Exporting the Buyers Health Care Action Group Purchasing Model: Lessons from Other Communities

Abstract

When first implemented in Minneapolis and St. Paul, Minnesota, the Buyers Health Care Action Group's (BHCAG) purchasing approach received considerable attention as an employer-managed, consumer-driven health care model embodying many of the principles of managed competition. First BHCAG and, later, a for-profit management company attempted to export this model to other communities. Their efforts were met with resistance from local hospitals and, in many cases, apathy by employers who were expected to be supportive. This experience underscores several difficulties that appear to be inherent in implementing purchasing models based on competing care systems. It also, once again, suggests caution in drawing lessons from community-level experiments in purchasing health care.

The buyers health care action group (BHCAG) is a well-known employer coalition located in the Twin Cities of Minneapolis and St. Paul, Minnesota (Bodenheimer and Sullivan 1998a, 1998b; Christianson et al. 1995; Toner 1993; Winslow 1993). Its efforts in the area of private-sector health care purchasing have received national attention and have been the subject of several research studies (Abraham, Feldman, and Carlin 2005; Christianson and Feldman 2002; Christianson et al. 1999; Feldman, Christianson, and Schultz 2000; Harris, Schultz, and Feldman 2002; Knutson 1998; Schultz et al. 2001). In particular, its purchasing model, based on tiered care systems and featuring risk adjustment and quality data at the care system level, could be viewed as private employers’ “best effort” to date to incorporate the principles of managed competition in their purchase of health benefits. In many ways, the BHCAG model also presaged the current employer movement toward “paying for performance” and insurance innovations incorporating tiered hospital networks.

The employer-innovators of the BHCAG approach, which was called “Choice Plus” in its early years and later “Patient Choice,” were enthusiastic about their creation and actively “marketed” it to employers outside the Twin Cities. (Although the name change from “Choice Plus” to “Patient Choice” did not become official until 2001, in order to avoid confusion we will use the new name to refer to the health plan even in pre-2001 references.) In 2000, they transferred the responsibility for maintaining and disseminating the product to a for-profit entity, Patient Choice Healthcare (PCH). Some of the participating BHCAG employers supported this change because they saw no advantage in continuing to be engaged in the day-to-day operation of Patient Choice, while others saw it as an opportunity to raise the private capital required to establish Patient Choice in other communities.

In this article, we analyze the diffusion experiences of Patient Choice and, more generally, the ideas on which it was based. Our analysis could be interpreted as a cautionary tale at several levels. First, it underscores the difficulties that already are being reported in constructing tiered hospital products, a more straightforward benefit design than the BHCAG care system model. Second, it again highlights the practical problems that purchasers have encountered in implementing managed competition approaches on a market-by-market basis. Third, at a more general level, our analysis suggests that policymakers should be skeptical about the usefulness of innovative, apparently successful, community-level “experiments” as models for health system reform at the national level.

This article has four parts. In the first section, drawing on previous work, we summarize the evolution and design features of the BHCAG Patient Choice model, the experience of the model in the Twin Cities, and the transition to PCH. The second section of the article describes BHCAG's efforts, before PCH's involvement, to work with other communities interested in creating a product based on the Patient Choice principles. A common feature of these efforts was the active involvement and endorsement of Twin Cities’ employers and their enthusiasm for the product. The third section addresses the challenges that PCH faced in maintaining and enhancing these early diffusion efforts while also attempting to enter other markets. The article's second and third sections are based on interviews that we conducted when visiting four of these communities and on telephone interviews with local respondents and PCH staff regarding three others. In the final section, we synthesize the diffusion experience across communities and over time and place it in the context of health policy.

BHCAG's Patient Choice Model

BHCAG was formed in 1991 by 14 large, self-insured Twin Cities employers. Shortly afterward, in 1992, BHCAG contracted with a single health maintenance organization, HealthPartners, which partnered with the Mayo Clinic and the Park-Nicollet Medical Center to provide a health benefit plan as an option to, or replacement for, the companies’ existing health plans. From its inception, BHCAG was committed to quality improvement, although many employers became increasingly concerned that their efforts at group purchasing were contributing to the consolidation of the health insurance market (in fact, HealthPartners merged with the MedCenters Health Plan sponsored by Park-Nicollet). BHCAG members also were concerned that their quality-improvement efforts were directed at the wrong organization—the health plan—rather than the provider groups actually responsible for delivering medical care (Christianson et al. 1999; Robinow 1997).

Beginning with open enrollment in 1996 for the 1997 benefit year, BHCAG embarked on a radically different strategy for purchasing health insurance. Instead of contracting with a single health maintenance organization (HMO), the employers decided to offer a broad choice of “care systems” built around groups of primary care physicians and affiliated specialists and hospitals (Robinson 2002). These care systems would be responsible for providing a full continuum of care, with primary care physicians being allowed to participate in only one care system. Each care system submitted a bid for covering an employee of standard risk. Based on the bids, sets of care-system specific provider fees were developed. This feature of the BHCAG model was attractive to providers because they were able to “set their own prices” for care, as opposed to reacting to the fee schedules and capitated payments dictated by large health plans. If actual expenditures exceeded the bid amount, adjusted for enrollee risk, the fees could be reduced (Knutson 1998). BHCAG employers were able to operate this “virtual capitation” program without obtaining an insurance license because they were self-insured and exempted by the Employee Retirement Income Security Act from the state's insurance regulations.

The care systems’ bids served another critical purpose in the BHCAG program: They were used to determine employees’ out-of-pocket premiums. Most BHCAG employers adopted the policy of passing on to their employees 100 percent of the incremental cost of more expensive bids. The feature of employee choice, coupled with a financial incentive to choose low-cost care systems, was modeled on Alain Enthoven's (1978a, 1978b, 1979, 1988) recommendations for creating “managed competition” among health care provider systems. In order to promote more-informed choices, BHCAG also required standardized benefit packages and distributed the results of surveys of enrollees’ satisfaction and comparisons of the care systems’ performance. The early indications were that many employees used this information to make their choices (Feldman, Christianson, and Schultz 2000).

Enrollment in Patient Choice received a boost when several large employers decided to offer it as a “total replacement” for their current plans. Patient Choice's enrollment increased from 116,000 in its first year of operation to 150,000 in 1999, out of 250,000 eligible employees and dependents. It became widely touted as a potential national model for health care purchasing and health care reform (Appleby 1995; Jossi 1998; O'Reilly 1998; Slomski 1997; Winslow 1998) and seemed to show some potential for controlling employers’ health benefit cost increases (Lyles et al. 2002).

In October 2000, BHCAG announced a major restructuring in which Patient Choice was spun off to PCH. This was expected to increase Patient Choice's market penetration of small and medium-size employers, under the assumption that PCH would obtain an insurance license or partner with an existing insurer. It also was expected that PCH would collaborate with employers and employer-purchasing coalitions in other communities to develop versions of Patient Choice that fit their markets (Christianson and Feldman 2002).

The formation of this independent company explicitly recognized what many observers already realized: that Patient Choice was not simply a new benefit design. It was becoming a significant competitor to established managed care organizations in the community. HealthPartners, in particular, viewed Patient Choice as a market rival and, through very aggressive premium bids, was successful in luring several large BHCAG employers away from the product. These withdrawals of employers from Patient Choice underscored a problem noted in previous assessments of employer coalitions: maintaining employers’ commitment to the coalition's principles and activities over time (Brown 1993; Christianson 1997; Pauly 1999). At the same time that Patient Choice was facing the problem of replacing exiting employers, some employers that continued to offer Patient Choice were becoming dissatisfied with the cost of collecting quality information on the care systems, especially in light of a weakening economy and internal pressures to cut costs. Responding to these concerns, BHCAG dropped the employee survey after 2001 (Abraham, Feldman, and Carlin 2005). An additional problem that contributed to employers’ dissatisfaction was the transfer in 2000 of responsibility for administering the program from HealthPartners to a different entity. Although the new administrator was selected a year in advance to facilitate a smooth transition, the care systems reported major delays in payment and poor service. These problems also affected participating employers, leading several more employers to drop Patient Choice.

In this context, maintaining care system participation and support for Patient Choice became increasingly important to the product's long-term success. Since its inception, Patient Choice had enjoyed varying degrees of support from Twin Cities providers. Most providers appreciated that by organizing as care systems, they could set their own prices through their risk-adjusted bids. They also valued BHCAG's efforts to reward quality at the care system level. However, over time they became increasingly frustrated by Patient Choice's enrollment growth, which was slower than they had anticipated, and by the complexity of the payment system, in contrast to the ways that they typically were paid by third-party administrators. These frustrations were amplified by the problems with the new administrator just noted. Nevertheless, care systems generally supported Patient Choice at the time that PCH assumed responsibility for its management.

In summary, by the time PCH entered the picture, Patient Choice had experienced early success in the Twin Cities and attracted considerable national attention. The participating care systems’ support for Patient Choice seemed to be relatively solid, despite concerns about specific operational issues, but some early employer advocates were beginning to distance themselves from the product and others were no longer offering it at all. Clearly, PCH would be challenged to maintain and grow Patient Choice in the Twin Cities at the same time it was attempting to export the model to other communities.

The First Stage: Diffusion before PCH

The initial diffusion of Patient Choice to other communities began before the formation of PCH, driven largely by the enthusiasm of a subset of BHCAG employers. These employers, along with the BHCAG staff, frequently presented the “BHCAG model” at regional and national conferences and professional meetings. This often attracted the interest of employers in other communities, followed by a visit to the Twin Cities and a presentation in their communities by BHCAG representatives. In four communities, employers took further steps to implement an approach based on the Patient Choice model. The problems they encountered foretold the problems PCH subsequently would face in these same communities, as well as in the new markets it would enter.

Des Moines, Iowa

Supported by strong executive leadership, an employer health care coalition in Des Moines, Iowa, had pursued a joint purchasing approach quite similar to that used by BHCAG in the Twin Cities during the early 1990s. In 1996, the coalition expanded its offerings to two care systems, each formed around one of the two major hospital-based health care systems in Des Moines. At one point, the total enrollment in the two options reached 10,000, but then it stalled for several reasons that were repeated in other communities. First, a dominant health insurer in the market pursued an aggressive pricing policy to lure employers away from the coalition product. Second, employers were not happy with the service provided by the third-party administrator (TPA) that administered the product. Third, the number of care systems available to compete for business was limited; hospitals were full; and physician practices were busy, causing employers to become increasingly skeptical about whether the bids submitted by the systems represented the “best available” prices. Fourth, some of the key employer leaders who were early supporters of the product retired, or their companies were purchased, thereby eroding the initial base of employer support. Fifth, the emergence of consumer-driven health plans as new, arguably lower-maintenance benefit alternatives attracted the attention of employers. We should note that some of the features of Patient Choice that contributed to its success in the Twin Cities market were modified or never implemented in Des Moines, including risk adjustment, the collection and dissemination of consumer ratings of care system features, and the development of other quality assessment data. An attempt by the Des Moines coalition to rejuvenate its purchasing effort by partnering with PCH was unsuccessful, and the coalition has now shifted its attention to consumer-driven health care, working with health systems to reduce waste through “Six Sigma” approaches and developing data-sharing models.

Sioux Falls, South Dakota

The Sioux Empire Employers Health Care Group was formed by 12 self-insured employers in late 1994. Shortly thereafter, it decided to work with the BHCAG staff to implement a Patient Choice model in Sioux Falls, South Dakota. From the beginning, this proved to be a difficult task. Local providers had trouble organizing into care systems in order to submit bids (even though they used a simplified bidding process based on fee schedules). They were concerned that their bids would be used to place them into cost groups and generally preferred the status quo to the version of care system competition envisioned by Patient Choice. Nonetheless, two care systems, formed around the two large hospitals in the community, eventually submitted bids. The employers had hoped for more care system participants, and the number of bidders did rise to three in 1998. Two employers offered these care systems as benefit options, but then the smallest of the three care systems withdrew. This situation continued until 2001, with no new employers offering the care systems. At that time, one of the remaining two hospital-based systems withdrew, and problems with the TPA's performance became a significant issue. The hospital system reentered the program in 2002, but one of the original employers then dropped the product. The coalition felt it could not, in good faith, encourage other employers to offer the product because of the perceived severity of the TPA's performance issues. One employer that continuously offered the care system product believed it had reduced health benefits costs. Overall, however, the competing care system model failed to attract the interest of most large Sioux Falls employers and was never fully implemented there. The care system choices were limited, and the “hassle factor” due to administrative problems was judged to outweigh the potential product savings.

St. Louis

The Gateway Purchasers for Health was incorporated in 1994. It initially implemented a purchasing model based on managed competition principles and patterned after the efforts of the Pacific Business Group on Health. In the beginning, Gateway offered nine fully insured products to its members, but this number had dropped to two by 2002, and one of those options was not accepting new members. Many of the founding employers had moved from offering insured products to being self-insured, tempering their enthusiasm for a joint purchasing approach based on competing HMOs offering insured products. In the fall of 1998, Gateway's board began to explore other purchasing models. Representatives from BHCAG described the Patient Choice model to the board, and six companies committed to undertake a pilot project. As part of this project, physicians were asked to organize into networks to receive risk-adjusted payments with rewards for quality. A preferred provider organization (PPO) would be offered to fill in any provider gaps caused by decisions by physicians’ networks not to submit bids. The pilot uncovered several problems relatively quickly. For instance, because employers were moving from insured to self-insured products, they were not able to supply the historical data that providers needed to construct bids. Also, because the new product was untested, the employers planned to offer it along with two or more other options, thus limiting the number of new patients that low-cost provider networks might expect to attract. In addition, the product was perceived by the provider networks and employers to be administratively complex. Primary care physicians resisted being in only one group—a product feature considered to be a hallmark of Patient Choice by Twin Cities employers—and were concerned that the systems would be too small to negotiate good payment rates for specialty care. In the end, ten to 12 physician networks were constructed, and Health Market, a Connecticut-based, consumer-driven health plan start-up company, was chosen to administer the program. Momentum came to a halt, however, when two major systems that included “name brand” hospitals in St. Louis withdrew after seeing a demonstration of the type of cost and performance data that would be made available for consumers to use when choosing a care system. At the same time, some coalition employers were becoming uncomfortable with the tasks necessary to administer the program and worried about the effect that participating in the program might have on their existing contracts with national health plans. Finally, some physicians were having misgivings about revealing information about their fees through the bidding process. Confronted with this resistance, which was compounded by the lack of strong support from the St. Louis–based operations of Twin Cities employers, Gateway decided not to move forward with its plans. By that time, Gateway itself had shrunk from 140,000 covered lives to 80,000. None of the six founding member companies was still intact: Four had been bought, one had gone bankrupt, and one had left St. Louis.

Denver

In 1999, some employers in Denver became intrigued with the Patient Choice model when they heard a presentation by BHCAG staff. Subsequently, the Alliance, a Denver purchasing program that focused mainly on small employers, brought BHCAG staff and representatives of supportive Twin Cities employers to the community in late 1999 to speak to its board, inviting large employers who belonged to the Colorado Business Group on Health to attend as well. Several of these large employers had headquarters in the Twin Cities at that time, and it was assumed they would support the development of a Patient Choice approach in Denver. After that meeting, three large employers agreed to fund a feasibility study. They hired an actuarial firm to determine whether there were enough care systems in Denver that might be interested, the geographic coverage of the systems, possible pricing models, and likely implementation issues. By the time the study was completed, the most enthusiastic large employer had been acquired, and the firm's benefits manager had been replaced. Without his leadership, other large employers chose not to move forward with implementation. The Alliance continued to see Patient Choice as a potentially attractive option for its purchasing program targeted at smaller employers. But these small employers were not self-insured, suggesting that Patient Choice would need to be offered as an insured product, presumably through an existing insurer, if it were to enter the Denver market in partnership with the Alliance. The Alliance did not feel that it could bring Patient Choice to Denver on its own, but in what appeared to be good timing from its perspective, a potential partner then emerged: PatientChoice Healthcare was established in the Twin Cities, with the diffusion of Patient Choice to other communities as part of its mission.

The Second Stage: After the Diffusion of PCH

When PCH assumed responsibility for managing Patient Choice in the Twin Cities, its challenges were to stop employers from defecting from the product and to extend its reach there to include medium-size and small employers, presumably through partnerships with insurers or the development of its own insurance license. To do this and particularly to expand to other markets, PCH felt it needed to attract venture capital funding. Its vision at this time was to “look like a health plan” as it entered new markets, but it did not have an internal capacity to process claims or to insure against losses. It saw its core competencies as the development and management of provider networks, and therefore it intended to partner with TPAs and insurers to deliver the product. Because they were unhappy with the services being provided by the TPA under contract for Patient Choice in the Twin Cities, PCH managers felt they could not simply enlarge this contract to cover new communities. Increasing enrollment by entering new communities was seen as necessary in order to show the revenue growth that could attract and retain venture capital support. To attract enrollees, PCH intended to price Patient Choice about 10 percent below the most popular PPO options. PCH managers hoped that insurers and TPAs in those communities would be willing partners because they would view Patient Choice as a way to be more competitive in a marketplace that increasingly valued consumer choice supported by comparative performance data. They also counted on the support of the providers in those communities who wanted a more direct relationship with purchasers. PCH anticipated entering new communities by first connecting with an existing employer coalition, or possibly a small number of large employers, to obtain instant visibility in, and, it hoped, the endorsement of, the employer community. With that visibility, it would approach providers and identify entities that had the capacity to become contracting care systems. It then would assess the level of competition in the health plan market and the availability of possible insurer partners. At that point in each community, a final decision would be made concerning whether or not to move forward. With this process in mind, Denver appeared to be promising. Initial visibility for the product had been secured, and a purchasing coalition had publicly expressed its support, suggesting the potential to secure enrollees relatively quickly. Then PCH encountered a series of obstacles in Denver that it was not able to overcome and that surfaced in other communities as well.

Denver

When PCH decided to enter the Denver market, the Alliance agreed to sponsor it. It rented PCH space for its Denver operations, promoted PCH in the community, and allowed PCH to use the Alliance's existing PPO provider contracts. These contracts were useful in “filling in” any potential gaps that might emerge if some providers declined to participate in Patient Choice as care systems. Care systems were recruited, and by the middle of 2001, the provider network consisted of three tiers (11 systems) and the Alliance PPO, with a January 2002 start-up date anticipated. However, the Alliance was facing its own financial problems, and in late 2001 it sold its PPO to a competing PPO, leaving Patient Choice without a PPO option. The care system contracts also found problems as the start date approached. The Denver hospital market was dominated by two large systems. One refused to be offered as a care system, and it would not participate in a PPO offered by Patient Choice. Lack of access to this system clearly limited the product's attractiveness to Denver employers. PCH also could not establish an insurer partnership in Denver. It did identify a product administrator, although this firm had little experience in marketing products locally. The lack of an insurer partner was a major problem, because the Alliance employers were small and wanted to offer an insured product. In general, the Denver employer community contained a relatively large proportion of middle-size to small firms, suggesting that with or without Alliance support, an insurer partner was necessary to achieve significant enrollment.

While PCH continued to pursue employer contracts in Denver, it was unable to recruit a highly visible champion among large employers. Interestingly, no champion emerged from the Twin Cities–based employers with Denver operations, possibly because of the poor experience of these employers at that time with the Patient Choice TPA. Other obstacles arose as well. The physician groups comprising the contracting care systems lacked brand-name recognition in Denver that might have attracted consumers. In addition, a previous negative experience accepting capitation from a large, out-of-state HMO made physicians’ groups wary of new payment arrangements. Finally, PCH could not locate a reinsurer that would work with small and mid-size employers interested in contracting with PCH on a self-insured basis.

As a result of this early experience in Denver and influenced by its implementation experience at that time in other cities (see later discussion), PCH began to reexamine its initial vision. It decided not to function as a “health plan” when entering communities. Instead, it would focus on establishing provider networks, consisting of care systems and their prices, which could be incorporated into the business of multiple TPAs or offered as part of an insurer's product line. PCH hoped that this reorientation would reduce the resources necessary to enter communities and result in faster enrollment and revenue growth. In Denver, PCH eventually negotiated a partnership with the PPO that had bought the Alliance provider contracts. The goal was for this PPO to offer the care systems as distinct options in its PPO network. The initiative for marketing the product was to come from the PPO, with marketing to begin in 2004. Unfortunately, the PPO was losing market share at the time to a competitor that used one of the two large hospital systems in Denver. It was unable to convince a large employer or the state of Colorado to offer the product, and eventually the Denver office of PCH was closed.

Portland, Oregon

Along with Denver, the most promising new market for PCH in 2001 appeared to be Portland, Oregon. A new employer group, the Oregon Health Care Quality Coalition, had been in operation for two years. Its director supported the concept of Patient Choice and, in particular, its potential to collect and display information on care system quality. But the coalition had no experience with joint purchasing and looked to PCH to provide this expertise. During 2001, PCH established an office in Portland and made contacts with providers to explain the care system model. It found that physicians were receptive, and by the end of 2001 it had signed contracts, based on fee-for-service bids, with ten care systems. Taking a different approach than it had used in the Twin Cities, PCH decided to contract with hospitals directly rather than asking each care system to negotiate its own hospital contracts. It found that the hospital market in Portland was quite consolidated, making it difficult to negotiate what it considered to be reasonably priced contracts. Also, one prominent hospital system had its own health plan and refused to contract with PCH. Another contracting issue arose when a large medical group affiliated with an HMO in Portland decided not to bid as a care system.

At the same time that PCH was developing its network, it also established relations with a local insurer and a TPA that agreed to serve the Portland area. But the insurer's experience was primarily in the small to mid-size employer market, and the TPA had no local presence. With these provider network and collaborative relationships largely in place, PCH began marketing to employers in 2002, hoping to build enrollment throughout that year. It was able to recruit only one group, however, an 85-person, self-insured employer that offered PCH as a “total replacement” product. The PCH staff attributed this disappointing start to several factors. They found that the decision-making process in large firms with respect to adding new benefits was longer and more complicated than they had expected. Also, none of the coalition employers emerged as a “product champion”; all, including three Twin Cities BHCAG employers with a presence in Portland, were hesitant to be the first employer to offer what they perceived to be a relatively untested benefit option. Unlike in the Twin Cities, many large employers in Portland offered only fully insured products. The collaborating insurer did not have a significant history with these employers and, given the perceived uncertainties associated with the product, was reluctant to pursue an aggressive pricing strategy. The facts that the care system network was limited to the Portland area at that time and that most large Portland employers had employees statewide also hampered marketing efforts.

In early 2003, PCH had to decide whether to continue to devote resources to the Portland market or to focus on opportunities in other communities. It renewed its care system contracts but limited its efforts in Portland to discussions with major health plans about partnership possibilities. This approach was consistent with the revised overall vision for PCH and the limited resources that it could devote to direct marketing efforts focused on large employers. The care system contracts were allowed to expire at the end of 2003, and PCH left Portland.

Milwaukee

Milwaukee was the only community that PCH entered without the encouragement and support of an employer coalition. It did, however, have a partnership with a major insurer in Wisconsin and a working relationship with a Wisconsin TPA, as well as the public support of the city's mayor, who was concerned about the high cost of health care in Milwaukee and its role in manufacturing firms’ decisions to shift their operations elsewhere. PCH established an employer advisory group in Milwaukee, but the major coalition of large employers in the community was focused on other issues, and it never became an advocate for PCH. Consequently, many large employers were relatively uninformed about the product, skeptical of its ability to control costs quickly in the face of rapid medical cost escalation, and (in some cases) resistant to the adoption of an “external” solution to this problem. Some benefit consultants in Milwaukee supported the care systems model, especially its potential to generate comparative data on quality. But they saw the new consumer-driven health plans as attractive alternatives to PCH for their clients.

As in other communities, many providers in Milwaukee were receptive to PCH, and by July 2002, eight care systems were under contract. Two large hospital-based systems had not agreed to participate, however, and they were major providers for specific geographic subsections of Milwaukee. Without one or the other, PCH was reluctant to move forward. In its opinion, one of these systems believed its high costs would place it at a disadvantage in a care system model and did not want to risk being the only system in the highest-cost tier. The other system cited a lack of interest by its physicians and, because it already was part of other broad networks, saw only a small growth potential from participating in PCH. Both systems were directly contracting with employers to a limited extent and were considering expanding their contracting efforts. Securing participation from the second system was particularly important because it was a major provider for city employees, and the mayor supported the inclusion of PCH in the city's health benefits program. Without this system, it was unlikely that PCH would be able to attract enrollees among city employees. The start-up date for PCH was delayed as negotiations with these two systems continued. By October 2003, agreement with the second system had been reached in principle, but it was too late for marketing to large employers for the 2004 benefit year.

PCH's Milwaukee operations were affected by the change in its business model, which, by that time, focused on finding a health plan partner. In Milwaukee, PCH collaborated with Wisconsin Physicians Service Insurance Corporation (WPS), an insurer based in Madison. A network was constructed with PCH owning the contracts and WPS serving as the exclusive marketing agent. The first employer (a mid-size firm in greater Milwaukee) offered PCH as an option on July 1, 2004, and 200 employees signed up. According to PCH, the city of Milwaukee was committed to offering PCH on January 1, 2005, and the state of Wisconsin to offering PCH to its 9,000 employees in the Milwaukee area on the same date.

Boston

PCH's involvement in Boston began early in 2002 when the Massachusetts Healthcare Purchasing Coalition and the Association for Industry and Manufacturing (AIM) invited it to conduct a feasibility study. This study revealed interest in the care system model among providers, who liked that they would be reimbursed on a fee-for-service basis without withholds, could specify their fees, and could be rewarded for quality of care. After determining that providers were interested, PCH solicited and received bids from 12 care systems in late June. But because some providers requested extra time, the bidding deadline was extended for all. By January 2003, 17 care systems had responded, but the largest system had declined to participate. PCH and the Boston employer groups were faced with the question of whether the product would attract enough employers and employees to be financially viable without this system. After meeting with a group of employers in April 2003, PCH decided to proceed. Unlike in Milwaukee, several large employers actively advocated PCH, meeting with providers and expressing their support for the product. The care systems under contract were asked to submit new bids, which were grouped into three pricing tiers. A TPA was in place to serve large, self-insured employers, with a local TPA also under contract to serve mid-size employers. Even though negotiations with an insurer were taking place, there was no insured product to offer employers. Another important concern of the PCH staff was that there was no reinsurer willing to collaborate with self-insured employers offering PCH.

At this point, marketing efforts were scaled down, and PCH began to explore the possibility of partnering with an existing health plan in Boston. Without a partner, PCH had concluded that given the entrenched health plans in Boston, it would be difficult and expensive to build enrollment quickly enough to achieve financial viability in a reasonable length of time. It began to work with a specific health plan, analyzing the plan's data and exploring the possibilities for “tiering” its network of providers. This approach was consistent with PCH's new vision for its future, which was to partner with purchasers and plans, providing expertise in network development, but not to enter markets as a competitor to existing plans. A three-tiered model was constructed based on existing provider contracts within this health plan, and marketing for the product began. According to a PCH manager, the Commonwealth of Massachusetts will offer the product to state employees in July 2005.

Interpreting the Patient Choice Experience and Its Implications

In March 2004, Patient Choice Healthcare issued a press release saying that Medica, the largest health plan in the Twin Cities, had agreed to acquire PCH's Minnesota and Dakota operations. The press release reiterated that PCH would continue to operate in these areas as a separate division within Medica, with its current staff. Existing provider contracts would be honored, and the operations of Medica and Patient Choice would be kept distinct.

But the press release also clearly stated the new vision and “business plan” that had evolved from PCH's experience in attempting to enter communities across the country as an alternative to existing health plans. It described PCH as an organization that “works with health plans, preferred provider organizations, government agencies, employers, and payers across the country to measure variations in health care cost and quality and develop innovative programs that enable value-based health care purchasing.” Under this approach, PCH would still “market” variations of the purchasing principles and technical expertise developed as part of BHCAG's experience with Patient Choice in the Twin Cities, but the new marketing target would be existing purchasing entities and the “product” would be, essentially, consulting services. In contrast, when PCH was created, communities were targeted for entry, and the product was a health plan.

This shift in emphasis was driven by unanticipated events, largely external, that affected the demand for Patient Choice, as well as factors that affected the cost of introducing the model into new communities. With respect to product demand, PCH was created at a time when the economy was weakening and health care costs were escalating, placing health benefits managers under intense pressure to improve the corporate bottom line through cost reductions. One could argue that the rising premiums of traditional plans presented an opportunity for PCH, which positioned itself as an alternative to these plans. However, employers were skeptical of its ability to deliver the immediate cost savings they demanded. An essential part of the PCH model was that providers submitted bids, in effect setting their own prices. Indeed, this was an important selling point with providers. Many employers failed to see how this would deliver short-run cost savings even if, by creating competition at the care system level, it would be able to moderate future cost increases. New benefit designs embodied in “consumer-driven health plans” appeared to offer a better opportunity to generate immediate savings, as did simply increasing employees’ share of the cost of existing PPO benefit options.

The hesitancy of large employers to embrace PCH made it difficult for PCH to generate revenues as quickly as planned. PCH had hoped to enter communities with the endorsement of large companies that would quickly agree to offer PCH as part of their benefit packages. Targeting markets in which local employer coalitions endorsed PCH presumably would lead to this result. Local visibility for the plan and momentum in the employer community would be created, and the growing enrollment would generate enough premium revenue to support additional marketing efforts. When this did not happen, PCH was forced to seek alternative funding sources to sustain its community development and early marketing activities. Although it was able to attract some venture capital funds, even in the face of a virtual collapse of the venture capital market, considerably more money was needed to cover the costs of entering the market. The reason was that the PCH model was not simply a change in employee health benefit design; it required building an attractive set of competing care system networks. The need to generate a significant amount of capital in a difficult venture capital market was a drain on the PCH administrators’ time and a significant factor in PCH's decision to alter its business strategy.

While network development was difficult and time-consuming, so was the establishment of the contractual relationships with TPAs and insurers that were necessary to enter the market. A TPA relationship was needed to administer the product for self-insured employers, and PCH had hoped to extend its TPA relationship in the Twin Cities to other communities. The contracting employers and care systems in the Twin Cities were unhappy with this TPA's performance, however, forcing PCH to seek a new TPA at the same time it was trying to enter new markets. This situation was further complicated because even though BHCAG employers in the Twin Cities were self-insured, many large employers in other communities were not. To enter these communities, PCH faced the daunting task of obtaining insurance licenses in multiple states, which meant meeting the states’ financial reserve requirements or finding an insurer that would support PCH by offering it as part of the insurer's “product line.” The latter approach was seen as requiring less capital, but insurers were generally unwilling to offer a product, in the face of an uncertain demand by community employers, for which they had no underwriting experience.

In addition to contracting with TPAs, insurers, and care systems, PCH had to negotiate hospital and PPO contracts in most markets. The Patient Choice product design was based on grouping providers into different cost tiers, with accompanying quality information, and financial incentives that rewarded consumers for selecting a care system from a lower-cost tier. As such, it could be viewed as more advanced than tiered hospital products, because the Patient Choice tiers were based on care systems consisting of physicians and affiliated hospitals, not simply hospitals. In the form that Patient Choice was originally implemented in the Twin Cities, the care systems executed their own hospital contracts. The involvement of the largest Twin Cities employers in Patient Choice encouraged hospitals to contract with care systems. In addition, several of the care systems were constructed around existing hospital systems.

As PCH entered other communities, it was forced to change its approach to hospitals. In some cases, in order to secure care system participation, it had to assume responsibility for establishing a set of hospital contracts that would be available to all care systems. Negotiating hospital contracts at what PCH considered a “reasonable price” became a major issue. The hospitals clearly held the upper hand in these negotiations. Rather than agreeing to participate at high rates, some hospitals declined to be offered as part of Patient Choice. Often the hospitals that declined had very strong reputations in their communities and, as has been the experience with the formation of some tiered hospital products, saw little to gain from participating. In the PCH staff's opinion, these hospitals may have been concerned that their costs were relatively high and that they would be called on to justify these costs within a tiered model that revealed the cost differences. To overcome this problem in most communities, PCH had to enter into a contract with an existing PPO that included the noncontracting hospitals in its network. Some hospitals responded by saying they would not be part of the PPO under these circumstances.

The successful resistance of key hospitals to the Patient Choice model was a major factor in delaying implementation and, ultimately, in causing PCH to abandon its efforts in some communities. One interpretation is that the consolidation of local hospital markets created market power for the dominant institutions (Capps and Dranove 2004; Devers, Brewster, and Casalino 2003; Robinson 2004), which they used to forestall the entry of a health plan that could be disadvantageous to them. While this might be accurate, it does not go far enough. In this instance, the source of hospital market power related not only to hospital consolidation in local markets but also to employer preferences that allowed hospitals to effectively exercise their power (Christianson and Trude 2003). In most communities, employers were hesitant to offer any product, including Patient Choice, unless it permitted access to virtually all providers. Employers with national or statewide workforces posed even greater problems for PCH, as they typically wanted benefit options to provide coverage for all employees in order to minimize the number of their health plan offerings. These preferences played a very important role in employer selection of benefit options, and hospitals were well aware of this. The result was that key hospitals could make PCH entry into their communities very difficult.

Implications for Tiered Hospital Networks

Some health plans are constructing tiered hospital networks in response to the demands of payers to reduce inpatient costs (Robinson 2003, 2004). These networks place hospitals in cost “tiers” and create financial incentives for employees to choose less expensive hospitals. Some tiered models also offer information to consumers about quality of care so that they can consider “cost/quality” tradeoffs when buying inpatient services. Conceptually, because tiered hospital networks can be created within existing PPO options, they are appealing and seem to be feasible, although establishing credible quality comparisons is obviously technically complicated. In practice, implementation has been more difficult than anticipated. Hospitals have resisted the formation of these networks (Robinson 2003), which would reveal comparative costs and, in some designs, outcomes for community hospitals. In particular, entrenched hospitals with strong reputations and patient bases in their communities believe they have little to gain in the form of new patients and potentially much to lose.

There are obvious parallels in the PCH experience. In theory, both Patient Choice and tiered hospital network products appeal to employers’ stated desires to provide financial incentives for employees to consider costs and quality when choosing health care providers (Robinson 2002, 2003). The attractiveness of both to employees also partly depends on the widespread participation of community providers. In practice, unless employers are willing to offer tiered products that do not encompass all available community providers, these products face an uncertain future.

Implications for Purchaser-Driven Managed Care Reforms

PCH's experience in trying to implement a model firmly grounded in the principles of managed competition, as espoused by Enthoven (1988), McClure (1978), Ellwood and colleagues (1971), Havighurst (1970), and others, can be viewed in the context of previous, similar efforts. For example, in reflecting on his experience over 25 years of trying to restructure the health care system under a managed competition model, Enthoven (2003, W3-239) observed that physicians have been very resistant to change in the organization and delivery of care and that employers’ purchasing policies have not “rewarded economical health care.” Regarding the Patient Choice product, he noted that “theoretically a single carrier could offer a suite of different delivery systems. As far as I know, this has happened in only one case, BHCAG. This model implements the principles of managed competition” (W3-244). Enthoven now sees greater promise in other approaches to implementing managed competition principles.

From 1995 to 1999, Medicare tried to demonstrate a new “competitive pricing” model for contracting health plans in four communities. This model required health plans to submit bids, with reimbursement related to the bids. In this key respect, the demonstration design was similar to the Patient Choice model. But at each demonstration site, Medicare was blocked by the opposition of local providers and health plans, which were able to mobilize political support for their views (Dowd, Coulam, and Feldman 2000).

The Medicare demonstrations underscore the practical barriers to implementing managed competition principles, whether undertaken by public or private purchasers. In general, competition does not necessarily benefit the entrenched interests of health plans and providers in local communities. The benefits of implementing a competitive approach vary widely and are likely to occur gradually, if at all, over time. In contrast, the costs are more immediate and likely to be concentrated on a few “losers” who therefore have strong incentives to resist change. Regarding the demonstration of Medicare competitive pricing, Cooper and Vladeck stated: “Real markets have losers. Without them, it is difficult to achieve much efficiency. In a political system, losers, potential losers, and even those who feel that they might someday be losers, often seek redress from their elected officials” (2000, 53). The same might be said for the PCH experience, during which potential losers under an effective care system model were able to prevent PCH's entry into local markets. Unlike the Medicare experience, in which the competitive pricing demonstrations were stopped by the political process, PCH's entry was effectively blocked at the local level through the market power of providers, amplified by employers’ reluctance to offer a product without very broad provider access. Both experiences illustrate that the tradeoff between diffuse and uncertain future benefits versus immediate and concentrated losses is often decided in favor of the latter.

Implications for Exporting Health System Change

It would not be stretching matters too much to characterize the health care system in the United States as a very large set of community-based, ongoing “experiments,” tied together by federal programs and regulations, with each reflecting a unique history and configuration of key actors: employers, health plans, and providers. Wholey and Burns noted that the unique conditions in the Twin Cities market supported the Patient Choice product and speculated about the potential barriers to implementing Patient Choice in other places:

One question of obvious interest is whether the BHCAG experiment can be replicated in other geographical markets. An analysis of network structure and interdependence in the Twin Cities suggests it may not. One factor facilitating BHCAG's emergence is the social network structure of corporate headquarters in the Twin Cities … another unique characteristic is the large number of multispecialty groups in the area. (2003, 133)

Both pre- and post-PCH diffusion efforts were predicated on the assumption that Twin Cities’ members of BHCAG would encourage their affiliates in targeted cities to adopt the Patient Choice model, as well as the hope that other large employers would emerge to champion the product both individually and through purchasing coalitions. That this did not happen speaks to unique features of the Twin Cities market and of BHCAG that were not present in other markets. In 1994, 78 percent of the 3.2 million Minnesotans in managed care plans were enrolled in the “Big 3” HMOs—Medica, HealthPartners, and Blue Cross—leading to demands by both purchasers and providers for more health plan options. The BHCAG coalition itself was the successor to collective action efforts on the part of Twin Cities employers that stretched back into the 1980s and earlier. These conditions were not found in other markets.

Moreover, as the diffusion efforts unfolded, expected employer support was eroded through mergers, downsizing, and changes in corporate headquarters. As mentioned earlier, the Gateway Purchasers in St. Louis lost all of its founding members, and the experience in Denver showed that the acquisition of a single large employer supporter could delay or derail implementation of the Patient Choice purchasing model.

More generally, Wholey and Burns (2003) emphasized that community health care markets have their own histories and developmental paths that are important to understanding the present configuration of these systems and, by extension, their evolution. In framing their arguments, they drew from the sociological literature on the “embeddedness of economic behavior” and the considerable literature on “institutional economics.” Health services research provides numerous empirical examples supporting this view. For example, Burns and colleagues (1997) found that contrary to conventional wisdom, community health care systems did not evolve in parallel ways in response to the development of managed care. Furthermore, the success of a managed care organization in one market does not guarantee that it will be successful in others. For example, Gitterman and colleagues (2003) recounted in detail how local market factors in North Carolina influenced and ultimately led Kaiser Permanente to close its operations in that state. In fact, Kaiser has met with varying degrees of success as it has attempted to expand to communities. Recent cross-community collaborative initiatives also highlight the importance of local market characteristics in understanding variation in outcomes (Bazzoli et al. 1997; Shortell et al. 2002; Wickizer et al. 1998). It appears that the adage “all health care is local” is not only useful in reminding us of substantial differences among health care systems across geographic areas (Kerr et al. 2004; Lesser, Ginsburg, and Devers 2003); it also underscores the difficulties that are likely to be encountered in transporting those “experiments” that have been successful in one community to other communities, whether through private or public initiatives.

In addition, one of the key insights from BHCAG's and, subsequently, PCH's experience is that health system change does not occur in a static environment. Market conditions can change quickly, with the implication that attempts to duplicate successful local interventions may fail because the initial supportive conditions are not present in the target market (and possibly no longer in the “home” community either). For example, tiered care system networks were sold to employees and employers in Minneapolis as an expression of consumer choice because, in comparison to HMOs, they would provide greater access to providers. But, by the time the Patient Choice model moved to other cities, broad provider networks had become the norm, and tiered networks were seen as offering less choice. On the supply side, hospitals were becoming more consolidated, making it more difficult to form provider tiers compared with the experience in the Twin Cities. In addition, there were relatively few multispecialty groups in other communities that could serve as building blocks for the care system model or feel invested in its success. The emergence of consumer-driven health plans as an alternative to Patient Choice also dampened purchasers’ enthusiasm for tiered provider networks.

In a sense, by the time Patient Choice reached other cities in 2001 and 2002, providers already had “won the war” with managed care. In order to gain providers’ acceptance in other communities, Patient Choice had to drop its expenditure target system and revert to simple fee-for-service reimbursement. Providers waited to sign up until Patient Choice could promise to deliver enrollees to them, which created a “chicken and egg” problem with employers, who wanted a quick fix to the problem of rising health care costs.

In conclusion, what is a reasonable interpretation of the experience of BHCAG and Patient Choice? We believe the BHCAG model provided important evidence that a purchasing initiative that (1) is based on direct contracting with providers, (2) employs innovative payment methods featuring risk-adjusted bids submitted by the providers themselves, and (3) incorporates consumer incentives for making price- and quality-conscious choices, can be developed and fielded with considerable local success. Furthermore, BHCAG's experience with the individual features of its purchasing model has already provided important information for health policy relating to risk adjustment, the development of comparative quality and consumer satisfaction data at the provider level, and pay-for-performance initiatives. The evidence also suggests, however, the importance of the health care environment in shaping community-level reform initiatives, as well as the impact that changes in that environment can have on their success over time.

In the opinion of one interview respondent, greater commitment by public employers and the CEOs of large private companies is important to the adoption and growth of a competing care system model. But it appears to us that unless employers can give up their short-term thinking about health insurance, as exemplified by their seemingly perpetual search for a “quick fix” to the problem of rising health care costs, they are not likely to become health insurance innovators. Stated another way by the same respondent, “Employers are wimps.” It may be the case that, for better or worse, consumers will have to be the “drivers” of change. For example, Patient Choice or a similar tiered care system product might be grafted onto a consumer-driven health plan, instead of using a conventional PPO as is now the case in most consumer-driven health plans. Patient Choice's strategy of differentiating providers with respect to cost and quality differences and providing information for consumers to use in making informed choices fits nicely with the philosophy of consumer-driven health plans and with employer pay-for-performance initiatives. It remains to be seen whether a health plan featuring a combination of Patient Choice's approach and the consumer-driven philosophy will emerge and attract consumers.