All but one of the plans identified at least one significant disparity in these baseline analyses. (In that plan, an apparent disparity turned out to be a problem in accuracy of claims data at some clinic sites.) In 10 of the 12 plans finding a disparity, more than one disparity was identified. An example of the types of disparities noted within participating plans is presented in Figure A.

Figure A. Disparities in rates of four specific quality of care measures, single managed care plan. (D-link)

Based on these analyses, the health plans chose to focus on the areas of disparity in quality as summarized in Table III.

Table III. Number of health plans and projects addressing each major clinical area.
Numbers add to more than 13 because some plans did more than one project.

Nine of the thirteen plans were able to implement their QI projects and carry through to at least some form of follow-up evaluation. For most of the plans, the follow-up analysis came in the form of follow-up HEDIS analyses and reports (usually 2004 reports for the 2003 year). For one of the plans, follow-up analyses involved measures and data from quarterly or monthly reports on events like prenatal care visits or laboratory tests for members with diabetes. These analyses have been useful to indicate whether or not a QI project was “on track” but the health plan doing these analyses generally viewed them as interim analyses rather than final assessments of project success. Initiatives begun in the fall of 2003 and carried through calendar year 2004 would be expected to show effects in HEDIS data for 2004, analyzed and reported in mid-2005.

Summary of Results

Of the eight plans reporting follow-up data that had a disparity at baseline, it appears that two (E, L) made clear progress toward the goal of reducing or eliminating disparities observed at baseline, and three other plans (B, C, I) made measurable improvements in quality of care for the target minority population without necessarily reducing or eliminating a disparity.

Only three plans (F, G, H) saw little or no change in disparity in follow-up analyses. In four plans (A, D, K, M) the short time frame of the project did not allow for quantitative follow-up analyses. (The scope of the project did not allow for detailed analysis of reasons for the relative success of different projects; such an analysis would probably require a larger sample of projects in order to make comparisons.)

In three health plans (A, J, M), it became clear that an apparent disparity at baseline was at least partially a function of data completeness and accuracy and not necessarily a true disparity in quality. Two of these plans (A, M) made improvements in the data collection and reporting process, and in both cases, the apparent disparity was still seen in the more accurate reports. The time taken to improve the data collection and analysis processes inevitably delayed the design and implementation of QI efforts, but even in these two instances, the plans made significant progress on the path toward quality improvement and disparity reduction by improving their measurement processes. Preliminary results of the effects of QI initiatives in those two plans will probably not be available until late fall of 2004 (after the publication date of this report).

Table IV. Summary of Plan-Level Project Outcome

Benefits Identified by Health Plan Participants:

Toward the end of the project, the participating plans were asked to provide feedback regarding their respective QI projects (e.g., benefits, barriers, resource use, cost, etc). Plan representatives reported positive benefits to participating in a project of this nature. Most felt that the project provided them with an opportunity to analyze one or more important subsets of their member populations to focus on specific health care issues and outcomes measurement. The establishment of collaborative relationships between state agencies, participating health plans, and local community agencies/coalitions was seen as a strong positive benefit for plan members. In the two instances where plans within a State focused on the same health care topic and target population, it provided an opportunity to share knowledge, ideas, and combine plan/state efforts. Plans also stated that the topic of racial/ethnic disparities aligned well with ongoing disease/case management activities and worked well with stratification efforts for high-risk conditions. It was important that the project meet multiple objectives and priorities, and be aligned with other identified plan QI initiatives.

Several plans indicated that a coordinated approach by the State Medicaid agency across health plans to address disease specific issues such as asthma, diabetes, depression and cardiovascular disease would be beneficial. Plans supporting this concept suggested that Medicaid agencies are in a unique position to facilitate coordinated efforts among their contracted health plans. Two States participating in this project followed this approach requesting that their plans focus on a similar clinical topic (e.g., diabetes).

Conclusions

The model being tested here is clearly generalizable to other state Medicaid programs (at least to those that have reasonably complete and accurate race/ethnicity data and can provide data to plans in a format that allows for integration with health plan membership files) and to Medicare+Choice plans. State Medicaid programs were able to provide data on race/ethnicity to managed care plans, and with relatively few and minor exceptions, plans were able to analyze HEDIS data and identify disparities, organize QI projects to reduce or eliminate disparities, and assess the effects of those interventions. Some of the initiatives were able to produce reductions in disparities or improvements in quality for target minority groups within one cycle of follow-up data analysis. Based on early experience in the project, the State of Michigan has already begun to expand the project from the original three participating plans to most of the managed care plans involved in Medicaid.

This approach may also serve as a model for the private sector if employers are willing and able to provide race/ethnicity data on employees to managed care plans. If commercial health plans are able to obtain data on members’ race/ethnicity directly from members or through other means, then the sequence of steps being followed in this project, from initial data analysis, to identification of disparities, to planning and implementing of QI projects, to assessment of impact of those projects, can be followed by commercial plans just as easily as by public sector plans.

The QI projects represented a diverse and interesting mix of patient education, provider education or reminder, disease management, and system improvement interventions. Clinical target areas include diabetes care, prenatal care, and well-child care. Health plans and state Medicaid programs should find useful ideas and models to follow as they organize similar disparities initiatives.

Next Steps

Some of the States participating in this project, and some other States as well, are expanding the work on disparities and culturally and linguistically appropriate services to include attention to these issues as a regular part of ongoing quality measurement and quality improvement activities. We can expect to see more States, and more health plans within those States, taking on the kinds of projects reported here in the very near future.

In order to enhance the number and quality of future quality improvement initiatives aimed at reducing or eliminating disparities, state Medicaid programs can consider taking some or all of the following steps:

• improving the completeness and accuracy of race/ethnicity data collected at the point of program eligibility/enrollment;
  
  
• where appropriate, adding racial/ethnic “sub-categories” (e.g., Mexican or Dominican as sub-categories of “Hispanic”) in order to facilitate use of culturally and linguistically appropriate quality improvement approaches;
  
  
• including data on primary language in the process of collection and transmission of race/ethnicity data;
  
  
• making participation in disparity-reduction initiatives one of the contractual requirements for health plans’ participation in the Medicaid managed care program;
  
  
• encouraging plans, particularly those with overlapping provider networks, to collaborate in terms of QI initiative planning, outreach to providers and community partners, and data analysis for program evaluation;
  
  
• providing technical assistance workshops for health plan quality measurement and quality improvement staff to assist in development and evaluation of QI initiatives;
  
  
• using state-level quality of care data to supplement plans’ own data for purposes of QI program evaluation.

The experience of health plans in this project has encouraged the development of at least two similar projects that will expand the number and type of plans involved. A project funded by the Robert Wood Johnson Foundation and organized by the Center for Health Care Strategies (CHCS) is currently accepting applications from Medicaid managed care plans to participate in a project using the Best Clinical and Administrative Practices (BCAP) methodology to address disparities in quality of care. The project will involve essentially the same structure as that described here, but the BCAP approach to quality improvement will be more explicitly part of the project and there will be more attention paid to opportunities for state-level analyses of data for purposes of identifying and understanding disparities and assessing effects of plan-level interventions.

A second project that is being considered for funding by a major private foundation is extending the experience of this project to large private sector health plans and private employers. Again, the key steps of baseline data analysis to identify disparities, planning and implementation of QI initiatives, and follow-up data analysis to assess impact of interventions will be used to determine whether plans can have a significant effect on reducing or eliminating disparities. A key methodological challenge is the lack of race/ethnicity data coming from a single purchaser; plans will have to use a variety of methods to obtain the data that in this project came from the state Medicaid programs.

(Back to: Table of Contents)