Health Affairs, 23, no. 1 (2004): 52-63 doi: 10.1377/hlthaff.23.1.52 © 2004 by Project HOPE	New Online   Pay Cuts For Medicare Docs   Access To Care Woes   Public Coverage More Efficient   Empowering Consumers

This Article Abstract Figures Only Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (13) Citing Articles via Google Scholar Google Scholar Articles by Carpenter, D. P. Search for Related Content PubMed PubMed Citation Articles by Carpenter, D. P. Related Collections Consumer Issues Pharmaceuticals Research And Technology Politics

Environment

The Political Economy Of FDA Drug Review: Processing, Politics, And Lessons For Policy

Abstract

 
U.S. Food and Drug Administration (FDA) drug review bears a structural similarity to many decisions made by other regulatory agencies: high uncertainty, low reversibility, avoidance of observable error, and high political stakes that induce lobbying by interested parties. This paper explores the policy lessons to be learned from viewing FDA drug review as a politically shaped exercise in information processing. I argue that the incentives facing regulators induce limits on the degree to which drug review can be accelerated, that the same incentives could render privatization initiatives problematic, and that political pressures could play a useful role in identifying priority drugs.

Not so with pharmaceuticals. The marketplace for pharmaceuticals is one of the most highly regulated industries in the U.S. economy.¹ To use any new pharmaceutical product, the patient must secure the approval of two agents: a licensed physician and the U.S. Food and Drug Administration (FDA).²

No discussion of the past, present, or future of the pharmaceutical industry can ignore the critical role played by the FDA in its evolution. The agency’s drug review decisions are essentially final (contesting them is extremely difficult and costly) and immensely consequential (regulators in other nations frequently cue off of the FDA’s decisions). If the FDA so chooses, it can materially impede the flow of new products to the pharmaceutical marketplace, or it can help accelerate that flow. Just as important, major changes are now occurring at the FDA—including the merger of its drugs and biologics review divisions—that will shape the future of the marketplace.

In this essay I use some lessons from political science to illuminate the way the FDA operates. My central claim is that FDA drug review is an exercise in learning shaped by organized interests. Both the learning and the politics have some peculiar features. The learning—more specifically, "optimal stopping" (deciding not just if but when to terminate drug review and approve an application)—is driven by the FDA’s desire to safeguard its reputation for protecting the public’s health.³ The politics involves the mobilization of drug-specific lobbies—mainly the firm submitting the drug and the patients for whom the drug is intended.

The learning incentive and the politics incentive combine to form a trade-off. The FDA will highly prize new information on a drug and will often delay approval to obtain that information. In most cases, however, there is a political cost to delay, and in recent years that cost has been shaped even more by patients and their lobbies than by pharmaceutical firms.

FDA officials seem keenly aware of these trade-offs and of the political difficulties that rejection of a new drug application (NDA) can cause, especially when few or no existing therapies exist for a given disease. In 1991 Paul Leber (then director of the FDA’s Division of Neuropharmacological Drug Products) read the initial drug application for Burroughs-Wellcome’s Lamictal (lamotrigine) for partial epileptic seizures. He found the NDA to be "disorganized" and poorly keyed to protocols for clinical trials. He considered a "refuse-to-file" (RTF) action, a major embarrassment for the pharmaceutical firm in which the FDA returns the NDA without reviewing it. Yet Leber decided against an RTF, reasoning that since "no new anti-epileptic product had been marketed over the previous 12 years...a refuse to file action, although justified, could have untoward political consequences."⁴

In this essay I consider the lessons and policy implications of this trade-off. My argument and evidence here are both taken from a larger project on the evolution of pharmaceutical regulation at the FDA over the past half-century. This project entails a massive empirical enterprise, including collection of data on more than 17,000 pharmaceutical products (and more than 2,000 new chemical entities, or NCEs) developed over the past fifty years; more than 250 companies that have submitted NDAs; the epidemiology of more than 250 medical conditions ("primary indications") for which NDAs have been submitted; more than 3,000 disease and patient advocacy groups representing medical conditions for which NDAs have been submitted; and more than 4,000 personnel who have served in the FDA from 1980 to the present.

Protecting Reputation

Top Editor's Notes Protecting Reputation Drug Review As A... The Politics Of Review:... What Does The Future... Notes

 
I posit that the FDA behaves in ways that enhance its reputation for protecting consumer safety and public health. As gauged by public opinion polls, the FDA remains one of the most popular agencies in government, regularly securing 70 percent or greater "approval" of its performance among sampled respondents.⁵ Buttressing this popularity are powerful symbolic lessons of history: The FDA is widely credited with saving thousands of American lives in its response to the sulfanilamide tragedy of 1937 and the thalidomide scandal of 1959–1961.⁶ Congress dramatically strengthened pharmaceutical regulations after each of these events. In short, whether or not the agency deserves it, the FDA clearly possesses a reputation for protecting public health and consumer safety.

This reputation did not arise by accident but is the result of refined bureaucratic strategies. FDA officials have labored for years to craft and guard their organizational legitimacy. They have done so through frequent contact with their various constituents: attending professional meetings; giving lectures to ensure that the FDA’s perspective on a given issue is heard; and cultivating the advice of academic and medical scientists, particularly on FDA advisory committees.⁷

Bureaucratic reputations often have multiple audiences. The FDA’s reputation is no exception; it has scientific, political, and popular facets. FDA officials want good press (or no bad press) in the news media but also want to preserve a reputation for scientific rigor among academics and medical professionals. As Alison Lawton, a long-time FDA observer, noted recently, "The FDA is very responsive to what I would call ‘opinion leaders’ in the scientific and medical communities. It cares very much about what these people think as to how the agency is doing."⁸

Another critical audience lies in Congress, particularly in the committees that oversee the FDA. For much of the FDA’s history, these bodies have launched criticisms at the agency for approving drugs too frequently and too quickly, and the FDA appears to sense these criticisms acutely. Commissioner Alexander Schmidt offered some testimony to this effect in 1974:

In all of the FDA’s history, I am unable to find a single instance where a Congressional committee investigated the failure of FDA to approve a new drug. But, the times when hearings have been held to criticize our approval of new drugs have been so frequent that we aren’t able to count them... The message to FDA staff could not be clearer. Whenever a controversy over a new drug is resolved by its approval, the Agency and the individuals involved likely will be investigated. Whenever such a drug is disapproved, no inquiry will be made.⁹

Contrary to the claims of some analysts, there is nothing inherently "selfish" or "inefficient" about reputation protection.¹⁰ Reputation is simply a currency of bureaucratic politics. Agencies with strong reputations can more easily attract desired personnel; fend off budget cuts; and lobby for the programs, funds, and other things they desire. There are other things that bureaucracies protect and "maximize," but for many agencies such as the FDA, reputation protection serves as the simplest and most powerful dynamic governing their behavior.¹¹

Drug Review As A Stopping Problem

Top Editor's Notes Protecting Reputation Drug Review As A... The Politics Of Review:... What Does The Future... Notes

 
In reviewing NDAs, the FDA must choose not only whether to approve, but when to approve. Every time the FDA reviews a new drug, it "invests" (takes a chance with) its reputation. There are three critical aspects of this decision.

Inherent uncertainty. FDA officials know that even the most successful clinical trials cannot eliminate the possibility that a drug will turn out to be unsafe or inefficacious. Consider, for example, the 1996 review of SmithKline Beecham’s Requip (ropinerole) for idiopathic Parkinson’s disease. In his summary memorandum, FDA official Paul Leber discussed Requip’s safety data and added an important cautionary note:

Because no pharmacologically active drug substance is entirely free of risk, the conclusion that a drug has been shown to be "safe for use," is actually no more than an opinion... Accordingly, risk to benefit assessments are inherently arguable, all the more so because each turns not only on personal sentiments about the nature of risks and benefits of a drug, but upon incomplete and imperfect information concerning the drug’s risks.¹²

Similarly, in 1997 FDA official Rudolph Widmark summarized safety data from Wyeth’s Duract (bromfenac sodium) for postoperative pain relief and cautioned:

In our safety review of NDA study we usually do not get definitive answers based on unequivocal data but are forced to interpret "flagging" events. We think that in the case of bromfenac, we have seen a "liver flag" that can be only fully explored through responsible marketing of the drug.¹³

Some uncertainty will always remain in drug review, and the marginal benefit of more trials and more delay tends to decline as the drug review gets longer.

Asymmetric observability of error. In the language of decision theory, a "Type I error" occurs when a decisionmaker accepts as true a hypothesis that is in fact false. A "Type II error" occurs when a decisionmaker rejects a hypothesis that is in fact true. The FDA, then, may be said to commit a Type I error when it approves a "bad" drug and a Type II error when it fails to approve a drug that should have been approved. For most of the FDA’s history, Type I errors have been more visible than Type II errors.¹⁴ As the remarks from former Commissioner Schmidt illustrate, the FDA has often been excoriated for approving a bad drug (or approving it too quickly) and only recently has been criticized for approving drugs too slowly.

Low (reputational) reversibility. Finally, the damage of a faulty approval decision is difficult to undo. Of course, the FDA can always secure a recall of a faulty product or compel the manufacturer to attach a "black box warning." Yet these steps will only publicize the error that the agency has made. Even though drug approvals are procedurally reversible, the FDA views drug approval as irreversible from the standpoint of reputation.

My model of drug review, then, is one in which agency reviewers are conducting a "cost-benefit" analysis of the drug at each moment in the review process. The costs of approval can be thought of as the publicly observable hazards of the drug (the FDA rarely gets criticized for approving a safe but inefficacious drug, again because safety errors are much more visible than efficacy errors). The benefit of approval is to reduce the political pressure that patients, medical professionals, and pharmaceutical firms can apply directly or indirectly upon the FDA (witness the AIDS lobby of the late 1980s and early 1990s).

There is a hitch, however, and the problem is not as simple as it might seem. When the FDA sees approval as irreversible, probability decision theory predicts that the FDA will approve a drug only when the benefits of approval exceed both the costs of approval and the benefits of waiting further.¹⁵ We can think of the benefits of waiting as the amount of information gleaned from another look at the file, or perhaps another test conducted by the drug company. As it turns out, this "value of waiting" is highest during the earliest stages of review, when the least is known about the drug.¹⁶

The Politics Of Review: New Factors In The Policy Arena

Top Editor's Notes Protecting Reputation Drug Review As A... The Politics Of Review:... What Does The Future... Notes

 
If neither doctors, patients, nor drug firms could apply public pressure for FDA approvals, the agency would find it much easier to delay drugs indefinitely. The costs of delay are not simply scientific; they are political. That is, they are subject to how well those who demand drugs can press their case before the agency, Congress, the media, and other public fora. Before the 1980s it was rare for anyone outside of clinical or academic circles to criticize the FDA for delay. Put differently, few in the media or in Congress were complaining of the agency’s Type II errors.

Today the situation is much different, and the political power of patients matters as much as or more than the political power of firms. The best evidence for this proposition comes in two strategies that are now widely adopted by pharmaceutical firms: (1) Firms themselves have in the past six to eight years created, fostered, and subsidized a number of patient advocacy groups; and (2) firms regularly seek alliances with patient advocates in pressing the case for priority status, accelerated approval, or simply approval before the FDA. The second of these is a much more common, and much more successful, strategy. Put differently, politically strategic pharmaceutical firms know that industry lobbying is less successful than patient advocacy, and their regulatory behavior adapts to this fact.

FDA drug approval has in recent years been powerfully shaped by two related factors: the increase of patient advocacy groups and the increasing visibility of Type II errors.

Patient advocacy groups. The past two decades have witnessed an explosion of interest groups and in no field more dramatically than in health.¹⁷ Several studies have shown that the increase is largely attributable to nonprofit and citizens’ groups. As part of the FDA project, my research team and I have tracked the evolution of disease-advocacy groups over the past half-century. As of 2000 we aggregated more than 3,100 disease-specific advocacy groups with at least some involvement in political issues. We were able to find founding dates for more than half of these groups. We found evidence of an explosion of high-specificity health groups in the 1970s and especially the 1980s.¹⁸

Type II errors. The rise in patient advocacy has led to a balancing of the visibility of Type II versus Type I errors. Before the 1980s it was rare for the public’s attention to be drawn to a drug that the FDA had not approved or was reviewing slowly. The AIDS epidemic changed this, less because AIDS protestors changed the FDA than because FDA officials foresaw the extraordinary politics of AIDS and rushed HIV treatments to approval.¹⁹ Yet AIDS was only the beginning of a much larger story of disease-based political mobilization in the United States. To a degree never before witnessed, disease-specific lobbies now press Congress for medical research funding, insurers and state governments for favorable coverage rulings, and the FDA for quick approvals (Exhibit 1).

Yet arthritis receives far more media coverage than asthma does. If we examine coverage in the Washington Post for a given year in the 1990s, there were 105 (nonobituary) stories mentioning asthma that year, but almost twice as many (204) mentioning arthritis. Similarly, during 1967–1997 there were 117 arthritis-related stories on the nightly newscasts of the three major networks (ABC, NBC, and CBS), but only 78 asthma-related stories.²¹ Perhaps this is not surprising, since the early 1980s arthritis drugs have consistently been approved with much greater speed (an average of twenty months) than have drugs for asthma (an average of thirty-two months, or a full year more).

Certain cancers. Consider also the most common and deadly forms of cancer—breast, lung, and prostate. Prostate cancer is the most prevalent of these conditions, and lung cancer is by far the deadliest, but breast cancer has far more organizations, research dollars, media coverage, and quick drug approvals (Exhibit 1).

Advocacy groups’ slow rise. One puzzle is why it took so long for patient and disease advocacy groups to become better organized, given that some have been around a long time. I can only hazard two educated guesses. First, political scientists have characterized the period before the 1970s as one dominated by industry, labor, and trade associations. In the 1980s and 1990s citizens’ groups came to the fore. Disease and patient advocacy groups thus might be part of a larger trend. Second, the rise of disease advocacy groups could present an example of organizational learning. Many unorganized disease communities, witnessing the political and economic successes of the AIDS and breast cancer advocacy coalitions, have been motivated to form their own groups and to enter the political arena.²²

Examples of patients’ power. As illustrative evidence of the power of patient advocates in publicizing possible Type II errors, consider the recent approval of AstraZeneca’s Iressa (gefitinib) for Stage III non-small-cell lung cancer patients. Despite strong skepticism from the Center for Drug Evaluation and Research (CDER) statistical reviewer and from industry watchers, the FDA approved Iressa in May 2003, in part for two reasons. First, the Wall Street Journal ran several editorials urging the FDA to approve it (the most vocal on 24 September 2002). So strong and visible was this pressure that FDA officials are reported to have complained to AstraZeneca about the editorials, worried of a link between the company and the editorial page. Second and more important, lung cancer patient advocates strongly supported approval, and several representatives of the groups offered robust and emotional testimony for the drug at a critical FDA advisory committee meeting in September 2002. As one journalist wrote following the meeting, "These patients—all of whom took Iressa through a compassionate use program—seemed to be the wild card that really helped AstraZeneca in the end. The company has allowed more than 18,000 patients access to Iressa outside its clinical trials, creating a very vocal and persuasive lobbying voice in the drug’s favor."²³

Resources. Perhaps the most pervasive influence upon FDA drug approval times has been the presence or absence of plentiful FDA staff to review new applications. One important reason that FDA drug approval times slowed in the 1970s, engendering complaints of a "drug lag," is that the 1962 Amendments to the 1938 Food, Drug, and Cosmetic Act piled many new responsibilities onto the FDA without a proportionate increase in personnel. The median FDA review time for new molecular entities (NMEs) submitted in 1978 was 30.8 months, and 30.0 months in 1983. In other words, the average drug was taking two and a half years to get approved; many others took considerably longer.

In recent years this situation has begun to change, particularly with the enactment of the Prescription Drug User Fee Act (PDUFA) of 1992 and subsequent renewals of this legislation. PDUFA is a more complicated law than I can assay here, but its most important provisions create (1) a system of per application "user fees" that fund increases in the reviewer staff at CDER, and (2) an incentive structure whereby the legislation is renewed only if the FDA meets specified performance goals.²⁴ In part because of PDUFA and in part because CDER staff totals began to rise five years before PDUFA, the average review time for NMEs has greatly declined, to thirteen months in 2002. The key here is that FDA staff began to increase appreciably five years before PDUFA was enacted and that the long-run effect of resources on review times is negative and large (with an elasticity of –1.6 to –2.2, meaning that a 10 percent increase in CDER personnel yields a 16–22 percent decrease in expected drug approval time).

Some statistical evidence. Iressa’s approval points to two factors: organized patients and media coverage of the disease. One way of assessing the influence of media and patient groups is to conduct a duration analysis of drug review times in which measures for these constructs are included as explanatory variables. Using data on 540 NCEs reviewed by the FDA over the past twenty-five years, I conducted a set of maximum likelihood duration analyses, regressing the approval time for a drug upon (1) a measure of the wealth of advocacy groups that existed for the primary indication disease of the drug; (2) a measure of the amount of news coverage the primary indication disease received in the three years before the NDA was submitted; and (3) a set of controls, including epidemiological variables, fixed effects for the submitting firms, a set of "shared frailties" (akin to random effects) for the primary indications of the drugs, and proxies for the number of previously existing therapies for the drug’s primary indication and the staff resources of the FDA.²⁵

Exhibit 2 shows the reduction in expected FDA approval time from a one-standard-deviation increase in the variable in question. Begin with the "baseline" predictions of the two models estimated, lognormal and gamma. These are 23.7 and 24.1 months, respectively. If all other covariates of FDA review times are held at their means, the marginal effect of a one-standard-deviation increase in Washington Post stories ranges from 4.2 to 8.4 months’ reduction in expected approval time from the baseline. Similarly, marginal effects of a standardized increase in group wealth range from 3.7 to 7.1 months’ reduction in expected review time. Finally, a one-standard-deviation increase in CDER staff (200 full-time equivalents, or FTEs) yields a reduction of three to four months in approval time for all drugs. Note that since the models include disease-specific frailty parameters, these observed associations cannot be attributed to unmeasured disease-level heterogeneity in drugs.

View larger version (29K): [in this window] [in a new window]   EXHIBIT 2 Marginal Effects Of FDA Staff, Advocates’ Wealth, And Media Coverage On New Molecular Entity (NME) Approval Times

What Does The Future Hold?

Top Editor's Notes Protecting Reputation Drug Review As A... The Politics Of Review:... What Does The Future... Notes

 
Further cuts in approval times? While approval times for NMEs (especially priority drugs) have become shorter during the past fifteen years, one implication of the perspective presented here is that further reductions (if they can be generated at all) are likely to come at a much higher marginal cost. Trimming two months from the median approval time when yearly averages were thirty to thirty-six months was much easier; now that CDER reviewers commonly crank through priority drugs in six months or less, further reductions will likely come only through dramatic procedural change. As CDER official Kenneth Edmunds recently cautioned, "My fear is that there is a law of diminishing returns setting in. It may be expensive to wring that last 5 percent of improvement out of the FDA without some things giving way. All the easy water has been drained out of the system. All the fat is near the bone now."²⁶

Third-party review and privatization? There have been several proposals to rely on non-FDA ("third party") reviewers or to privatize the FDA’s review processes. Among these are actual provisions for occasional third-party review in recent legislation (for medical devices in the 1997 FDA Modernization Act, or FDAMA) and proposals from the Hoover Institution to make the FDA a certifier of private reviewers.²⁷ In both cases, reformers wish to exploit interorganizational competition among reviewers to reduce approval times and increase the likelihood of approval.

The form and likely effects of privatization are far too complex a subject to be tackled here. Still, it is difficult to believe that the incentives facing FDA drug reviewers would not materialize under privatization. In the simplest case, third-party reviewers would rationally decide to compete not just over quickness in review but also over minimization of visible errors. In the presence of any tort liability, moreover, the procedural conservatism of third-party reviewers might be greater than that under the FDA status quo.²⁸ Whether privatization would still be preferable to the status quo is another issue, but the limits posed by the incentives of drug review ought to be kept in mind.

Disease politics as usual? Patient advocacy groups and media attention can accelerate drug review, particularly for diseases that have well-organized advocates and are newsworthy. The problem, as often in politics, is one of inequality: Better-organized and more publicly salient groups get a disproportionate share of the "benefits" (quicker approvals). From a policy standpoint, there are at least two ways of looking at this pattern. One steady conclusion is that it represents an unfortunate (perhaps indefensible) result of distributive politics. Another interpretation would see these political patterns as more benign. With scarce resources, the FDA cannot avoid privileging one disease over another in its drug approval decisions. Moreover, more highly organized and "newsworthy" medical conditions are likely to be (but might not always be) more prevalent, deadlier ones. In short, the FDA must make tough (moral) choices, and disease politics could provide crucial information and guidance in doing so.

Pharmaceutical regulation is an immensely complex process, and no combination of theoretical modeling and empirical analysis can ever do it full explanatory justice. There is also much that the FDA does in drug regulation that I have not covered here, including labeling and advertising regulation, the promotion of good manufacturing practices, and the regulation of clinical trials. Nonetheless, viewing FDA drug review as a learning exercise shaped by organized interests sheds illumination upon the process and its policy implications. The FDA protects its reputation, views its approval decisions as irreversible, and responds dramatically to patient advocacy groups and their coverage in the media because they make the consequences of delay and rejection more visible.

Editor's Notes

Top Editor's Notes Protecting Reputation Drug Review As A... The Politics Of Review:... What Does The Future... Notes

 
Dan Carpenter is a professor of government at Harvard University, in Cambridge, Massachusetts, and a fellow in residence at the Center for Advanced Study in the Behavioral Sciences, Stanford University, in Stanford, California.

The author acknowledges the National Science Foundation (Grant no. SES-0076452) for research support and thanks the editor and three anonymous reviewers for criticisms and suggestions. He also acknowledges responsibility for all characterizations, errors, and omissions.

Notes

Top Editor's Notes Protecting Reputation Drug Review As A... The Politics Of Review:... What Does The Future... Notes

 

1. The stringency of regulation is arguably higher in other nations, because the United States does not (at this writing) directly constrain pharmaceutical prices.
2. This is a very simplified representation of the pharmaceutical marketplace. For more, see the FDA’s Web site, www.fda.gov/cder.
3. For a general theory and empirical investigation of how agencies engage in "reputation-maximization," see D.P. Carpenter, The Forging of Bureaucratic Autonomy: Reputations, Networks, and Policy Innovation in Executive Agencies, 1862–1928 (Princeton, N.J.: Princeton University Press, 2001).
4. "Lamictal Efficacy Comparable to Carbamazepine in First-Line Epilepsy, Glasgow Study; Lamotrigine in Phase III for Monotherapy, Pediatrics," Pharmaceutical Approvals Monthly, F-D-C Reports (January 1996): 29.
5. See Pew Research Center, Deconstructing Distrust: How Americans View Government (Washington: Pew Research Center, 1998), 33. This is an imperfect measure, used here only for heuristic purposes. Such "approval ratings" could simply signify public agreement with the agency’s mission and not its performance. Still, it is worth noting that all but a handful of other agencies score materially lower in these polls.
6. P.J. Hilts, Protecting America’s Health: The FDA, Business, and One Hundred Years of Regulation (New York: Alfred A. Knopf, 2003), chaps. 5 and 10.
7. I acknowledge Susan Moffitt, a doctoral candidate in the University of Michigan Department of Political Science who is writing a dissertation about federal advisory committees, for some of the insight on FDA advisory committees.
8. Alison Lawton, vice-president for regulatory affairs, Genzyme Corporation, interview, 11 June 2003.
9. A. Schmidt, "The FDA Today: Critics, Congress, and Consumerism" (Speech given at the National Press Club, Washington, D.C., 29 October 1974), quoted in H. Grabowski, Drug Regulation and Innovation (Washington: AEI Press, 1976), 76.
10. In their book on FDA "founder" Harvey Wiley, Hugh Coppin and Jack High (The Politics of Purity [Ann Arbor: University of Michigan Press, 1999]) seem to equate reputation enhancement with selfish, inefficient behavior, and Hilts (Protecting America’s Health, p. 346) disparages them for criticizing Wiley as selfish. All of these authors fall into the same trap of assuming that altruism cannot possibly have any relation to self-interested behavior. In regulation as in many other fields, reputation enhancement incentives may lead to cooperative, altruistic behavior, and more efficient outcomes.
11. Analysts such as William Niskanen have argued that bureaucracies attempt to maximize their budgets (Bureaucracy and Representative Government [Chicago: Aldine-Atherton, 1971]), while others (including myself) have argued that agencies maximize their discretion or autonomy. Although agencies such as the FDA will usually prefer more budget to less, they will generally value reputation over resources. Among the reasons for this are that (1) regulators’ personal income is only weakly related to the agency’s budget; and (2) budget increases can increase the workload or task diversity of agencies in a way that leaves them "worse off"See J.Q. Wilson, Bureaucracy: What Government Departments Do and Why They Do It (New York: Basic Books, 1989), 118–119, 180–181. In addition, statistical tests of the budget maximization hypothesis have not supported Niskanen’s theory. See D.P. Carpenter, "Adaptive Signal Processing, Hierarchy, and Budgetary Control in Federal Regulation," American Political Science Review (June 1996): 283–302.
12. Memorandum from Paul Leber, director, Division of Neuropharmacological Drugs, to Robert Temple, director, Office of New Drug Evaluation I, Subject: "NDA 20-658, Requip [ropinerole HCl tablets]," 6–7, NDA Public File 20-658, FDA Center for Drug Evaluation and Research.
13. R. Widmark, "Memo regarding Hepatotoxicity of Bromfenac," undated [December 1995], 3, NDA File 20-535, FDA CDER. See also "Wyeth-Ayerst Duract Hepatotoxicity Warning Was Suggested during NDA Review," Pharmaceutical Approvals Monthly, F-D-C Reports (April 1998): 34.
14. This is not necessarily the case. If the hazards of drug products are easily discerned and newsworthy, then the assumption is safe. But solid knowledge about product hazards often emerges only many years after market entry and is disseminated more in academic discussions than in the popular news media (the thousands of lost lives attributable to malprescription of arrhythmia drugs such as Tambocor and Encaid). See Hilts, Protecting America’s Health, 231–232; and C.F.L Heimann, Acceptable Risks: Politics, Policy, and Risky Technologies (Ann Arbor: University of Michigan Press, 1997).
15. See D.P. Carpenter, "Why Do Bureaucrats Delay? Lessons from a Stochastic Optimal Stopping Model," in Policy, Politics, and Organizations: Scientific Approaches to Bureaucratic Politics, ed. G. Krause and K. Meier (Ann Arbor: University of Michigan Press, forthcoming).
16. See D.P. Carpenter, "Protection without Capture: Product Approval by a Politically Responsive, Learning Regulator," Robert Wood Johnson Foundation Scholars in Health Policy Working Paper no. 13 (Princeton, N.J.: RWJF, 2000).
17. F.R. Baumgartner and B.D. Jones, Agendas and Instability in American Politics (Chicago: University of Chicago Press, 1993); and J. Walker, The Mobilization of Interest Groups in America (Ann Arbor: University of Michigan Press, 1991).
18. As far as one can discern from the data (including Jack Walker’s careful 1991 study), the vast majority of these groups neither are for-profit groups nor are funded by the pharmaceutical industry; indeed, most groups diligently avoid the label of "front group" for drug companies. In their analysis of Walker’s data, Baumgartner and Jones find that just 26 percent of groups in the health policy field were "for-profit," 43 percent were nonprofit, and another 31 percent were citizens’ groups. Only foreign affairs, education, and civil rights had lower ratios of profit-to-citizen involvement. Baumgartner and Jones, Agendas and Instability in American Politics, 183.
19. Hilts, Protecting America’s Health, 246.
20. Agency for Healthcare Research and Quality, Statistics from the HCUP-3 Nationwide Inpatient Sample for 1997: Principal Diagnoses (Rockville, Md.: AHRQ, 1998).
21. The Washington Post and the Vanderbilt TV News database both have electronically searchable archives. For the methodology used to aggregate stories, see D.P. Carpenter, "Groups, the Media, Agency Waiting Costs, and FDA Drug Approval," American Journal of Political Science (July 2002): 490–505.
22. I thank an anonymous reviewer for suggesting this puzzle.
23. A. Feuerstein, "AstraZeneca Scores Comeback Victory on Iressa," www.thestreet.com/_yahoo/tech/adamfeuerstein/10044113.html (12 November 2003). I gathered similar impressions from an interview with Philip Crooker, regulatory affairs director, AstraZeneca, 19 May 2003.
24. R.A. Merrill, "Modernizing the FDA: An Incremental Revolution," Health Affairs (Mar/Apr 1999): 96–111.
25. One drawback of the estimations here is that we "observe" only those drugs that have been submitted to the FDA. Strategic firms are likely to anticipate likely regulatory outcomes, and so the sample is highly selected. I am trying to address this problem in ongoing theoretical and empirical research. See D.P. Carpenter and M.M. Ting, "Product Approval with Endogenous Submissions" (Unpublished manuscript, Harvard University, 2003).
26. Kenneth Edmunds, director of electronic submissions, CDER Information Technology Group, interview, May 2002.
27. H.I. Miller, To America’s Health: A Proposal to Reform the Food and Drug Administration (Stanford, Calif.: Hoover Institution Press, 2000).
28. Consider the possibility that when third-party reviewers commit a Type I error, private or corporate parties could bring suit against them. Whereas Type I errors are more observable than Type II errors now, they would undoubtedly become much more observable under a tort system because the financial payoff to revealing such errors would rise. Hence, privatized review might well suffer from a greater procedural conservatism than FDA review. Although well-functioning insurance markets could smooth this risk, privatized reviewers would still face strong incentives to limit Type I errors.

                      What's this?

This article has been cited by other articles:

				W. M. Sage Legislating Delivery System Reform: A 30,000-Foot View Of The 800-Pound Gorilla Health Aff., November 1, 2007; 26(6): 1553 - 1556. [Abstract] [Full Text] [PDF]

				K. A. Phillips and S. L. Van Bebber Regulatory Perspectives on Pharmacogenomics: A Review of the Literature on Key Issues Faced by the United States Food and Drug Administration. Med Care Res Rev, June 1, 2006; 63(3): 301 - 326. [Abstract] [PDF]

				R. Lee FDA and Drug Safety: New Tufts Study Challenges Critics of the Prescription Drug User Fee Act J. Law Med. Ethics, March 1, 2006; 34(1): 131 - 134. [PDF]

				S. Keyhani, M. Diener-West, and N. Powe Are development times for pharmaceuticals increasing or decreasing? Health Aff., March 1, 2006; 25(2): 461 - 468. [Abstract] [Full Text] [PDF]

				A. C. Gelijns, L. D. Brown, C. Magnell, E. Ronchi, and A. J. Moskowitz Evidence, Politics, And Technological Change Health Aff., January 1, 2005; 24(1): 29 - 40. [Abstract] [Full Text] [PDF]

				F. A. Sloan, S. Berman, S. Rosenbaum, R. A. Chalk, and R. B. Giffin The Fragility of the U.S. Vaccine Supply N. Engl. J. Med., December 2, 2004; 351(23): 2443 - 2447. [Full Text] [PDF]

				P. B. Fontanarosa, D. Rennie, and C. D. DeAngelis Postmarketing Surveillance--Lack of Vigilance, Lack of Trust JAMA, December 1, 2004; 292(21): 2647 - 2650. [Full Text] [PDF]