Chapter II. Description of the Collaborative
Evaluation of a Learning Collaborative's Process and Effectiveness to Reduce Health Care Disparities Among Minority Populations
This chapter describes the organizations sponsoring, supporting, or participating in the Collaborative, what they said (in round one) about their initial motivation to participate, the structure and content of the Collaborative's work and future plans, and issues that bear on judging the Collaborative's success. Readers seeking more insight into the origins of the Collaborative will find a review of its historical development in Appendix B.
A. Organizations Participating in the Collaborative
AHRQ and RWJF cosponsored the Collaborative. AHRQ was the overall convener and contracted with RAND to conduct an initial needs assessment and to work with the participating firms to obtain and analyze racial and ethnic data. RWJF contracted with and paid the supporting organizations that provided guidance to the participating firms. CHCS was the central support organization responsible for organizing the Collaborative process and meetings, and for collecting firms' quarterly update reports (although firms' response to this reporting requirement remains incomplete). CHCS was also the main repository of the documentation on the work undertaken by the Collaborative. Under subcontract to CHCS, IHI provided a limited amount of support to the leadership team (i.e., the sponsors and organizations providing support to the Collaborative) of the Collaborative as well as advice based on its extensive work in provider-based quality improvement. In July 2005, RWJF entered into a contract with GMMB to support the Collaborative's communications objectives. Although RAND's work was initially viewed as distinct from that of the support organizations, CHCS, RAND, and IHI staff worked together to support the participating firms.
The Collaborative originally comprised 10 firms, but two—Anthem and WellPoint—merged during the first year of the Collaborative, leaving nine firms. Five of these nine are large national firms that operate health plans in many regions: Aetna, Cigna, Kaiser Permanente, United Healthcare, and WellPoint; Kaiser Permanente is unique because it is built on integrated delivery systems. Four are regional firms: Harvard Pilgrim Healthcare of Massachusetts, HealthPartners of Minnesota, Highmark Blue Cross-Blue Shield Organization in Pennsylvania, and Molina Healthcare, Inc., headquartered in California. With the exception of Molina, whose business is largely in Medicaid, these commercial firms offer a variety of products geared to groups and individuals. Many also participate in Medicare. Table II.1 summarizes the key characteristics of each firm.
The Collaborative was structured in ways consistent with many such collaboratives. Over the period of a year, the Collaborative was to meet in person three times in a general meeting. These collaborative meetings occurred over a 21-month period: on September 10, 2004; March 17 and 18, 2005; and June 20 to 21, 2006. In addition, the Collaborative held a final in-person meeting on NHPC key principals in Chicago on September 15, 2006, to determine the focus of Phase II. Table II.2 shows the activities and timeline for the Collaborative.
Two main support organizations received contracts to support the Collaborative's activities (RAND and CHCS). We summarize the way the main support organization contracts were structured, the commitments sponsors perceived the firms to have made to the Collaborative, the support provided by the Collaborative for collecting racial and ethnic data and developing pilot interventions, and the structure of support and plans for disseminating and communicating results.
1. Support Organization Contracts
RAND received two rounds of funding from AHRQ to support the Collaborative. The first contract was awarded on November 30, 2003, for assistance to begin in 2004. A second contract was awarded in spring 2005. Each contract totaled $200,000 to $225,000 and, although each was intended for a one-year period, they appear to have been extended to match the Collaborative's flow of work.
The scope of work for the first year's contract called for RAND to (1) recruit participants and convene the first meeting of the Collaborative; (2) interview participants to assess their capacity, readiness, and interest in working on disparities data (building on earlier interviews for the California Endowment); and (3) adapt the tool developed by RAND with United Healthcare as part of AHRQ's Integrated Delivery System Research Network (IDSRN) to help firms start measuring disparities. The document RAND prepared to support the award envisioned that baseline measures of disparities for at least some of the plans would be available during the first year and that AHRQ would fund additional tool development and pilot projects. To facilitate progress, RAND's work was to be complemented by a separately contracted "learning organization" expert. Interventions would not be tested until after the first year.
RWJF's contract with CHCS involved two years of funding (from September 1, 2004, through August 31, 2006), for a total of about $500,000, a portion of which was allocated to support from IHI. CHCS's project proposal envisioned a mix of participants, including those serving Medicaid and commercial markets, and work that would build on CHCS's Best Clinical and Administrative Practices (BCAP) typology.4 CHCS expected challenges in applying the models they had historically used for quality improvement in Medicaid plans to collaboratives involving firms based in the commercial market because (1) the fragmentation among purchasers in the marketplace made it challenging to harness their power in ways that allowed health plans to send a consistent and effective message to providers about the importance of quality improvement and disparities reduction; and (2) their multi-location and product organization made focusing the intervention more difficult. As CHCS expected, its approach to the Collaborative departed substantially from the BCAP-like models originally proposed because firms had different views of the Collaborative's mission and their own needs; some perceived a need for less focus on small-scale improvement than is traditionally the case in learning collaboratives.
2. Participant Commitments and Motivation
Each participating firm in the Collaborative had its own goals derived from its organizational context and priorities. In addition, many of the key actors had been involved in nationally focused work to address disparities (Appendix B). Based on the round one interviews, we summarize their goals as follows.
Sponsors. For AHRQ, supporting the Collaborative was consistent with the agency's emerging emphasis on the use of research to drive quality improvements and the active involvement of users. AHRQ's history of work on disparities and its new responsibilities for the National Healthcare Disparities Report (NHDR) made the agency particularly interested in working with health plans in a visible way to address perceived needs. Although RWJF staff say that they view AHRQ as the dominant and driving partner in the Collaborative, they also note that co-sponsorship of the Collaborative served important internal needs. In particular, RWJF's new leaders had significant interest in disparities, and involvement in the Collaborative allowed them to move while internal plans for funding were still under development. In addition, staff perceived that the foundation could work more actively with health plans given their leverage over large populations.
National Firms. In our round one interviews in summer 2005, the national firms—Aetna, Cigna, Kaiser Permanente, United Healthcare, and WellPoint—were not explicit about their objectives for participating in the Collaborative. It appears that they were motivated by perceived needs, both internal and external. They indicated they were using the Collaborative for a mixture of purposes, including making changes in delivery and dealing with political concerns. By allowing firms to work together, the Collaborative could reduce the risks perceived in addressing issues related to disparities. Some firms also felt that if they did not participate, they risked falling further behind the rest of the industry. Negative perceptions resulting from failure to participate were a concern for some of the organizations that were industry leaders.
Our round one interviews with national firms suggested that their interests focused mostly on developing organizational commitments to improve data infrastructure for addressing concerns over disparities. Although they might ultimately improve the quality of local care, organizations seemed more interested in learning how to employ knowledge internally than sharing what they were doing with other organizations. Nor were they interested in taking small local steps to improve quality (by relying on the rapid-cycle techniques that are a traditional part of learning collaboratives).
Regional Firms. While regional firms' objectives did not necessarily differ from those of the national firms, the impetus for participation was more distinct. Participating regional firms were typically large, well-established organizations that wanted to use the Collaborative to expand in areas they were already pursuing. Two of the four regional firms were recruited through their ties to RAND staff. Compared with national firms, regional firms found pilot interventions more relevant, although they were constrained by limited resources and competing priorities. At least one looked to the Collaborative primarily for insight on how to capture disparities data for its members.
Support Organizations. The support organizations are contractors that receive payments for carrying out a specified scope of work. However, given that the organizations have earned high regard and face many competing demands, their involvement also reflects particular organizational and staff interests. RAND's interest in the Collaborative was a natural outgrowth of its staff's earlier work on racial and ethnic disparities; Dr. Nicole Lurie, who served in the Clinton Administration as a federal government appointee in the area of racial/ethnic disparities, used her contacts and experience to move the Collaborative forward. RAND staff were also experienced in using geocoding and surname analysis to examine racial and ethnic disparities through AHRQ's Integrated Delivery Systems Research Network (IDSRN), in which it participated as a subcontractor on the Center for Health Care Policy and Evaluation's team based at United Healthcare.
AHRQ and RWJF divided responsibilities, with RWJF responsible for arranging for a support contractor to coordinate and guide the Collaborative's efforts. Few organizations that are involved in guiding quality improvement collaboratives are experienced with health plan (versus provider) collaboratives. After considering a range of firms, RWJF selected CHCS because although it works primarily in the Medicaid area, it is perhaps the only learning organization with a history of work on health plan collaboratives. RWJF and CHCS also have a history of successfully working together. RWJF asked CHCS to involve IHI because the latter brought knowledge and recognized leadership in provider-based initiatives. A key staff member at AHRQ familiar with IHI's work with community health center-based collaboratives encouraged the organization's involvement. The lead staff from CHCS and IHI had worked together as senior staff at RWJF, which gave them a good basis for establishing a partnership. Under the contract, CHCS is responsible for most activity; IHI staff provide targeted, substantive support in selected areas. Firms participating in the Collaborative may not necessarily distinguish between the support IHI and CHCS provide, because of the way the two work together.
3. Firms' Initial Commitments
From the Collaborative's inception, participating firms struggled to varying degrees with how open to be about their internal processes and concerns, whether to share data, and whether to commit to shared activities. In an effort to secure clear commitments, AHRQ discussed an agreement with firms in 2003 and again on July 8, 2004, when the major organizational stakeholders—the sponsors, support organizations, participating firms, and their affiliated trade associations—held a two-and-a-half hour conference call to agree on how to proceed. An important area of discussion involved the commitments firms were making to the Collaborative. The Memorandum of Participation Principles stated that:
- Improving overall quality and reducing disparities are important national and plan objectives. Participating plans will commit senior leadership to attend three Collaborative meetings and intervening calls to report on progress.
- Data are needed to assess performance and assess quality. Participating plans agree to obtain the necessary data to move forward, with technical support from RAND as required.
- Workgroup measurement will focus on one or more accepted evidence-based measure. Participating plans agree to the common measurement expectations they define for the Collaborative.
- The workgroup will balance efforts to achieve consistency of measurement with flexibility reflecting varied plan market conditions.
- The workgroup will balance its efforts to share data, pilot designs, and results with requirements for maintaining privacy, confidentiality, and proprietary interest.
Participants also agreed that disparities in diabetes can provide a starting place for mutual work and that they would build on existing measurement efforts and thus involve HEDIS measures. Firms were not asked to formally approve or sign the memorandum. The way the final two principles dealt with consistency versus flexibility and sharing versus proprietary interest suggest that some lack of consensus about what firms would do existed from the start.
4. Support Activities During Phase I of the Collaborative
In addition to structuring and leading formal meetings, learning organizations supported the Collaborative in Phase I by providing assistance to firms by telephone. Several rounds of such calls were completed. While CHCS originally hoped to group firms for joint assistance calls, it found that firms preferred communicating separately. The calls were important for documenting activity, as firms provided only limited detail in their progress reports, which were often missing information or submitted late. To coordinate their support to the Collaborative, key staff from each support and sponsoring organization participated in periodic conference calls—termed operational workgroup meetings—convened by CHCS. Firms were supposed to submit quarterly progress reports to CHCS. Compliance was spotty and CHCS ultimately put less emphasis on this activity, asking firms instead to prepare slides and other tools for briefing others in the Collaborative about their progress. As highlighted in the framework, the Collaborative structure could help firms advance their ability to deal with issues of racial and ethnic disparities, support overall firm and leadership commitment to addressing racial/ethnic disparities, and help firms better measure and assess disparities and take action to address them. Ultimately, the Collaborative can generate learning about disparities that can be shared with those in the Collaborative and others.
Measuring Disparities. In the Collaborative's initial year, most firms' focused on developing insights into disparities within the firm. RAND provided support for geocoding and surname analysis of firm data on members with diabetes, thereby helping firms to generate estimates of racial and ethnic disparities.5 RAND recognized that firms had limited internal data on the racial and ethnic composition of their membership and that data improvements would take time (see Chapter IV). To that end, RAND formed a workgroup that appears to include all firms except the two that were already getting needed data. The hope was that developing such data would reinforce firms' sense that disparities were a problem warranting attention. While there was less active support to firms in collecting their own racial and ethnic data, geocoding/surname analysis helped firms appreciate the value of such collection and spurred them to consider how primary data could be collected. The Collaborative set up sessions for firms to learn about member organizations' work—particularly, that of Aetna, whose decision to capture member data was an important impetus for the Collaborative, and HealthPartners, whose affiliated clinics actively collect racial and ethnic data from patients who seek care. Support organizations also requested firms to submit common measures based on HEDIS diabetes indicators; however, firms did not prioritize this effort and response was varied (Chapter IV).
Intervening to Reduce Disparities. From the inception of the Collaborative, firms disagreed about how much effort should be spent in developing and testing specific pilot interventions to reduce disparities. The organizations brought in to support the Collaborative were experienced in this area—one of the two AHRQ senior staff guiding the Collaborative's development had experience working with community health centers, and was very interested in pilot interventions. Round one interviews revealed an uneven interest among firms in testing pilot interventions. Support organizations reported a "push back" from firms to following a traditional learning collaborative model, especially with respect to using tools developed by CHCS and IHI for Medicaid plans or provider groups. Firms wanted to pursue strategies that made the most sense to them. Some perceived small scale pilots too narrow an approach, unnecessary given their existing investments in quality improvement, or inappropriate to the extent they had a provider emphasis if they perceived their health plan's strength favored member-based interventions. CHCS responded by clarifying that pilot interventions included a variety of activities: data collection/refinement, provider- and member-directed strategies, community-based strategies, and work on organizational assessment and capacity building. At the first group meeting, plans presented details of their existing initiatives. Many indicated that future intervention would follow the results from geocoding and other data analysis, an approach consistent with RAND's original concept that interventions would begin in Year 2.
In the second year of the Collaborative, firms further developed their interventions, some of which were new, while others built on existing activity (Chapter V). As complex organizations with several ongoing activities, firms did not distinguish Collaborative-specific activity from other firm work.6 Because many of the activities and interventions were not initiated until late in the Collaborative, most are ongoing, and there is limited information thus far on their impact. Firms said that these activities would, for the most part, continue after the formal end of Phase I of the Collaborative.
Building Communication and Dissemination Infrastructure. Although communications was not a part of the initial Collaborative infrastructure, the need to disseminate information about the Collaborative and what it was learning was always an important goal. To support that goal, RWJF entered into an 18-month, $160,000 contract with GMMB in summer 2005. The contract called for GMMB to coordinate all public communications related to the Collaborative. Since then, GMMB developed relationships with the communications staff at each firm, developed a logo and other material to create an identity for the Collaborative, and hosted the National Health Plan Collaborative Roundtable Briefing to publicize the work of the Collaborative (Chapter VI). Currently, GMMB is drafting a Phase I report on NHPC activities, which will include a "call to action." While it is likely that communications will have a more substantial role in Phase II of the Collaborative, to date, RWJF has not yet decided exactly how that function will be handled and where the focus will lie.
C. Context for Judging the Success of the Collaborative
The Collaborative involved nine diverse firms whose interests and operational styles needed to be coordinated. Because the Collaborative's model of engagement was new to many participants, the tools for structuring the Collaborative had to be developed. In view of participants' varied interests, support organizations found that they had to modify their proposed strategies substantially. A key point of contention involved whether to emphasize broad-based efforts to build national and firm infrastructure for improved measurement of disparities or specific interventions designed to reduce such disparities and, if so, on what scale. Because of participants' varying views on this subject, the goals of the Collaborative were not necessarily well defined or interpreted the same way by all participants.
In today's environment, firms face a wide range of competing demands—for example, two firms in this study were involved in a merger, and another two were dealing with recent and severe financial stress. Leadership changes are common, and the market continues to pose challenges for all firms. In our interviews, we typically heard that work on disparities was a high priority for quality improvement, but that each firm's ability to proceed depended on a range of considerations and market demands.
Initial interviews with senior leaders at AHRQ and RWJF revealed that both organizations were aware that participating firms cannot always influence care delivery directly, although they are responsible for millions of covered lives. Sponsor interviewees saw value in the Collaborative's ability to influence such organizations to make disparities a more legitimate focus of quality improvement work, to understand the value of relying on information to measure disparities, and to motivate "silo" components of firms to talk with one another. That is, the Collaborative's scale means that even small effects may be influential in enhancing work to address disparities in ways that will potentially affect many people.