Finding the Balance in Physician Rankings

September 2010

Recently, the American Medical Association (AMA) and 47 state medical societies criticized the efforts of some of the nation’s largest health insurers to rank physicians on cost and quality measures.1 These rankings are used to develop tiered networks and steer participants toward higher-performing physicians. In a letter to insurers, the physician groups called for greater transparency in the development of physician profiles. They contend that the methodologies used to develop these profiles lack scientific rigor and consistency, and fear that they will be not be ranked accurately.2 For their part, insurers state that they are only responding to the concerns of employers and other payers who want to incentivize more efficient and lower-cost health care. In light of health reform, tiered networks are likely to continue to generate interest from employers as they seek out affordable sources of insurance coverage.

Cost, Quality, or Both?         

Tiered physician networks are typically not separate insurance products, but are an option for employers inside an existing plan such as a preferred provider organization (PPO).3 Visiting a higher-ranking physician may be incentivized through lower copayments or premiums. These plans have gained traction in markets where large employers have actively pursued their implementation—companies in Boston, Seattle, and Milwaukee have led the charge.4 Physicians are ranked on cost, quality, or a combination of both. The basis for these rankings is contentious. In terms of cost, insurance plans want to reward providers who deliver efficient care. This cost-effective care could mean less utilization of expensive procedures and diagnostics, or rating as low-cost compared to peers in the same network. Physicians fear that they will be ranked on costs only and will be disadvantaged if their patient population carries a heavy disease burden. The AMA opposes ranking physicians on costs alone, and encourages the use of quality data to guide patients and improve care delivery. Insurance groups agree on this point—in a letter to the AMA, America’s Health Insurance Plans (AHIP) CEO Karen Ignagni states that the goal of ranking systems is to encourage quality improvement.5 Despite an agreement in principle from both sides of this issue, physicians continue to be wary of being subject to cost-only rankings, and see these rankings as reminiscent of unpopular aspects of managed care.6

While there is agreement on the need to utilize quality data, the application is difficult. There are numerous possible indicators and a lack of consensus between physicians and insurers on quality measures that paint the most accurate picture of physician performance. Due to concerns about the accuracy of public information and potential physician defamation, policymakers in several states have taken steps to regulate physician rating and the presentation of quality data to consumers. In Colorado, the Physician Designation and Disclosure Act requires that any quality measures must be evidence-based and endorsed by National Quality Forum, a national physician-specialty organization, or the Colorado Clinical Guidelines Collaborative.7

Although there are challenges with both cost and quality measures, insurers state that they are currently using a combination of both in determination of physician rankings. The nature of this combination, however, is not clear, and leads to the chief complaint of physicians—lack of transparency and standardization in ranking programs.8 They argue that if physicians are not clear on how they are being rated and are potentially subject to different methods by different plans, how can they achieve quality improvement?

Cost Profiles: Reliability and Validity Concerns

Research suggests that physician concerns regarding transparency and standardization may be legitimate. In a HCFO-funded study on methods of evaluating the cost-efficiency of specialty physicians, J. William Thomas, Ph.D. discovered problems with both the accuracy and reliability9 of physician cost profiles. Using commercial software to construct episodes of care and cost profiles, he showed that approximately 22 percent of physicians could be misclassified in a two-tiered system. In addition, the majority of physician profiles did not meet reliability standards.10 

Dr. Thomas’s HCFO-funded work also led to an exploration of “episode attribution” and its role in the development of cost profiles. In creating cost profiles, insurers must develop a method to assign episodes of care to physicians and attribute responsibility for the care provided. There are several methods to assign an episode to a physician—two common ways are to use the highest percentage of patient visits, or alternatively the highest percentage of costs. Dr. Thomas and his colleagues tested 12 attribution rules and their impact on the classification of physicians as low- or high-cost. They found that between 17 and 61 percent of physicians could be assigned to different cost categories by using different attribution rules.11 This lack of standardization means that the same physician could be categorized differently by different health plans, ultimately leading to bad information for consumers. As discussed above, Dr. Thomas’s work shows that this is not a simple process—different attribution methods can yield different outcomes. The future of well-designed ACOs depends on the development of effective methods of attributing responsibility for care.

Future Directions: Consumer Behavior and Accountable Care Organizations

In addition to concerns regarding the methodology used to develop tiered networks, there is the issue of whether these cost incentives have the desired effect of steering patients toward more efficient providers. In a current HCFO study, Meredith B. Rosenthal, Ph.D. is examining the effects of tiered physician networks introduced in six health plans offered through the Massachusetts Group Insurance Commission (GIC). The study will assess whether tiering affects consumers’ initial choice of physician as well as their decision to stay with a current physician, especially when faced with financial incentives to change doctors. The study will identify the patient, plan, and physician characteristics associated with patient responsiveness and will shed light on the value of tiered networks as a benefit design feature.

Physician tiering systems also have implications for the development of Accountable Care Organizations (ACOs). ACOs are networks of providers that work together to promote quality care for a defined population while reducing costs and potentially enjoying shared savings. A key issue for the future of ACOs is the assignment of patients to these networks. One of the basic features of these organizations is “invisible enrollment.” This means that patients would be assigned to an ACO if they received most of their care from an affiliated provider.12 As discussed above, Dr. Thomas’s work shows that this is not a simple process—different attribution methods can yield different outcomes. The future of well-designed ACOs depends on the development of effective methods of attributing responsibility for care.

Physician tiering and ranking systems will continue to be a feature in the health insurance market as health plans and employers look to bend the cost curve. HCFO studies are producing valuable information to inform the design of these plans. Details on the studies led by Dr. Thomas, Dr. Rosenthal, and other HCFO grantees are available at

Impact of a Tiered Physician Network on Consumer Behavior
Grantee Institution: Harvard School of Public Health
Principal Investigator: Meredith B. Rosenthal, Ph.D.
Grant Period: July 1, 2010-June 30, 2011

The researchers will examine the effects of tiered physician networks introduced in six health plans offered to beneficiaries of the Massachusetts Group Insurance Commission (GIC). Specifically, they will (1) determine whether tier status of a regular physician is associated with choice of plan; (2) determine whether tier status affects the probability that a patient will visit a physician for the first time; (3) determine whether tier status affects the probability that a patient will continue seeking care with a physician seen prior to tiering; and (4) identify patient, plan, and physician characteristics associated with greater responsiveness on aims 1-3. The objective of this project is to assist employers in assessing the impact of physician tiering on consumer behavior and to assess the value of this benefit design feature.

Title: Evaluating Cost Efficiency of Specialty Physicians
Grantee Institution: University of Southern Maine
Principal Investigator: J. William Thomas, Ph.D.
Grant Period: April 1, 2007-March 31, 2009

The researchers will analyze alternative strategies for measuring specialist physician cost efficiency. Accurate measures of physician cost efficiency allow consumers to make more informed decisions, while helping health plans make better choices about which physicians to include in their networks. Specifically, the researchers will explore the: (1) feasibility of using multi-plan claims databases for cost efficiency measurement; (2) need for risk adjusting episode expected costs to account for patients' comorbidities; (3) methodology for dealing with cost outlier episodes; (4) methodology for attributing responsibility for individual episodes to individual physicians; (5) minimum episode sample sizes required for cost efficiency measurement; (6) methodology for case-mix standardization; (7) influence and availability of pharmaceutical claims on cost efficiency measurement; and (8) suitability of ratio of observed to expected cost as a cost efficiency metric. The objective of the study is to develop standard methods for measuring the cost efficiency of specialist physicians, providing a broad range of stakeholders with reliable means for developing physician networks, assigning tiers, and implementing better public reporting.


1. Mathews, A.W. “Doctors Slam Insurers Over Their Rankings,” The Wall Street Journal, Vol. 256, No. 16, July 20, 2010, p. D2.
2. Elliott, V.S. “Medical societies demand insurers rethink doctor cost ratings,” American Medical News, August 2, 2010. Also see
3. Draper, D.A. et al. “High-Performance Health Plan Networks: Early Experiences,” Issue Brief, No. 111, Center for Studying Health System Change, May 2007. Also see
4. ibid.
5. Elliott, 2010.
6. Draper et al, 2007.
7. Cartwright-Smith, L. and S. Rosenbaum. “Fair Process in Physician Performance Rating Systems: Overview and Analysis of Colorado’s Physician Designation Disclosure Act,” Aligning Forces for Quality, The Robert Wood Johnson Foundation, September 2009.
8. Mathews, 2010.
9. Reliability refers to the consistency of measurement and the amount of variability that is due to actual differences in performance. Dr. Thomas and his colleagues constructed reliability ranges from 0 to 1; 0 indicated that all variation was due to measurement error, and 1 indicated that the variation was due to actual performance differences.   Adams, J.L., Mehrota, A., Thomas, J.W., and E.A. McGlynn, “Physician Cost Profiling—Reliability and Risk Missclassification,” New England Journal of Medicine, Vol. 362, No. 11, March, 18, 2010, pp. 1014-1021.
10. Adams, J.L. et al. “Physician Cost Profiling—Reliability and Risk Missclassification,” New England Journal of Medicine, Vol. 362, No. 11, March, 18, 2010, pp. 1014-1021.
11. Mehrota, A. et al. “The Effect of Different Attribution Rules on Individual Physician Cost Profiles,” Annals of Internal Medicine, Vol. 152, No. 10, May 18, 2010, pp. 649-654.
12. Merlis, M. et al. “Health Policy Brief: Accountable Care Organizations,” Health Affairs, July 27, 2010.