360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 1

View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by IssueLab

No. 632 February 6, 2009

A Better Way to Generate and Use Comparative-Effectiveness Research by Michael F. Cannon

Executive Summary

President Barack Obama, former U.S. Senate industry will prevent the agency from conduct- majority leader Tom Daschle, and others propose ing useful research and prevent purchasers a new government agency that would evaluate the from using such research to eliminate low-value relative effectiveness of medical treatments. The care. need for “comparative-effectiveness research” is The current lack of comparative-effectiveness great. Evidence suggests Americans spend $700 research is due more to government failure than billion annually on medical care that provides no to market failure. Federal tax and entitlement value. Yet patients, providers, and purchasers typ- policies reduce consumer demand for such ically lack the necessary information to distin- research. Those policies, as well as state licensing guish between high- and low-value services. of health insurance and medical professionals, Advocates of such an agency argue that com- inhibit the types of health plans best equipped to parative-effectiveness information has character- generate comparative-effectiveness information. istics of a “public good,” therefore markets will A better way to generate comparative-effective- not generate the efficiency-maximizing quantity. ness information would be for Congress to elimi- While that is correct, economic theory does not nate government activities that suppress private conclude that government should provide com- production. Congress should let workers and parative-effectiveness research, nor that govern- Medicare enrollees control the money that pur- ment provision would increase social welfare. chases their health insurance. Further, Congress Conservatives warn that a federal compara- should require states to recognize other states’ tive-effectiveness agency would lead to govern- licenses for medical professionals and insurance ment rationing of medical care—indeed, that’s products. That laissez-faire approach would both the whole idea. If history is any guide, the more increase comparative-effectiveness research and likely outcome is that the agency would be com- increase the likelihood that patients and providers pletely ineffective: political pressure from the would use it.

______Michael F. Cannon is director of health policy studies at the Cato Institute and coauthor of Healthy Competition: What’s Holding Back Health Care and How to Free It. 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 2

Americans product, on medical services of no discernible spend more than Introduction value. 4 Data on the relative effectiveness of differ - $700 billion Economists describe medicine as a “cre - ent modes of care can reduce uncertainty and each year, or dence good” because patients have difficulty help purchasers, providers, and patients avoid 5 5 percent of judging its value even after consuming it, and unnecessary expenditures. Research shows therefore must rely on the advice of doctors, there is much less variation in medical spend - gross domestic who know more about such things. 1 Yet doc - ing when there is a consensus about the best product, on tors themselves frequently have difficulty course of treatment. 6 making accurate judgments about the quali - Unfortunately, current institutions appear medical services ty of their services, both before and after they to underprovide such data. In terms of med - of no discernible have provided them. Doctors (and nurses, ical interventions, estimates of the share of value. pharmaceutical manufacturers, etc.) may existing interventions that have a solid evi - think they were responsible for a good out - dence base vary, though many researchers come, or not responsible for a bad outcome, believe the share is “well below half.” 7 David but it is often impossible to know for sure. Eddy, a leading advocate of evidence-based That uncertainty guarantees that patients medicine, estimates the share to be as low as 15 will receive some services that provide little or percent. 8 In terms of overall medical spending, no value, and even some services that prove the Institute of Medicine estimates that “less harmful. than 0.1 percent is invested in assessing the A growing body of evidence suggests that comparative effectiveness of available inter - the problem of low-value medical care is ventions.” 9 That seems small relative to the much larger than it need be—that Americans estimated 30 percent of expenditures lost to spend hundreds of billions of dollars each services of no discernible value. As discussed year on medical care that delivers no value— below, a number of government activities and that many of those expenditures could reduce incentives for private entities to gener - be identified and eliminated without harm - ate comparative-effectiveness research, provid - ing health or reducing patient satisfaction. ing further reason to believe that the current Much of that evidence comes from Medi- level of spending on such research is subopti - care, the federal health insurance program for mal. the elderly and disabled, which is the single In theory, additional spending on com - largest purchaser of medical care in the nation. parative-effectiveness research could pay for Examining Medicare records, researchers have itself by reducing spending on low-value ser - found that per-beneficiary spending varies vices. To that end, many policymakers seek to widely from one area of the country to the next. boost the production of comparative-effec - In some areas, Medicare spends twice as much tiveness research. per senior as it does in other areas. Researchers Comparative-effectiveness information has have also found that beneficiaries in high- characteristics of a “public good.” Economists spending areas do not start out sicker, do not argue that markets often do not generate the end up healthier, and are no happier with the efficiency-maximizing quantity of such goods. care they receive, than beneficiaries in low- Many observers therefore propose creating a spending areas. 2 That suggests that a signifi - new federal agency devoted to generating com - cant amount of Medicare spending provides parative-effectiveness research, on the assump - no discernible benefit to the program’s intend - tion that doing so would improve economic ed beneficiaries. Those researchers estimate efficiency. 10 For example, the Medicare Mod- that as much as 30 percent of total U.S. medical ernization Act of 2003 provides funding for the spending provides no discernible value. 3 If so, Agency for Healthcare Research and Quality to then Americans spend more than $700 bil- conduct comparative-effectiveness research rel - lion each year, or 5 percent of gross domestic evant to Medicare, Medicaid, and the State

2 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 3

Types of Medical-Effectiveness Research

1. Clinical effectiveness “Does Treatment A work?” 2. Comparative effectiveness “Does Treatment A work better than Treatment B?” 3. Cost-effectiveness “Treatment A works better and costs more than Treatment B. Is the added benefit worth the added cost?”

Children’s Health Insurance Program. Recent cludable, since anyone who can look skyward proposals to expand SCHIP would create such can enjoy them without paying. As a result, an agency. 11 President Barack Obama pro - many will effectively free ride on the fireworks posed a comparative-effectiveness agency dur - displays purchased by others. If pyrotechni - ing his campaign. 12 Former U.S. Senate major - cians could exclude nonpayers, those free rid - ity leader Tom Daschle proposes a “Federal ers would have to pay in order to watch. With Health Board” that would conduct such that additional revenue, the pyrotechnicians research and use it to make coverage deci - could then produce more (and more impres - sions. 13 sive) fireworks displays. Data on The case for government provision of pub - Economic theory does not suggest that the relative lic goods in general, and comparative-effec - markets will provide no public goods. Some effectiveness of tiveness research in particular, is not so clear- people are willing to pay for fireworks dis - cut. This paper examines the factors that plays. Markets also devise innovative strate - different modes determine whether government provision of gies for boosting production of nonexclud - of care can reduce comparative-effectiveness research would able goods, such as bundling them with uncertainty and increase economic efficiency. It also examines excludable goods. Examples include the fol - state and federal policies that discourage the lowing: help avoid private generation of such information, and unnecessary that block its use. Finally, it suggests reforms Lobbying groups face a free-rider prob - that would encourage the private sector to • lem because legislative victories that expenditures. produce more comparative-effectiveness benefit members also may benefit non - research and develop innovative ways of over - members. Lobbying groups get around coming the public-goods problem. that problem by bundling additional (excludable) services, such as insurance and information, with (nonexcludable) The Public-Goods Problem lobbying services. 14 Broadcast television and radio signals Economists define public goods as those • are often nonexcludable. Broadcasters that are nonexcludable and nonrivalrous in get around the free-rider problem by consumption. A good is nonexcludable if bundling their nonexcludable program - producers cannot exclude nonpaying con - ming with advertisements, which offer sumers from enjoying it, and nonrivalrous in an excludable benefit to advertisers. consumption if one consumer can enjoy it Though charitable contributions and without diminishing others’ ability to enjoy • medical research have public-good char - it. Classic examples of public goods include acteristics, corporations and philan - national defense and fireworks displays. thropists can “purchase” an excludable Unlike most goods, markets have difficulty reputation for compassion and civic- producing the efficiency-maximizing quantity mindedness by donating to such causes. of public goods due to the free-rider problem. “No doubt,” writes Pulitzer Prize-winning For example, fireworks displays are nonex - sociologist Paul Starr, “the Rockefellers

3 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 4

sought to gain public credit and good will by supporting research approved by med - Is Government ical authorities.” 15 the Solution? Countless runners participate in road • races (e.g., the Susan G. Komen Race for Health economist and former Medicare the Cure) for amusement and exercise, administrator Gail Wilensky writes, “Economic only to see their entrance fees donated theory argues that goods or services that meet to charities they otherwise might not this [public-good] definition will be underpro - have supported. The firms sponsoring duced by the private sector and should there - those races probably would not have fore be financed by government.” 17 Wilensky’s given as generously to those charities first claim is correct; the second is not. A had their donations not also purchased descriptive science, economics makes no value them advertising and goodwill. judgments or normative statements about what government should do. Economic theory Markets increase the quantity of nonexclud - no more argues that government should pro - able goods (lobbying, research, charity), be- vide public goods than nuclear physics argues yond the amount that people are willing to that government should build atomic bombs. purchase directly, by bundling them with Economic theory can tell us whether govern - excludable goods (insurance, advertising, ment provision of public goods would increase reputation, recreation). efficiency. It goes no further. Insofar as producers can exclude nonpay - And it may not even go that far. Despite ing consumers, markets can further increase many confident assertions that government production of public goods toward the effi - provision of public goods increases efficiency, ciency-maximizing level. Consumer Reports economic theory is equivocal on that question. generates information on the quality of con - Government provision suffers from the same sumer goods. The organization excludes free-rider problem that markets do, and cre - nonpayers, albeit imperfectly, by making that ates additional problems associated with the information available only to subscribers excess burden of taxation, politicization, and who pay a fee. The better Consumer Reports crowd-out of private provision. (The appendix can exclude nonpayers—that is, the better offers a graphic explanation of the economics they can collect money from the people who of public goods under both market and gov - use their research—the more research they ernment provision.) Moreover, even if govern - can produce. ment provision would improve economic effi - Lobbying groups also try to get around the ciency, there may be other competing values at free-rider problem by excluding nonmembers stake. from enjoying the benefits of legislative victo - The object of government provision of pub - Economic theory ries. Nobel Prize-winning economist George lic goods is to increase efficiency by boosting no more argues Stigler argues that if free riders “are not repre - quantity from the market-supplied level toward that government sented in the coalition, they may find that the efficiency-maximizing level. Like the mar - their cheap ride is to a destination they do not ket, government faces a challenge in determin - should provide favor. The proposed tariff structure may ing that optimal quantity. Government could public goods than neglect their products; the research program try to approximate that quantity by asking peo - may neglect their processes; the labor negotia - ple how much they value a public good, taxing nuclear physics tion may ignore their special labor mix.” 16 people according to their preferences, and using argues that Markets create incentives for private the revenue to fund production. However, gov - government actors to overcome the challenges posed by ernment also encounters a free-rider problem: public goods. Innovators who develop ways individuals could try to reduce their own tax should build to solve the free-rider problem can capture burden by pretending not to value such atomic bombs. the money that others leave on the table. research, hoping instead to free ride on the

4 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 5

research “purchased” by those who honestly majority. Mancur Olson observes: Government reveal their preferences. Nobel Prize–winning is no better economist Paul Samuelson explains: The small oligopolistic industry seek - ing a tariff or a tax loophole will some - equipped than One could imagine every person in the times attain its objective even if the vast the market to community being indoctrinated to majority of the population loses as a determine the behave like a “parametric decentralized result. The smaller groups . . . can often bureaucrat” who reveals his preferences defeat the large groups . . . which are “right” amount by signaling in response to price para - normally supposed to prevail in a of a public good. meters . . . to questionnaires, or to other democracy . . . because the former are devices. But . . . by departing from his generally organized and active while indoctrinated rules, any one person can the latter are normally unorganized hope to snatch some selfish benefit. 18 and inactive. 21

In other words, government is no better Particularly well-organized and effective equipped than the market to determine the interest groups could conceivably boost the “right” amount of a public good. The effi - quantity of a government-provided public ciency-maximizing quantity is unknown and good beyond the (new, lower) efficiency- unknowable .19 The market may produce less maximizing level, which would create eco - than the efficiency-maximizing quantity, but nomic losses. government might produce less or more than More likely, however, such groups could that amount—either of which would involve obtain government funding for public goods economic losses. that they otherwise would have funded At the same time, government faces unique themselves. Insofar as government provision challenges. Government spending on public “crowds out” market provision, that too goods incurs what economists call the “excess imposes losses on society. Society must pay burden” of taxation, or the reduction in eco - not just the cost of those public goods, but nomic output that results from increasing tax - also the excess burden of the taxes required es. The excess burden imposes real costs on to have government provide them. society. Some economists estimate that due to For government provision of a public good the excess burden, it may cost society more to increase efficiency, (a) the gains from any than two dollars to raise just one additional net increase in supply must outweigh (b) the dollar of government revenue. 20 The excess losses stemming from the excess burden of burden therefore could make the actual cost of the taxes needed to fund any crowded out government-provided public goods as much public goods (which the market would have as twice the apparent cost, and (all else being supplied anyway), plus (c) the losses stemming equal) twice as much as the cost of market pro - from any quantity supplied in excess of the vision. By increasing the marginal cost of pro - new efficiency-maximizing level. Since it is ducing public goods, the excess burden also unknown whether the gains from (a) would reduces the optimal quantity. That shift outweigh the losses from (b) and (c), it is the - reduces the potential gains from government oretically ambiguous whether government provision and makes it more likely that gov - provision of public goods will increase or ernment would boost production beyond that reduce economic efficiency. quantity (see Appendix). Finally, economic efficiency is only one Unlike markets, government decisions among many values, such as liberty and the about providing public goods must pass rule of law. For government provision to be through the political process, where small desirable, economic efficiency must trump any groups with an intense interest in the out - conflicting values. For example, federal provi - come can override the will of a disinterested sion of many public goods conflicts with the

5 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 6

rule of law because the U.S. Constitution Evaluation Center, Hayes, Inc.; the ECRI grants Congress the power to provide or pro - Institute; the Tufts–New England Medical mote only specific public goods. Those include Center; the HMO Research Network; and “the common Defence” and “the Progress of InfoPOEMs gather and compile such informa - Science and useful Arts.” Moreover, the Con- tion for those willing to purchase it. 25 More stitution specifically enumerates the powers important, the Congressional Budget Office granted to provide or promote those public notes that “private health plans—most com - goods—such as the power “To raise and sup - monly larger or more integrated ones—conduct port Armies” and the power of “securing for their own reviews of evidence and sometimes limited Times to Authors and Inventors the undertake new analyses of comparative effectiveness exclusive Right to their respective Writings and using claims data for their enrollees.” 26 Discoveries.” 22 The Constitution’s silence with To some extent, producers of compara - regard to other public goods indicates that the tive-effectiveness information can exclude people have not granted Congress the power to nonpaying consumers. In its discussion of provide them. 23 That conclusion is reinforced private health plans that generate and com - by the Tenth Amendment: “The powers not pile such research, the CBO writes: delegated to the United States by the Consti- Government tution . . . are reserved to the States respective - Health plans may choose to publicize provision of ly, or to the people.” 24 To argue that Congress the results, or they may decide to keep comparative- should provide such public goods anyway is to their findings confidential and use argue that economic efficiency is more impor - them to shape their policies regarding effectiveness tant than the rule of law. coverage of and payment for the treat - information ments in question. For example, health may do little or plans usually have an entity known as a What about pharmacy and therapeutic committee nothing to Comparative-Effectiveness that considers the evidence regarding increase efficiency Information? the relative effectiveness of different prescription drugs and makes recom - compared to a Government provision of comparative- mendations about which ones should policy of laissez effectiveness information may do little or be covered (that is, included on formu - 27 faire. nothing to increase efficiency compared to a laries) or given preferred status. policy of laissez faire. As suggested above, mar - kets already create incentives for private actors The better that health plans become at exclud - to produce comparative-effectiveness infor - ing (and the more they choose to exclude) mation. Producers frequently can exclude nonpayers, the closer the market-supplied nonpayers from using that information. Even quantity of comparative-effectiveness research when producers do not exclude nonpayers, will come to the efficiency-maximizing quan - markets boost production by bundling com - tity, and the less likely it is that government parative-effectiveness information with ex- provision would improve efficiency. cludable goods. Finally, government provision Markets also bundle comparative-effec - would be particularly susceptible to political tiveness information with excludable goods. manipulation and would crowd out much pri - Wealthy individuals and charitable founda - vate research. tions may fund such research not only because they value the expected health gains Market Provision but also to purchase a reputation for altru - Markets do provide some comparative-effec - ism or civic-mindedness. Universities and tiveness information despite its public-good budding scholars likewise perform such characteristics. Private firms such as the Blue research to enhance their academic reputa - Cross Blue Shield Association’s Technology tions. 28

6 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 7

Most important, private health plans may ducted assessments of only two tech - gain an advantage over competitors by gaining a nologies—the artificial heart and end- reputation for generating comparative-effective - stage renal dialysis—before it expired in ness research. For example, Kaiser Permanente is 1989. Again, the voices of organized a leader in the field of comparative-effectiveness medicine and the drug and device indus - research (see below). In effect, Kaiser purchases tries were influential in achieving an excludable reputation for quality by investing CHCT’s demise. 31 in less-excludable comparative-effectiveness research. Such bundling strategies resemble the In 1995, AHCPR produced an unflattering “private clubs” approach to public-goods provi - assessment of the efficacy of many back surg - sion expounded by Nobel Prize–winning econo - eries. Back surgeons and the medical-device mist James Buchanan, 29 and further push the manufacturer Sofamor Danek stood to suffer market-supplied quantity toward the efficiency- financially, and nearly succeeded in defunding maximizing quantity. the agency. Instead, Congress cut AHCPR’s budget by a mere 21 percent. 32 According to Government Provision: Politicization author Shannon Brownlee, the retributions did Unlike market-generated research, a federal not end there: “The AHCPR was given a new comparative-effectiveness agency would be name, the Agency for Health Care Research and subject to political manipulation, which could Quality, and stripped of its authority to recom - block the generation of any useful research. mend payment decisions to Medicare and The purpose of comparative-effectiveness Medicaid.” 33 The agency got the message. After research is to demonstrate which modes of care 1995, it abandoned controversial research activ - provide value to patients and which do not. If ities that were likely very useful. 34 AHRQ never - it is to be at all useful, such research necessarily theless fell under political attack again. In 2002, poses a direct threat to the incomes of phar - the House of Representatives voted to eliminate maceutical manufacturers, medical device AHRQ’s funding, though the agency ultimate - manufacturers, and millions of providers. If a ly survived. 35 government agency produces unwelcome Princeton economist Uwe Reinhardt argues research, those groups will spend vast sums on that “AHRQ’s disturbing history and continued lobbying campaigns and political contribu - precarious existence has shown [that a similar] tions to discredit or defund the agency. approach would make [a comparative-effective - Industry groups have done so repeatedly. 30 ness agency] vulnerable to lobbying by interest Congress created the congressional Office of groups, because one or a few members of Technology Assessment in 1972 and the execu - Congress could easily imperil the [agency’s] exis - tive-branch National Center for Health Care tence through the appropriations process.” 36 Technology in 1978, charging both agencies The United Kingdom’s National Institute for with assessing the effectiveness of medical tech - Clinical Evaluation and Excellence is likewise nologies. In due course, both agencies produced under constant assault from the industry and If Congress funds 37 research that offended the health care industry. individual patients. Even if a federal compara - comparative- According to John Eisenberg and Deborah tive-effectiveness agency temporarily survives Zarin of the former Agency for Health Care the inevitable industry-led assaults, its contin - effectiveness Policy and Research, industry opposition led to ued existence, its ability to produce useful research, politics the elimination of the National Center for research, and its influence on medical practice Health Care Technology in 1981, and the Office will be highly uncertain. will govern that of Technology Assessment in 1995. Eisenberg Supporters acknowledge the problem research, imperil and Zarin continue: that political pressure creates for the agenda, its existence, credibility, and survival of a federal compara - In 1984 Congress created the Council on tive-effectiveness agency. 38 They therefore and limit its Health Care Technology [which] con - propose various approaches to insulate the usefulness.

7 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 8

The Federal agency from political influence, none of and employers, 43 or all medical expenditures. 44 Reserve Board’s which is likely to be effective. Daschle would But as MedPAC writes, “an agency that relies model the agency on the Federal Reserve on such a mandatory funding source would be mythic reputation Board, which many believe to be insulated accountable to policymakers because for independence from political influence. 39 Yet according to Congress always has the option to alter or end 45 is undeserved. economist Allan H. Meltzer, the author of a its funding.” two-volume history of the Fed, the Federal Whatever structure Congress gives to a fed - Reserve Board’s mythic reputation for inde - eral comparative-effectiveness agency, the pendence is undeserved: industry will ultimately convince Congress of what the late Sen. William Proxmire enjoyed We talk about an independent Federal reminding his colleagues about the Fed: the Reserve, but in reading and writing the agency is a creature of Congress, and Congress history of the Federal Reserve, there are may direct it at will. 46 If Congress funds com - very few occasions since the 1930s parative-effectiveness research, politics will when the Fed actually practiced inde - govern that research, imperil its existence, and pendence. . . . [current chairman Ben] limit its usefulness. Bernanke is anything but an indepen - Conservatives warn that a federal compar - dent central bank governor. He is being ative-effectiveness agency would lead to gov - leaned on by the Congress, and he ernment rationing of medical care. 47 Indeed, accedes to them. . . . In reading the min - that’s the whole idea. Ironically, the more utes of the Fed and watching what they likely outcome is that the agency will be com - do, the Fed has always been very much pletely ineffective. Compounding that irony, afraid of Congress. . . . The idea of hav - government provision of comparative-effec - ing a really independent agency in tiveness research enables opponents, such as Washington, that’s just not going to the back surgeons and their Republican happen. . . . The Federal Reserve derives allies, to cast their opposition as an effort to its power from Congress. . . . The Fed’s limit government—even as they guarantee power is delegated, and they are very greater government spending on low-value much aware that Congress could medical care. always change that. 40 Government Provision: Crowd-Out Other presumably independent federal agen - Moreover, it is likely that much of the cies, such as the Securities and Exchange comparative-effectiveness research actually Commission, face similar pressures. 41 Politi- funded by a government agency would mere - cization appears not to imperil the existence of ly crowd out research that the private sector the Fed or the SEC, perhaps because those would have funded anyway. 48 agencies have “customers” whose support is To survive, a federal comparative-effective - broad and deep enough to protect the agen - ness agency must necessarily cater to the needs cies from political attacks by disaffected of its core political constituents. When Clifton groups. In contrast, experience suggests that Gaus took the helm of AHCPR in 1994, he government agencies conducting compara - explained to agency staff (according to Brad- tive-effectiveness research do not have an ade - ford Gray and colleagues) “that the agency had quate counterbalance to attacks by the indus - to consider its customers to be those who try. would make use of the products of its work Some supporters argue that a dedicated, and . . . on whose goodwill the agency’s support mandatory funding source would provide would depend.” 49 Gray and colleagues argue more stable funding than annual appropria - that after the industry-led assault on AHCPR tions. Proposals would variously tax pharma - in 1995, the agency focused even more intently ceutical expenditures, 42 private health plans on its “customers”: “Recognizing the impor -

8 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 9

tance of engaging in activities that are valued PGPs: An Engine of Comparative- by those who directly or indirectly might affect Effectiveness Research the agency’s resources, the agency undertook In a laissez-faire environment, health plans energetic efforts to establish ongoing contacts and providers would profit from delivering and liaisons to learn what activities and types health improvements at a lower cost than their of information might be important, to whom, competitors. Competition would push plans in setting priorities.” 50 and providers to invest in comparative-effec - The primary constituency for any federal tiveness research, because that research would comparative-effectiveness agency will be pri - enable them to abandon unnecessary services vate health plans and employers. Private pur - and find less costly ways of improving health. chasers are most likely to fund comparative- Plans and providers would compete on the effectiveness research on their own, and will basis of who generates comparative-effective - have the most intense interest in the funding ness information, how well they incorporate levels and research agenda of the agency. For new information into their practice styles, and example, enthusiastic proponents of federally who has the best approach to deviating from funded comparative-effectiveness research clinical guidelines when doing so is in the include the Blue Cross Blue Shield Associ- patient’s interest. ation, 51 whose Technology Evaluation Center As noted above, a number of private entities For 100 years, collects and disseminates such research, and currently provide comparative-effectiveness federal and state Kaiser Permanente, a leader in generating such research. Among private health plans, however, governments research (see below). 52 integrated prepaid group plans (PGPs), such as If a federal comparative-effectiveness agency Kaiser Permanente and Group Health Coop- have suppressed is to survive the inevitable political attacks from erative, appear to be uniquely suited to gener - the generation providers, it must maintain a positive relation - ate and deploy comparative-effectiveness and use of ship with private health plans. As a result, those research. Those plans are integrated in the plans would likely obtain government funding sense that all the doctors and other clinicians comparative- for research they otherwise would have funded generally work for the same corporate entity. effectiveness themselves. As discussed above (and in the They are “prepaid” in the sense that the insur - Appendix), such crowd-out represents a net loss ance carrier is also part of the same corporate research. for society. 53 entity, thus the enrollees’ premiums more or less comprise the providers’ entire budget for the year. In contrast, most Americans receive Market Failure or medical care from a fragmented collection of Government Failure? providers whose incomes rise with the volume of services they provide. The current lack of comparative-effective - The combination of integration and pre - ness research is due more to government fail - payment uniquely gives PGPs the incentive ure than to market failure. For 100 years, fed - and the means to generate and use compara - eral and state governments have suppressed tive-effectiveness information. Prepayment the generation and use of comparative-effec - ensures that if a health plan delivers low- or tiveness research. Interventions on both the zero-value services, the cost comes directly out supply and demand sides push private provi - of the health plan’s bottom line. PGPs there - sion below what markets would produce in a fore face enormous financial incentives to laissez-faire environment. Moreover, the conduct research that will enable them to dis - politicization of coverage and reimbursement tinguish high-value from low-value services. decisions prevents purchasers from using Integration gives PGPs the means to mea - comparative-effectiveness research, and will sure effectiveness by tracking all services limit the utility of even government-provided received by their enrollees and those patients’ research. health outcomes. Since the mid-20th century,

9 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 10

PGPs have collected and used data from Randomized, controlled trials are the gold enrollees’ medical records to improve patient standard of medical-effectiveness research. care. 54 PGPs were the first to develop and PGPs are also well-equipped to overcome deploy electronic medical records, and they the public-good challenges inherent to com - continue to lead the industry. 55 For example, parative-effectiveness information. Investing in one Kaiser Permanente publication boasts of comparative-effectiveness research does more “a research program with millions of subjects”: than enable a PGP to avoid low-value services; it can earn the plan a reputation for quality. As Our integrated care delivery model and discussed above, PGPs therefore boost the pro - 45-year history of electronic records give duction of a nonexcludable good (compara - us the ability to analyze and leverage tive-effectiveness information) by bundling it decades of data. In 2005, Kaiser with an excludable good (reputation). Staff- Permanente had under way or complet - model PGPs are also well-equipped to keep the ed approximately 2,250 different re- findings of their research confidential, because search projects in a number of vital clinicians generally work solely for the plan. As areas, including cancer, heart disease, discussed above, the ability to exclude nonpay - diabetes, women’s health, obesity, de- ers will further encourage PGPs to boost com - pression, genetics, and disparities in parative-effectiveness research toward the effi - health care. . . . Our physicians and ciency-maximizing level. researchers also conduct clinical trials of Finally, any health plan that generates com - new drugs, medical devices, behavioral parative-effectiveness research has an advan - interventions, and other therapies. 56 tage in implementing it. PGPs may have the greatest advantage. Providers are more likely to When federal Food and Drug Administration resist efforts to change their practice style if reviewer David J. Graham sought to establish those efforts are imposed upon them by multi - whether the anti-inflammatory drug Vioxx ple and distant purchasers. When PGPs trans - (generic name: rofecoxib) increases the risk of late comparative-effectiveness information serious coronary heart disease, he turned to into practice guidelines, however, they do so in Kaiser Permanente. Their study of 1.4 million collaboration with the physicians who will use Kaiser enrollees’ medical records established those guidelines. At a minimum, that dynamic that Vioxx does increase the risk and led to the has the potential to reduce friction between drug’s withdrawal. 57 purchasers and providers, and increase the like - The ability to track and measure patient lihood that providers will use comparative- outcomes even enables PGPs to conduct ran - effectiveness research. 59 Moreover, unlike The combination domized, controlled trials of medical treat - providers who bill multiple payers, staff-model of integration ments. According to researchers Raymond PGPs need comply with only one set of clinical Fink and Mitch Greenlick: practice guidelines, which increases the likeli - and prepayment hood that comparative-effectiveness research uniquely gives PGP integrated information systems will influence medical practice. also permit the generation of study and Yet PGPs command a tiny share of the pri - PGPs the control groups, using member files for vate health insurance market—an estimated incentive and randomization based on personal char - 11 million Americans in 2004, or 4 percent of 60 the means to acteristics. In addition, these systems the insured population. That is due largely can create matched control groups for to a century’s worth of state and federal gov - generate and use members identified with a target illness ernment interventions—often enacted at the comparative- in order to observe differences between behest of the medical profession—which have the groups over time or to study the blocked the growth of PGPs, and with them, effectiveness effect of medical interventions. 58 the market’s ability to generate comparative- information. effectiveness research.

10 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 11

Suppressing Supply Finally, Medicare and other government PGPs command a Supply-side obstacles to PGPs date as far interventions favor fee-for-service payment tiny share of the back as the medical profession’s efforts to over prepayment. Fee-for-service discourages eliminate prepaid practice in the early part of providers from adopting the electronic med - private health the 20th century. The profession used the ical records that facilitate comparative-effec - insurance market 67 powers it gained under physician licensing, tiveness research. Medicare puts PGPs at a due largely to corporate-practice-of-medicine laws, and market-wide disadvantage by giving providers other measures to drive integrated, prepaid a highly remunerative alternative. Meanwhile, a century’s worth plans from the market. 61 Researchers Jon Medicare beneficiaries’ access to PGPs has of state and Christianson and George Avery write, “orga - been “highly variable,” 68 rising and falling with nized medicine also accused PGPs of being the perceived adequacy of Medicare’s pay - federal under communist influence and later used ments to private Medicare Advantage plans. In government control of local health planning bodies to January 2009, president Barack Obama pro - interventions— deny PGPs permits to construct facilities.” 62 posed eliminating the Medicare Advantage Whether the medical profession noticed the program. 69 That step would eliminate seniors’ often enacted at irony, Christianson and Avery do not say. access to PGPs, diminish PGPs’ ability to con - the behest of Licensing of medical professionals contin - duct comparative-effectiveness research, and ues to hamper PGPs. PGPs face unique incen - further distort the market toward fee-for-ser - the medical tives to employ mid-level clinicians, such as vice payment. profession. nurse practitioners and physician assistants, when doing so will reduce costs without sac - Suppressing Demand rificing quality. 63 According to professor of Government intervention has taken away health policy Jonathan Weiner, nonphysician almost any financial incentive for consumers clinicians comprise 14 percent of primary to demand comparative-effectiveness infor - care providers nationally, but 17 percent at mation. Americans pay for only a small frac - Kaiser Permanente and 25 percent at Group tion of medical services directly (13 percent). Health. 64 The scope-of-practice rules that are More importantly, consumers control a sim - part of every state’s licensing regime prevent ilarly small fraction of the money that pur - PGPs from employing mid-level clinicians to chases their health insurance (15 percent). their full potential. 65 Licensing laws therefore Due to a large federal tax break for employer- further undercut PGPs’ ability to compete on sponsored health insurance and government the basis of price. Scope-of-practice rules also health-insurance programs, employers and vary from state to state. That variation forces government control the vast majority of the PGPs to devise new, state-specific workflows $1.6 trillion spent on insurance schemes in if they seek to expand into new markets. the United States (see Figure 1). 70 Consumers State insurance regulations likewise place have little reason to demand comparative- disproportionate burdens on PGPs. States effectiveness information, because on average typically regulate PGPs and other managed they would see only 15 percent of the savings care organizations more heavily than other that result from avoiding unnecessary med - insurance carriers. 66 The fact that insurance ical spending. 71 regulations vary from state to state also pos - According to Stanford health economist es an obstacle to PGPs. Though every insur - Alain Enthoven, “less than 5 percent of the ance carrier must contend with a new set of insured workforce can both choose a health regulations when expanding into a new state, plan and reap the full savings from choosing the marginal cost of compliance is greater for economically.” 72 Consumers’ indifference to PGPs because it comes on top of PGPs’ the cost of their health insurance inhibits uniquely high start-up costs (e.g., acquiring PGPs, whose primary advantage is that they facilities, hiring a large clinician workforce, offer more affordable coverage, 73 with appar - etc.). ently no adverse effects on health outcomes. 74

11 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 12

Figure 1 Health Insurance: Who Controls the Money (2006)?

Consumers 15%

Employers 33%

Government 52%

Source: U.S. Centers for Medicare and Medicaid Services and author’s calculations.

If consumers do not enjoy the premium sav - spending for AHRQ to conduct comparative- First and ings, they will see PGPs as offering nothing but effectiveness research, it limited the federal reduced access to services. Consumers’ lack of government’s ability to use that information. foremost, cost-consciousness helped kill Kaiser The act prohibits Medicare from using that Congress should Permanente’s attempt to enter the North research to deny coverage of relatively ineffec - allow workers Carolina market. 75 tive prescription drugs. The act also prohibits In their role as health care purchasers, Medicare from limiting payments for certain to capture employers and government express what lit - services to that of the lowest-cost alternative 100 percent of the tle demand remains for comparative-effec - that is equally effective. 77 The extent of the savings from tiveness information. As discussed below, industry’s ability to block the use of effective - however, those purchasers are ill-equipped to ness research can be seen further in the fact eliminating low- make use of that information. that Congress forbids Medicare to use cost- and zero-value effectiveness in coverage decisions. Suppressing Deployment medical care. Government interventions also ensure that whatever comparative-effectiveness informa - Reforms that Would Promote tion exists will scarcely be put to use. By giving Comparative-Effectiveness employers control over the portion of workers’ Research earnings that purchases the workers’ health insurance, 76 government all but guarantees that Rather than create yet another ineffective workers will rebel when employers attempt to government agency, a better way to generate use comparative-effectiveness information to comparative-effectiveness information would reduce unnecessary services. The managed care be to undo the series of government missteps backlash of the 1990s is a case in point. that suppresses the market’s ability to create Government is even less likely to employ and use this important research. comparative-effectiveness information itself. First and foremost, Congress should roll Since the formation of Medicare, providers back government activities that insulate con - have used their political influence to prevent sumers from the cost of their health insur- Medicare from doing so. At the same time the ance, as those activities reduce consumer Medicare Modernization Act authorized new demand for comparative-effectiveness infor -

12 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 13

mation. In the Medicare program, that would and boards of medicine to compete with their mean giving enrollees a fixed, risk-adjusted counterparts in other states to provide the best voucher that enables them to purchase a basic balance between quality assurance and access. level of coverage, and letting enrollees face the It would allow consumers to shop nationwide full cost of any additional benefits. 78 For those for insurance- and provider-licensing protec - under age 65, Congress should level the play - tions, yet it would retain a strong role for local ing field between employer-sponsored cover - regulators to enforce those protections and for age and other sources of health insurance, local courts to provide quality assurance which would make the cost of their health through contract and medical-malpractice law. insurance more apparent to workers. With a Finally, regulatory federalism would reduce level playing field, markets would return to barriers to competition for those health plans workers the portion of their earnings that (i.e., PGPs) that are most likely to generate employers currently control—but only over the comparative-effectiveness research. long term. Congress should therefore endeav - To guarantee competition among regula - or to give workers more immediate control tors, Congress must itself relinquish any role in over those dollars. 79 regulating medical professionals or health Those reforms would allow workers to cap - insurance. Otherwise, the health care industry ture 100 percent of the savings from eliminat - will use federal regulation to block research- ing low- and zero-value medical care. They generating health plans, just as the industry would therefore give an enormous boost to the used state regulations for that purpose. Once demand for comparative-effectiveness infor - those protectionist regulations creep into fed - mation, and to the health plans that generate eral law, they will be much harder to dislodge. and use it. Those reforms would also reduce Unless Congress relinquishes that role, reform the health care industry’s ability to block the may not be worth the effort. generation and use of such information. Second, Congress should eliminate the reg - ulatory obstacles that inhibit comparative- Conclusion effectiveness research. State licensing of insur - ance and medical professionals creates barriers To economists, the term “public good” is to entry for new, more economical forms of not a trump card that ends debate over the health care delivery. These regulations particu - merits of a government activity. Advocates of larly burden the types of health plans most government-funded comparative-effective - likely to generate comparative-effectiveness ness research make the facile assumption research. Congress should recognize these reg - that because such research has public-good ulations for what they are—barriers to trade characteristics, government provision would among the several states—and use its power increase social welfare. In reality, they have no under the Constitution to sweep those trade idea whether the benefits of government pro - barriers away. 80 vision would outweigh the costs. The most promising approach would have The case for government provision has Congress should Congress require each state to recognize the many hurdles it must clear. Supporters must insurance and provider licenses issued by other demonstrate (1) that the pursuit of economic eliminate the states. That approach, known as regulatory efficiency should trump any competing values; regulatory federalism, would have a number of salutary (2) that existing government obstacles to pri - effects. 81 It would make basic medical care vate provision must be preserved; and (3) that obstacles that more affordable by allowing mid-level clini - after taking into account the additional costs inhibit cians to practice to their full competence. It of government provision—the excess burden of comparative- would generate much greater competition taxation, the losses due to crowd-out, the among insurers, which would drive premiums politicization and uncertainty, and the dimin - effectiveness downward. It would force insurance regulators ished incentives for the private sector to “solve” research.

13 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 14

To economists, the public-good problem—that the benefits of to society of each additional unit of a public the term “public government provision would exceed the costs. good. The downward-sloping demand curve, Supporters have yet to acknowledge these Ds = MV s, shows the marginal value to society good” is not a issues, much less build that case. of each additional unit. Because people can trump card that Comparative-effectiveness research is consume public goods without paying for ends debate over unlikely to have a serious impact on dubious them, however, we need a second demand medical expenditures, or the growth in med - curve, Dp, to represent the amount that con - the merits of a ical expenditures, until Congress removes the sumers are willing to pay for each additional government perverse financial incentives it has created for unit. providers. Since replacing those distorted To maximize efficiency, producers should activity. incentives with market incentives would also increase production whenever the next unit’s enable markets to boost production of com - value to society exceeds that unit’s cost. parative-effectiveness information, those Producers should stop only when the value of reforms must take center stage. the last unit equals its cost. In Figure A-1, the market-supplied quantity that maximizes efficiency is represented by Qm*, the point Appendix where marginal cost ( Sm) equals marginal val - ue ( Ds). Figure A-1 illustrates how the market pro - Figure A-1 shows that although markets vides a public good in the absence of govern - will produce some quantity of a public good, ment intervention. The upward-sloping sup - they likely will produce less than the efficien - ply curve, Sm = MC m, shows the marginal cost cy-maximizing amount. The reason is that

Figure A-1 The Public-Good Problem e c i r P

Quantity

14 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 15

some people will free ride on public goods Sg curve lies above Sm because government purchased by others, because producers can - provision incurs the excess burden of taxation, not exclude nonpayers. Markets will there - which may increase the cost of each addition - fore only provide a public good up to the al unit by as much as 100 percent. point where the payers —whose willingness to It is likely, though not certain, that govern - pay is represented by Dp—are no longer will - ment provision will increase the overall supply ing to pay the cost of producing the next of a public good. If so, there will be efficiency unit. In Figure A-1, that point is represented gains. Due to the excess burden, however, the by Qm'. gains would not be as great as they would be if Figure A-1 also illustrates the economic markets supplied that increase. In Figure A-2, losses caused by the free-rider problem. The Qt(g) represents the total quantity supplied in shaded triangle represents the potential bene - the presence of government provision. The fits were markets to increase production from gap between Qt(g) and Qm' represents the net the actual, market-supplied quantity ( Qm') to increase in the quantity supplied. For those the optimal, efficiency-maximizing quantity additional units, the efficiency gains are repre - (Qm* ). Equivalently, the shaded triangle shows sented not by the gap between Ds and Sm, but the “deadweight economic loss” society suffers the gap between Ds and Sg, shown as the shad - The sole when the market supplies only Qm'. ed area A. The size of the gap between actual and When government provision crowds out justification for optimal market production ( Qm* minus Qm') private provision, it imposes new losses on government depends on how well producers can exclude society. In Figure A-2, Qm(g) ' represents the nonpayers. The better producers are at exclud - market-supplied quantity in the presence of provision of ing nonpayers, the closer the market-supplied government provision. Again, Qm' represents public goods is quantity will come to the efficiency-maximiz - the quantity the market would have supplied to improve ing quantity. Importantly, the greater the in the absence of government provision. The potential gains from increasing production, difference between those quantities ( Qm' economic the greater the incentive for producers to find minus Qm(g) ') represents the amount of efficiency. mechanisms to exclude nonpayers. crowd-out. The shaded area B between Sg and The sole justification for government pro - Sm represents the excess burden associated vision of public goods is to improve econom - with having government provide those crowd - ic efficiency. Equivalently, it is to increase ed-out units. production such that the benefits outweigh An important implication of the excess bur - the costs. den of taxation is that when government sup - Figure A-2 illustrates the dynamics of gov - plies a public good, the efficiency-maximizing ernment provision. It reproduces the supply quantity is less than under a laissez-faire policy. curve for market production of public goods In Figure A-2, the efficiency-maximizing quan - (Sm) and the demand curve showing the total tity is not Qm*, but the (unmarked) intersec - value to society of each additional unit of a tion of the Ds and Sg curves. Up until that public good ( Ds). It also reproduces the point, government can increase efficiency by efficiency-maximizing quantity when supplied adding to the market-supplied quantity. If gov - by the market (Qm*). Figure A-2 omits the ernment increases the quantity supplied demand curve representing the amount that beyond that point—for example, if it increases consumers are willing to pay for each addi - production to the optimal laissez-faire quanti - tional unit ( Dp), though it retains the actual, ty Qm*—those additional units would reduce suboptimal market-supplied quantity ( Qm'). efficiency, because their cost would exceed When government supplies a public good, their expected value. The shaded area C repre - the relevant supply curve is no longer Sm but sents those potential losses. By reducing the Sg, which represents the total cost of having efficiency-maximizing quantity of a public government provide each additional unit. The good, the excess burden of taxation makes it

15 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 16

Figure A-2 The Government “Solution” e c i r P

Quantity

more likely that government would boost pro - burden reduces the gains from A and increases duction beyond that quantity. (Markets are less the losses from B and C. A large initial gap likely to commit that error; private entities tend between the laissez-faire quantity ( Qm') and the The not to invest their own resources where costs efficiency-maximizing market-supplied quan - economic case exceed expected benefits.) tity ( Qm*) implies smaller losses from B and for government Figure A-2 shows that the necessary con - larger gains from A. A large degree of crowd- dition for government provision of a public out implies greater losses from B and smaller provision of good to increase efficiency is: gains from A. A large government investment public goods in increases the likelihood of losses from C. If pro - A > B + C ducers can partially exclude nonpayers, such as general, and by bundling nonexcludable goods with exclud - comparative- That is, the efficiency gains from increasing able goods, that will increase B and reduce A. If effectiveness the total supply of a public good A must out - the groups interested in a public good are orga - weigh the combined losses due to crowd-out nized and influential, that will likely increase B. information in of private effort B and any government over- The economic case for government provi - particular, provision C. sion of public goods in general, and compar - Figure A-2 also shows how various parame - ative-effectiveness information in particular, is neither obvious ters affect the likelihood that government pro - is neither obvious nor simple. Analysts who nor simple. vision will increase efficiency. A large excess present it as such do the public a disservice.

16 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 17

tiveEffectiveness.pdf.

Notes 7. Institute of Medicine, Learning What Works Best: The author thanks Shannon Brownlee, Tyler The Nation’s Need for Evidence on Comparative Cowen, Trapier Michael, and Peter Van Doren for Effectiveness in Health Care , p. 2, http://www.iom. their helpful comments. edu/CMS/28312/RT-EBM/41137.aspx.

1. See Winand Emons, “Credence Goods and 8. Shannon Brownlee, Overtreated: Why Too Much Fraudulent Experts,” RAND Journal of Economics 28, Medicine Is Making Us Sicker and Poorer (New York: no. 1 (Spring 1997): 107–19; and “Sawbones, Bloomsbury, 2007), p. 237. Cowboys, and Cheats: Is Your Doctor, Mechanic, or Taxi-Driver Cheating You? Economics Can Help,” 9. Institute of Medicine. The Economist , April 12, 2006, http://www.econo mist.com/finance/displaystory.cfm?story_id=6800 10. Arnold Kling, Crisis of Abundance (Washington: 750. Cato Institute, 2006), pp. 87–90. For a list of orga - nizations and individuals who “have reached a 2. Elliott S. Fisher et al., “The Implications of similar conclusion,” see U.S. Medicare Payment Regional Variations in Medicare Spending. Part 1: Advisory Commission, Report to the Congress: The Content, Quality, and Accessibility of Care,” Reforming the Delivery System , June 2008, p. 112, Annals of Internal Medicine 138, no. 4 (February 18, http://www.medpac.gov/documents/Jun08_Enti 2003): 273–87; and Elliott S. Fisher et al., “The reReport.pdf. Implications of Regional Variations in Medicare Spending. Part 2: Health Outcomes and Satisfaction 11. See, for example, H.R. 3162, the Children’s with Care,” Annals of Internal Medicine 138, no. 4 Health and Medicare Protection Act of 2007. See (February 18, 2003): 288–98. Regional variations in also U.S. Congressional Budget Office, “Research medical spending or large expenditures on low- or on the Comparative Effectiveness of Medical zero-value care, by themselves, would not be a public Treatments.” policy concern. Such expenditures are a public poli - cy concern, however, when purchased with tax dol - 12. Obama ’08, “Barack Obama’s Plan for a lars (as in Medicare) or when government encour - Healthy America: Lowering Health Care Costs ages their prevalence in the private sector. On and Ensuring Affordable, High-Quality Health government encouraging the consumption of low- Care for All,” p. 7, http://www.barackobama.com/ value medical care in the private sector, see Michael pdf/HealthPlanFull.pdf. F. Cannon, “Large Health Savings Accounts: A Step toward Tax Neutrality for Health Care,” Forum for 13. Tom Daschle, Scott S. Greenberger, and Jeanne Health Economics and Policy 11, no. 2 (Health Care M. Lambrew, Critical: What We Can Do about the Reform), Article 3 (2008). Health-Care Crisis (New York: Thomas Dunne Books, 2008). Senate Finance Committee Chair-man Max 3. Elliott S. Fisher, “Expert Voices: More Care Is Baucus (D-MT) and Budget Committee Chairman Not Better Care,” National Institute for Health Kent Conrad (D-ND) have introduced legislation Care Management, no. 7, January 2005, http:// that would create a nonprofit “Health Care www.nihcm.org/~nihcmor/pdf/ExpertV7.pdf. Comparative Effectiveness Research Institute” funded by federal taxes. John Reichard, “Baucus, 4. U.S. Centers for Medicare and Medicaid Services, Conrad Offer Bill Creating Comparative Effective- “National Health Expenditure Projections 2007– ness Institute,” CQ HealthBeat , August 1, 2008, 2017,” p. 3, http://www.cms.hhs.gov/National http://www.commonwealthfund.org/healthpolicy HealthExpendData/Downloads/proj2007.pdf; and week/healthpolicyweek_show.htm?doc_id=698085 author’s calculations. &. See also S. 3408.

5. On the difficulties in applying evidence-based 14. Mancur Olson, The Logic of Collective Action: clinical practice guidelines to large groups of Public Goods and the Theory of Groups (Cambridge, patients, see Michael F. Cannon, “Pay-for-Perform- MA: Harvard University Press, 1977), pp. 132–67. ance: Is Medicare a Good Candidate?” Yale Journal of Health Policy, Law, and Ethics 7, no. 1 (Winter 2007): 15. Paul Starr, The Social Transformation of American 10–14, 37, http://www.cato.org/pubs/papers/can Medicine (Basic Books, 1982), p. 228. non_p4p.pdf. 16. George J. Stigler, “Free Riders and Collective 6. U.S. Congressional Budget Office, “Research on Action: An Appendix to Theories of Economic the Comparative Effectiveness of Medical Treat- Regulation,” The Bell Journal of Economics and ments,” December 2007, p. 12–14, http://www.cbo Management Science 5, no. 2 (Autumn 1974): 362. .gov/ftpdocs/88xx/doc8891/12-18-Compara Emphasis in original. “Whatever the theory says,

17 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 18

there is no doubt of the existence of thousands of on the Comparative Effectiveness of Medical trade associations in the United States,” 359–65. Treatments.” Emphasis added.

17. Gail R. Wilensky, “Developing a Center for 27. Ibid. Comparative Effectiveness Information,” Health Affairs Web Exclusive (November 7, 2006): w583, 28. On scholars producing comparative-effective - http://content.healthaffairs.org/cgi/reprint/hlth ness information to enhance their reputations, see aff.25.w572v1.pdf. See also similar statements in Uwe Reinhardt, “An Information Infrastructure for U.S. Medicare Payment Advisory Commission, p. the Pharmaceutical Market,” Health Affairs 23, no. 1 111; and (more tentatively) U.S. Congressional (January/February 2004): 111, http://content.heal Budget Office. thaffairs.org/cgi/reprint/23/1/107.pdf. (“Doctoral candidates might seek to make a name for them - 18. Paul A. Samuelson, “The Pure Theory of Public selves by critically examining the [agency’s] research. Expenditure,” Review of Economics and Statistics 36, Alternatively, they might wish to make their career no. 4 (November 1954): 389. Emphasis in original. by working for the [agency], which could easily attain the academic status of a first-rate university. 19. Ibid. Since Samuelson’s article was published, One could imagine, for example, that distinguished economists have devised innovative, complex, yet academicians might wish to spend a few sabbatical still-imperfect strategies for solving the demand- years with equally accomplished colleagues at [an revelation problem. See Dennis C. Mueller, Public agency].”) Choice III (Cambridge, UK: Cambridge University Press, 2003), pp. 160–68. Whatever advantages 29. James M. Buchanan, “An Economic Theory of those strategies may offer, governments rarely (if Clubs,” Economica 32, no. 125 (February 1965): ever) use them, relying instead on truly hopeless 1–14. strategies such as majority rule. 30. Indeed, political manipulation dogs federal 20. Martin Feldstein, “How Big Should Govern- activities of far lesser consequence. According to ment Be?” National Tax Journal 50, no. 2 (June the Medicare Payment Advisory Commission 1997): 197, http://ntj.tax.org/wwtax%5Cntjrec. (MedPAC): “In 2007, three former Surgeons nsf/36CFE3E5BCCB188C85256863004A5939/ General testified before the House Committee on $FILE/v50n2197.pdf. Oversight and Government Reform about their lack of independence from executive branch offi - 21. Mancur Olson, p. 127–28. See also pp. 165–67. cials. . . . from administrations of both political parties. . . . These individuals reported that 22. U.S. Constitution, Article I, Section 8. administration officials discouraged them from speaking about certain public health topics. They 23. One can argue that the “general welfare” clause also noted the declining role of the office in deal - of Article I, Section 8, grants Congress the power to ing with key issues. . . .” U.S. Medicare Payment provide other public goods, since nonexcludable Advisory Commission, p. 119. goods may be enjoyed by all and would improve the general welfare. However, if that were the meaning 31. John M. Eisenberg and Deborah Zarin, “Health of the general-welfare clause, there would be no Technology Assessment in the United States,” need for explicit grants of power pertaining to oth - International Journal of Technology Assessment in Health er public goods. Such an expansive interpretation Care 18, no. 2 (2002): 195. of the general-welfare clause would render redun - dant and meaningless the grants of power relating 32. Bradford H. Gray, Michael K. Gusmano, and to national defense, patents, and copyrights. See Sara R. Collins, “AHCPR and the Changing generally Robert A. Levy and William Mellor, The Politics of Health Services Research,” Health Affairs Dirty Dozen: How Twelve Supreme Court Cases Web Exclusive w3–w283, June 25, 2003. Radically Expanded Government and Eroded Freedom (New York: Sentinel, 2008), pp. 19–36, which 33. Brownlee, p. 294. quotes James Madison describing that interpreta - tion of the general-welfare clause as “an absurdity” 34. Bradford H. Gray, Michael K. Gusmano, and (pp. 21–22). Sara R. Collins, W3-301–W3-303.

24. U.S. Constitution, Amendment X. 35. U.S. Congressional Budget Office, “Research on the Comparative Effectiveness of Medical 25. U.S. Congressional Budget Office, “Research Treatments.” on the Comparative Effectiveness of Medical Treatments,” and Shannon Brownlee, Overtreated . 36. Uwe Reinhardt, “An Information Infrastruc- ture for the Pharmaceutical Market,” Health Affairs 26. U.S. Congressional Budget Office, “Research 23, no. 1 (January/February 2004): 111, http://con

18 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 19

tent.healthaffairs.org/cgi/reprint/23/1/107.pdf. published in the Journal of Public Economics ; and Jonathan Gruber Daniel M. Hungerman, “Faith- 37. Robert Steinbrook, “Saying No Isn’t NICE— Based Charity and Crowd-Out During the Great The Travails of Britain’s National Institute for Depression,” Journal of Public Economics 91, no. 5–6 Health and Clinical Excellence,” New England (June 2007): 1043–69. Journal of Medicine 359, no. 19 (November 6, 2008): 1977–1981, http://content.nejm.org/cgi/content/ 49. Bradford H. Gray, Michael K. Gusmano, and short/359/19/1977. Sara R. Collins, “AHCPR and the Changing Politics of Health Services Research,” Health Affairs 38. See, for example, Gail R. Wilensky, w583; U.S. Web Exclusive w3–w293, June 25, 2003. Medicare Payment Advisory Commission, p. 119. 50. Ibid., w3–w302. 39. Ibid. 51. “BCBSA: Time Is Right for Comparative 40. Allan H. Meltzer, “Meltzer on the Fed, Money, Effectiveness Research Institute; Blues’ Support and Gold,” EconTalk podcast, May 19, 2008, http: New Legislation That Builds on Healthcare Re- //www.econtalk.org/archives/2008/05/meltzer_o form Proposal,” Blue Cross Blue Shield Associ- n_the.html. Author’s transcription. ation news release, August 1, 2008, http://www. bcbs.com/news/bcbsa/time-is-right-for-compara 41. Uwe Reinhardt, “An Information Infrastructure tive-effectiveness-research-institute.html. for the Pharmaceutical Market,” Health Affairs 23, no. 1 (January/February 2004): 111, http://content 52. See, for example, “Letter from the Coalition for .healthaffairs.org/cgi/reprint/23/1/107.pdf. Health Services Research to Sens. Arlen Specter and Tom Harkin,” June 6, 2006, http://www.chsr. 42. Ibid., p. 110. org/060606.htm; “Coalition Letter to Regula and Obey on Comparative Drug Effectiveness,” Sep- 43. U.S. Medicare Payment Advisory Commission, tember 22, 2005, http://www.aafp.org/online/en p. 128. /home/policy/federal/legislation-endorsements- and-letters-to-congress/regulaobeyletter.html; 44. Alan M. Garber and Victor Fuchs, “Medical and “Letter to Chairman and Ranking Members Innovation: Promises and Pitfalls,” Brookings of Senate HELP and Budget Committees in Review (Winter 2003), http://www.brookings.edu Support of Providing Funding for Clinical Com- /articles/2003/winter_technology_fuchs.aspx. parative Effectiveness Research,” April 30, 2004, http://www.aafp.org/online/en/home/policy/fed 45. U.S. Medicare Payment Advisory Commission, eral/legislation-endorsements-and-letters-to-con p. 110. gress/2004-archives/clinicalletter.html.

46. George F. Will, “What the Fed’s Job Isn’t,” 53. If an agency fails to conduct valued research, Washington Post , April 20, 2008, p. B07. whether due to industry pressure or poor cost-ben - efit projections, then private purchasers might 47. See, for example, David Gratzer, “Not So fund such research on their own, mitigating crowd- NICE,” National Review Online , October 27, 2008, out. Nevertheless, the uncertainty created by polit - http://article.nationalreview.com/?q=ZGExMjFh ical wrangling over the agency’s agenda would MDBkZmE0MWQ4M2RhZGY3NWJkZDgyN2 delay private development of such research. U5MmQ. 54. Raymond Fink and Merwyn R. Greenlick, 48. Different economic models make divergent “Prepaid Group Practice and Health Care Re- predictions about the propensity of government search,” in Toward a 21st Century Health System: The provision of public goods to crowd out private pro - Contributions and Promise of Prepaid Group Practice , ed. vision. See, for example, Karine Nyborg and Mari Alain C. Enthoven and Laura A. Tollen (San Rege, “Does Public Policy Crowd Out Private Francisco: Jossey-Bass, 2004), p. 160. Contributions to Public Goods?” Public Choice 115, no. 3–4 (June 2003): 397–418. For empirical evi - 55. U.S. Congressional Budget Office, “Evidence on dence of crowd-out, see James Andreoni, “An the Costs and Benefits of Health Information Experimental Test of the Public-Goods Crowding- Technology,” May 2008, p. 3, http://www.cbo.gov/ Out Hypothesis,” The American Economic Review 83, ftpdocs/91xx/doc9168/05-20-HealthIT.pdf. George no. 5 (December 1993): 1317–27; Kenneth S. Chan C Halvorson, “Epilogue: Prepaid Group Practice and et al., “Crowding-Out Voluntary Contributions to Computerized Caregiver Support Tools,” in Toward Public Goods,” Journal of Economic Behavior and a 21st Century Health System , pp. 249–63. Organization 48 (2002): 305–17; Daniel M. Hunger- man, “Are Church and State Substitutes? Evidence 56. Kaiser Permanente, “We Believe in the Power from the 1996 Welfare Reform,” October 2004, of Change,” 2005 Annual Report, p. 18, http://

19 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 20

xnet.kp.org/newscenter/aboutkp/annual-reports Jessica C. Smith, “Income, Poverty, and Health /kpreport_2005.pdf. Insurance Coverage in the United States: 2007,” U.S. Census Bureau, August 2008, p. 61, http:// 57. Rita Rubin, “Scientist Says FDA Called Journal www.census.gov/prod/2008pubs/p60-235.pdf; to Block Vioxx Article,” USA Today , November 28, and author’s calculations. 2004, http://www.usatoday.com/news/health/20 04-11-28-fda-vioxx_x.htm; David J. Graham et al., 61. See, for example, Mancur Olson, pp. 137–41; “Risk of Acute Myocardial Infarction and Sudden Paul Starr, pp. 198–232; and H. E. Frech, Competition Cardiac Death in Patients Treated with Cyclo- and Monopoly in Medical Care (Washington: American Oxygenase 2 Selective and Non-Selective Non- Enterprise Institute, 1996), pp. 69–71. Regarding Steroidal Anti-Inflammatory Drugs: Nested Case- corporate-practice-of-medicine laws, see Michael E. Control Study,” The Lancet 365, no. 9458 (February Porter and Elizabeth Olmstead Teisberg, Redefining 5, 2005): 475–81, http://www.thelancet.com/jour Health Care: Creating Value-Based Competition on Results nals/lancet/article/PIIS0140-6736(05)17864- (Boston: Harvard Business School Press, 2006), p. 7/abstract; and David J. Graham to Paul Seligman 358; and Mary H. Michal, Meg S. L. Pekarske, and memorandum, “Risk of Acute Myocardial Infarc- Matthew K. McManus, “Corporate Practice of tion and Sudden Cardiac Death in Patients Treated Medicine Doctrine: 50 State Survey Summary,” with COX-2 Selective and Non-Selective NSAIDs,” National Hospice and Palliative Care Organization/ September 30, 2004, http://www.fda. Center to Advance Palliative Care, September 2006, gov/CDER/DRUG/infopage/vioxx/vioxxgraham. http://www.nhpco.org/files/public/palliativecare/ pdf. corporate-practice-of-medicine-50-state-summary. pdf. 58. Raymond Fink and Merwyn R. Greenlick, “Prepaid Group Practice and Health Care Re- 62. Jon B. Christianson and George Avery, “Prepaid search,” in Toward a 21st Century Health System , p. Group Practice and Health Care Policy,” in Toward 163. For example, “Group Health Cooperative and a 21st Century Health System , p. 66. Kaiser Permanente–Northern California used their diabetes registries for selection and randomization. 63. Kenneth H. Chuang, Harold S. Luft, and R. . . . Study subjects received additional training from Adams Dudley, “The Clinical and Economic Per- nurses, health educators, behaviorists, or pharma - formance of Prepaid Group Practice,” in Toward a cists in improving self-care and in seeking appropri - 21st Century Health System , p. 47. See generally Paul ate medical care. . . . [B]oth found among study sub - Starr, p. 225. jects higher levels of patient satisfaction and physical functioning. . . . Changes in laboratory 64. Jonathan P. Weiner, “Prepaid Group Practice findings were in a favorable direction, as were and Medical Workforce Policy,” in Toward a 21st changes in utilization in the six months following Century Health System , pp. 128–55. the study period” (p. 166). “Randomized controlled trials aimed at reducing smoking and alcohol use 65. On scope-of-practice rules generally, see have demonstrated reductions in the use of both Shirley Svorny, “Medical Licensing: An Obstacle substances as a result of brief physician interven - to Affordable, Quality Care,” Cato Institute Policy tions during the primary care visit” (p. 167). “PGPs Analysis no. 621, September 17, 2008, http:// also continue to play a critically important role as www.cato.org/pubs/pas/pa-621.pdf. valuable research platforms for population-based health care, generating the kinds of clinical data on 66. Alain C. Enthoven, “Open the Markets and defined populations that are difficult or impossible Level the Playing Field,” in Toward a 21st Century to obtain anywhere else.” See also, “Preface,” p. xxxii. Health System , p. 238; Daniel P. Gitterman et al., “The Rise and Fall of a Kaiser Permanente Ex- 59. See generally Kenneth H. Chuang, Harold S. pansion Region,” The Milbank Quarterly 81, no. 4 Luft, and R. Adams Dudley, “The Clinical and Eco- (2003): 574–80. nomic Performance of Prepaid Group Practice,” in Toward a 21st Century Health System , pp. 48–50. 67. U.S. Congressional Budget Office, “Evidence Collaboration can reduce friction between pur - on the Costs and Benefits of Health Information chasers and providers, but is unlikely to eliminate it. Technology,” p. 8. Such “tension” persists between the Permanente Medical Groups and the Kaiser Foundation Health 68. Jon B. Christianson and George Avery, “Pre- Plan. Donald M. Berwick and Sachin H. Jain, “The paid Group Practice and Health Care Policy,” in Basis for Quality of Care in Prepaid Group Practice,” Toward a 21st Century Health System , p. 72. in Toward a 21st Century Health System , p. 38. 69. ABC News, This Week with George Stephanopoulos , 60. Toward a 21st Century Health System , p. xxix; Monday, January 12, 2009, http://media.bulletin Carmen DeNavas-Walt, Bernadette D. Proctor, and news.com/playclip.aspx?clipid=8cb4275f6a44ad3.

20 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 21

70. Data are for 2006. U.S. Centers for Medicare 76. In a recent survey, 91 percent of health econo - and Medicaid Services, “Sponsors of Health Care mists agreed with the statement, “Workers pay for Costs: Businesses, Households, and Governments, employer-sponsored health insurance in the form 1987–2006,” Tables 1, 2, and 4, pp. 6, 7, 9, http: of lower wages or reduced benefits.” Michael A. //www.cms.hhs.gov/NationalHealthExpend Morrissey and John Cawley, “Health Economists’ Data/down loads/bhg08.pdf; and author’s calcula Views of Health Policy,” Journal of Health, Politics, tions. Policy, and Law 33, no. 4 (August 2008): 712. That implies that rather than encourage employers or 71. Eliminating unnecessary spending would reduce shareholders to spend their own money on work - the cost of employer-sponsored and government ers’ health benefits, the federal tax exclusion for health insurance. While those savings should pass to employer-sponsored insurance instead gave consumers in the form of higher wages and lower employers control over a significant portion of taxes, they may not. Even if they do, the benefits are workers’ earnings. not salient to consumers. See Peter R. Orszag, “Health Care and Behavioral Economics: A Pres- 77. The limitation on “functional equivalence” entation to the National Academy of Social Insur- pricing pertains to Medicare payments to hospital ance,” May 29, 2008, pp. 6–7, http://www.cbo.gov/ outpatient departments. U.S. Congressional ftpdocs/93xx/doc9317/05-29-NASI_Speech.pdf; Budget Office, “Research on the Comparative and David M. Cutler, Your Money or Your Life: Strong Effectiveness of Medical Treatments.” Medicine for America’s Health Care System (Oxford: Oxford University Press, 2004), pp. 93–95. 78. See David A. Hyman, Medicare Meets Mephistoph- eles (Washington: Cato Institute, 2006). 72. Alain C. Enthoven, in Toward a 21st Century Health System , p. 232. 79. See Michael F. Cannon, “Large Health Savings Accounts.” 73. See generally Alain C. Enthoven and Laura A. Tollen, “Preface,” in Toward a 21st Century Health 80. U.S. Constitution, Article I, Section 8. System , p. xxxv. Kaiser Permanente claims its pre - miums are “some 10 percent or more below those 81. For an overview of regulatory federalism as of competitors.” Francis J. Crosson, “The Chang- applied to health insurance, see David A. Hyman, ing Shape of the Physician Workforce in Prepaid “Health Insurance: Market Failure or Government Group Practice,” Health Affairs Web Exclusive , Failure?” Illinois Law and Economics Research (February 4, 2004): w4–w61, http://content.health Papers Series, Research Paper No. LE08-003, pp. affairs.org/cgi/reprint/hlthaff.w4.60v1.pdf. 11–12; and David A. Hyman, “The Massachusetts Health Plan: The Good, the Bad, and the Ugly,” 74. See Joseph P. Newhouse et al., Free for All? Cato Institute Policy Analysis no. 595, pp. 8–9, Lessons from the RAND Health Insurance Experiment http://www.cato.org/pubs/pas/pa-595.pdf. With (Cambridge, MA: Harvard University Press, 1993), regard to medical professionals, see Michael E. pp. 261–306; and David M. Cutler, p. 91. Porter and Elizabeth Olmstead Teisberg, p. 362; and Shirley Svorny, http://www.cato.org/pubs/ 75. Daniel P. Gitterman et al., p. 577. pas/pa-621.pdf.

21 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 22

STUDIES IN THE POLICY ANALYSIS SERIES

631. Troubled Neighbor: Mexico’s Drug Violence Poses a Threat to the United States by Ted Galen Carpenter (February 2, 2009)

630. A Matter of Trust: Why Congress Should Turn Federal Lands into Fiduciary Trusts by Randal O’Toole (January 15, 2009)

629. Unbearable Burden? Living and Paying Student Loans as a First-Year Teacher by Neal McCluskey (December 15, 2008)

628. The Case against Government Intervention in Energy Markets: Revisited Once Again by Richard L. Gordon (December 1, 2008)

627. A Federal Renewable Electricity Requirement: What’s Not to Like? by Robert J. Michaels (November 13, 2008)

626. The Durable Internet: Preserving Network Neutrality without Regulation by Timothy B. Lee (November 12, 2008)

625. High-Speed Rail: The Wrong Road for America by Randal O’Toole (October 31, 2008)

624. Fiscal Policy Report Card on America’s Governors: 2008 by Chris Edwards (October 20, 2008)

623. Two Kinds of Change: Comparing the Candidates on Foreign Policy by Justin Logan (October 14, 2008)

622. A Critique of the National Popular Vote Plan for Electing the President by John Samples (October 13, 2008)

621. Medical Licensing: An Obstacle to Affordable, Quality Care by Shirley Svorny (September 17, 2008)

620. Markets vs. Monopolies in Education: A Global Review of the Evidence by Andrew J. Coulson (September 10, 2008)

619. Executive Pay: Regulation vs. Market Competition by Ira T. Kay and Steven Van Putten (September 10, 2008)

618. The Fiscal Impact of a Large-Scale Education Tax Credit Program by Andrew J. Coulson with a Technical Appendix by Anca M. Cotet (July 1, 2008)

617. Roadmap to Gridlock: The Failure of Long-Range Metropolitan 360451_PA632_1stClass:3603_PA632_1stClass 2/6/2009 3:44 PM Page 23

Transportation Planning by Randal O’Toole (May 27, 2008)

616. Dismal Science: The Shortcomings of U.S. School Choice Research and How to Address Them by John Merrifield (April 16, 2008)

615. Does Rail Transit Save Energy or Reduce Greenhouse Gas Emissions? by Randal O’Toole (April 14, 2008)

614. Organ Sales and Moral Travails: Lessons from the Living Kidney Vendor Program in Iran by Benjamin E. Hippen (March 20, 2008)

613. The Grass Is Not Always Greener: A Look at National Health Care Systems Around the World by Michael Tanner (March 18, 2008)

612. Electronic Employment Eligibility Verification: Franz Kafka’s Solution to Illegal Immigration by Jim Harper (March 5, 2008)

611. Parting with Illusions: Developing a Realistic Approach to Relations with Russia by Nikolas Gvosdev (February 29, 2008)

610. Learning the Right Lessons from Iraq by Benjamin H. Friedman, Harvey M. Sapolsky, and Christopher Preble (February 13, 2008)

609. What to Do about Climate Change by Indur M. Goklany (February 5, 2008)

608. Cracks in the Foundation: NATO’s New Troubles by Stanley Kober (January 15, 2008)

607. The Connection between Wage Growth and Social Security’s Financial Condition by Jagadeesh Gokhale (December 10, 2007)

606. The Planning Tax: The Case against Regional Growth-Management Planning by Randal O’Toole (December 6, 2007)

605. The Public Education Tax Credit by Adam B. Schaeffer (December 5, 2007)

604. A Gift of Life Deserves Compensation: How to Increase Living Kidney Donation with Realistic Incentives by Arthur J. Matas (November 7, 2007)

603. What Can the United States Learn from the Nordic Model? by Daniel J. Mitchell (November 5, 2007)

602. Do You Know the Way to L.A.? San Jose Shows How to Turn an Urban Area into Los Angeles in Three Stressful Decades by Randal O’Toole (October 17, 2007) 3603_PA632_no indicia:3603_PA632_no indicia 2/6/2009 9:23 AM Page 24

601. The Freedom to Spend Your Own Money on Medical Care: A Common Casualty of Universal Coverage by Kent Masterson Brown (October 15, 2007)

600. Taiwan’s Defense Budget: How Taipei’s Free Riding Risks War by Justin Logan and Ted Galen Carpenter (September 13, 2007)

599. End It, Don’t Mend It: What to Do with No Child Left Behind by Neal McCluskey and Andrew J. Coulson (September 5, 2007)

598. Don’t Increase Federal Gasoline Taxes—Abolish Them by Jerry Taylor and Peter Van Doren (August 7, 2007)

597. Medicaid’s Soaring Cost: Time to Step on the Brakes by Jagadeesh Gokhale (July 19, 2007)

596. Debunking Portland: The City That Doesn’t Work by Randal O’Toole (July 9, 2007)

595. The Massachusetts Health Plan: The Good, the Bad, and the Ugly by David A. Hyman (June 28, 2007)

594. The Myth of the Rational Voter: Why Democracies Choose Bad Policies by Bryan Caplan (May 29, 2007)

593. Federal Aid to the States: Historical Cause of Government Growth and Bureaucracy by Chris Edwards (May 22, 2007)

592. The Corporate Welfare State: How the Federal Government Subsidizes U.S. Businesses by Stephen Slivinski (May 14, 2007)

591. The Perfect Firestorm: Bringing Forest Service Wildfire Costs under Control by Randal O’Toole (April 30, 2007)

590. In Pursuit of Happiness Research: Is It Reliable? What Does It Imply for Policy? by Will Wilkinson (April 11, 2007)

589. Energy Alarmism: The Myths That Make Americans Worry about Oil by Eugene Gholz and Daryl G. Press (April 5, 2007)

PA Masthead.indd 2 2/9/06 2:08:35 PM