William M. Sage Professor Sage joined the Columbia faculty in 1995. He teaches health law, regulatory theory, antitrust, and professional responsibility. In the fall of 2006, be will become vice provost for health affairs and' professor of law at the University of Texas at Austin. He served as an intern at Mercy Hospital and Medical Center in San Diego and as a resident in anesthesiology and critical care medicine at the Johns Hopkins Hospital .. In addition, be practiced corporate law at O'Melveny & Myers in Los Angeles, and served on President Clinton's Task Force on Health Care Reform. From 2002 to 2005, be was the principal investigator for The Pew Charitable Trusts' Project on Medical Liability in Pennsylvania. His awards and honors include election as Fellow of the Hastings Center, receiving a Robert Wood Johnson Foundation Investigator Award in Health Policy Research, serving on the editorial board of Health Affairs, serving on the Institute ofMedicine' s Committee on Rapid Advances in Health Care, and serving on the JCAHO Tort Resolution and Injury Prevention Roundtable. He has published widely in the area ofbealth care law and is co-editor of Medical Malpractice and the U.S. Health Care System (Cambridge University Press, 2006) and Uncertain Times: Kenneth A"ow and the Changing Economics of Health Care (Duke University Press, 2003). Professor Sage received his AB. from Harvard and his M.D. and J.D. degrees from Stanford, where he was a note editor for the Stanford Law Review. PAYFORPERFORMANCE: WILLITWORK IN THEORY? William M. Sage· And can you imagine fifty people a day? I said FIFTY people a day ... walkin' in, singin' a bar of "Alice's Restaurant" and walkin' out? Friends, they may think it's a MOVE­ MENT, and that's what it is: THE ALICE'S RESTAURANT ANTI-MASSACREE MOVEMENT! ... and all you gotta do to join is to sing it the next time it comes around on the guitar. Arlo Guthrie, Alice's Restaurant The title of my lecture about pay for performance in health care (often abbreviated as "P4P") echoes a French management saying: "It's all very well in practice, but it will never work in theory." There is no doubt that medical pay for performance is popular. P4P initiatives are everywhere; a search of online news reveals nearly 2,500 stories about pay for performance in the last two years alone. Folk legend Arlo Guthrie would have called P4P a movement-a "Pay for Performance Medical Quality Movement." But before you join in, you might want to know what you are joining. So let's talk about the theory of paying for performance, to help you recognize the dangers as well as the opportunities in this particular movement as it gathers steam. I. MOVEMENTS WITHOUT THEORIES Agreement as to the theory of any particular policy prescription is not a prerequisite for politicians endorsing it. To the contrary, American democracy seldom, if ever, generates broad consensus on the purpose of an action to be taken that results in that action. As they say, politics makes strange bedfellows. Unexpected political coalitions form all the time. Back in the 1970s, there was an unusual alliance between clean air environmentalists and dirty coal producers in eastern states to make sure that coal-burning power plants had to install expensive scrubbers to reduce sulfur dioxide emissions. 1 It was in the interest of the coal producers to have power plants continue to use high­ sulfur Eastern coal rather than unscrubbed, cleaner coal mined in the West. 2 * Professor of Law, Columbia University. As of September 2006, Vice Provost for Health Affairs and Professor of Law, University of Texas at Austin. A.B. 1982, Harvard University; M.D. and J.D. 1988, Stanford University. 1. BRUCE A ACKERMAN & WIIliAMT. HAssLER, CLEANCOAiiDIRTY AIR 65-78 (1981 ). 2. /d. 306 INDIANA HEALm LAw REviEW [Vol. 3:303 Around the same time, and closer to health care, a coalition formed between social liberals and fiscal conservatives to deinstitutionalize the mentally ill. 3 In the 1980s, a coalition formed between safety advocates and automobile manufacturers to induce states to enact mandatory seatbelt use laws, which the latter had resisted for decades.4 The auto makers became enamored with mandatory seatbelt laws as soon as the federal government had threatened them with mandatory airbag requirements if sufficient states did not adopt seatbelt laws. s On the other hand, fundamental disagreement on the purpose of a policy change can be crippling and destructive. In 1988, Medicare's newly enacted Catastrophic Coverage. Act was repealed before it was ever implemented because the people it affected had not understood what they would be paying for. 6 In a nutshell, those inside the Beltway were solving an actuarial problem with acute care Medicare using means-tested financing that made sense on fiscal grounds and on fairness grounds, while those outside did not realize that seniors would shoulder unaccustomed fmancial responsibility for an expansion in program benefits that was not meaningful to them. The fiasco that followed significantly reduced the risk tolerance of both Congress and Washington-based interest groups, which helped kill the Clinton administration's universal coverage proposal in 1993-94.7 Later in the 1990s, debate . in Washington over ''patient protection" legislation-that is, regulation of managed care-foundered on the question of whether people should have a right to sue HMOs. 8 It made sense for the federal government to play a larger role in managed care regulation because of ERISA preemption. However, there was a fundamental disagreement between those who supported extending liability in order to make managed care function more effectively, and those (such as physicians) who hoped that 3. See, e.g., NEW DIRECTIONS FOR MENTAL HEALTH SERVICES, DEINSTITUTIONAUZATION: PROMISE AND PROBLEMS (H. Richard Lamb & Linda E. Weinberger eds., 2001); E. Fuu.ER TORREY, OUr OF THE SHADOWS: CONFRONTING AMERICA'S MENTAL IUNESSCRISIS (1996); RAELJ. ISAAC& VJRGINIAC.ARMAT,MADNESSINTHE STREETS: How PSYCHIATRY AND THE LAW ABANDONED THE MENTALLY ILL (1990). 4. Jerry Louis Mashaw, The Story ofMotor Vehicle Manufacturers Association of the U.S. v. State Farm Mutual Automobile Insurance Co.: Law, Science and Politics in the Administrative State, in Al>MlNISTRATIVE LAW STORIES 396-97 (Peter Strauss ed., 2006). 5. ld. at 352. 6. Pub. L. No. 1()()..360, 102 Stat. 683 (1988), terminated by Medicare Catastrophic Repeal Act of 1989, Pub. L. No. 101-234, 103 Stat. 1979. See RICHARD HIMELFARB, CATASTROPIDCPOLITICS:THERisEANDFALLOFTHEMEDICARECATASTROPIDCCoVERAGEACT OF 1988 (1995); Theodore R. Marmor & Gary J. McKissick, Medicare's Future: Fact, Fiction, and Folly, 26 AM. J.L. & MED. 225, 232-34 (2000). 7. See HAYNES JOHNSON &DAVIDS. BRODER, THE SYSTEM (1997) (offering a political history of the Clinton health care plan). 8. See David A. Hyman, Regulating Managed Care: What's Wrong with the Patients' Bill of Rights?, 73 S. CAL. L. REv. 221,230-33 (2000). 2006] P4P: Wru... IT WORK IN THEoRY? 307 greater liability would make managed care disappear entirely.11 The resulting gridlock left the federal-:-state balance in managed care oversight to the courts, which failed to answer many of the most pressing questions regarding cost­ containment and accountability.10 At present, the biggest health care issue in Washington, DC, is the newly enacted Medicare Modernization Act ("MMA "),including the Medicare Part D drug benefit. 11 This legislation, which represents the largest expansion of Medicare spending since the program began in 1965, was supported by a bipartisan coalition with different views of what the statute was designed to accomplish. Democrats voted for the bill hoping and believing that its dominant effect would be to expand access to prescription drugs for seniors. Republicans voted for the bill with the long-term goals of privatizing the entire Medicare program through contracts with commercial insurers and increasing individual responsibility using health savings accounts.12 My guess is that the next few years will demonstrate how difficult it is to make the MMA a success given this fundamental disagreement about its purpose. Pay for performance is at a similar crossroads today. Rewarding physicians for health rather than illness has a long history of academic speculation going back at least to a footnote in the 1963 American Economic Review, in which economist Kenneth Arrow attributed the practice to the Chinese. 13 These have been sporadic, tentative proposals from health care outsiders-in essence, theories without movements.14 In the last five years or so, by contrast, quality improvement advocates of various stripes within the medical mainstream have coalesced around P4P, a phrase that simultaneously channels the futurism of Internet commerce (B2B, Y2K) and the glamour of Fortune 500 executive compensation.15 9. See Amy Goldstein & Helen Dewar, Senate Kills "Patients' Rights" Bill, WASH. PosT., Oct. 10, 1998, at Al. 10. See Pegram v. Herdrich, 530 U.S. 211 (2000); Aetna Health Inc. v. Davila, 542 U.S. 200 (2004); WilliamM. Sage, UR Here: The Supreme Court's Guide for Managed Care, 19(5) HEALTHAFF. 219,219 (2000). 11. Pub. L. No. 108-173, 117 Stat 2065 (2003). 12. Theodore R. Marmor &JacobS. Hacker,MedicareReformandSocialinsurance: The C/ashesof2003andTheirPotentia/Fallout,5YALEJ.HEALTHPOL'YL.&BTIDcs475,478-80 (2005); see also Robin Toner, Rival Y'uions Led to Rocky Start for Drug Benefit, N.Y. TIMEs, Feb. 6, 2006, at Al. 13. Kenneth J. Arrow, Uncertainty and the Welfare Economics of Medical Care, 53 AM. ECON. REv. 941, 961 n.35 (1963). 14. See, e.g., DavidA. Hyman & Cbarles Silver, Just What the Patient Ordered: The Case for Result-Based Compensation, 29 J.L. MED. & ETIDCS 170 (2001); David A. Hyman & Charles Silver, You Get What You Pay For: Result-Based Compensation for Health Care, 58 WASH.&LEEL.REv.1427 (2001). 15. See, e.g., Meredith B. Rosenthal et al., Payingfor Quality: Providers 'Incentives for Quality Improvement, 23(2) HEALTH AFF. 127, 127 (2004); Sheila Leatherman et al., The Business Case for Quality: Case Studies and an Analysis, 22(2) HEALTHAFF. 17, 17 (2003). 308 INDIANA HEAL1H LAw REVIEW [Vol. 3:303 But the "Pay for Performance Medical Quality Movement, lacks a common principle or purpose beyond frustration with conventional reimburse­ ment mechanisms and a general commitment to quality improvement. Moreover, many rationales that one can articulate for P4P--and that seem to be percolating in the public and among policy makers-are either internally inconsistent or incompatible with one another. There are three fundamental questions that need to be asked about pay for performance in health care. First, will pay for performance incentives make buyers of medical care smarter? Second, will pay for performance incentives make sellers of medical care more skillful? Third, is pay for performance primarily a private health care innovation or primarily a governmental one? With the goal of resolving these uncertainties, let us turn to a description of the multiple, potentially conflicting explanations that have been explicitly articulated for P4P or are implicit in endorsement by particular groups. Then let me suggest to you three plausible unifying theories of pay for performance, and how P4P systems might be designed to further each. I will conclude by making the case for strong government leadership in pay for performance, a need I believe to be compelling notwithstanding its secondary place in the history of the movement. II. P4P's FELLOW TRAVELERS To begin to understand the tensions fu pay for performance, it is useful to compare physicians' incentive compensation to seemingly similar payment methods for lawyers, who comprise America's other "sovereign profession., Litigators' contingent fees-a pre-agreed percentage of their clients' financial recovery, paid only if the lawyer prevails in court c;>r by settlement-are without doubt pay for performance. Why do we pay these lawyers only for success, and not for the tasks they perform or the time they spend? Two reasons are immediately relevant to a discussion of pay for performance in medicine. The first reason is to inspire lawyers to work harder for clients in pursuit of victory. The second reason is promote cost-effective work, rather than just "running up the meter." We will return to a third reason-encouraging lawyers only to take winnable cases--and a fourth-ensuring the availability of lawyers to poor clients-in a discussion of the government's role in medical P4P. One would think that medicine and law would have similar views as to how success fees relate to their underlying professional missions. Not true. The notion that lawyers will work harder and more cost-effectively for clients if they are paid for getting results sends a pretty clear message that success in law is uncertain and depends on effort, which can be increased through incentive compensation. Physicians, however, seem to have a very different view of the relationship between professional uncertainty and conditional payment. Opinion 6.01 of the American Medical Association's code of medical ethics opposes the notion of paying doctors for results on a relatively 2006] P4P: WILL IT WORK IN THEORY? 309 strange ground: that patients who pay only if they get better will be misled into believing that they will get better. 16 This is exactly opposite the reason for paying lawyers for success: that litigation is risky and harder-working lawyers have a better chance of winning. In fact, critics of contingent fees in law complain that lawyers often demand risk-based fees for cases that routinely generate quick, effortless settlements. That the medical profession ascribes failure primarily to chance, not lack of effort, and that it regards failure as the rule and success the exception, constitute a fairly stunning indictment of medical quality. It is also a challenging starting point for pay for performance. Policymakers, providers, payers, and the public conceive of P4P as serving a variety of worthy objectives (Table 1 ), each of which is identified with a movement and a vocabulary of its own. Many of the buzzwords in modern health policy are superficially compatible with P4P, which explains why we already have fifty people in the room, jumping up and down and singing a bar of the "Pay for Performance Medical Quality Movement" song. P4P is an umbrella movement, like the World Council of Churches or the AFL-CIO; it attracts widespread support but may not be able to fulfill the individual aspirations of its boosters. A. Patient Safety The Institute of Medicine got into the P4P business incrementally. In 1999-2000, the 10M published To Err Is Human, a report estimating that nearly 100,000 deaths occur from medical errors in the United States every year. 17 A year later, the 10M issued its Crossing the Quality Chasm report, 16. AM. MED. Ass'N, COUNCIL ON ETHICAL AND JUDICIAL AFFAIRS, CODE OF MEDICAL ETHICS Op. 6.01 (2002-03 ed.). 17. INST. OF MED., To ERR IS HUMAN: BUILDING A SAFER HEALTH SYSTEM (Linda L. Kohn et al. eds., 1999). 310 INDIANA HEALTH LAW REVlEW [Vol. 3:303 in which it described six general quality goals for American health care and proposed ways to achieve them, including incentive compensation.18 In 2005, the IOM released a detailed report on pay for performance, which one might consider the capstone of its efforts to connect P4P to patient safety. 19 Medical errors present an obvious paradox for traditional fee-for-service reimbursement. Health care services do not come with warranties. Consequently, providers who deliver suboptimal care may end up earning more for subsequent consultations, hospitalizations, and procedures than those whose skill yields a quick, definitive diagnosis and cure. Patient safety advocates have made much of this contradiction in their support for P4P. Beyond rhetoric, some private health insurers and, more recently, Medicare, have indicated that they will not pay the same providers for care involving serious mistakes.20 But can pay for performance reduce medical errors? The IOM's approach regards safety as a subset of quality. It ignores the fact that safety events are rarer, more salient, and more threatening to a physician's professional reputation than background changes in incremental quality.21 Payment policies that publicize errors may drive error reporting underground, and may deprive provider organizations of funds needed for improvement. The IOM approach also sidesteps a related question: if society learns by paying for performance that some physicians simply do not perform well, what do we do with them? How does P4P tie to licensing and discipline, to weeding out the bad doctors that the public, and for that matter the profession, still very much believe exist, no matter how often patient safety gurus repeat the mantra that most medical errors are committed by good doctors rather than by bad doctors?22 B. Competition As befits its moniker, P4P has a business pedigree, with some of its earliest and strongest proponents representing corporate health care 18. lNST. OF MED., CROSSING THE QUAUTY CHASM: A NEW HEALTH SYSTEM FOR THE 21ST CENTURY (200 1) [hereinafter CROSSING THE QUAUTY CHASM]. 19. INST. OFMED., PERFORMANCE MEAsUREMENT: ACCELERATING IMPROVEMENT (2005). 20. Chen May Yee, HealthPartners to Withhold Payment for Surgical Errors, MINNEAPOliS STAR TR.rn., Oct. 6, 2004, at lA; Statement of Mark B. McClellan, Adm'r, Ctrs. for Medicare & Medicaid Servs. to the Comm. on Senate Fin., May 17, 2006 ("As a necessary step toward encouraging better care and lower overall heath care costs, we support further steps such as eliminating payments for 'never events' and want to work with the Congress to take such steps."). 21. William M. Sage., Reputation, Malpractice Liability, and Medical Error, in ACCOUNTABIUTY: PATIENT SAFETY ANDPOUCYREFORM 159 (Virginia A. Sharpe ed., 2004). 22. Robert J. Blendon et al., Views of Practicing Physicians and the Public on Medical Error, 347NEWENG.J.MED. 1933, 1935-36, 1938-39(2002); RobertPear,Pane/ Seeks Better Disciplining of Doctors, N.Y. TIMEs, Jan. 5, 2005, at A2l (summarizing the results of a study which found that disciplining incompetent physicians would reduce malpractice litigation). 2006] P4P: WITL IT WORK IN THEoRY? 311 purchasers.23 Unsurprisingly, the payor community expect$ pay for performance to make the health care system more competitive. on this view' conventional reimbursement represents a combination of misguided regulation and the exercise of market power by physicians who have large informational advantages over those who buy their services. Breaking this stranglehold by switching to P4P, some might argue, will increase competition and generate the things competition is supposed to bring: lower prices, higher quality, and closer matching of the services provided to the preferences of the people who are going to receive them. However, competition on quality does not model cleanly in economic theory, and competition on medical quality is fraught with additional uncertainty.24 P4P may indeed channel competition in particular quality­ related directions, but it will not necessarily promote overall competition. One concern is that P4P might reduce competition based on price, which is something that consumers also care about. Another concern is how much competition will be lost in the process of establishing P4P systems that rely on consistency and uniform application among health care providers for their effectiveness-structures that invite collusion as well as collaboration. That leading safety and quality advoeates within the medical profession are ambivalent about competition in health care-neither 10M report so much as mentions competition oversight through. antitrust enforcement-provides little reassurance that these risks will be taken into account as pay for performance advances. C. Consumer-Directed Care · P4P, notwithstanding its catchy name, trails "consumer-directed care" as the darling oftoday's health policy solipsists. Those who believe that the solution to waste in health care spending is to let individuals take charge of their purchasing decisions find pay for performance attractive because it connotes consumer empowerment. It is not clear, however, that substituting formal performance incentives for traditional reliance on physician competence and beneficence will improve the ability of individual patients to make coherent decisions about their treatment. That will happen only ifP4P reliably captures information desired by consumers but otherwise unavailable to them. Moreover, consumer-directed care is ·commonly used as a 23. Leathermanetal.,supranote 16; VrrroRIOMAIOETAL., VAWE-~PuRCHASING: A REVIEW OF THE LITERATURE (Commonwealth Fund Report No. 636, May 2003), available at http://www.cmwf.org (1asi visited May 17, 2006); Roger Lowenstein, The Quality Cure?, N.Y. TIMEs MAG., Mar. 13, 2005; at 46. 24. William M. Sage & Peter J. Hammer, Competing on Quality of Care: The Need to Develop a Competition Policy for Health Care Markets, 32 MicH. J.L REFoRM 1069,.1109-18 (1999); William M. Sage, David A. Hyman, & Warren Greenberg, Why Competition Law Matters to Health Care Quality, 22(2) HEALTHAFF. 31,31 (2003). 312 INDIANA HEALTH LAw REVIEW [Vol. 3:303 euphemism for consumer-financed care-people spending their own money, not that of an insurance pool or th.e public.25 A belief in the benefits of self help will not make everyone financially independent. What about people who do not have money of their own to spend? Performance is likely to be defined quite differently by a third-party payer than by the actual recipient of services. D. Patient-Centered Care A core attribute ofhigh-quality health care identified by the 10M is that it is ''patient-centered."26 Patient-centered care sounds similar to consumer­ directed care, and therefore superficially compatible with P4P, but its connotations are non-financial. Patient-centered care can be understood as health care that the patient wants and that has the patient's subjective welfare as the focus of the care delivery experience. Patient-centered care contains its own ambiguities. Is the notion of patient-centered care an ethical commitment to patient autonomy or an ethical commitment to professional beneficence? One can imagine P4P aligning incentives so as to serve either objective, but not without specific effort and acknowledgment of tensions between them. Furthermore, the technical manipulations necessary to devise and implement pay for performance . seem remote from the human touch that motivates interest in patient-centered care. E. Evidence-Based Medicine For thought leaders in the medical quality movement generally, pay for performance is attractive because it can promote evidence-based medicine.27 Individual physicians have not responded with alacrity to requests to remedy widespread quality deficiencies in health care that the requesters have documented.28 This has bred cynicism among the policymaking community 25. ·James C. Robinson, Consumer-Directed Health Insurance: The Next Generation, HEALTH .AFF. SUPPL. WEB ExCLUSIVB; W5 583-W5 590 (2005); Phil Gramm, Why We Need Medical Savings Accounts, 330 NEW ENG. J. MED. 1752, 1752-53 (1994) (claiming that waste in health care is primarily attributable to the moral hazard of costlessness at the point of service); Gerald L. Musgrave et al., Lunch Insurance, 15 REGuLATION 257 (1992) (postulating a growing and wasteful "lunch system." including Lunchicare and Lunchicaid, if noontime meals were subject to the same moral hazard as health care). But see JOHN A. NYMAN, THE THEORY OF DEMAND FOR HEALTH INsURANCE (2002); John A. N~ Is "Moral Hazard" Inefficient?: The Policy Implicatiom of a New Theory, 23(5) HEALTHAFF. 194, 194 (2004) (raising the possibilitY that precommitments to health care spending through insurance are not wasteful but shift dollars tO more higbly valued uses). 26. CROSSING THE QuAUTY CHAsM, supra note 19. 27. Stephen M. Shortell et al., Implementing Evidence-Based Medicine: The Role of Market Pressures, Compensation Incentives, and Culture in Physician Organizations, 39 MED. CARE 1-62 (2001). 28. Blendon, supra note 22; Lucian L. Leape & Donald M. Berwick, Five Years After To Err Is Human: What Have We Learned?, 2931AMA 2384, 2387-88 (2005). 2006] P4P: WILL rr WORK IN THEORY? 313 that dollars speak louder than words, and has prompted a search for quality levers with financial consequences. By supplying both buyers and sellers with a pre-fabricated quality purchasing strategy, P4P represents a refinement of the general "business case" that quality gurus had urged, with only limited success, on the private payer community. Evidence-based medicine is a worthy goal, as are the others I have mentioned. But the core premise of evidence-based medicine is that scientific best practices exist that need to be more widely applied to clinical care. P4P that advances evidence·based medicine may not serve other constituencies. Scientific best practices are not necessarily market--competitive best practices. Nor are they necessarily patient--centered best practices. Performance-based payment will also need to be calibrated to address each of the three ways in which clinical practices fail to conform to scientific ideals: overuse of medical services, under use of medical services, and misuse of medical services. 29 F. Selective Contracting Pay for performance can be seen as an attempt to resurrect health insurers' and employers' cost--control capacity under cover of the afore­ mentioned "business case" for quality. Managed care failed spectacularly in the 1990s because neither consumers nor providers wanted care managed using the blunt instruments available at the time. But the reasons for managing care-rising· cost and declining private sector access-have persisted. P4P may allow private purchasers to manage care in both a politically more acceptable and a substantively more effective fashion. Utilization review-refusing to cover what your doctor says you need­ becomes quality review-determining that what your doctor has done is not good enough to warrant full payment. Capitation-a form of financial incentive that shifted insurance risk to provider groups that could not bear it without either their solvency or their commitment to quality being compromised-becomes pay for performance that takes explicit account of quality. Selective contracting-limits on choice of physician imposed for the financial benefit of insurers-evolves into coverage arrangements that allow access on preferential terms to physicians who are measurably better. G. Public Purchasing Reform Pay for performance also gives new life to efforts to make government a more effective purchaser of health care in the Medicare and Medicaid programs. It is well known that Medicare has never been able to obtain the 29. Mark R.. Chassio & Robert W. Galvin, The Urgent Need to Improve Health Care Quality: Institute ofMedicine National Roundtable on Health Care Quality, 280 JAMA 1000, 1000 (1998). 314 INDIANA HEAL Til LAW REviEW [Vol. 3:303 price and quality of services for its beneficiaries that its size and sophistication would suggest because government faces political obstacles and legal barriers that do not similarly handcuff private parties. 30 Using objective measures of medical quality to determine payment amounts under P4P is likely to provoke less public and interest group resistance than attempting to exert pure financial leverage through administered pricing. Government may even be able to back into selective contracting using P4P. Historically, it has been nearly impossible for public purchasers to exclude physicians from participation barring fraud or other extreme circumstances. But it might be politically palatable if measurably worse physicians simply receive less money. Eventually; however, a difficult challenge for government arises: what to do about physicians (or hospitals) who really do not measure up, and who should get paid nothing rather than just less. H Information Technology The dominant trend in American health policy has been to increase the supply of medical services, with other social goals such as ensuring broad access largely an afterthought31 P4P is easily appropriated by traditional supply-side forces for a straightforward if limited purpose: financing information technology in the health care system. This account of pay for performance regards the details of P4P as secondary, so long as its implementation requires sponsor organizations and participating providers to invest in electronic health records, computerized patient order entry systems, and other hi-tech tools. These resources may indeed pay long-term dividends in quality and even cost-effectiveness, but the primary objective of those who support P4P on these grounds is simply to induce capital investment in information technology. ill. UNIFY1NG THEORIES OF P4P The eight sources of enthusiasm for pay for performance outlined above come from different camps for different reasons with different manifestations. Together, they constitute a movement, but a movement without a theory. Can one sort their motives and effects into unifying principles, so that supporters from different groups can coalesce around a core notion ofP4P that society might actually carry into operation? To my mind, there are three plausible theories of pay for performance: a measurement theory, a loyalty theory, and a productivity theory. These 30. Len M. Nichols & Robert D. Reischauer, Who Really Wants Price Competition in Medicare Managed Care?, 19(5) HEALTHAFF. 30,30(2000); Bryan Dowd etal.,A Tale ofF our Cities: Medicare Reform and Competitive Pricing, 19(5) HEALTH AFF. 9, 9 (2000). 31. Lawrence R Jacobs, Politics of America's Supply State: Health Reform and Technology, 14(2) HEALTHAFF.I43, 143 (1995). 2006] P4P: WnL IT WORK. IN THEORY? 315 theories are plausible in the sense that if we as a nation, or as a profession, or as a community, decided that this is what we really want to do with P4P, we would have a pretty good sense of bow to do it and of the pros and cons involved. A. Defining "Performance" In order to appreciate the distinct implications of the three approaches, it is necessary to construct a basic typology of performance metrics. The types of rewardable performance one observes in emerging P4P systems mirror the ambiguities and tensions in the movement. Pay for performance consists of financial incentives for measurable quality, but measurable quality can take many different forms. By mentally matching points along the spectrum of potential incentives to the various motivations that exist for P4P, one begins to see the challenges involved in operationalizing a national commitment to paying for performance. Applying the traditional Donabedian classification ofhealth care quality demonstrates that performance incentives can be structured to further each aspect of quality that Donabedian laid out.32 First, one can reward "interpersonal quality," meaning the ability of doctors, hospitals, and other health professionals to connect with patients and make them feel subjectively better. One can reward interpersonal quality using measures of patient satis­ faction, continuity of care, and perhaps patient compliance with recommended treatment. Second, one can reward "technical quality," which Donabedian breaks down further into structural measures, process measures, and outcome measures. Adoption of information technology is one of the structural measures commonly incorporated into pay for performance as it has been evolving. 33 Rewards might be based on acquisition of particular computer systems or capacities like electronic medical records or computerized physician or pharmacy order entry, or on telecommunications capability such as instant messaging. Process measures are the quality metrics most frequently used in pay for performance systems to date. Many P4P programs focus on disease screening and early intervention for preventable illness. Some focus on actual treatment of disease in accordance with some notion of"best practice,'' though whether it is conformity with a science-based clinical practice guideline or a cost­ effectiveness standard depends on the context Outcome measures also can 32. A VEDIS DoNABEDIAN, THE DEFINITION OF QuALITY AND APPROACHES TO AsSESSMENT: ExPWRATIONS 1N QuALITY AsSESSMENT AND MONITORING (1980); Avedis Donabedian, Evaluating the Quality of Medical Care, 44 MILBANK MEMoRIAL FUND Q. 166 (1966). 33. Rainu Kaushal et al., The Costs of a National Health Information Network, 143 ANNALS OF INTERNAL MED. 165, 170-72 (2005). 316 INDIANA HEALTH LAW REviEW [Vol. 3:303 be associated with pay for performance systems. Some involve avoiding particular complications of medical care such as hospital-acquired infections or readmissions. Others assess longer-term; events such as restoration of functional capacity or patient survival. B. Measurement Theory What I call the measUrement theory of pay for performance is fairly common in practice but is not labeled as such. It simply asserts that the goal ofP4P is to make quality measurable. The measurement theory is founded on a view of health care inefficiency that derives from both lack of information and asymmetric information. 34 Lack of information means that health care providers do not know how well they are doing for patients, and therefore cannot improve. Asymmetric information means that health care providers know how well they are doing, but patients and health care purchasers do not, and therefore cannot choose sensibly among them. Structural pay for perf01'Dl8DCe systems linked· to the adoption of electronic medical records and other sorts of information and communication technologies may begin a cascade of me8surability.3$ Measurable quality probably means better quality, although the mechanism for achieving it may vary.36 Measurement, particularly measurement that reduces information asymmetry, is a key component of competition-based visions of American health care. In economic terms, competition involving the quality of a good or service is no different than competition based on price, except for the fact that it is harder to measure. If improved measurement capacity allows providers to make Credible assertions regarding the quality of the care they deliver, buyers can act on those assertions. So if one can measure quality better, arguably one can channel competition into fostering quality improvement. Another reason one· might want to measure quality is to benchmark quality so there can be aggressive price competition. Back in those heady days oftheClintonhealthplan,managedcompetitiongurusbelievedthatastandard benefit package with quality metrics and government oversight would make it possible for consumers to comparison shop for health insurance. If one wants better price competitimi for medical care, therefore, one also needs to 34. See WiUiam M. Sage, Reguloting Through Information: Disclosure Laws ond American Health Care, 99 CowM. L. REv. 1701, 1715-20, 1771-80 (1999). 35. See,e.g.,PeteWelch&H.GilbortWelch,Fee-jor-Data:AStrategytoOpentheHMO Black Box, 14(4) HEAI.m AFF. 104, 104 (1995) (proposing that Medicare pay HMOs more for sharing claims and encounter data).· 36. Sage, supra note 34, at 1826 (distinguishing eonsumersovereignty ftom ends-forcing regulatory or self-regulatory oversight); see also Anne-Marie J. Audet et al., Measure, Learn, ondlmprove:Physicians'lnvolvememmQualil)'lmprovement,24(3)HEAI.mAFF.843,843-44 (2005) (contrasting physicians desire ·for quality-related practice information with their resistance to external disclosure). 2006] P4P: WILL IT WORK IN THEORY? 317 be able to measure quality. This rationale for P4P, however, must be imple­ mented carefully. One caution is complexity. Adding layers of pay for performance incentives to already complicated payment methodologies in health care makes benchmarked price even less transparent to buyers. It may be good to have more information about quality, but it may not be good to combine traditional payment systems with P4P supplements. The other problem with a competition-based rationale for P4P is collusion. If quality measurement is to improve, measurement techniques need to be consistent from provider to provider and payer to payer. But the process of getting everyone on the same page in terms of what they are measuring and how much of a reward they are offering or receiving for an increment of quality may end up dampening competition rather than improving competition. Private insurers have been alert to the antitrust risks of agreeing on specific P4P payment amounts, but the health policy and medical professional literature on pay for performance by and large views the development of quality incentives as a cooperative enterprise and seems unaware of the anticompetitive potential of uniformity. The measurement theory ofP4P is also compatible with non-competitive mechanisms for quality improvement. In fact, competition on quality eventually should render explicit P4P superfluous. At the point where providers can make credible quality commitments to purchasers, the need for supplemental P4P recedes and payment terms simply integrate quality considerations with price considerations. However, P4P that furthers measurement can reinforce government regulation or professional self­ regulatory processes even if competition never develops. Whether or not the buying public values quality, once regulators or the medical profession can measure quality, they can do something about it directly. If that is the mechanism by which P4P connects measurement to quality improvement, of course, the principal benefits are going to be collective rather than accruing to individual buyers (or to individual sellers anticipating the behavior of those buyers). Consequently, economically self-interested parties will not invest enthusiastically in P4P systems because they cannot capture exclusive returns from them. A national commitment to make quality measurable through pay for performance may require public investment. C. Loyalty Theory Another plausible theory of pay for performance, though one that is harder to implement, can be called the loyalty theory. The loyalty theory grows out of the managed care experiences of the 1990s, which introduced into the health care system a host of what economists and legal scholars call 318 INDIANA HEALTH LAW REVIEW [Vol. 3:303 "agency costs.'m Instead of a dyadic relationship between a single consumer (the patient) and a single provider (a physician or perhaps a hospital), with care usually financed by a passive third party payer (a health insurer), there are various intermediaries offering or receiving financial incentives that create competing loyalties. 38 Insurers, now called managed care organizations, actively negotiate fees with providers and monitor care ~t is given to beneficiaries. "Usual and customary'' fee-for-service reimbursement gives way to discounts, withholds, capitation, and a host of mixtures. 39 Some providers join fQrces with payers to form integrated enterprises, while others create consolidated entities or common contracting vehicles to improve their financial leverage. A consequence of· this activity was to erode, or at least create the appearance of eroding, longstanding fiduciary obligations running from doctor to patient. Paying physicians for performance, like paying lawyers contingent fees, is a superficially appealing way to restore traditional ethical values by aligning incentives between provider and patient. Think about it as a Goldilocks story- the search for the '~ust righf' compensation package. If fee-for-service induces too much treatment, and capitation induces too little treatment, where should Dr. Goldilocks sleep? Policymakers continue to seek a payment formula that will reinforce both medical science and fiduciary obligation. In the midst of the national health reform debate in 1993, for example,.JAMA published an article proposing "fee for time" as an "incentive-neutral" system. 40 I assign the article to my professional responsibility students because it is so charmingly naive. Any lawyer (or client) held hostage by billable hours understands the perverse incentives that a fee for time system can create. Is pay for performan~ the holy grail for physician payment? Will it perfectly align physician behavior with patient desires? Probably not. Rewarding physicians for improving interpersonal quality using metrics such as patient satisfaction surveys does seem loyalty-enhancing. Defenders. of patient autonomy would agree that what the patient subjectively experiences from medical care is an appropriate basis for physician payment. P4P based on hard measures of clinical outcomes (technical quality) also furthers loyalty to patients. If we reward physicians for curing patients, the patients who 37. MARc A. RODWIN, MEDICINE, MONEY & MORALS: PHYsiCIANS' CONFLICTS OF INTEREST (1995). Agency costs have long been a concern of scholars Qf corporate governance. Michael C. Jensen & William H. Meckling, Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure, 3 J. FIN. EcoN. 305 (1976). 38. Lawrence Casalino, Managing Uncertainty: Intermediate Organizations as Triple Agents, 26 J. HEALTH POL. POL'Y & L. 1055, 1055-57 (2001 ). 39. Jon Gabel, Ten Ways HMOs Have Changed During the 1990s, 16(3) HEALTH AFF.l34,140-41 (1997); Robert A Berenson, Beyond Competition, 16(2) HEALTHAFF. 171, 177-78(1997). . 40. TomJ. Wachtel & Michael D. Stein, Fee-for-Time System: A Conceptual Framework for an Incentive-Neutral Method of Physician Payment, 270 JAMA 1226, 1226-29 (1993). 2006] P4P: WILL IT WORK IN THEoRY? 319 receive treatment from those physicians will be well cared for (we will return to the issue of patients who have trouble finding a physician willing to serve them under those financial conditions). On the other hand, process-based P4P-the most common form of pay for performance currently being implemented-has ambiguous implications for physician loyalty. Process-based P4P is basically compliance with rules. Someone has to set those rules, and it is not going to be the patient. Maybe third-party payers will set the rules, maybe scientific and professional bodies will set the rules, maybe government will set the rules. But the patient won't. Physicians who earn process-based P4P rewards, therefore, will likely be furthering collective interests of public and private payers in cost-effective care for populations of enrollees.41 If these types of metrics dominate, the loyalty effects of pay for performance are not going to be that different from the loyalty effects of managed care. I was a cautious supporter of managed care in the 1990s, and P4P may bring out some of the better qualities of managed care. However, it will do so by refining managed care, not by superimposing a payment formula that obviates conflicts of interest or obligation.42 D. Productivity Theory The third plausible theory of pay for performance is the one that I think has the. greatest potential payoff. Predictably, it is also the one that presents the greatest challenges for the next ten years. The idea of what I call the productivity theory of pay for performance is that setting up the incentives correctly will induce American medicine to get better at its job. Increasing productivity through P4P can be slow or rapid. One approach is to produce productivity gains without massive industry restructuring.43 That is probably what we are attempting right now. Most P4P programs seek to make doctors practice better but not fundamentally change the way that medical practice is organized. They reward the acquisition of basic information technologies. They make incentive payments for compliance with professionally determined, consensus best practices that are broadly acceptable to physicians. They key payment to process measures that can be verified using standard insurance claims data, and do other things that are not 41. A compromise approach to P4P intended to reward comprehensive, complete care rendered to individual patients has been proposed. See Thomas Nolan & Donald M. Berwick, All-or-None Measurement Raises the Bar on Petformance, 295 JAMA 1168 (2006). 42. E. Haavi Morreim, Benefits Decisions in ERISA Plans: Diminishing Deference to Fiduciaries and an Emerging Problem for Provider-Sponsored Organizations, 65 TENN. L. REv. 511, 523-34 (1998). 43. See Robert Cunningham, Professionolism Reconsidered: Physician Payment in a Small-Practice Environment, 23(6) HEALTH AFF. 36, 46 (2004) (urging the development of quality-based payment systems that accommodate the persistence, and perhaps the nonnative desirability, of small-group practice). 320 INDIANA HEALTH LAW REvlEW [Vol. 3:303 too intrusive or too. costly· to participating physicians in the hope of stimulating incremental improvement.'" But the dividends from this type of productivity-based P4P. will be limited. The problem is that there are trade-off's between how easily we can get the information on which we are basing performance incentives and how useful that information really is. We also will not be able to incorporate many outcome-based P4P measures because outcome measurement requires pooling large numbers of patients to achieve statistical validity and being able to confidently attribute higher or low~ performance accurately to the person or entity actually responsible for it. The more one segments a patient's experience among different providers offering different services at different times, the less likely it is that outcomes can ·be measured reliably and usefully.45 A related problem is that ·individual physicians have limited authority to command resources for which they do not directly pay. Unlike lawyers receiving contingent fees, who beyond the opportunity cost of their own time have at most to advance expenses for expert witnesses, a physician who truly wanted to improve outcomes would have to front hospital, pharmaceutical, and other hard costs that far exceed the potential personal financial gain from success. How might medicine improve more rapidly? Primarily by becoming more integrated, more coordinated, and larger scale than it currently is.46 The persistent fragmentation of medical practice in this country is its greatest barrier to improvement. American health care is the world's largest cottage industry. Changing this-something that pay for performance may be able to do if its proponents attempt it-is likely to have the greatest measurable benefits in terms of quality improvement and cost effectiveness.47 44. See PACIFIC BUSINESS GROUP ON HEALm, ADVAWCJNQ PHYSICIAN PERFoRMANCE MEAsuREMENT: USINGADMINISTAATIVEDATA TOAssESSPHYSICJANQuAIIIY ANDEmCIENCY (Sept. 2005), available at http://www.pbgb.org (last visited May 17, 2006). 45. A weak compromise is to reward participation in quality improvement programs regardless of short-term measurable change. See Nancy J.O. Birkmeyer & John D. Birkmeyer, Strategies/or Improving Surgical Quality-Should Payers Reward Excellence or Effort?, 354 NEW ENOL J. MED. 864 (2006). 46. See CROSSING THE QuAU1Y CHASM, supra note 19 at 61-88 (bighligbting coordination and cooperation as keys to improvement); Joseph R. Newhouse, Why Is There a QualityChasm?,21(4)HEALTHAFF.l3,22(2002)(11gR'eingwitbthe(lualityChasmreportbut describing "even more fundamental" barriers to good performance tban lack of organized systems); Diane R. Rittenhouse et al.. Physician Organization and Care Management in Califomia:FromCottagetoKaiser,23(6)HEALTHAFf.SI,S8-S9(2004)(reportinggreateruse of quality tools in HMO practice settings). But see Judith Smitb & Kieran Walshe, Big Business: The Corporatlzation of Primary Care in the UK and the USA, PuB. MONEY & MGMT., Apr. 2004, at 87 (asserting tbat tightly organized primary care is unsuccessful). 47. See. e.g., Robert S. Huclanan & Gary P. Pisano, The Firm Specificity of Individual Performance: Evidence/rom Cardiac Surgery, 52 MGMT. SCI. 473,484-85 (2006) (finding that surgeon performance is largely dependent on hospital-specific characteristics). 2006] P4P: WILL IT WORK IN THEORY? 321 For this purpose, outcome-based pay for performance is clearly superior. Outcome measurement with significant rewards attached to improvement will force providers to coordinate and combine their practices so that they can take collective responsibility for success and collectively share the rewards. 48 It remains to be seen, however, whether the political influence of fragmented provider organizations will retard the implementation of this type of P4P if proposed by government payers, or whether the public will accept integrated organizations bmught into existence through private P4P more readily than they did traditional HMOs. One unanswered technical question for P4P intended to induce organiza­ tional change, beyond the practice restructuring itself, is how financial rewards for productivity gains get distributed internally within the organiza­ tions that receive them.49 The California experience with pay for performance is instructive because capitated medical groups made much greater headway there in the 1990s than elsewhere.5° California P4P programs have two distinct components: the method by which payers structure incentives for the medical groups, and the manner in which the medical groups divide those rewards among their employed or affiliated physicians. This observation has a further implication for the design of P4P programs. For productivity gains that do not involve practice consolidation and industry restructuring, modest amounts of cash may go a long way and unanticipated consequences should be minor. Individual professionals, like all small businesspeople, respond well to small cash incentives. This seems to be the experience in England, where general practitioners enthusiastically agreed to various process measures in exchange for a pay increase. By contrast, performance rewards for corporate entities must be orders of magnitude greater to attract participation by senior executives, who then must be willing to create internal incentives for middle managers and front-line care providers who are precluded from accepting direct payment from outside the organization. Experience with incentive stock options and other performance­ based executive compensation packages in Fortune 500 corporations suggests that these programs have both benefits and risks. Major corporate scandals involving fraudulent "earnings management" and share price manipulation are 48. This is why federal antitrust enforcers regard pay for performance as a potential indicator of either clinical or financial integration, sufficient in many cases to immunize provider collaboration from per se condemnation under the Sherman Act. William M. Sage & Dev N. Kalyan, Horses or Unicorns: Can Payingfor Performance Make Quality Competition Routine?, 31 J. HEALTH POL. POL'Y &L. 529,536-39 (2006). 49. INST. OF MED., CROSSING THE QUAUTY CHASM, supra note 19, at 28-30, 111-44; Meredith B. Rosenthal et al., Transmission of Financial Incentives to Physicians by Intermediary Organizations in California, 21(4) HEALTIIAFF.197,197-98 (2002). 50. JAMBS C. ROBINSON, THE CORPORATE PRACTICE OF MEDICINE: COMPETITION AND INNOVATION IN HEALTII CARE (1999); Robin R. Gillies et al., How Different is California? A Comparison of U.S. Physician Organizations, HEALTHAFF. SUPPL. WEB EXCLUSIVE, W3 492- W3 502 (2003). 322 INDIANA HEAL1H LAw REviEW [Vol. 3:303 clear evidence that rewarding senior executives for short-term gains in market capitalization did not align management and shareholder incentives exactly as intended. E. The Government's Role Pay for performance is often described as an innovation in private health care purchasing-an example of managed care learning how to do what it wants to do better. Nonetheless, I believe government has a central part to play, and I am heartened by the fact that the Bush administration has been aggressively promoting P4P in the Medicare program. There are three aspects to the government's rQle. There is the govern­ ment's role as purchaser through Medicare and Medicaid, the government's role as regulator, and-though there is not a perfect way to describe this-the government's potential role as guarantor of access to health care and health insurance. Each aspect has implications for the design and implementation of pay for performance. Government can do very well as a purchaser because of the scale and visibility of Medicare. First, Medicare is so big that it can produce modal change in health care quality rather than marginal change .. Economists focus on differences at the margin-turning a lower quality provider into a higher quality one because of the new incentive structure-but quality control is so poorly developed in medicine that P4P must improve the average experience of patients. Only very large payers can accomplish that. Second, Medicare can encourage provider integration without completely compromising physician morale. Empirical research suggests that large organizations deliver better medical care than small organizations. st However, it also appears that physicians are less satisfied practicing in settings that limit their clinical autonomy, which may include large organizations.s2 Medicare can ease this tension through its support for medical education and training. Over the long term, only improved education and generational change in the professions will allow practice restructuring to happen in American medicine in a beneficial rather than a counterproductive fashion. Third, government can pay for public goods and partial public goods. As described above, many benefits of establishing the infrastructure to respond to P4P are collective rather than individual. In the United Kingdom, the NHS obtained buy-in for its general practitioner pay for performance requirements by offering physicians an overall increase in compensation.s3 By 51. Edward L. Halinan, The Relation Between Volume and Outcome in Health Care, 340 NEWENG.J.MED.l677, 1677-79(1999). 52. Bruce E. Landon et al., Changes in Career Satisfaction Among Primary Care and Specialist Physicians, 1997-2001,289 JAMA 442, 442 (2003). 53. Peter C. Smith, Petformance Management in British Health Care: Will It Deliver?, 21(3) HEALTHAFF. 103, 103 (2002). 2006] P4P: WnL IT WORK 1N THEoRY? 323 contrast, a private insurer would be hard pressed to justify increased compensation simply because providers are now demonstrably delivering the results they have always promised. A significant caution is that health care is such a huge slice of the federal budget that additional appropriations tied to P4P may encounter political resistance. Large-scale changes in Medicare payment are less driven by health care politics than by general fiscal and tax politics. For example, hospitals have had only modest success defending their Medicare reimbursement because even· small changes in hospital payment have significant budgetary consequences. 54 What can government do as regulator? Pay for performance is a fairly standard type of regulatory intervention. It sits somewhere in between a mandatory information disclosure law ("Tell us and the world how you are doing but nothing more is required'') and a command-and-control standard ("Tell us how you are doing and we'll tell you if you need to do better, and how"). In P4P, financial inducements make disclosure more likely, and quality metrics make disclosure more meaningful, but do not constitute mandatory standards. For example, financial incentives to adopt information technology may prove more effective, and more politically achievable, than a direct mandate. Another function ofP4P will be to focus government on the detrimental consequences of some of its prior regulation. Medical regulatory policy in this country has always preserved physicians' prerogatives both as ethical professionals and as small businesspeople. Pay for performance will force government at various levels to rethink that JeffersOnian commitment. Medical regulation, notably professional licensing but also managed care regulation, has also discouraged consumers from accepting tradeoffs between price and quality in medical care. 55 P4P clearly contemplates paying more for better medicine and less for worse medicine, a bow to economic reality that may spill over and loosen restrictions on private contracting for medical care. Finally, pay for performance should attract greater scrutiny from the federal antitrust enforcement agencies. Both private and government sponsors ofP4P have considered its regulatory interactions with federal fraud and abuse law, and with federal and state tax-exempt organization law. It is equally important for government regulators to assure that pay for performance does not evolve so as to reduce competition in American medical care. Let me conclude by mentioning government's role in assuring broad public access to medical care. Lack of health insurance and health care is a much larger and more pressing problem in American society than poor medical quality for those who are well insured. Unlike lawyers' contingent 54. Bruce C. Vladeck, ThePoliticalEconomyofMedicare,l8(l)HEALrnAFF.22,22 (1999). 55. CLARK C. HAVIOHURST, HEALTH CARE CHoiCES: PRivATE CONTRACTS AS INSTRUMENTS OF REALm REFoRM (1995); James F. Blumstein, Health Care Law and Policy: Whence and Whither?, 14 HEALTIIMATRIX 35, 38-39 (2004). 324 INDIANA HEAL1H LAW REviEW [Vol. 3:303 fees, medical pay for performance does not automatically improve access to professional services~ and may have the opposite effect.56 Iil addition to providing incentives to work bard and to work cost effectively for clients, contingent fees in litigation induce lawyers to take winnable rather than non­ winnable cases, and allow poor people who could not afford to pay hourly fees to receive legal services. Medical P4P does not generate funds for indigent care. A poor patient "recovers" health, not money, and any payment that rewards recovery must come from elsewhere. Moreover, pay for performance creates uncomfortable selection dynamics with respect to the types of cases that physician and hospitals will be willing to undertake if payment depends on successful outcomes. In law, contingent fee payment discourages frivolous litigation (setting aside its risk of provoking lawyers to act unethically when pursuing settlement) because a lawyer does not get paid for losing. However, society's interest in health care is not limited to the curable case. Nor should physicians be rewarded for treating the healthy or those with self-limiting disease. Society's commitment is to do the best for every patient, particularly patients with grave illnesses who crave hope and compassion. Government is likely to be in a stronger position than private payers to assert these interests when designing and monitoring pay for performance systems. N. CONCLUSION Since the spectacular failure of national health reform in the early 1990s, it is rare than any health policy prescription generates widespread enthusiasm. Pay for performance makes a very short list of innovations with broad-based support. With planning, and luck, P4P may contribute significantly to improving the quality of medical care. To do so, however, it needs a more clearly articulated theory than has been the case thus far. That theory, moreover, must assert a leading role for government. Performance-based payment will not automatically link financial flows in health care to desirable social outcomes. In particular, neither active quality competition nor quality­ related productivity gains apart from competition will resolve daunting problems of access and affordability in American health care. It has been my pleasure to welcome you to the Pay for Performance Medical Quality Movement. And remember---all you have to do to join is to sing it the next time it comes around on the guitar. 56. See William M. Sage, Physicians As Advocates, 13 Hous. L. REv. 1529, 1624-25 (1999).