Do off-label drug practices argue against fda efficacy requirements? a critical analysis of physicians' argumentation for initial efficacy requirements

FEATURED ARTICLES
Do Off-Label Drug Practices Argue Against
FDA Efficacy Requirements?
A Critical Analysis of Physicians’ Argumentation
for Initial Efficacy Requirements
By DANIEL B. KLEIN and ALEXANDER TABARROK* The amended Food, Drug and Cosmetics Act requires efficacy certification for a drug’s initial uses (“on-label”), but does notrequire certification before physicians may prescribe the drug forsubsequent uses (“off-label”). Does it make sense to require FDAefficacy certification for new drugs but not for new uses of old drugs?Using a sequential online survey, we carried on a “virtual conversa-tion” with some 500 physicians. The survey asked whether efficacyrequirements should be imposed on off-label uses; almost all physi-cians said no. It asked whether the efficacy requirements for initialuses should be dropped, and most physicians said no. We then askedrespondents whether opposing efficacy requirements in one case butnot the other involved an inconsistency. In response, we receivedhundreds of written commentaries. We organize and discuss thesecommentaries with an eye to understanding how the medical marketcertifies off-label drug uses and how this compares to FDA certifica-tion. Does off-label medicine use suggest that efficacy requirementsshould be placed on new uses of old drugs? Does it suggest thatefficacy requirements on new drugs should be lifted? We explore these *Daniel Klein is at the Department of Economics, George Mason University, Fairfax, VA, 22030, and Ratio Institute, Stockholm, Sweden; e-mail: AlexanderTabarrok is at Department of Economics—MSN 1D3, George Mason University, Fairfax,VA, 22030; e-mail: The authors thank Dr. Robert Klein for helpingin formulating the survey, Robert Higgs, Tyler Cowen, Mats Ekelund, and seminarparticipants at George Mason University, University of Uppsala, and the MedicalProducts Agency (Uppsala, Sweden) for valuable comments, and Santa Clara Universityfor a grant to conduct the survey.
American Journal of Economics and Sociology, Vol. 67, No. 5 (November, 2008).
2008 American Journal of Economics and Sociology, Inc.
The American Journal of Economics and Sociology questions, and ask whether the response of many of the doctorsexhibits the familiar behavior bias toward the status quo.
Introduction
THE FOOD, DRUG AND COSMETICS ACT of 1938 with amendments in 1962forbids new drugs from being sold unless they have passed FDA-approved tests for safety and efficacy in a specified use, called the“on-label” use. Physicians are allowed, however, to prescribe anFDA-approved drug not only for its on-label use but also for other,“off-label,” uses. It often happens that physicians and researchersdiscover new uses for a drug after it has been permitted, so off-labeluse is quite common.
Amoxicillin, for example, has an on-label use for treating respiratory tract infections and an off-label use for treating stomach ulcers. For theon-label treatment of respiratory tract infections, amoxicillin has beentested and certified in all three phases of the FDA’s InvestigationalNew Drug clinical study; phase I trials for basic safety, and phase IIand phase III trial for efficacy. For the treatment of stomach ulcers,however, amoxicillin has not gone through FDA phase II and phase IIItrials and thus is not FDA certified for this use. Amoxicillin will nevergo through FDA efficacy trials for the treatment of stomach ulcersbecause the basic formulation is no longer under patent. Yet anytextbook or medical guide discussing stomach ulcers will mentionamoxicillin as a potential treatment, and today a physician who didnot consider prescribing amoxicillin or other antibiotics for the treat-ment of stomach ulcers would be considered negligent.
Off-label prescribing is very common in all areas of medicine. It is not uncommon for a drug to be prescribed more often off-label thanon-label. Thalidomide has been approved for use in treating leprosybut is much more commonly used to treat multiple myeloma andAIDS. Most cancer and AIDS patients are given drugs that are not FDAcertified for the prescribed use (GAO 1991; Brosgart et al. 1996). In alarge number of fields, a majority of patients are prescribed at leastone drug off-label (Tabarrok 2000: 26).
Off-Label Drug Practices vs. FDA Efficacy Requirements To explore the policy lessons of off-label usage, we tapped the knowledge and judgment of actual practitioners. We asked physicianswhether the FDA should hold drug uses to efficacy requirements, bothas the question applies to initial (on-label) uses and as it applies tosubsequent (off-label) uses. Virtually all opposed imposing efficacyrequirements on subsequent uses. But the majority of respondentssupported the FDA efficacy requirements on initial uses.
Is it inconsistent to favor efficacy requirements for new drugs but not for new uses of old drugs? We asked the physicians to justify theirresponses in an open-ended format, and received hundreds of justi-fications of their opinions about FDA policy. The key feature of ourstudy is asking physicians the justification question.
The Consistency Argument: If Off-Label Uses Should Not Require FDA
Efficacy Certification, Why Should On-Label Uses Require FDA
Efficacy Certification?
AGAIN, WHEN THE FDA EVALUATES a new drug, the evaluation of safety andefficacy is made with respect to a specified use. Once a drug has beenpermitted, physicians often come to prescribe the drug for other uses.
But there seems to be a logical inconsistency in allowing off-label uses and requiring proof of efficacy for the drug’s initial use. Logicalconsistency would seem to require that one either (1) be in favor of allowing physicians to prescribe off-label and allowing physicians to prescribe (and pharmaceutical compa-nies to make and sell) new drugs that have not been FDAefficacy certified, (2) be against allowing physicians to prescribe off-label and allowing physicians to prescribe (and pharmaceutical companies to makeand sell) new drugs that have not been FDA efficacy certified.
But even if logic dictates as such, it would not tell us whether to favor(1) or to favor (2). Tabarrok (2000) argues that off-label usage pro-vides a natural experiment. In a sense, off-label uses are regulated The American Journal of Economics and Sociology according to the pre-1962 rules, under which the FDA held new drugsonly to safety requirements, whereas on-label uses are regulatedaccording institutions—in the same country at the same time—are operatingunder dual systems of drug regulation. Off-label prescribing gives usan idea of how medical affairs would proceed in a world in whichnew drugs were allowed until banned, rather than banned untilpermitted. Since physicians appear to support off-label prescribing,Tabarrok argues in favor of consistent option (1).
Shapiro (1979) also recognizes the inconsistency in current regula- tion, but draws a lesson opposite from Tabarrok. Shapiro calls thefreedom to prescribe off-label “a regulatory anomaly which deprivessome drug consumers of the protection of the [Food, Drug andCosmetic] Act” (1979: 801). He argues for tightening restrictions onoff-label prescribing. Such a position has also been taken by the FDA,1although since 1982 the FDA has focused its efforts on limitingpharmaceutical manufacturers rather than physicians.2 The principleof consistency gives rise to dual arguments for reform, as shown inFigure 1.
Our survey drew physicians into this debate. One physician responded by quoting Emerson: “A foolish consistency is the hob-goblin of little minds.” Like more than 150 responding physicians,he gave his reasons for deeming the consistency argument forliberalization flawed or imperfect.
The Survey and the Main Quantitative Results
We drafted the questionnaire so as to pose the two main policyquestions and then the consistency argument. We hired HostedWare.com to host the survey online. The questions were presentedsequentially: each important question appeared on its own webimage, and the respondent had to provide his or her answer beforeviewing the next question. The survey limited responses to one percomputer. To get physicians to access and complete the survey, we Off-Label Drug Practices vs. FDA Efficacy Requirements hired Medical Marketing Services to send an e-mail message to 8,000physicians.3 The broadcast message invited the physician to aidacademic research on pharmaceutical regulation by accessing andresponding to the brief questionnaire at the URL provided. Themessage asked the recipient not to share the URL with others. Weinstructed Medical Marketing Services to randomly select physiciansin certain fields, including allergy/immunology, cardiology, endocri-nology, neurology, oncology, urology, internal medicine, geriatrics,and pediatrics. The e-mail broadcast yielded 504 physicians whoanswered at least one question and 492 who completed the surveyby answering at least one of the main policy questions (a responserate of about 6 percent).
When a responding physician clicked on the indicated URL, he or she came to a simple web image titled “Opinion Survey on Pharma-ceutical Regulation” and a brief welcoming message about responsesbeing anonymous and used only for purposes of academic research.
The respondent clicked “Begin Survey,” which led directly to sevenpreliminary questions about the respondent’s practice.
The American Journal of Economics and Sociology The seven preliminary questions are provided here (the numbering 1through 7 did not appear in the survey). The response-rate percent-ages or other summary information are indicated, as is the absolutenumber of respondents in square brackets.
1) What state do you practice in?
47 states in total were represented, with the five largest being: 2) How many years have you been in practice?
The distribution was described by the following: 3) What are your areas of clinical specialization?
After collapsing multiple responses into single responses (e.g.,hematology/oncology is listed under oncology) and subsuming pedi-atric [blank] into pediatrics (i.e., pediatric oncology and pediatricallergy are both listed under pediatrics), the top five categories were: Off-Label Drug Practices vs. FDA Efficacy Requirements 4) Are you employed at or affiliated with a teaching hospital?
physicians
principally
tioners and some are also involved in doing and publishing
medical research (some also teach, but let’s put that aside).
Of the following choices, how would you describe your
career?
a. Strictly practitioner, not a researcher b. Mainly a practitioner, only limited involvement in c. About half practitioner, half researcher d. Mainly a researcher (with, of course, some practice 6) When the FDA approves a drug, it does so for a certain
specified use. Often the drug is later found to have other uses,
known as off-label uses, for which physicians may also pre-
scribe the drug.
How often do you prescribe drugs for off-label indications?4
The American Journal of Economics and Sociology 7) In your medical practice, do you treat children?
b. Rarely
c. Sometimes
e. Always
Our survey was not intended to be a random sample of all physicians,nor was that necessary for our purposes. Instead, we wanted toreceive responses from a diverse group of physicians.
Next, the respondent encountered the two main questions. One askedabout the imposing of efficacy requirements on off-label uses.
What would be your position on a proposal to change FDA law
so that physicians could not prescribe drugs for off-label uses?
Would you favor or oppose such a change?
Don’t know/Not sure
Of 492 physicians answering the question, 460 opposed ending the freedom to prescribe off-label.5 Even though the respondent yieldfrom the broadcast e-mail message was low and the sample not fullyrandom across all physicians, we are nonetheless reasonably comfort-able in concluding that nearly all physicians favor being allowed toprescribe off-label. In addition to our survey, physician groups such asthe American Medical Association have long opposed restrictions onoff-label prescribing. Thus, George Lundberg, then the editor of theJournal of the American Medical Association, testified in Congress: Prescribing FDA-approved drugs for off-label (unlabeled) uses often isnecessary for optimal patient care. For a product to have the most effectivepotential benefits, law and regulation should and must follow, not precede,science. There are too many variations in clinical circumstances and toomuch time delay in regulations to allow the government to impede thephysician’s ability to practice in these regards when it is medicallyappropriate.6 Off-Label Drug Practices vs. FDA Efficacy Requirements In addition, several respondents volunteered strongly worded objections to the idea of banning off-label prescribing. Such a reformwould be “clearly naïve,” “stupid and unethical,” “dangerous,”“disastrous,” and “medicine would grind to a halt.”7 Given the strongsupport for off-label prescribing, we can predict that physicians willnot accept Shapiro’s consistency argument for further FDA control. Itdoes not follow, however, that they will accept the economist’sconsistency argument for liberalization.
The other main question asked about dropping the efficacy require- Under current law, when the FDA reviews an application for a
new drug, it holds the drug to both safety and efficacy require-
ments before permitting the drug.
What would be your position on a proposal to change FDA law
so that physicians could prescribe a new drug once the current
FDA safety requirements had been met?8 Under this system,
manufacturers and researchers could continue with efficacy
certification (from the FDA or some other institution) if they so
choose, but physicians would not be prevented from prescrib-
ing drugs that did not have efficacy certification from the FDA.
In brief, what would be your position on a proposal to make the
FDA efficacy standards an optional form of certification, rather
than a requirement as at present?
Don’t know/Not sure
Given how little the average American questions the FDA, it may be surprising that 42 percent of the respondents were not decidedly infavor of retaining initial efficacy requirements, and 27 percent favoredeliminating FDA efficacy requirements. The sample is small and notnecessarily representative (albeit not biased with respect to our ques-tions in any obvious way), but the numbers are consistent with previousstudies. The Competitive Enterprise Institute in Washington, D.C. has The American Journal of Economics and Sociology posed a very similar question in six nationally representative telephonesurveys of physicians conducted by the Polling Company. After askingseveral questions that do bring out the costs of drug restrictions, CEIconsistently has found that a majority of physician respondents“strongly favor” or “somewhat favor” making unapproved drugs anddevices “available to physicians as long as they carry a warning abouttheir unapproved status,” while a minority answer “somewhatopposed” or “strongly opposed.”9 Our results and those of CEI indicatethat there is not a strong consensus among physicians about thedesirability of initial efficacy requirements. Here, about 25 percent ofphysicians are in line with the consistency argument—they favor beingable to prescribe off-label drugs and drugs lacking initial FDA efficacycertification. The majority, however, gave “inconsistent” responses.
The majority who gave “inconsistent” responses next encountered thefollowing statement and question.10 I noticed that you answered in favor of physicians being
allowed to prescribe off-label but against physicians being
allowed to prescribe new drugs that had met FDA safety require-
ments but not FDA efficacy requirements.
Because off-label indications have not been FDA-certified for
efficacy, some people argue that off-label prescribing is equiva-
lent to prescribing a new drug that has been FDA safety-certified
but not FDA efficacy-certified. According to this argument, to be
consistent, one should either be in favor of allowing physicians
to prescribe off-label *and* allowing physicians to prescribe
new drugs that have not been FDA efficacy-certified, or against
both kinds of allowances.
How do the following choices best reflect your thoughts on this
argument?
It’s an interesting argument but I would need more time to
think about it before responding to it.
The argument makes me less inclined to support off-label pre-
scribing.
Off-Label Drug Practices vs. FDA Efficacy Requirements The argument makes me more inclined to support allowing
physicians to prescribe new drugs that have not been efficacy-
certified by the FDA.
I think the argument is invalid. Letting doctors prescribe off-
label differs from the proposed reform of letting them pre-
scribe new drugs that have not been efficacy-certified by
the
because:
This presentation of the consistency argument led 12 percent to reconsider their views—with almost twice as many revising in favor ofliberalization as opposed to expanding restrictions—but the majorityof respondents were unmoved by the argument. This is unsurprising,as few people quickly change their minds upon encountering anargument (especially in an impersonal web survey).
An assessment of the consistency argument was also solicited from the physicians who gave “consistent” responses.11 I noticed that you answered in favor of physicians being
allowed to prescribe off-label and in favor of allowing physi-
cians to prescribe new drugs that met FDA safety requirements
but not FDA efficacy requirements.
Preliminary results from the survey indicate that many other
physicians are in favor of off-label prescribing but are against
loosening FDA requirements. Since your response differs, we
would like to explore this in a little more detail.
In particular, we are interested in your evaluation of the follow-
ing argument:
Because off-label indications have not been FDA-certified for
efficacy, off-label prescribing is very much like prescribing a
new drug that has met FDA safety but not efficacy requirements.
Therefore, one should either be in favor of doctors being
allowed to prescribe off-label *and* being allowed to prescribe
new drugs that have met safety but not FDA efficacy require-
ments, or against both allowances.
The American Journal of Economics and Sociology How do the following choices best reflect your thoughts on this
argument?
I think the consistency argument makes a lot of sense; it agrees
with the reasons behind my responses. [Use other box for
further response.]
I think there’s merit to the argument, but other considera-
tions explain my responses. [Use other box for further
response.]
It is for other reasons that I have favored allowance in both of
my replies; the consistency argument is faulty because [Use
other box for further response.]
We saw that most “inconsistent” respondents rejected the consis- tency argument. Here we see that most “consistent” respondentsaccepted it. Indeed, 95 percent saw merit in the argument. However,one ought not count this as a clear endorsement of the consistencyargument because it is likely that a respondent is disposed to sign onto a justification of responses he or she just made.
Finally, all respondents came to a page inviting them to share any “thoughts or ideas about the questions in the survey,” againin an open-ended format. This provided yet another stream offeedback.
In the Appendix we discuss two correlations—support for FDA liberalization increases markedly among practitioners as opposed toresearchers, and support for liberalization increases with reportedrates of off-label prescribing. Now we turn to the written commentson the consistency argument.
The “Virtual Conversations”
OF THE 205 “INCONSISTENT” PHYSICIANS who explicitly deemed the con-sistency argument invalid, 176 wrote something in the “because” box.
We read, organized, and posted the complete set of comments online,along with final comments and other remarks from respondents.12 Wehave organized the comments into a series of three challenges to the Off-Label Drug Practices vs. FDA Efficacy Requirements consistency argument. We set out the idea as the respondents them-selves would approve; that is, we represent their idea fully andfaithfully to our best capability. We then discuss the arguments,drawing on the literature on FDA reform as well as material from otherphysician respondents.
The pharmacological mechanisms of off-label uses are closely
related to those of the on-label uses.
The most common challenge to the consistency argument involves the idea of related pharmacological mechanisms. The simplest case isout-of-age prescribing: Off-label use can mean using a drug under FDA age limits—for example,Zyrtec in a 1-year-old. (g57) Many off-label uses in my case are in children younger than the approvedages. The efficacy has been tested and proven for the given use, just notin these age groups. (g166) Indeed, 80 percent to 90 percent of pediatric patient regimens involve at least one off-label prescription (Jaffe 1994; Kauffman 1996).
In making the relatedness argument, 22 physicians specifically citedage classes or pediatrics. But many others presented the argument inmore general terms, speaking of related mechanisms, similarities ofdrugs within a given class, proven activeness of the drug, prescribing“by analogy,” and “extrapolating” from on-label to off-label. Onephysician illustrated the argument thusly: Some of the newer antihistamines were initially only indicated for thetreatment of seasonal allergic rhinitis, but not for perennial allergic rhinitis.
Well, there is no difference in the allergic cascade and mechanismof seasonal and perennial allergic rhinitis and their response toantihistamines.
Consequently, most allergists prescribed them for both forms of rhinitisbefore the FDA published its official approval of indications. (g93) The force of the consistency argument for liberalization comes from the premise that off-label prescribing is like prescribing a new drugthat has not been efficacy-certified by the FDA. The relatednessargument challenges this premise by arguing that off-label prescribing The American Journal of Economics and Sociology is not that different from on-label prescribing. In this view, prescribinga drug that has not been FDA efficacy-certified is like searching in thedark, while prescribing off-label is like searching in the dusky lightcast by the nearby lamp of the on-label use.
DiscussionIs it true that most off-label prescriptions are closely related to theon-label prescription? It’s difficult to say because observers can dis-agree about whether two treatments are “closely related” (see furtherbelow). But there is agreement, however, that many off-label uses arenot closely related to the on-label use. Thalidomide is used on-labelfor the treatment of leprosy but, as one physician wrote, “we found byserendipity that it was effective in myeloma and supported by the peerreview literature” (g134). Today some 99 percent of Thalidomideprescriptions are for off-label uses quite distinct from leprosy (Youngand Adams 2003). The role of serendipity in discovering off-label usestestifies to the unrelatedness of such uses. Minoxidil, for example, wasdeveloped as a drug for the treatment of hypertension, but after usersreported unusual hair growth, it later became much more widely usedoff-label and then under the brand name Rogaine as a treatment forbaldness.
We frequently find uses for drugs that the FDA has not realized yet. A goodexample is the use of verapamil for treatment of headaches. This wasinitially (and still is) primarily a cardiovascular drug, however patientsstarted reporting that their headaches had improved or gone away whileon this drug, so it was a simple step for physicians to begin trying this drugfor a different indication. I don’t know that the FDA has ever approved thisdrug for headaches [It has not.—Au.], but we use it, and it works. (g14) The unrelatedness of much off-label prescribing is acknowledged by both those who defend and those who oppose or defend off-label.
Consider, for example, the following defense written by Tabarrok:13 Decades ago, quinacrine was approved for malaria, and chlorpromazinefor schizophrenia. Both drugs have recently been found to be potentialtreatments for Bovine-Creutzfeld-Jakob disease (BJD), commonly knownas mad-cow disease. Because these drugs were already approved foranother use, BJD patients could begin taking them within months of the Off-Label Drug Practices vs. FDA Efficacy Requirements publication of scientific papers suggesting their effectiveness. If these drugshad been new they could not have been marketed until completion of FDAapproved clinical trials—a process that could have taken a decade or more.
Indeed, BJD is rare, so the cost of this process would almost certainly havekept any company from funding the necessary trials. The drugs could beprescribed as soon as physicians and patients evaluated the risk–returntradeoff favorably only because they had been permitted for other uses.
With respect to treating BJD, quinacrine and chlorpromazine were essen-tially new drugs. Why should other patients, not so lucky as to be in needof an old drug with new uses, not have access to new drugs on the sameterms? Now consider the following story from a Knight-Ridder investigative For the last three and a half months she was pregnant, Tammie Snyder hada small medical device strapped to her thigh. It pumped a drug calledterbutaline through her body to prevent her from going into labor toosoon. On Sept. 17, 2002, Snyder gave birth to two healthy girls. Withindays, however, her lungs filled with fluid, her heart began to fail and shewas told she might need a heart transplant. She recovered, but she’s beentold she can never have a baby again. Her heart wouldn’t stand the strain.
Terbutaline is an asthma drug, and the Food and Drug Administrationhasn’t approved its use to prevent premature labor. The FDA has warneddoctors that the treatment is “potentially dangerous” and may not beeffective. Snyder said her doctor never told her about the warning or thatthe FDA had approved terbutaline only to treat asthma. (Young and Adams2003) The first story focuses on the benefits of off-label prescribing and thesecond on the costs, but both stories acknowledge that off-labelprescribing can depart considerably from FDA-approved uses. Simi-larly, one physician wrote: “Most of the drugs I use for diseases suchas lupus, AS, Reiters, Behcet’s, vasculitis, etc etc etc are off-label”(g104). Surely, this doctor’s therapeutic arsenal is not based chiefly onsure-thing extrapolation from on-label indications. Throughout theresponses, physicians provided many examples—antileukotrianes,verapamil, Amiodarone, elavil, plaguenil, cyclobenzaprene, DepakoteER (g107, g14, g20, g58, f157, c11)—in which important off-label uses,though perhaps related to were not direct or certain extrapolationsfrom the on-label uses.
The relatedness challenge fails as a positive statement of how off-label prescribing works—in any reasonable metric, much off-label The American Journal of Economics and Sociology prescribing is unrelated to on-label prescribing.14 Although the relat-edness challenge fails as a positive statement of how off-label pre-scribing works, perhaps it can be understood as a statement abouthow off-label prescribing should work. Physicians who invoke therelatedness argument might agree that if an efficient way were foundfor an official body to deem off-label uses either “related” or “notrelated” to the on-label uses, then it would be desirable to have newpolicy rules that freely allowed related uses but imposed new efficacyrequirements on “not related” uses. That is the implication of therelatedness argument, but, once put this way, would the physiciansreally embrace it? Later, we examine in more detail whether theconsistency argument might better be resolved by favoring moreFDA control, contra the physicians’ stated preferences.
B. The Cognitive Saturation Challenge Dropping efficacy requirements would flood the market with
ineffective drugs; pharmaceutical companies would promote
ineffective drugs and push them on patients and doctors,
saturating their cognitive abilities to sort out good indications
from bad.
These physicians are concerned that dropping the initial require- ments would bring a “flooding” of the market with ineffective drugs.15“[T]hat’s what the makers of Aspercreme and Icy Hot are for!” (g87).
Several made reference to the “chaos” of dietary supplements orherbal remedies (e.g., g29, g69, g73). The job of medical science andpractice is to sort out good from bad, but the cognitive resources ofthe medical nexus are limited, of course. The greater the number ofnew drugs, and the greater the proportion of ineffective drugs, thethinner will these cognitive resources be spread. A rapid influx ofnew products could lead to treatment practices that are based on ashallower base of clinical experience, which would increase the errorrate and impair the collective learning process.
DiscussionThe physicians’ arguments are logical explanations for why theyoppose the dropping of efficacy requirements, but they do not addressthe heart of the consistency argument. Fear of flooding of the market Off-Label Drug Practices vs. FDA Efficacy Requirements and consequent cognitive saturation appears to be inconsistent withthe very extensive support that all physicians gave for off-label pre-scribing. Today, all the drugs that have been permitted are availableon the market and collectively constitute an ocean of potential off-label treatments for every possible ailment. Yet doctors do not ran-domly dip into this expanse and prescribe drugs without evidence ofeffectiveness. Thus, the possibility of many ineffective drugs beingavailable does not really work as a challenge to the consistencyargument because the consistency argument carries the implication:So, then, why not prohibit off-label uses? Some respondents, however, spoke of a flood of new drugs that [I]t is now commonplace for drug companies to directly market to thepublic which could bring unwanted patient pressure to bear on the MD toprescribe for the use not tested for efficacy. (g100) Many physicians prescribe drugs based on the “flashiest ads” and detailrepresentatives. (g109) [P]hysicians sometimes give in to patient requests for medications eventhough they may not think that the drug is effective. (g119) [P]hysicians and consumers alike often enjoy trying the newest, “best” thingon the market; this could allow a significant amount of prescribing ofpresumably safe pharmaceuticals with questionable benefit. (g148) Just being safe is deceptive to consumers (patients) and allows pharma-ceutical representatives, from whom most physicians seem to get most oftheir information, to twist information in all kinds of ways. (g124) Given that 40 percent of physicians are willing to prescribe whatever thepatient asks for, the result would be a mess. (f64) If medicines were approved without proof of efficacy, this could lead toworsening of the current problems brought on by overaggressive adver-tising without evidence. (g59) Physicians would be . . . subjected to barrages of claims from drug com-panies and would have a lot of difficulty evaluating them for accuracy.
Drug companies are notorious for misrepresenting their products. (g14) [The efficacy requirements prevent] the chaos that now exists with alter-native medicine “Natural Herbal Medications” which make unsubstantiatedclaims as to their potential benefit to the consumer. (g69) The American Journal of Economics and Sociology The reform proposal to make efficacy requirements optional, as put in the survey question, did not specify one way or the other howissues of drug promotion would be handled. The respondents pre-sumed that, under the reform, drug companies would enjoy the samepromotion privileges that a company today enjoys in promoting theon-label uses of an FDA-permitted drug. But it would actually be morein keeping with the consistency argument to suggest that drug com-panies that did not get FDA efficacy certification for an indicationwould only be allowed the same (limited) promotional freedoms thatthey enjoy today for off-label uses. With this provision, the floodingor cognitive saturation challenge does not answer the consistencyargument.
With similar restrictions on promotion, off-label prescribing and allowing sale of new drugs with FDA-approved safety but not efficacytrials appear to be similar. Even with this proviso, however, it seemsunlikely that the physicians quoted above would favor droppingefficacy requirements.16 If so, what these physician comments dosuggest is that they think the inconsistency of the status quo ought tobe resolved in favor of greater regulation of off-label prescribing. Wetake up the issue of how to resolve the consistency argument atgreater length below.
Efficacy requirements generate knowledge but because of dif-
ferential incentives arising from the temporal limit on patent
protection, efficacy requirements suppress fewer drugs when
placed on initial uses than they would if placed on subsequent
uses. This difference recommends opposite policies in the two
cases.
Many doctors responded to the consistency argument by saying, absent initial efficacy requirements, “companies would not have incen-tives to provide efficacy studies” (g90). (Physicians also pointed out thatFDA efficacy studies increased knowledge of safety; thus, removingefficacy requirements would diminish safety knowledge.) But: there are many instances where the market for a new indication for an old,off-patent drug is too small for a drug company to have any incentive to Off-Label Drug Practices vs. FDA Efficacy Requirements fund an FDA approval process. Would manufacturers of generic drugs haveany economic reason to fund such an approval? In many cases, the answerwould be no. (g95)17 FDA efficacy requirements have two effects. The good effect, the “knowledge effect,” is that the requirements induce the pharmaceuti-cal company to fund the requisite studies and thereby enhanceknowledge beyond the level otherwise attained. Better knowledgemay mean more information about specific drugs but also includes theeradication or avoidance of spurious and useless ideas; it alleviatescognitive saturation. The bad effect comes from the increases in costs,delays, and uncertainties in developing and getting the FDA’s permis-sion.18 The costs of gaining FDA approval will in some cases discour-age industry, science, and medicine from bringing forth the drugor indication. This is the “suppression effect.” Efficacy requirementsincrease the knowledge about the indications that do become avail-able, but suppress their number.
The balance of the knowledge effect and the suppression effect, these physicians say, favors initial efficacy requirements because thesuppression effect is not so large, since the company will begin sellingthe drug while the patent is young, and the knowledge effect is large,since apart from clinical testing of the new drug there would be littleexperience with it. But for subsequent uses, the balance opposesefficacy requirements because the suppression effect is large, since thepatent is old, and the knowledge effect is not so large, since medicineis learning from the drug’s initial indications.
DiscussionA pharmaceutical manufacturer’s willingness to pay for putting adrug through the FDA process depends on the market exclusivityafforded by a patent. If a drug is not under patent, then other firmsthat did not incur costs of drug development will compete and driveprices below the profit point. Since subsequent uses are discoveredafter a drug has been on the market for some time, they are dis-covered when the patent is winding down or has expired. Theincentives to fund efficacy studies for subsequent uses dwindle astime wears on—even if the subsequent uses are highly valuable tosociety as a whole.19 Thus the fact that the patent is winding down The American Journal of Economics and Sociology is a good argument in favor of not requiring FDA efficacy certifi-cation for off-label prescriptions. In this respect, the incentive chal-lenge supports the status quo but it does not per se argue againstthe dropping of efficacy requirements.
Insofar as the incentive challenge may be construed as a rebuttal to further liberalization, it should be taken as challenging the argu-ment that markets will supply assurance. The incentive challengesays patents are a necessary but not sufficient inducement for pro-ducing efficacy studies—to create sufficiency, we need FDA efficacyrequirements. Here we may quote one of the physicians whofavored liberalization by responding to this argument as follows: “Apharmaceutical company must support the efficacy of its drugs withclinical research to sell its product” (c22). The pharmaceuticalcompany could not hope for the medical community to adopt itsdrug into standard care (or, at least not for very long) without somedemonstration of its relative efficacy. It could not hope to reallyestablish the drug’s indication without serious evidence recognizedby the relevant professional and scientific communities. The dollarvalue of the patent on a superior new drug depends not merelyon legal permission to supply the drug and some measure ofexclusivity in doing so but also on demonstrating to the medicalcommunity that it is superior.
Moreover, physicians pointed out that medical research is per- formed and paid for by many parties other than pharmaceuticalcompanies—universities, large medical organizations such as HMOs,joint ventures among hospital groups, research nonprofits, govern-ment organizations such as the NIH, and others. Physicians mayunderestimate the sagacity of their own profession. Economist J.
Howard Beales (1996) found that off-label uses that later came to berecognized by the FDA appeared in the U.S. Pharmacopoeia onaverage 2.5 years before FDA recognition. That the U.S. Pharmaco-poeia recognizes off-label indications years ahead of the FDA dem-onstrates that physicians and scientists have certified thousands ofdrug indications quite independently of the FDA, even when thoseindications are not very closely related to the original indications. Hereare some physicians’ remarks on the wider forms of recognition andcertification used in medicine.
Off-Label Drug Practices vs. FDA Efficacy Requirements Often efficacy information is already available from studies done outsidethe USA. (g47) There is often data from Europe or in peer review journals. FDA efficacytrials are important, but they are not the only measure (except legally interms of company marketing) of a product’s efficacy for a certaincondition. (g28) Off-label use is very often based on valid smaller studies concerning otherthan the index medical condition; those studies may not be large enoughor the pharmaceutical company may not want to spend the $ it takes to getFDA approval. (g44) FDA approval on efficacy lags behind peer-reviewed data that may suggestefficacy. I favor off-label use only if there is reasonable data, or reasonableinference, of efficacy . . . (g50) Almost all cancer chemotherapy is off-label. There is no way 2 or 3 drugcompanies can expend the effort to get a combination regimen approved.
Oncologists use the peer-reviewed literature to decide therapy. Almostalways decisions are based on randomized clinical trials. (g53) Plaquenil was developed and FDA-approved as a malarial drug. Later itwas found to relieve Rheumatoid arthritis symptoms in the patients takingit for malaria. Studies show that it worked and was efficacious but shouldwe wait for the FDA to prolong the relief of pain and suffering for severalyears while the necessary drug company/FDA studies are done or just usecommon sense? Often there is no financial incentive for a drug companyto pursue off-label indications for conditions that wouldn’t generate suffi-cient income to offset the cost of FDA approved trials. But university-based, double-blind, highly powered studies show benefits that outweighrisks. (g58) Most of the drugs that I deal with are only approved for one form ofcancer. They are then put through trials in other diseases and these arerecorded in the literature. Those that show efficacy are then NON-FDAapproved but COMPENDIUM approved and are paid for by insurance.
(g64) The off-label experience testifies to the fact that much knowledge about efficacy (and about safety) is produced outside the FDA regu-latory apparatus. The natural incentives arising from economic inter-ests, the patient’s self-interest, liability risks, professional pride andesteem, scientific curiosity and competition, and basic human moralitycreate significant incentives to invest in knowledge creation (Klein The American Journal of Economics and Sociology 2002). Initial efficacy requirements, therefore, may induce less netknowledge about new drugs than one would at first suspect. To somedegree, initial efficacy requirements merely require what would bedone in any case. What matters in the final analysis, however, is howthe knowledge effect compares to the suppression effect.
The Dissidents SpeakA significant minority of physicians staked out a consistent position.
Recall that 32 percent of respondents with a definite opinion favoredthe elimination of initial efficacy requirements. And 76 percent of suchliberalizers said the consistency argument “makes a lot of sense,” plusanother 20 percent said “there’s merit” to the argument.
These dissidents from the status quo made many pro-liberalization The patients need my help and trust my judgment. If through my ownevaluation I find a use for a drug my patients need, I don’t care whatopinion of [it] the FDA has. (c6) I practiced for several years in CentroAmerica where the use of drugs iswithout any “FDA” approval and never had any problems with the newmedication, as a matter fact I remember when we first used Zythromax. (f79) You might have asked—Are there instances where you can documentpatient harm by the current process? STI571 for CML is a recent fineexample where efficacy and safety data appeared to be present for 6–9months before actual approval . . . (f140) There is a direct relationship between the physician and the patient andthis allows a more accurate choice of alternative medications to be used inthe medical treatment. The FDA is too distant to the reality of medicine thatthey need to reevaluate their procedures. (c9) The FDA must change the way drugs are currently approved. The currentprocess is too expensive, limited in scope, and of little benefit in clinicalpractice. (f145) [T]he FDA needs to get real and allow people who practice medicine to doso. (c17) Our hands are tied enough in medicine. Please don’t add more tether. (f74) Medicine is already bogged down in governmental regulation. (f63) Regulations are the bane of our practice. (f168) Off-Label Drug Practices vs. FDA Efficacy Requirements [O]ne does not want an official, politicized body like the FDA to control thepractice of medicine; scientific information should be the basis for deci-sions made by a free scientific community, not constrained by officialsanction. Not infrequently, the “official” view is wrong . . . Physicians, astrained practitioners applying the science of medicine, should have theequivalent of academic freedom. We are adequately constrained by con-siderations of liability risk and our professionalism. (c18) Although a significant minority of physicians staked out a consis- tent position, a majority strongly endorsed the status quo of off-labelprescribing but not removal of efficacy requirements. Our reading,however, is that the physicians’ arguments in favor of the status quodo not fully explain the apparent inconsistency between favoringoff-label prescribing and favoring efficacy tests on initial uses. Muchoff-label prescribing is not closely related to on-label prescribing,and dropping efficacy requirements would not substantially increasecognitive saturation beyond what occurs today because of alreadyextensive off-label prescribing. The incentive challenge does providea reason for thinking that the status quo is consistent but in and ofitself does not argue against dropping efficacy requirements. Indeed,in recognizing that the suppression effect is too large to justifyefficacy requirements for off-patent drugs, the incentive challengeraises the possibility that the suppression effect might also belarge relative to the benefits of efficacy requirements for newdrugs.
Status Quo BiasThe fact that a majority of physicians staked out a seemingly incon-sistent position in favor of the status quo is not surprising. Statusquo bias is natural and common. Samuelson and Zeckhauser (1988),for example, have demonstrated that even the suggestion of a statusquo can influence decision making. When their respondents wereasked to invest an inheritance from “your great-uncle” in a“moderate-risk company, a high-risk company, treasury bills ormunicipal bonds,” respondents made one set of choices, but whenasked to make this choice with the additional information that “[a]significant portion of this portfolio is invested in a moderate-riskcompany,” respondents made quite different choices with a strong The American Journal of Economics and Sociology bias toward the status quo. (Respondents were told: “The tax andbroker commission consequences of any change are insignificant.”) Most importantly, status quo bias has been shown to significantly reduce the efficacy of decision making in real-world settings. Beshearset al. (2006), for example, found that the status quo or default positionhas a “tremendous” effect on retirement planning. When the defaultsavings rate in one firm’s retirement plan was 3 percent of salary, morethan a quarter of the workers chose that as their savings rate, despite anemployer guarantee of a dollar-for-dollar match on contributions of upto 6 percent of salary. The same company later switched to a 6 percentdefault savings rate; in that setting, hardly any new workers chose the3 percent savings contribution rate. Employees could easily switch fromone plan to another and had strong incentives to choose the best planfor retirement, yet this trivial change in the status quo significantlyinfluenced behavior. We also see seemingly trivial differences in statusquo influence important choices in the medical field. It’s well-known,for example, that default procedures and standards of care differ acrossthe country and that these defaults are often more determinative of howa patient is treated than more objective factors (Weinstein et al. 2006;Baicker et al. 2004).
Toward a Liberal Consistency?We have not proven, of course, that physicians are biased toward thestatus quo or that the status quo should be changed in favor of a moreconsistent and liberal position. Nevertheless, it’s important that thesequestions be investigated; status quo bias in the field of policy towardthe FDA could be resulting in unnecessary morbidity and mortality(Peltzman 1973; Wiggins 1981; Grabowski and Vernon 1983; see alsoKlein and Tabarrok 2002 for a review).
It’s important to understand the arguments for and against the consistent liberal position because significant trends exist that aremoving policy in that direction. The FDA’s position on promotion ofoff-label prescriptions, for example, has liberalized in recent years.
After losing a series of court cases (e.g., Washington Legal Foundationv. Friedman 1998; Washington Legal Foundation v. Henney 2000;Pearson v. Shalala 1999; Pearson v. Shalala 2001), the FDA hasnow accepted that manufacturers of pharmaceuticals and dietary Off-Label Drug Practices vs. FDA Efficacy Requirements supplements have significant First Amendment freedoms. In the past,the FDA forbade dietary supplements from making health claims, suchas the benefits of using folic acid or of using aspirin after a secondheart attack (e.g., Rubin 1995; Keith 1995). The FDA can no longerforbid manufacturers from making health claims, but it can and willgrade claims according to their scientific merit. A new grading systemwill assign letter grades—ranging from “A” to “D”—on each claim adietary supplement company makes, indicating the quality of thescientific evidence supporting the claim. Competition and advertisingon health consequences has had beneficial effects in markets fromcereal to cigarettes, so we think this approach is promising (Ippolitoand Pappalardo 2002; Calfee 1997).
Pharmaceutical manufacturers now may disseminate off-label infor- mation in the form of peer-reviewed journal articles and medicaltextbooks. Could the new grading system for dietary supplementclaims be extended to off-label uses? That is, does information aboutnonapproved uses need to be “off-label”? A split-label approachwould allow manufacturers to present subsequent-use information onthe label and in their other media. As with the grading system, theFDA could keep speech and competition within bounds by gradingthe claims.
More recently, the Washington, D.C. Circuit Court of Appeals in Abigail Alliance v. Eschenbach (2006) ruled that dying patients have adue process right to access drugs once they have been through FDAapproved safety trials. The FDA’s refusal to allow firms to sell andpatients to buy these drugs “impinges upon an individual libertydeeply rooted in our Nation’s history and tradition of [respectingthe right of] self-preservation” (2006: 486). The court further notedthat off-label prescribing suggests that the FDA’s control over drugapprovals is not beyond question.
Government regulation of drugs premised on concern over a new drug’sefficacy, as opposed to its safety, is of recent origin. And even today, apatient may use a drug for unapproved purposes even where the drug maybe unsafe or ineffective for the off-label purpose. Despite the FDA’s claimsto the contrary, therefore, it cannot be said that government control ofaccess to potentially life-saving medication “is now firmly ingrained in ourunderstanding of the appropriate role of government.” (2006: 483) The American Journal of Economics and Sociology Concluding Comments
NEW DRUGS MUST BE TESTED for efficacy before being permitted, but nosuch requirement exists for new uses of old drugs. In a survey of some500 physicians, we found that most physicians favored the status quo,but in our judgment they were not able to really resolve the incon-sistency. The inconsistency could be resolved either by removing FDArequirements on efficacy or by increasing restrictions on off-labelprescribing.
In our judgment, this investigation provides support to the reform proposal that would resolve the inconsistency by dropping efficacyrequirements, so that people would be freer to produce, sell, andmarket drugs, even for initial uses, with the clearly stated caveat thatthe drug’s efficacy had not been certified by the FDA. Those with thebest knowledge of the particular circumstances and with the strongestincentives to do right by the patient would then have expandedoptions of utilizing therapies that may be very beneficial.
1. In 1972, for example, the FDA announced that there were no extant controls on off-label prescribing but that “when an unapproved use of anew drug may endanger patients or create a health hazard” it was “obli-gated” to act. Thus, it proposed a proceeding to create controls. It plannedto consider the following new rules: revoking the approval of any drugextensively used off-label; regulating off-label uses as experimental (just asif the drug was a new drug); and limiting distribution channels to hospitalsor physicians with special qualifications. The medical profession, includingthe AMA, objected vociferously, however, and the FDA backed down. Overthe next decade, the FDA asked Congress for similar powers but was notsuccessful. On the 1972 episode, see Shapiro (1979) and Christopher (1993).
David Kessler (1978), prior to becoming FDA commissioner, also supportedrestrictions on off-label prescribing; under his leadership in 1991 the FDAindicated that it was reexamining the off-label question, but no new rulesmaterialized.
2. In 1982, the FDA issued a bulletin formally stating that it condoned off-label use as “accepted medical practice” (12 FDA Drug Bulletin (U.S. Foodand Drug Admin., Washington, D.C.), Apr. 1982, at 4). However, this has notprecluded significant FDA impact on off-label usage, as the FDA tightlyrestricts manufacturers’ speech about off-label uses.
Off-Label Drug Practices vs. FDA Efficacy Requirements 3. The cost of HostedWare services was $710 and the cost of Medical 4. Reported rates of off-label prescribing should not be used as estimates of off-label prescribing because physicians often do not know whether anindication is off-label. The Appendix shows that reported rates correlate withsupport for FDA liberalization.
5. Analysis of the complete responses suggests that a number of the 32 physicians who answered either “Favor” or “Don’t know/Not sure” had actuallygotten confused and got “the sign” wrong when answering the question.
6. Statement of George Lundberg, Promotion of Drugs and Medical Devices for Unapproved Uses: Hearing Before the Human Resources and Inter-governmental Relations Subcomm. of the House Comm. on GovernmentOperations, 102nd Cong., 1st Sess. 103.
7. The remarks come from written comments f63, g58, f157, f47, and f51.
(“f63” means the 63th final comment; “g58” means the 58th challenge.
Elsewhere, we indicate “consistent” comments with a “c.”) Here and else-where we take the liberty of correcting spelling and occasionally improvingminor punctuation. The respondents’ written comments are available onlineat theiroriginal form.
8. (This note was not included in the survey.) Physicians in the United States may prescribe drugs that have not been FDA approved (such prescrip-tions might be filled by pharmacies abroad or domestic institutions engagedin drug trials). In practice, however, a physician’s ability to prescribe a drugis tied to the manufacturer’s right to market and sell the drug in the UnitedStates. To keep our survey questions from becoming overly complicated, weoften employed phraseology that would suggest that the FDA directly regu-lates prescribing. We are confident that this simplification did not bias or blurthe investigation, as not a single physician remarked on this technicality orappeared to be confused because of it.
9. Competitive Enterprise Institute (2007) summarizes all six CEI surveys 10. We varied the survey so that the posing of the two main questions was ordered one way in one survey and the reverse in the other. We found thatthe order of the questions did not make a significant difference. Thus theprecise wording of the consistency question varied slightly depending on theorder of the questions.
11. We did not get the idea of asking the “consistent” respondents what they thought of the consistency argument until after the survey was inprogress. Thus, not every consistent respondent encountered this question.
The consistency question always came last in the survey (save the finalsolicitation of final comments), so adding the question could not havechanged the distribution of answers to the preceding questions.
The American Journal of Economics and Sociology 12. The complete (and organized) set of challenges, “consistent” comments, and final comments can be accessed online at Each comment ismarked to indicate any relatedness to the challenges. Also, some commentscontained no clear theory that we could discern (such as commentsthat simply restated positions, justified only one of the positions takenby the respondent, or described the respondent’s own prescriptionpractices).
13. Drawn from an unpublished op-ed article written in 2001.
14. Furthermore, “relatedness” is a tricky concept, and its presence (or absence) may be more obvious ex post than ex ante. Prozac, for example, isused on-label to treat depression, but it is also prescribed off-label for thetreatment of alcoholism. Are these treatments related? Since the etiology ofneither depression nor alcoholism is well understood, one could not concludeon the basis of theory that these diseases were related. Indeed, one of the fewreasons to think that these diseases bear some relation to one another is thatProzac has had some limited success in treating both (Naranjo et al. 1988). Inthis case, relatedness, to the extent that it exists, is more suggested by off-labelprescribing than a cause of such prescribing. It’s also interesting to note thatthe FDA cautions against assuming that “related” drugs will have similar effects(e.g., Suydam 1999).
15. The term “flood” is used by g55, g87, g90, g148; the concern is similarly expressed by many other challenges (see especially those in the PromoHazsection of the listing online).
16. We do not take up this discussion here, but a substantial body of work by economists and others develops a respectable case for the self-correcting dynamics and social benefits of the freedom of speech in healthproducts and foods. See, for example, Leffler (1981), Ippolito and Mathios(1991, 1995), Ippolito and Pappalardo (2002), Masson and Rubin (1985),Rubin (1994, 1995), Keith (1995), Calfee (1997), and Tabarrok (2000). It isnot obvious, therefore, that the net effect of promotion of off-label therapieswould be negative.
17. Other challenges along these lines are g7, g20, g40, g58, g59, g79, g85, 18. DiMasi, Hansen, and Grabowski (2003) estimate that the average cost of getting a drug to legal status is $800 million. On the uncertainty of obtainingpermitting, see DiMasi (2002).
19. The FDA can and does grant what are in effect extensions to patents for producing new and valuable research, but such privileges are difficult ifnot impossible to enforce when patents have expired (see Tabarrok 2001 fora discussion of exclusivity privileges).
20. We also dropped respondents if there were missing or “not sure” answers on the independent variables.
Off-Label Drug Practices vs. FDA Efficacy Requirements 21. To run the regression, we set off-label use at the means of the respective intervals; thus, 10 percent–20 percent was set at 15 percent (40percent or more was set at 45 percent).
References
Abigail Alliance v. Eschenbach. (2006). 370 U.S. App. D.C. 391, 445 F. 3d470, Adams, C., and A. Young. (2003). “Risky Rx: A Knight Ridder Investigation.” Knight Ridder Newspapers November 2–4.
Baicker, K., A. Chandra, J. Skinner, and J. E. Wennberg. (2004, Oct. 7). “Who You Are and Where You Live: How Race and Geography Affectthe Treatment of Medicare Beneficiaries.” Health Affairs Beales III, J. H. (1996). “New Uses for Old Drugs.” In Competitive Strategies in the Pharmaceutical Industry. Ed. R. B. Helms. Washington, DC: AmericanEnterprise Institute.
Beshears, J., J. Choi, D. Laibson, and B. Madrian. (2006). “The Importance of Default Options for Retirement Savings Outcomes: Evidence from theUnited States.” Brosgart, C. L., T. Mitchell, E. Charlebois, R. Coleman, S. Mehalkol, J. Young, and D. I. Abrams. (1996). “Off-Label Drug Use in Human Immunodefi-ciency Virus Disease.” Journal of Acquired Immune Deficiency Syn-dromes and Human Retrovirology 12(1): 56–62.
Calfee, J. E. (1997). Fear of Persuasion: A New Perspective on Advertising and Regulation. Monnaz, Switzerland: Agora Association with AEI Press.
Christopher, W. L. (1993). “Off-Label Drug Prescription: Filling the Regulatory Vacuum.” Food and Drug Law Journal 48: 247–262.
Competitive Enterprise Institute. (2007). A National Survey of Orthopedic Surgeons Regarding the Food and Drug Administration and theAvailability of New Therapies. Washington, DC: Competitive EnterpriseInstitute.
DiMasi, J. A. (2002). “Uncertainty in Drug Development: Approval Success Rates for New Drugs.” In Clinical Drug Trials and Tribulations, 2nd ed.
Ed. A. Cato and L. Sutton. New York: Marcel Dekker.
DiMasi, J. A., R.W. Hansen, and H. G. Grabowski. (2003). “The Price of Innovation: New Estimates of Drug Development Costs.” Journal ofHealth Economics 22(2): 151–185.
GAO. (1991). “Off-Label Drugs: Initial Results of a National Survey.” GAO/ Grabowski, H. G., and J. M. Vernon. (1983). The Regulation of Pharmaceu- ticals: Balancing the Benefits and the Risks. Washington, DC: AmericanEnterprise Institute.
The American Journal of Economics and Sociology Ippolito, P. M., and A. D. Mathios. (1991). “Health Claims in Food Marketing: Evidence on Knowledge and Behavior in the Cereal Market.” Journal ofPublic Policy and Marketing 10(1): 15–32.
——. (1995). “Information and Advertising: The Case of Fat Consumption in the United States.” American Economic Review 85(2): 91–95.
Ippolito, P. M., and J. K. Pappalardo. (2002). Advertising Nutrition and Health: Evidence from Food Advertising, 1977–1997. Washington, DC: FederalTrade Commission.
Jaffe, S. Y. (1994). Statement Before the Subcommittee on Health and the Environment: House Committee on Energy and Commerce, February 8.
Kauffman, R. (1996). Off-Label Drug Use and the FDA: Review of Supple- mental Drug Applications. Testimony by the American Academy ofPediatrics Before the Subcommittee on Human Resources and the Inter-governmental Relations Government Reform and Oversight Committee,U.S. House of Representatives, September 12.
Keith, A. (1995). “Regulating Information About Aspirin and the Prevention of Heart Attack.” American Economic Review 85(2): 96–99.
Kessler, D. (1978). “Regulating the Prescribing of Human Drugs for Nonap- proved Uses Under the Food, Drug, and Cosmetic Act.” Harvard Journalon Legislation 15(4): 693–760.
Klein, D. B. (2002). “The Demand for and Supply of Assurance.” In Market Failure or Success: The New Debate. Eds. T. Cowen and E. Crampton.
Cheltenham, UK: Edward Elgar.
Klein, D. B., and A. Tabarrok. (2002). “Is the FDA Safe and Effective?” Leffler, K. B. (1981). “Persuasion or Information? The Economics of Pre- scription Drugs Advertising.” Journal of Law and Economics 24(1):45–74.
Lundberg, G. (1991). Promotion of Drugs and Medical Devices for Un- approved Uses: Hearing Before the Human Resources and Intergovern-mental Relations Subcommittee of the House Comittee or GovernmentOperations, 102 Cong., 1 Sess. 103.
Masson, A., and P. H. Rubin. (1985). “Matching Prescription Drugs and Consumers.” New England Journal of Medicine 313: 513–515.
Naranjo, C. A., E. M. Sellers, P. Sanhueza, et al. (1988). “The Serotonin Uptake Inhibitor, Fluoxetine, Reduced Alcohol Consumption in ProblemDrinkers.” Psychopharmacology 96(Suppl): 311.
Peltzman, S. (1973). “An Evaluation of Consumer Protection Legislation: The 1962 Drug Amendments.” Journal of Political Economy 81(5): 1049–1091.
Rpt. in G. J. Stigler (ed.) (1988) Chicago Studies in Political Economy.
Chicago: University of Chicago Press.
Rubin, P. H. (1994). “Are Pharmaceutical Ads Deceptive?” Food and Drug Law Off-Label Drug Practices vs. FDA Efficacy Requirements ——. (1995). “FDA Advertising Restrictions: Ignorance is Death.” In Hazard- ous to Our Health? FDA Regulation of Health Care Products. Ed. R. Higgs.
Oakland, CA.: Independent Institute.
Samuelson, W., and R. Zeckhauser. (1988). “Status Quo Bias in Decision Making.” Journal of Risk and Uncertainty 1(1): 7–59.
Shapiro, S. A. (1979). “Limiting Physician Freedom to Prescribe a Drug for Any Purpose: The Need for FDA Regulation.” Northwestern University LawReview 73(5): 801–872.
Suydam, L. A. (1999). Keynote Address. FDLI Conference on Advertising and Promotion in the New Millennium. Available at speeches/offlabel.html Tabarrok, A. (2000). “Assessing the FDA Via the Anomaly of Off-Label Drug Prescribing.” Independent Review V(1): 25–53.
——. (2001). “The Blessed Monopolies.” Regulation Winter: 1–4.
U.S. Food, and Drug Administration. (1982). “Use of Approved Drugs for Unlabeled Indications.” Food and Drug Administration Bulletin 12: 4–5.
Weinstein, J. N., J. D. Lurie, P. R. Olson, K. K. Bronner, and E. S. Fisher. (2006).
“United States’ Trends and Regional Variations in Lumbar Spine Surgery:1992–2003.” Spine 31(23): 2707–2714.
Wiggins, S. N. (1981). “Product Quality Regulation and New Drug Introduc- tions: Some New Evidence from the 1970s.” Review of Economics andStatistics 63: 615–619.
Appendix
Correlations Between Liberalization and Other Variables We investigated whether support for liberalization correlated withother variables. Table A1 reports a probit regression for which thedependent variable was 1 if the physician favored making FDAefficacy certification optional and 0 if he or she opposed that reform.
(We dropped respondents answering “Don’t know/Not sure.”)20 Inde-pendent variables included years of practice, whether the physicianworked at a teaching hospital, was a pediatrician, the physician’s careertype, and off-label prescribing history. Years in practice, working in ateaching hospital, and being a pediatrician had no discernible effect onsupport for FDA liberalization. The responses to the career question aredivided between “Strictly Practitioner,” “Mainly a Practitioner,” “AboutHalf Practitioner, Half Researcher,” and “Mainly Researcher.” Wedropped “Strictly Practitioner,” so read the coefficients on the othercareer variables as relative to physicians who are strictly practitioners.
The American Journal of Economics and Sociology Probit Regression of Support for FDA Liberalization Standard errors in parentheses.
*significant at 5%.
**significant at 1%.
We find that those who are mainly practitioners are about 5 percent lesslikely than strict practitioners to support liberalization, although theeffect is not statistically significant. Physicians who report splitting theirtime evenly between practice and research, however, are 22 percentless likely than strict practitioners to support liberalization, and thosewho mainly do research are about 25 percent less likely, with bothcoefficients statistically significant at the 1 percent level. One interpre-tation of the result might be that practicing physicians are more sensibleto the heterogeneity of patients’ conditions and are in closer contactwith patients who lose out because of FDA restrictions. Hence,practicing physicians would be more cognizant of the costs of FDArestrictions and less enamored with the FDA. Another interpretation is Off-Label Drug Practices vs. FDA Efficacy Requirements that physicians who do research have a stronger allegiance to officialinstitutions because they are more involved in the world of governmentdeterminations and research grants, and feel themselves part of anacademic or elite social stewardship.
We also find that physicians who report greater off-label prescribing are more likely to support making efficacy standards optional. Thecoefficient on off-label usage indicates that a 1 percent increase inreported off-label prescribing increases the probability of supportingFDA reform by 0.40 percent.21 Thus an increase of one standarddeviation, about 12 percentage points, would raise predicted supportfor liberalization by just under 5 percentage points. It seems thatphysicians who regularly prescribe off-label (or who are aware thatthey do so) are more likely to embrace private, voluntary forms ofefficacy certification.
Table A2 shows support for FDA liberalization by area of special- ization (for areas with at least 20 respondents). We find no statisticalsignificance between the rates.
Percent of Physicians Who Support Making FDA Efficacy Certification Optional, by Area of Specialization

Source: http://econfaculty.gmu.edu/klein/PdfPapers/AJESNov2008.pdf