High-Performance Health Plan Networks: Early Experiences

Issue Brief No. 111

May 2007

Debra A. Draper, Allison Liebhaber, Paul B. Ginsburg

Health plans began rolling out high-performance networks designed to steer enrollees toward network providers — primarily physician specialists — who demonstrated strong results on efficiency and quality metrics. The earliest adopters of these networks were large national employers, and although additional employers showed interest, actual uptake remained slow, according to research by the Center for Studying Health System Change (HSC). Enrollment in products built around high-performance networks was limited, and rigorous evidence regarding impacts on utilization, spending, and quality remained unavailable. Initial lessons from these efforts highlighted the importance of transparent communication between plans and providers, incorporating both efficiency and quality indicators, developing industry-wide performance standards, and securing active employer support.

Network Strategy to Lower Costs and Improve Quality

In the wake of the managed care backlash during the mid-1990s, health plans expanded their hospital and physician networks in response to consumer demand for greater choice. This approach weakened plans' ability to control costs, as hospitals gained leverage to negotiate higher payment rates as a condition of continued network participation, and plans found themselves less able to exclude physicians viewed as relatively inefficient or offering lower quality. More recently, however, health plans developed high-performance networks aimed at directing enrollees toward network physicians who scored well on efficiency and quality indicators. Plans analyzed claims data to evaluate network physicians based on efficiency using per-episode costs — such as the treatment of low back pain — and on quality metrics that could be derived from claims data, such as hemoglobin A1c blood testing for diabetic patients. At the market level, if these networks succeeded in shifting enough enrollees toward high-performing providers, physicians who lost market share could be motivated to enhance their efficiency and quality to remain competitive. HSC researchers first observed the emergence of high-performance networks in early 2005 during the fifth round of the Community Tracking Study (CTS) site visits. Because activity was insufficient to analyze at that time, an in-depth examination of high-performance networks was undertaken later (see Data Source).

High-Performance Network Specifications Differ

High-performance networks represented a recent addition to the toolkit that health plans and purchasers were deploying to try to reduce costs — frequently described in terms of efficiency gains — and elevate quality. The reach of these networks was limited, as plans had introduced them only in select markets. In markets where these networks were available, the number of enrollees participating depended largely on whether employers incorporated high-performance networks into the health insurance benefits they sponsored.

High-performance networks generally were not standalone products but rather an option layered across different product platforms, most commonly preferred provider organizations (PPOs).

The precise design of high-performance networks varied from plan to plan. The most prevalent model employed tiered-provider levels with corresponding enrollee cost-sharing differentials. The first tier comprised high-performing providers; the second tier included remaining in-network providers; and the third tier encompassed out-of-network providers. Employers frequently chose not to differentiate cost sharing between the first and second tiers, instead offering these networks purely as an informational resource for employees about which providers demonstrated superior performance.

Targeted Providers

Plans predominantly targeted physicians — generally specialists — for inclusion in high-performance networks; hospitals were typically excluded. Evaluations of which physicians to designate as high performing were carried out by specialty, though some multispecialty group practices pushed plans to include all of their specialties. The criteria plans used to select which specialties to include centered on those that:

- Accounted for a substantial portion of medical expenditures; - Exhibited meaningful variation in costs and quality; - Generated adequate claims volume to evaluate physician- or practice-level efficiency and quality; and - Had established quality measures and/or clinical guidelines against which to benchmark performance.

Most plans encompassed specialties like cardiology, gastroenterology, orthopedics, and obstetrics/gynecology, but beyond those, consistency diminished. For instance, CIGNA's high-performance network covered roughly 20 specialties, whereas Blue Cross of California included all network physicians, including primary care physicians (PCPs). Some plans, such as Aetna and UnitedHealthcare, initially concentrated on specific specialties but later expanded to incorporate others.

Most plans excluded PCPs from their high-performance networks. Plan executives explained that this was done to avoid disrupting patients' established relationships with their primary care doctors — a situation that arose less frequently with specialists. While Tufts Health Plan in Boston did not tier PCPs at the time, it did furnish enrollees with performance data on these physicians. Tufts identified "blue ribbon" PCPs — approximately 20 percent of the plan's PCPs who had achieved the highest scores in the plan's assessment of costs and quality.

Although most plans did not target hospitals for high-performance networks, hospitals were relevant to physician assessments because total claims costs per episode of care — including hospital costs and prescription drugs — were used to evaluate physician efficiency. A benefits consultant explained that the emphasis on physicians was primarily because, except for complex care such as transplants where patients more often select the facility, people choose their physician, who then directs them to a hospital. Another benefits consultant observed that because hospitals frequently possessed considerable market leverage, they could resist plan initiatives like high-performance networks that threatened their market share. Evidence of this was the limited traction gained by tiered-hospital networks, which plans had introduced several years earlier. Tufts Health Plan was an exception — its high-performance network originated as a tiered-hospital network and subsequently expanded to encompass physicians.

Profiling Methodologies

The methods plans employed to profile physicians for high-performance network inclusion typically combined some assessment of costs and quality. Nearly all plans measured costs using episodes of care attributed to the responsible physician. (1) Plans constructed these episodes using the total costs of care associated with an enrollee, encompassing inpatient and outpatient facility costs along with prescription drug expenses. As one plan executive put it, while they recognized that physicians were not entirely responsible for costs incurred at the hospital, they believed it was beneficial for physicians to consider the efficiency of the hospitals with which they practiced.

There was considerably more variation in the quality measures plans used for physician profiling. In general, however, quality assessments were grounded in physicians' adherence to evidence-based medical guidelines and consensus-based quality standards. These assessments tended to be constrained by what could be evaluated through claims data. For example, the Boston-based Group Insurance Commission (GIC), which managed health insurance benefits for public employees in Massachusetts, required all contracted plans to include high-performance networks and to utilize nearly 60 specified quality measures, including those from the Health Plan Employer Data and Information Set (HEDIS), the Agency for Healthcare Research and Quality, and specialty society best practices. The number of measures was expected to grow by approximately one-third that year. By contrast, Blue Cross of California relied on approximately 20 HEDIS-type measures to gauge quality.

Drawing on cost and quality data, plans applied an algorithm incorporating both dimensions to further evaluate physicians, though the precise specifications of these algorithms were typically proprietary. Some plans built their high-performance networks primarily around costs. For others, a physician first needed to satisfy a quality threshold before being evaluated on efficiency.

Profiling was conducted at the individual physician level, but individual results were then aggregated at the group level — the level at which contracting decisions, including high-performance network designations, were typically made. There were, however, complexities in profiling at the individual versus practice level. For example, one plan executive explained that they examined both the group and every doctor in it but did not selectively designate individual physicians as high performing while excluding others in the same group — since patients at group practices do not always see the same doctor and cross-coverage issues arise. A benefits consultant added that another complication was the potential for provider pushback at the practice level: a group practice might insist that all members of the group be designated as high performing or threaten to terminate its contractual relationship with the plan. Depending on how important the group was to the plan's provider network, such scenarios could present significant challenges.

Plans differed substantially in the share of network providers designated as "high performers." A benefits consultant noted that in terms of where they set the bar, carriers were all over the map — there was no common benchmark of 70/30 or 50/50, and none of the carriers went in with a predetermined percentage in mind. Plan executives reported a wide range — 25 percent to more than 80 percent — of physicians included in their high-performance networks compared with their broader networks.

Physician Incentives

Most plans reported that payment rates for physicians within their high-performance networks were identical to those for physicians in their larger networks. Plan executives indicated that while they paid the same for the present, they were considering some form of differentiated payment structure in the future. One plan, for example, was exploring modifications to its incentive program to better reflect quality outcomes. Another was examining how to integrate pay-for-performance metrics.

Not all incentives for physician inclusion in plans' high-performance networks were directly tied to payment. Enrollees could learn about high-performing physicians through plan directories, which often highlighted designated physicians. As one plan executive explained, their broader expectation was that by offering members a choice model and incentives to seek care based on the best available information, it would reward providers with increased membership or help them retain their existing market share. To the extent that plans profiled and ran these programs, physicians would become more aware of their performance and improve accordingly. Another plan executive described focusing on improving data capability, recognizing the top 10 percent of providers and providing the bottom tier with specific data for improvement — adding that if performance did not improve over time, the plan might revisit whether those physicians should remain in the network.

Enrollee Incentives

Employers determined how high-performance network benefit structures were configured, including any incentives to encourage employee enrollment. An initial decision employers faced was whether to offer the high-performance network as the sole network option or to also provide access to a broader provider network. Large employers tended to present high-performing networks as one among several network choices.

Employers then made decisions regarding enrollee cost-sharing requirements. Benefits consultants and plans frequently recommended cost-sharing differentials for high-performance networks, reasoning that such differentials were critical to steering enrollees toward higher-performing providers. One plan, for instance, generally suggested a 1.5 times differential for copayments — $20 for high-performing physicians versus $30 for other in-network physicians. Another plan made specialists in its high-performance network available at the lower PCP copayment level.

However, employers were frequently reluctant to institute substantial — or any — cost-sharing differentials between physicians designated as high performers and other network physicians. Some respondents believed that imposing large cost-sharing differentials might be premature until there was greater confidence in these networks.

Market Responses

Physicians

Market reactions to high-performance networks varied widely depending on how the networks were introduced. Physicians voiced some concern about these networks across most of the target markets. The most frequent complaint centered on insufficient communication by health plans. Physicians reported feeling uninformed about their designations when high-performance networks were launched. Moreover, physicians indicated that plans did not always fully explain how performance was assessed or share data comparing their performance to that of other physicians.

Physicians also raised concerns about the methodologies used to determine high-performance designations. They questioned data quality and whether adequate sample sizes were employed. They further questioned whether the methodologies were appropriately applicable across all physicians. In one prominent case, the Washington State Medical Association brought suit against Regence Blue Shield, alleging the plan relied on flawed methods and outdated information to exclude physicians from its high-performance network. (2)

Physicians additionally objected to the lack of methodological standardization. Because plans defined high-performance networks differently, certain physicians received a high-performing designation from one plan but not from another. This inconsistency bred general skepticism regarding these networks and the approaches used to construct them.

Plans' market share also influenced physicians' responses. Plans with low market share worried that physicians would more readily dismiss their data due to smaller sample sizes, or that physicians dissatisfied with their designation might simply drop out of the network with minimal economic consequences.

Despite resistance to high-performance networks in certain markets, plan executives reported monitoring those situations closely and adjusting their approaches to reduce conflict. The experience in St. Louis was cited as one such example — in that market, several large providers resisted UnitedHealthcare's effort to introduce a high-performance network after they were excluded from the plan's high-performance tier. (3) Most plan executives emphasized the necessity of effective communication with physicians to minimize friction, and they acknowledged that engaging physicians in the process sooner rather than later was an important way to accomplish this. Plans that incorporated physician feedback while developing their methodology and shared data with physicians in advance reported a more favorable reception to their high-performance networks.

Employers

A key driver behind the development of high-performance networks was a limited number of large, national employers. While other employers expressed interest in these networks, not all were ready to handle potential employee pushback. As a result, actual adoption had not yet matched the level of interest. Employers that had championed these products aggressively reported feeling isolated, as if everyone else was merely "window shopping" while waiting to see what happened with the early adopters.

High-performance networks gained a foothold in markets only where large employers had been proactive in their interest and implementation. This was the case in Boston, Milwaukee, and Seattle. In Boston, for example, the GIC required contracted plans to develop and offer high-performance networks. In Milwaukee, the business coalition, representing a large number of employers, was instrumental in bringing Humana's high-performance network into the market. In Seattle, Aetna had sizeable national accounts — including Boeing, Costco, Nordstrom, and Starbucks — that played an active role in launching a high-performance network.

Even in these markets, however, enrollment in high-performance networks had not expanded much beyond the employers that initiated them.

Benefits consultants and plans observed that adoption of high-performance networks might increase if employers saw more significant premium savings. Preliminary savings estimates at the time ranged between 3 percent and 5 percent. Achieving larger savings would require more substantial benefit design changes, such as wider cost-sharing differentials, which plans and employers appeared unwilling to implement at that point.

Other concerns also slowed employers' adoption of high-performance networks. Large multistate employers wanted to offer consistent health insurance benefits to employees across different locations, which proved difficult when plans defined their high-performance networks differently. For instance, if some plans' high-performance networks included primary care physicians while others did not, this made high-performance networks less appealing to national employers as part of a unified strategy.

Employers were also concerned that adopting high-performance networks could create geographic access problems for employees. Employers with employees spread across a wide area worried that high-performing providers might not be evenly distributed geographically. Because of these concerns, some plan executives and benefits consultants noted that while these networks had been designed for large, national employers, they might ultimately gain more traction among local employers — like the membership of the Milwaukee coalition. Looking ahead, high-performance networks could prove more appealing to small employers, who tend to be more price sensitive and less deterred by issues such as network disruption.

Regulators

Up to that point, because high-performance networks had been offered predominantly by large, self-insured employers, they had largely escaped state regulators' attention. However, one plan executive reported that regulators had raised concerns about these networks potentially creating access problems, particularly in rural areas. The respondent noted that the traditional access perspective — whether sufficient providers exist in a given area — runs counter to the objective of high-performance networks. He observed that while reasonable access should be maintained, the question for regulators is whether plans should lower quality standards simply to have "high-performing" physicians in every zip code. He added that if people want care from the best doctor, they may need to travel to the city.

Early Lessons Learned

Respondents universally agreed that it was premature to evaluate the impact of high-performance networks on service utilization, costs, and quality because enrollment volume in these networks remained insufficient. Respondents also indicated it was too early to identify any shifts in physician market share attributable to high-performance networks, though one health plan executive cited early anecdotal evidence of shifting market share among some physicians based on the presence or absence of a high-performing designation. Some respondents expressed hope that high-performance networks would help re-engineer the care delivery system, pushing it toward greater efficiency and higher quality.

While evidence on the effects of high-performance networks continued to accumulate, evidence from analogous initiatives suggested the potential for savings associated with care delivered by physicians who were more efficient and of higher quality. The experience of the UNITE-HERE Labor Management Trust Fund — a Taft-Hartley trust providing health care to 120,000 hotel workers and their families in Las Vegas — offered a compelling illustration. In 2003, the Trust screened all 1,800 physicians in its network on the basis of efficiency and quality and, as a result, excluded 50 physicians. Physicians deemed as providing higher quality of care also received a special designation in the Trust's provider directory. These network changes reportedly contributed to a substantial reduction in the Trust's medical cost trend, and the resulting cost savings enabled workers to receive salary increases for the first time in several years. (4) Despite the evolving nature of high-performance networks, respondents identified several important lessons from their early experiences.

Communication with providers essential

Respondents emphasized the critical importance of communicating openly and transparently with providers during the development and implementation of high-performance networks. For instance, one plan executive urged communicating with the provider community about intentions and giving them a forum for feedback — rather than operating in a black box without willingness to explain the rationale. He added that there is no value in gathering data if it is not being shared with physicians and giving them opportunities to become more efficient.

Several plan executives highlighted the importance of working with and educating physicians months before entering a market or launching a new product or initiative. As one physician representative commented, if you do not include the frontline providers who deliver care during the development and vetting of the product, you lose from the outset. Plans that engaged physicians from the beginning fared considerably better than those that simply confronted them with the final product.

Costs and quality both important

Nearly all respondents asserted that the success of high-performance networks depended on evaluating both costs and quality. It is possible for high-quality care to be delivered inefficiently. A plan executive described networks as typically depicted on a 2x2 graph with costs on one axis and quality on the other, with the goal of progressively moving networks into the low-cost/high-quality quadrant.

In these initial iterations of high-performance networks, much of the attention focused on the cost side, largely due to limitations in the integrity and sophistication of quality data. Respondents uniformly agreed that improved data were necessary. A plan executive highlighted the issue, saying his organization was cautious about differentiating on the basis of quality because they believed datasets needed to be as valid as possible before making such assertions. A benefits consultant, however, cautioned that avoiding quality measurement was perilous for plans, warning that no one would want to see a headline reading that a carrier had switched to cheap doctors.

Standardized industry measures needed

Respondents discussed the need for the industry to progress toward more uniform standards of provider performance measurement. One plan executive, for example, expressed concern that physician quality determinations should not rest solely on data held by a single plan. He further commented that collaboration from both public and private sectors was necessary to make informed decisions about the quality of a practicing physician. He stated that the most damaging outcome would be for competitors to have competing definitions of quality, which would erode credibility in the eyes of consumers.

Movement toward any form of industry standardization, however, was not without complications. A potentially significant obstacle involved antitrust concerns. A plan executive noted that in a litigious environment, his plan was wary of collaborative activities among plans to compare methods used to define networks, as these could be perceived as violating antitrust laws. He added that while employers reported some consistency among plans regarding which providers were deemed high performing — at least in larger markets — there remained ample opportunity for inconsistencies. Employer groups, such as the Business Roundtable, had advocated for legislation granting insurers access to Medicare Part B data with physicians identified, to enable more accurate assessments of physician costs and quality through larger patient samples. (5) Several respondents observed that without greater industry uniformity, any performance information provided to physicians and consumers would be confusing and of limited utility. Physicians would be less likely to respond positively — by improving efficiency and quality — if different health plans rated them differently.

Purchaser support needed

One plan executive explained that when building high-performance networks, the plan sought to bring leading employers along to talk to providers, underscoring that the health care system is grounded in employer-sponsored insurance. The respondent added that providers were more receptive when they understood the choices were coming from employers. A benefits consultant said that if employers truly wanted high-performance networks, they needed to be front and center with plans, demanding and supporting them — they could not stay behind the scenes. Respondents noted, however, that employers were often reluctant to get involved. As one plan executive remarked, the most fascinating dynamic was that the employer wants it, the health plan builds and implements it, and then the health plan takes the heat while the employers who wanted it duck and cover.

It was too early to determine whether, or to what extent, high-performance networks would lead to lower-cost, higher-quality health care. Initially, attention focused on the potential gains from shifting some enrollees to higher-performing providers, but systemwide gains were constrained by the capacity of designated high-performing providers. The greater potential lay in redirecting sufficient volume away from lower-performing providers to motivate them to improve, thereby raising the overall performance of the system. For this to materialize, health plans needed to furnish physicians with detailed and accurate information about where their performance fell short. The potential also hinged on whether employers would institute benefit structures that motivated employees to shift to high-performing providers — which in turn depended on whether employees viewed high-performance networks as something valuable or as an unwelcome reminder of what they disliked about managed care.

Notes

1. The most commonly cited data tool for carrying out this measurement process was the Symmetry ETG. This technology combines inpatient, outpatient, ancillary, and pharmacy claims to construct a treatment episode from the onset of care through completion of treatment.

2. See "Regence BlueShield sued over 'select' network," Seattle Post-Intelligencer (Sept. 21, 2006).

3. See, for example, "BJC warns it may drop United Healthcare," St. Louis Post-Dispatch (March 18, 2005).

4. This information is drawn from testimony of Peter V. Lee, Pacific Business Group on Health, before the U.S. House of Representatives, Committee on Ways and Means, Subcommittee on Health, "Promoting Quality and Efficiency of Care for Medicare Beneficiaries" (March 15, 2005).

5. This advocacy is reflected in S.3900 (Gregg) "Medicare Quality Enhancement Act of 2006."

Data Source

To study high-performance networks, HSC researchers gathered data at both the national level and in selected markets with experience in these networks, including Boston, Milwaukee, Seattle, and California. Information was collected between January and June 2005 as part of Round Five of the CTS site visits to help identify target markets. The findings are based on semi-structured interviews conducted between May and September 2006 with approximately 20 respondents. Respondents included representatives of national health plans and regional plans offering high-performance networks in the target markets, as well as representatives of providers, employers, and benefits consultants with knowledge of these local markets. Interviews were carried out by a two-person interview team either in person or by telephone.

Acknowledgment: This research was funded by the California HealthCare Foundation.

ISSUE BRIEFS are published by the Center for Studying Health System Change.

Sources and Further Reading

Agency for Healthcare Research and Quality — National Healthcare Quality and Disparities Reports — AHRQ publishes annual reports tracking progress on health care quality, disparities, and access across the United States, offering benchmarks relevant to provider performance measurement and network quality standards.

Centers for Medicare & Medicaid Services — Medicare Advantage Plans — CMS administers Medicare Advantage, the federal managed care program that applies star ratings and quality metrics to health plan networks, providing a large-scale framework for evaluating network performance.

Kaiser Family Foundation — Employer Health Benefits Annual Survey — KFF's annual survey documents trends in employer-sponsored health insurance, including network design, cost-sharing structures, and the prevalence of tiered and high-performance network products.

Health Affairs — A leading peer-reviewed journal covering health policy and research, including studies on provider network adequacy, value-based care models, and the intersection of cost containment and quality improvement.

The Commonwealth Fund — The Commonwealth Fund conducts independent research on health care system performance, insurance coverage, and access, with particular attention to how network structures and plan design affect cost efficiency and patient outcomes.