Objective: To examine health plan strategies, planning, development, and implementation of pay-for-performance programs (financial incentives for hospitals and physicians tied to quality and efficiency) at the community level, focusing on differences across markets.
Study Design: A fifth round of site visits to 12 nationally representative metropolitan areas between January 2005 and June 2005, based on more than 1000 protocol-driven interviews with representatives from health plans, provider organizations, employers, and policy makers.
Methods: In each of 12 communities, we interviewed several executives from 35 health plans, including chief executive officers, marketing executives, and network contracting directors. Additional perspectives were obtained from representatives of employers, large medical groups, and hospital systems.
Results: Growing numbers of health plans are developing and implementing pay-for-performance programs for physicians and hospitals. Although in their early stages, plans' customized programs show substantial design variation within and across markets. This design variation reflects local conditions that include information technology capabilities, data availability, relative leverage of health plans and providers, willingness of providers to participate, and employer influence. The concerns of providers include the administrative burden of health plans' customized programs and the potential for conflicting financial incentives.
Conclusions: Most health plans are committed to pay-for-performance programs. Although providers would prefer health plans in their communities to use a single standardized set of measures and methods, this is unlikely given local market environments. A national effort directed at standardization might significantly reduce the extent of customization but also may limit the opportunities for local collaboration with providers.
(Am J Manag Care. 2006;12:537-542)
Health plans are increasingly turning to financial incentives for providers to control costs, encourage provider efficiency, and improve quality of care in the American healthcare system. The number of studies evaluating cost-effectiveness and quality improvement for specific pay-for-performance programs for physicians and hospitals is growing.1-4 In addition, issues of programmatic challenges and physicians' perspectives have been raised.5-7 However, less is known about how these programs are being received and implemented at the community level, especially during the early stages of development.
Three years ago, during the 2003 site visits of the Center for Studying Health System Change (HSC), health plans in 7 of 12 communities had implemented pay-for performance programs. At that time, there was substantial variation in the ways in which health plans measured performance and quality, as well as in the amount and structure of provider incentives.8 Health plans' efforts to improve quality and provider efficiency have expanded, so that health plans in all 12 communities are developing or have implemented pilot or full pay-for-performance programs for physicians or hospitals.
In this article, we examine health plans' strategies and approaches for developing and implementing pay-for-performance programs in the 12 communities. Rather than examine best practices or the experience of high-profile national pay-for-performance programs, we instead interviewed several health plans within the 12 nationally representative communities and describe the range of approaches to pay-for-performance programs in the context of different market environments. We describe health plans' choices of performance measures, financial incentives, and program design, as well as provider and employer views of these efforts and potential barriers to their viability within some markets. This contrasts with the study by Bodenheimer et al,9 who used the same research data but restricted their analysis to fully implemented pay-for-performance programs for physicians, excluding hospital-based programs, pilot programs, and programs under development.
This study used qualitative interview data from HSC site visits in the following 12 nationally representative metropolitan areas: Boston, Mass; Miami, Fla; Orange County, Calif; Northern NJ (Essex, Morris, Sussex, Union, and Warren counties); Cleveland, Ohio; Indianapolis, Ind; Phoenix, Ariz; Seattle, Wash; Lansing, Mich; Syracuse, NY; Greenville, SC; and Little Rock, Ark.10 During the most recent (fifth) round of site visits, carried out between January 2005 and June 2005, more than 1000 protocol-driven interviews were conducted with representatives from provider organizations, health plans, employers, and policy makers.
Our study elicited viewpoints about pay for performance from representatives of commercial health plans, large employers (with >500 employees at the HSC site), hospitals, and large physician groups, using neutral open-ended questions about health plans' use of financial incentives to influence the quality or efficiency of care delivery. Therefore, the findings reflect respondents' current priorities and concerns rather than their reaction to a predetermined set of issues. However, given the standardized nature of the protocols, comparable open-ended questions were asked across communities. As a result, each effort (from a fledgling pilot development to a well-established program) received equal weight, resulting in potential overemphasis on modest efforts and too little detail for large well-established programs. Furthermore, because the findings reflect respondents' perspectives, they are taken at face value but are characterized within the context of the key stakeholders (plans, providers, and employers). Therefore, findings can speak to the interplay between stakeholders within a community but not to the effectiveness of any particular program or approach.
In each community, we conducted interviews at the 3 largest health plans, based on enrollment, typically including a regional Blues plan, a national plan, and a local plan. Exceptions included 4 health plans in Orange County and 2 health plans each in Little Rock and Syracuse, communities in which a single plan dominated the market. We discussed health plan pay-for-performance activities with several executives from 35 health plans, including chief executive officers, marketing executives, and network contracting directors.
Because our interviews were conducted at the community level, we did not systematically capture national pay-for-performance efforts. This is especially true for efforts sponsored by the largest national or global employers.
Strategies for Program Development
Most health plans (77% of 35 plans) had hospital- or physician-based pay-for-performance strategies that were being actively developed or had pilot or full programs that had already been implemented. Most of the health plan efforts were new, with about one third of all reported efforts being in the planning or developmental stages. As noted by Bodenheimer et al,9 large ongoing pay-for-performance programs that directly target physicians are less common. Among our study sites, health plans in Boston, Orange County, and Lansing had the longest-standing and most comprehensive efforts.
The health plan pay-for-performance efforts ranged in size from small pilot programs targeting particular diseases, such as diabetes mellitus, to comprehensive efforts comprising separate components targeting primary care physicians, specialists, and hospitals. Similar to findings by Rosenthal et al,5 the programs varied in size in terms of the number of providers or medical groups that participated in the program, as well as in the total covered lives that the program affected. For example, some health plans limited their programs to physicians or medical groups that served a high volume of their enrollees.
Representatives of health plans investing in pay-for-performance programs uniformly reported that their goal is to reduce costs through improved quality and provider efficiency. In contrast, plans choosing not to develop pay-for-performance programs gave a range of different reasons for their reluctance. Three health plans, in Miami, Little Rock, and Cleveland, chose to develop provider performance-based measures for consumers rather than to implement provider financial incentives. Two other health plans, in Phoenix and Greenville, reported that they lacked sufficient resources. One of these plans had previously tried a pay-for-performance program but abandoned it because of the administrative burden. Two recently acquired health plans were waiting for direction from their national corporate offices before developing programs. One health plan reported that financial incentive programs did not fit well within a staff-group model health plan in which physicians had less discretion in their practices.
Local Market Adaptations
Local market conditions led health plans in the 12 HSC communities to customize their pay-for-performance programs to reflect the willingness of providers to participate, employer interest and influence, data availability, information technology capabilities, and relative leverage of health plans and providers. The health plans in Boston implemented sophisticated pay-for-performance programs that include primary care physicians, medical groups, specialists, and hospitals and that have stronger financial incentives compared with programs in the other HSC communities. The relative sophistication of the programs in Boston reflects previous experience in hospital and physician contracting, the commitment of physicians to information technology development, and the area's longstanding involvement in quality initiatives. In contrast, pay-for-performance efforts in Syracuse remained in the pilot and planning stages, even though one health plan has roughly 70% of health plan enrollment, based on our findings. In Syracuse, where indemnity health insurance products dominate, that health plan was converting its physician claims system from revenue centers that provide limited information to medical claims that itemize the services provided. Health plan respondents also noted that Syracuse physicians have been reluctant to invest in new technology infrastructure. As a result they have limited their pay-for-performance efforts to hospitals, where portions of annual rate increases for hospitals with more than 100 beds were tied to the Centers for Medicare & Medicaid Services (CMS) quality measures.
The leverage of a hospital system or a physician group in a community can influence how health plans develop their pay-for-performance programs. Because of past contentious relationships with area providers, health plans in Seattle have taken a cautious approach. In 1999 and 2000, some prominent Seattle providers terminated their contracts with health plans in publicized disputes.11 In response, health plans made improved provider relations a top priority, and 1 plan developed its program collaboratively with area providers. In the Cleveland area, the clout of the Cleveland Clinic Health System has limited health plans' pay-for-performance efforts. Area health plans have chosen instead to develop their pilot pay-for-performance programs in other Ohio markets. One health plan has a small pilot program for physicians through the Cleveland Clinic Health System. Although this program targets physician performance, payments are made to the participating Cleveland Clinic Health System hospitals, and physicians reportedly are largely unaware of the program.
Employer interest and influence within a market also affected how health plans designed and developed their pay-for-performance programs. For example, representatives from a few health plans reported that it is difficult to convince self-funded employers to pay extra to reward performance of providers. As a result, a health plan in Seattle in which about 40% of enrollment is self-funded chose not to use bonus payments because this would have required negotiating with employers over providing additional funding. Instead, this plan chose to use performance information in negotiations with providers regarding annual payment rate increases. Medical groups must justify requests for large payment rate increases in light of their performance scores relative to those of other medical groups in the market.
In several HSC communities, pay-for-performance programs were being developed as community-based efforts, typically led by employer-based organizations. In Seattle, the Puget Sound Health Alliance was formed to develop evidence-based protocols, a data warehouse, and a standardized approach to performance measures. A public employer in Seattle assembled a task force to identify potential long-term solutions to rising healthcare costs. This led to the formation of a partnership, including private and public employers, health plans, and providers, with the goal that broad participation will lead to a unified approach across stakeholders as the effort progresses. For example, 2 insurers in the Seattle market have different sample size thresholds for quality measurement.
In Phoenix, respondents reported that some large national employers and the Human Resources Policy Association support development of standardized performance measures across plans for their Bridges to Excellence and LeapFrog Group efforts. Given that more than 12 health plans compete in the Phoenix market, getting health plans to collaborate is considered crucial to obtaining sample sizes large enough to measure quality and efficiency for individual physicians. In another collaborative effort, the Employers' Forum in Indianapolis is calling for the development of pay-for-performance programs and a tiered network product.
Group efforts to identify and implement a core set of performance measures within a market may prove problematic. In Orange County, health plan respondents described their participation in California's Integrated Healthcare Association pay-for-performance effort launched in January 2002. This initiative provides a structure for performance measures, such as having 50% of the performance based on clinical measures, 40% on information technology, and 10% on patient satisfaction. Despite this, participating health plans are diverging in the measures that they use. One plan added the requirement that providers report performance with respect to medical management, regulatory compliance, and resolution of grievances and appeals. Another health plan incorporated a financial incentive for medical groups based on the hospital to which the physicians admit their patients. Another health plan in Orange County added access measures, such as whether a patient is seen within a certain length of time for a routine problem.
Customization of Measures and Financial Incentives
Health plans' adaptation of their pay-for-performance programs to local market conditions led to substantial variation in their types of measures and financial incentives, often reflecting plans' differing expectations about the ability of providers to meet performance thresholds. The approaches for rewarding providers roughly fell into the following 3 categories: comparing performance with that of their peers, reaching absolute targets of performance, and demonstrating improvement over previous scores. A few health plans used a combination of approaches, with the first part of the incentive tied to performance relative to that of peers and the second part tied to improvement over previous scores.
In addition, health plans' choice of financial incentive may depend on the performance measure and expectations for changing provider practices. One health plan reported using a different financial incentive for each of the following 3 hospital quality initiatives: (1) hospitals receive a higher reimbursement rate for 1 year if they rank above the 90th percentile in 4 of 10 quality measures submitted to the CMS; (2) hospitals receive a higher reimbursement rate if they report all of the Leapfrog Group measures; and (3) hospitals receive a lump-sum grant if they implement a computerized physician order entry.
The amount of money that health plans put at stake in their pay-for-performance programs ranges from nominal amounts to substantial sums. For example, a Lansing health plan holds back 4% of a participating hospital's reimbursement for use in its financial incentive program. One health plan in Miami provides mean rewards of $4000 per physician, with a maximum possible reward of $12 000 per physician. An Orange County health plan negotiates its performance-based incentive separately from the capitation rate, with the incentive ranging from 1% to 5% of the capitation rate. A few representatives of health plans said they believed that plans must tie at least 10% of provider compensation to performance to change physician practices.
The proportions of participating providers who receive a bonus vary among programs. Some health plans reward only a small percentage of physicians achieving the highest level of performance, while other health plans give most physicians a bonus, but with larger payments going to the highest-rated physicians. For example, an Orange County health plan provides a per-member-per-month bonus to 60% of the medical groups participating in the program. In Boston, a plan provides a per-member-per-month bonus to about 80% of participating physicians, with 31% of participating physicians receiving the maximum bonus.
Views of Providers
Representatives of health plans and providers were asked about potential challenges implementing financial incentives or pay-for-performance programs. With most efforts at their early stages, respondents typically raised issues that were relevant to program development. Future interviews might identify new concerns as these programs progress, such as the amount and type of financial incentive to employ and how to distribute the monies, as in the description by Bokhour et al12 of a Rewarding Results demonstration.
Health plan representatives reported that their key challenge was gaining acceptance from providers. One respondent said that providers are supportive of pay-for-performance programs in theory, but that the sticking points are in the rules of the game. As a result, most health plans developed their pay-for-performance programs cautiously and often collaboratively with providers. Collaborative approaches include having a group of community physicians consult on the program design or implementing a pilot program with local medical groups or hospitals that are interested in pay for performance.
Providers reported concerns about plans' choice of measures, potential administrative burden imposed by the measurement process, and inconsistencies stemming from customized programs. Providers who were not involved in the development of a program were often skeptical of health plan decisions regarding program design. Physicians were concerned that many of the health plan measures depended on too few observations of performance, especially from health plans with limited market share. Developing valid measures for medical specialists was considered particularly challenging because of the need for different performance measures for each medical specialty, and because of the treatment by medical specialists of less common conditions, with treatment tailored to the individual needs of the patient.
Physicians also thought that financial incentives should be tied only to performance measures that they could influence through changes in their medical practice, not to patient behavior or to decisions stemming from a patient's health insurance benefit design. For example, in Syracuse, which has a large indemnity market, a patient might decline to seek preventive care because of out-of-pocket costs, or a preferred provider organization member might access specialty care providers directly (without a primary care physician who coordinates the care), resulting in several different physicians independently treating the patient.
Some providers described the need for consistent measures across health plans, noting that they had received conflicting scores and incentives from health plans within the same community, scoring high by one set of standards and scoring low by another set. In addition, the lack of standardized measures was said to increase the reporting burden for providers in the absence of common data sources such as medical claims. Health plan representatives reported that a few hospitals asked them to use only the 10 CMS performance measures, but the plan representatives thought that this would unduly limit the scope of their programs. An Orange County respondent expressed concern that, while the number of performance measures used in pay-for-performance programs is increasing over time, the money available to the programs is not.
Providers questioned other stakeholders' interests, noting that health plans could potentially choose measures that are better suited to selling their product to employers, but that such measures would represent little clinical value among providers.
Role of Employers
Local benefit managers showed limited interest in plan efforts to implement pay-for-performance programs and typically considered these efforts as part of health plans' traditional role in contracting with network providers. Benefit managers were more interested in financial incentives for consumers to improve healthcare decisions through participation in disease management and wellness programs or via consumer-oriented health plan products.
Local benefit managers who were aware of health plans' use of provider financial incentives gave mixed reviews. A few managers expressed their belief that providers should not be paid more for what they already should be doing, while another manager noted that these programs are hard to implement in a preferred provider organization environment if a patient is not tied to a single physician. Another benefit manager thought that providers should lead the effort instead of health plans, so that providers could develop a single standardized set of measures.
The growing numbers of health plans developing and implementing pay-for-performance programs and committing resources to these efforts demonstrates health plans' expectations that these programs will improve quality and moderate cost growth. However, several strategic and operational issues could dampen this growth. For example, health plans' customization underscores the lack of consensus across plans regarding the most effective approaches. Furthermore, health plan decisions about program design often reflect local market conditions.
Provider concerns about the administrative burden of the reporting requirements of these customized approaches and about the potential for conflicting financial incentives may work against effective implementation within communities.13 Eddy6 calls for a single national core set of measurements as the only viable solution that would meet the needs of large national corporations and health plans. At a minimum, Forrest et al7 argue that a common set of measures and methods should be implemented within each medical marketplace. Epstein et al14 observe that developing common measures is important but requires the cooperation of multiple health plans. They suggest that this cooperation may be difficult to achieve because many plans seek a competitive advantage through their pay-for-performance programs. As our findings suggest, there is a small number of HSC communities in which groups have formed to address these issues. However, this type of collaboration seems unlikely in communities in which a single stakeholder exerts considerable leverage, purchasers have little clout, or contentious relationships among stakeholders persist. Furthermore, despite California's Integrated Healthcare Association efforts to standardize pay-for-performance measures, health plans in Orange County have continued to add individualized measures.
Although challenging and unlikely to lead to complete standardization across health plans and within communities, a national effort to develop, coordinate, and update pay-for-performance measures and methods might reduce the extent to which health plans expend resources inefficiently or implement less effective approaches. A national set of measures and methods could serve as a gold standard for health plans in less sophisticated markets and as a minimal standard in others. Some movement in this direction seems to be occurring. For instance, the CMS, American Medical Association Physician Consortium, and National Committee for Quality Assurance developed an ambulatory care performance measurement set that was submitted to the National Quality Forum (a nongovernmental entity) for review.15 The Ambulatory Care Quality Alliance work group, including the American Academy of Family Physicians, American College of Physicians, America's Health Insurance Plans, and Agency for Healthcare Research and Quality, also is seeking to develop a set of measures of quality for ambulatory care.16
The call for a national effort to coordinate pay-for-performance efforts raises the question of the role of the Medicare program. Berwick et al17 note that whatever the Medicare program does in terms of pay for performance is likely to have a significant effect on the broader healthcare system and advocate that the CMS should implement a major pay-for-performance initiative to spark similar efforts by private payers. Three years after this strategy was proposed, our findings suggest that additional CMS leadership is probably not needed to convince the private sector to pursue pay-for-performance programs; in the 12 HSC communities, pay-for-performance initiatives seem to be proliferating. However, sustained leadership by the CMS might improve standardization and allow systematic evaluation of the effect on patient outcomes, taking advantage of the large denominator of Medicare beneficiaries.17 The current administration in Washington says it is committed to developing payment systems that reward quality, has initiated a voluntary program for physicians to report quality data on 36 measures, but opposes sharing Medicare data with the private sector because of privacy concerns for physicians.18,19
Regardless of which entity ultimately assumes leadership in developing national standardized performance measures for physicians and hospitals, our research suggests that implementing these measures may meet with resistance in some communities. Although providers may value standardization to reduce the costs of responding to the diverse reporting demands of multiple health plans, they may question the legitimacy of measures constructed and implemented without local input.
From the Center for Studying Health System Change (ST and JBC) and Mathematica Policy Research (MA), Washington, DC; and the Division of Health Policy and Research, University of Minnesota School of Public Health, Minneapolis (JBC).
This research was conducted as part of the Center for Studying Health System Change Community Site Visit project, which is funded by The Robert Wood Johnson Foundation.
Address correspondence to: Sally Trude, PhD, Center for Studying Health System Change, 4078 Slam Gate Rd, Crozet, VA 22932.