Vous êtes sur la page 1sur 11

A Critique of Clinical Equipoise: Therapeutic Misconception in the Ethics of Clinical Trials Author(s): Franklin G.

Miller and Howard Brody Source: The Hastings Center Report, Vol. 33, No. 3 (May - Jun., 2003), pp. 19-28 Published by: The Hastings Center Stable URL: http://www.jstor.org/stable/3528434 Accessed: 03/03/2010 13:48
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=hastings. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

The Hastings Center is collaborating with JSTOR to digitize, preserve and extend access to The Hastings Center Report.


in the Misconception







A predominant ethical view holds that physician-investigatorsshould conduct their research with therapeutic intent. And since a physician offering a therapy wouldn't prescribe second-rate treatments, the

is intervention and the best proventherapyshouldappearequallyeffective."Clinical experimental equipoise" different, necessary.Butthis perspectiveis flawed.The ethics of researchand of therapyare fundamentally
and clinicalequipoise should be abandoned.

Trial he Hypericum Study Depression Group

publishedin 2002 the resultsof a randomized trial comparinghypericum(St. John'sWort), sertraline(Zoloft), and placebo in the treatmentof In the study, funded by the Namajor depression.1 tional Institutesof Health, 340 subjectsfrom twelve participatingcenters were randomizedto three trial armsfor an eight-weekperiod, with carefulmonitoring to assurethat patientswho worsenedsignificantly or who becamesuicidalwere removedfrom the study and receivedadequatetreatment.Neither hypericum
FranklinG. Miller and Howard Brody, "A Critique of Clinical in the Ethicsof ClinicalTriMisconception Equipoise: Therapeutic Center 33, no. 3 (2003): 19-28. als,"Hastings Report 2003 May-June

on wasfoundto be superior to placebo norsertraline The authors the primary outcomemeasures. noted, "From a methodological pointof view,thisstudycan of inof the importance an example be considered in testtrials inactive and active comparators cluding effects of medications. the antidepressant ing possible could easily In fact, withouta placebo,hypericum as sertraline."2 as effective havebeenconsidered What can we concludeaboutthe ethicsof this ethics trial?One dominantviewpointin research On thisviewpoint, the study. wouldhaveprohibited of a randomized trialis ethical onlyin circumstances within "clinical uncertainty equipoise"-a genuine as to whether(in this case) the medical community to the three treatment armsaresuperior of the any

other two. No such uncertaintyexists. Approximately twenty-fivecliniavailable includantidepressants, cally have been shown to be ing sertraline, to superior placebo.3Moreover, the majority opinion within psychiatry probablyholds that sertralineis definitely superior to hypericum for major depression,even if hypericum has potential for the treatment of mild to moderatedepression.But another widespread viewpoint would hold that the trial was ethically sound. Depressedindividualswidely use hypericum,a "natural" agent, deof the lack spite proven efficacy.Aca cordingly, rigorous evaluation offered scientific, clinical, and social value.Accordingto the reportof trial results,the studywas approvedby institutional review boards (IRBs) at twelve sites and subjects provided writteninformedconsent. But if clinical equipoise is a basic how for ethical research, requirement could all these reviewboardsbe blind to the unethical nature of this trial? diAnd how could two such radically rewithout exist, viewpoints vergent as searchethics being widely regarded in a state of crisis? Therapeutic Misconceptions

he prevailingethical perspective on clinicaltrialsholds that physican dischargetheir cian-investigators to patientsin "therapeutic obligation" clinicaltrithe contextof randomized als (RCTs)as long as treatments being tested scientifically satisfy clinical equipoise.We contend that this ethical perspective is fundamentally flawed. An ethical framework that providesnormativeguidanceabout a practice should accuratelycharacterThe prevailing ethical ize the practice. this test: All sound fails perspective ethical thinking about clinical reframework search,and the regulatory for reviewof protocolsfor clinicalinvestigation, depends on a basic distinction between researchand therapy. But the claims in the prevailing ethical perspective on clinical trials conflate researchand therapy.These

claimsarethat the ethics of the physimust govern cian-patientrelationship RCTs, that physicianswho conduct these trialshave a "therapeutic obligation" to patients enrolled in them, and that RCTs must be compatible with some form of equipoise. Certainly, investigatorsand ethicists recognizethat clinical trials are scientific experiments, which differ from standard medicalcare.They also recognizethat they aresubjectto regulatory requirementswhich do not apply to routine medical practice. However, the prevailing ethical framework views clinical trials through a therapeutic lens. The mainstream ethicalapproachto clinical trials attempts to have it both ways:to view the clinicaltrialas a scientific experiment,aimed at producing knowledgethat can help improve the care of future patients, and as treatment conducted by physicians who retainfidelityto the principlesof therapeutic beneficence and therapeutic non-maleficence that govern the ethics of clinical medicine. The doctrine of clinical equipoise has emergedas the bridge between medical care and scientific experimentation, allegedlymaking it possible to the conduct RCTswithout sacrificing of therapeutic obligation physicians to provide treatmentaccording to a scientifically validated standard of care. This constitutes a "therapeutic misconception" concerningthe ethics of clinicaltrials,analogousto the tendency of patient volunteers to confuse treatmentin the context of RCTs with routine medical care.4As Paul Appelbaum has recently observed, "In fact, this confusion between the ethics of research and of ordinary clinical care appearsrampant in the world of clinicaltrials."5 The therapeuticmisconceptionin the ethics of clinical trialsis reflected in the languagecommonly usedwithin the clinical research enterprise. Clinical trials are often describedas and investigaresearch," "therapeutic tors are regardedas having a "therapeutic intent." Researchparticipants who are being studied because they

have a medical condition under investigation are referred to as "patients," and investigatorsas "physicians"or "doctors," without qualification. To demonstrate our contention about the mainstream approach to the ethics of clinical trials, we will of offer an intellectualreconstruction ethics some of the history of research since the 1970s. This history is characterized by incoherence resulting from commitment to two incompatible positions, each approaching researchethics in a fundamentallydifferent way. The therapeuticmisconception about the ethics of clinical trialshas emergedfrom the "similarity position," which argues that ultimately,the ethics of clinicaltrialsrest on the same moral considerations that underliethe ethics of therapeutic medicine. The "differenceposition" arguesthat the ethics of clinicaltrials must start with the realizationthat medical researchand medical treatment are two distinct forms of activity, governedby differentethicalprinciples. The reigningethical paradigmfor clinicaltrialshas coexistedwith clinical trialspracticethat departsfrom its guidance.Clinicalequipoise,the cornerstone of the similarity position, rules out placebo-controlled trials whenever there is a proven effective treatmentfor the disorderunder investigation.6 However, IRBs have such placebo-conroutinelyapproved trolled trials.These two anomaliesunappreciated theoretical incoherence and conflict between the theoretical paradigmand the practice of ethical review of clinical trials-call for criticalexaminationof the similarity position and the doctrine of clinical equipoise. The Distinction between Research and Therapy n 1979, RobertLevinesummarized "the most important achievements of the National Commission"for the Protectionof Human Subjectsof Biomedical and BehavioralResearchin
2003 May-June


"correctingthe conceptual and semantic errors that had undermined virtuallyall previousattempts to develop rational public policy on research involving human subjects."7 Two portions of Levine's summary capture the essential ingredients of the difference position: recognizing the distinction between researchand therapyand, accordingly, abandoning the distinction between therapeutic research. and nontherapeutic Clinical research shareswith medical care the fact that both are performed by physiciansin clinical settings, and both often use similardiagnostic and treatment interventions. When the commission began its work, physicianscommonly regarded clinical researchand medical therapy as inextricably connected. One authorityquoted by Levineclaimedthat a time a physicianadministers "Every is in a sense he to a perpatient, drug forming an experiment." But the commission recognized the importance of determiningthe boundaries betweenroutinemedicalpracticeand research. For Levine, the commiscame sion'sconceptualbreakthrough with the realizationthat the physicians of the day were thinking about clinical researchin the wrong way, and that the boundary between research and therapy was clear rather than fuzzy.The commission came to hold that cinical researchis fundamentallydifferentfrom medicalpractice.8 Clinical medicine aims at providing optimal medical carefor individual patients. Ethically,it is governed by the principles of therapeutic beneficence and therapeutic nonmaleficence.Therapeuticbeneficence directs physicians to practice medicine with primaryfidelityto promoting the health of particularpatients. According to therapeuticnonmaleficence, the risks of medical care to which a patient is exposed are to be justified by the prospectof compensating medical benefits for that patient. The physician uses scientific knowledgeto carefor the patientand engages in therapeuticexperimenta2003 May-June

tion with the aim only of finding optimal treatment.It is not part of the role of the physician in providing medical care to develop scientific knowledge that can help future patients. Clinical research, in contrast, is not a therapeuticactivity devoted to the personalcareof patients.It is dea scientificquessignedfor answering with the aim of tion, producing"genThe investigaeralizable knowledge." tor seeksto learnabout diseaseand its treatmentin groupsof patients, with the ultimate aim of improvingmedical care.Scientificinterestin any particularpatient concernswhat can be learnedthat is applicableto other patients. In view of the natureand pur-

In clinical research, by contrast, the interestsof investigatorsand patient volunteersare likely to diverge, even when the investigatoracts with complete integrity. Patient volunteers,especiallyin clinicaltrials,typically seek therapeuticbenefit, though they also may be motivatedby altruism.?1 are interestedpriInvestigators in marily developingscientificknowledge about groups of patients. Remotivations, gardlessof investigators' at risk of havare volunteers patient their well-being compromisedin ing the course of scientific investigation. involvesan inherent Clinical research tension between pursuing rigorous science and protecting researchparticipantsfrom harm.1

How could two such radically divergent exist, without researd

viewpoints ely


as in a state of crisis?

pose of clinical research,the principles of beneficence and nonmaleficence applicable to clinical research lack the therapeutic meaning that guides their application to medical care. Clinical research is dedicated primarilyto promoting the medical good of future patients by means of scientificknowledgederivedfrom experimentationwith current research participants-a frankly utilitarian purpose. A major reasonfor distinguishing research from therapy is to underhas an inscore that clinical research herent potential for exploiting researchparticipants.9 Exploitationalso in clinical medicineoccur may venal physicians sometimes perform medicallyunnecessary proceduresfor the sake of profit, for example. Yet when physiciansof integritypractice medicine, physicians' and patients' interestsconverge.The patientdesires to regainor maintainhealth or to relieve suffering;the physicianis dedicated to providing the medical help that the patient needs.

the ethicaldistinction Historically, between research and therapy emerged out of concern about exploitive abuses of patients in clinical research.Reflectionon this dark history gaveriseto a majordevelopment the in the ethics of clinical research: for independent, requirement of rereviewand approval prospective Priorindependent searchprotocols.12 review was considered necessaryfor becauseof the diverclinicalresearch gence betweenthe interestsof the investigator and the research particiby physician-inpant. Self-regulation not be trustedin the could vestigators researchcontext to the same extent that self-regulation by physicianswas in the therapeutic conappropriate text. The basic rationalefor prospecreviewdetive, independentresearch rebetween the on distinction pends searchand therapy. The point of distinguishing researchand therapyis not to make an invidious comparison,implying that clinical trials are more risky or ethically problematicthan routine clinical practice.Indeed, there is some evHASTINGS CENTER REPORT 21

idence that patients receivemore favorable medical outcomes in many clinical trials,'3and clinical medicine is certainlyrifewith ethicalproblems. Further,since researchis more carefully regulatedthan medicalpractice, it is quite likely that fewerethicalviolations occur in research. To say that two activitiesare ethicallydifferentis not to say that either is inherently betterthan the other. Abandoning the Distinction e|he distinction between research and therapyis most likely to be obfuscatedin the context of clinical trials,which test the safetyor efficacy of investigational and standardtreatments. Since patients may derive medical benefit from trial participain phaseIII RCTs(the tion, especially

expect to find considerablevariation in the treatment administered to those 340 patients after eight weeks or so. Fromthe vantagepoint of therapy, this is what it means to provide careto patients. From the vantage point of research, such variation would wreak havocon experimental designand the of findand validity generalizability ings. So when patients are randomized to one or another experimental drug,and aretreatedaccordingto relativelyinflexibleprotocols,the activity is very different from therapeutic medicine. In many other ways, too, routine deviatefrom what aspectsof research would be requiredby the duties of therapeutic beneficence and nonmaleficence. Volunteer patients and areoften ignophysicianinvestigators

A major reason for distinguishing therapy is to underscore inherent potential participants. the inter


from a an .

that clinical rarch research

for exploiting

In clinical research,

by contrast,

ors and patient volunteers acts

are likely to diverge, even when the investigator with complete integrity.

final stageof testing,which many investigational drugsnevereven reach), clinical trials are often characterized as "therapeutic research." Nonetheless, the process of treatment in RCTs differs radicallyfrom Consider routine clinical practice.14 the contrastbetween the hypericumsertraline trial and routine medical care for depression. If a physician treated340 patientsfor majordepression, she would not decide which drug to administerby flippinga coin. If the physicianelected to use sertraline, she would judge each case individually to determine dose, when to change the dose, and whetherto preor recscribea second antidepressant ommend other treatment.We would

rant of assignmentto the experimental or control treatment,which may be a placebo.Trialsoften include interventionssuch as blood draws,lumbar punctures,radiationimaging, or biopsies that measuretrial outcomes but in no way benefit participants. RCTsoften containa drug "washout" phase before randomizationto avoid confoundingthe evaluationof the investigational treatment with the effects of medicationthat patientswere receivingpriorto the trial.These various featuresof researchdesign promote scientific validity; they carry risks to participants without the prospectof compensatingtherapeutic benefit.

For these reasons, Levine argued that the second majorcontributionof the commission was to abandon the distinction between thera"illogical" and peutic nontherapeuticresearch, which previouspolicymakers thought was essentialto the properregulation of research and the protection of human subjects.15Because research and therapyaredistinctactivities,and the ethics of therapeutic medicine exthereforecannot be automatically it is mistaktended to guide research, en to label researchas "therapeutic" or "nontherapeutic," as if that made any fundamental ethical difference. trialsconsistof a comMany research mix of therapeuticand nontherplex elements-the placebo-conapeutic trolled trial being only one obvious example-such that labelingthe trial or "nonas a whole as "therapeutic" is therapeutic" misleading. In addition, the therapeutic-nontherapeutic distinctiondivertsattention from key ethical issues. Consider a nontherapeutic trial in which one interviews subjectsand takessalivasamples,and a therapeutic trial in which one is testing a new cancer drug that has some promise for creatingremission, but also has potentiallylife-threatening toxicity. Is the latter trial less in need of stringentregulatory oversight Or does because it is "therapeutic"? the therapeutic-nontherapeuticdistinction distract the observer from those aspectsof the trialsthat assume far greatermoral weight, such as the level of risksand the potentialvulnerabilityof subjects? the distincOnce one understands tion between research and therapy, one realizes that "therapeutic"research is still research,and that the ethical rules appropriate to it are for clinicalresearch those appropriate Even though the patient generally. benefit from treatment derive may the basic goal of the being evaluated, is not activity personal therapy,but ratherthe acquisitionof generallyapplicable scientific knowledge. The basic goal and nature of the activity determinesthe ethical standardsthat ought to apply.
2003 May-June

Writingin 1993, JayKatzaffirmed the vital importanceof the distinction betweenresearch and therapyand deplored its blurringin practice:"The astronomicalincrease in clinical research has, in practice, not led to a clear demarcation between therapy and research, bioethical theories notwithstanding. This vital distinction remainsblurredwhen physicianview subjectsas patients, investigators and then believe that patients'interests and not science'sarebeing served by participationin randomizedclinical trials that are so commonly conducted in today's world."16One of the reasonsinvestigators (and bioethito the dishave failed cists) appreciate tinction between researchand therapy is that the similarityposition has conceived the ethics of clinical trials within the context of the physicianpatient relationship. Charles Fried and the Similarity Position 1974, Fried published Medical PersonalIntegrity Experimentation: and SocialPolicy,which launchedthe similarityposition within bioethics.'7 Friedassumedthat answersto ethical dilemmas in researchwould have to be found within the ethics of therapeutic medicine.He defendedfidelity to the interestsof the individualpatient againsta model in which "medicine is to be viewed as caring for What made the RCT populations."'8 ethicallysuspectwas that it seemedto him a prime exampleof populationfocused-rather than individualmedicine. ized-and utilitarian Frieddevoted most of his book to in personal defendingpatients'"rights care."19Returning to medical research, he took issue with trials in which patients were randomized to receiveeither the experimentalintercare.Friedcoined vention or standard to describethe the term "equipoise" for concondition ethicallynecessary ducting an RCT:physician-investigators must be indifferentto the therapeutic value of the experimentaland control treatments evaluated in the


trial.The basic idea of equipoise had previouslybeen articulatedby Bradford Hill, a pioneer in the developBut what Fried obment of RCTs.20 jected to primarilyin RCTs was not randomization per se, but the fact that no informed consent had been obtained.Friedsaw the threatof "care for groups"(insteadof "carefor individuals")as residingprimarilyin the idea that it was legitimate to enroll subjectsin an RCT without explicit, informed consent becausethe results of the trial would provide new medical knowledge that would improve the lot of future patients.21Because Friedwas concernedchieflyabout informed consent, an essentialingredient of both medical research and therapeutic medicine, he saw no problem in applying the ethics of medicaltherapyto medicalresearch. for paIn the 1970s, the "respect tient autonomy"movementwas gainfor the old ing steamas a replacement ethic of paternalistic Hippocratic beneficence.Since both Friedand the National Commissionseemed on the surfaceto be championingpatientautonomy, it was easy to miss the point that they were proposingtwo fundamentally different strategies for approachingthe ethics of clinical trials. Put another way, so long as the bioethics debate of the moment has ethics reto do with whetherresearch all quires competent subjectsto give fully informed consent, any fundamental divergencebetween the similarity and the differencepositions is likely to be obscured. The Emergence of Clinical Equipoise uring the 1980s, philosophers ethics recinterestedin research the between tension a obligaognized tion of physicians to offer optimal careto theirpatients("thetherapeutic obligation") and the provision of medical treatment in the context of clinicaltrials.Don Marquisaddressed this problem in a 1983 essay,"LeavThe title is ing Therapyto Chance."22 significant,suggestingthat the RCT

is a form of therapy rather than an ethically distinct activity. Marquis beganhis essay,"Considerthis dilemma: accordingto an argumentthat is hardto refute,the procedurefor conducting randomizedclinical trials of anticancer drugsis incompatiblewith rethe ethics of the physician-patient lationship.If this problemis to be resolved, then either a key procedure for achievingscientific knowledge in medicine must be given up or unethical behavior by physicians must be In framing this "RCT tolerated."23 dilemma,"Marquisassumedthat the ethic for clinicaltrialswas appropriate that of the (therapeutic) physicianpatient relationship. FredGifford,following the lead of Marquis,examined the RCT dilemma in greater depth: "The central dilemma concerning randomized clinical trials (RCTs) arises out of some simple facts about causal methodology (RCTsare the best way to generatethe reliablecausalknowledge necessary for optimally-informed action) and a prima facie plausible principle concerning how physiciansshould treat their patients (alwaysdo what it is most reasonable to believe will be best for the patient)."24Neither Marquis nor Gifford found what they regardedas a solution, and neitherconsatisfactory sideredthe possibilitythat the difference position could dismissthe "RCT as misguidedto begin with. dilemma" In a landmark 1987 article, Benjamin Freedmanoffereda solution to the RCT dilemma that gained widespread acceptance within bioethics. He argued that the tension between ethically legitimate scientific experimentation and the therapeuticobligation of physicians could be overcome by the principle of "clinical Freedman agreed with equipoise."25 Fried and Marquisthat ethical clinical trials had to be compatible with therapeutic beneficence and nonmaleficence. But he argued that Fried'sformulation of equipoise was too constraining. Freedman called Fried's original concept "theoretical equipoise" (sometimes called "indiHASTINGS CENTER REPORT

2003 May-June


vidual equipoise")and contrasted it with his favoredconcept of "clinical equipoise"(sometimescalled "collective equipoise").In the lattersense of equipoise,any individualinvestigator or physician might have reasons to believe that one arm of the RCT offers a therapeutic benefit over the other arm,but the medicalprofession as a whole remainsdivided. Accordan RCT is ethicalso ing to Freedman, as the community professional long has not yet reached a consensus, is sothat "medicine which recognizes cial rather than individual in nature."26When, and only when, clinical equipoise is satisfiedwill patients enrolledin a clinicaltrialbe assured that they will not be randomized to treatmentknown to be inferior. Freedmanthus assertedin a later article that clinical equipoise is in the normativenatureof "grounded clinical practice,the view that a patient is ethically entitled to expect treatment from his or her physician-an entitlement that cannot be sacrificedto scientificcuriosity."27 The bioethics community perceived Freedman's concept of clinical a both theoreticaland a as equipoise it apadvance. Theoretically, practical a more to offer intellectually peared inicompellingargumentthan Fried's would it tial formulation.Practically, permitusefulRCTsthatwould otherto go forwise be ethicallyproscribed ward. Since it appearedto solve the RCT dilemma by accommodating the conduct of clinical trialswith the therapeutic obligation of physicians to offer optimal medicalcare,clinical equipoise gained wide currencyas a fundamentalconcept of the ethics of The persuasive clinicaltrials.28 way in fortifiedthe similarwhich Freedman ity position diverted attention from the fact that clinical equipoise collapsed the distinction between researchand therapy. The similarityposition and clinical equipoise have been popular not only among bioethicists, but also among investigators. We speculate that this ethical perspectivehelps to address investigators' psychological

needs. Physician-investigators,after all, went to medicalschool, not investigator school. To think of research with patients outside the ethical framework of the physician-patient as the differenceposition relationship, requires,may be difficult and threatening to them. Clinical equipoiseoffers a formula that seems to allow them to mix both physician and investigatorroles-even if the psychological comfort is purchased at the price of ethicalobfuscation. The anomaly thereforeexists that much of today'sbioethical thinking accepts clinical equipoise as an outgrowth of the similarity position, while the Federal regulations grew out of the work of the NationalCommission, which largely endorsed the differenceposition. One would imagine that sooner or laterproponentsof clinical equipoise would realize the need to defend this doctrinefrom the charge that it conflates the ethics of clinical trialswith the ethics of medwhat has ical care.But this is precisely not yet happened.

for a wide range of chronic conditions-including mood and anxiety disorders,asthma, stable angina, hypertension, and migraine headaches -all of which can be treated with medicationof provenefficacy. for this Therearetwo explanations incoherence between theory and practice.First, the FDA has encouraged the use of placebo controls in trials concerning these and other chronic conditions.30 Active-controlled trials designed to test the treatequivalenceof the experimental treatmentsuffer ment with a standard from serious methodological limitatritions. Wheneveractive-controlled als show no statistically significant differencebetweenthe investigational treatmentand an active comparator, two conclusions are possible. Either both were effectivein the trialsample of patients, or neither was effective. Without the use of a placebocontrol, such trials lack internal validity.Accordingly,the FDA has insisted that companies use placepharmaceutical bo controlsin trialsof new treatments for conditions characterized by flucrates of and The Case of Placebohigh symptoms tuating the U.S. Controlled Trials Second, placebo response.31 federalregulationsgoverninghuman Ithough the similarity position, subjectsresearchdo not provide any tlbolstered by clinical equipoise, explicit guidanceon the use of placeIRBs have been free to became the reigningparadigmin the bo controls.32 such its dominion clinical ethics of trials, placebo-controlledtriapprove that This diover practice was limited. als, provided they meet regulatofor vorcebetweentheoryand practicehas ry requirements a favorableriskbeen particularly pronounced in the benefit ratio, including the potential case of placebo-controlled trials. value of knowledgeto be gained and Freedmanand his colleaguesargued informedconsent. For the most part, this lack of fit that the use of placebocontrolsis unethical whenever proven effective between theory and practicereceived treatmentexists for the medical con- little criticalattention until the publidition under investigationin a clini- cation in 1994 of an article in the cal trial becausethose randomizedto New England Journalof Medicineen"The titled treatment receive would Continuing Unethical placebo Kenneth Use of Placebo Controls."33 known to be inferior.29 Michels Karin and Rothman of clear castigatimplications Despite the clinical equipoise for the ethics of ed the practiceof placebo-controlled placebo-controlled trials, numerous trials in the face of proven effective trials, such as the hypericum-sertra- treatmentand the role of the FDA in line trial, continued to use placebo encouraging these trials. They cited controlsdespiteproveneffectivetreat- the Declaration of Helsinki, which ment. Placebocontrolshave typically relies heavily on the similarityposibeen used in trialsof new treatments
2003 May-June

tion, as prohibiting this widespread "unethical" practice. Their article stimulated a lively debate over the ethics of placebocontrolled trials. Freedmanand his colleaguesattacked "the placebo orthodoxy" in a two-part article that challenged the scientific value of trialsand reiteratplacebo-controlled ed that they are unethical when proven effective treatmentsexist because they contravene clinical equipoise.34 Other commentators, writing in leading medical journals, defended more or less extensive use of placebo-controlled trials on ethical and methodological Without directlychallenggrounds.35 ing the doctrineof clinicalequipoise, they implied that clinical equipoise provides erroneous ethical guidance trials.Accordfor placebo-controlled over the debate placebo-coningly, the reigning trolled trialsjeopardizes ethicalparadigmof the similarityposition and clinicalequipoise. Critique of the Similarity Position and Clinical Equipoise ur reconstructionof the recent history of the ethics of clinical trials has traced the emergence and dominanceof the similarityposition. This historyalso revealscracksin the foundation of this ethical paradigm. Simultaneous endorsement of the difference position, reflected in the federal regulatory system and the Belmont Report, and the similarity position, which invokes the doctrine of clinical equipoise, has left the ethics of clinicaltrialsin a stateof incoherence. Although this incoherence has not receivedcritical attention, it becomes apparentonce the assumptionsunderlyingthe similarity position and clinical equipoiseare challenged. In addition, the divorce between researchethics theory and clinical trials practicein the case of trialssuggeststhat placebo-controlled a critique of the similarityposition and clinicalequipoiseis overdue. We contend that clinical equipoise is fundamentallymistaken
2003 May-June

because "the RCT dilemma," for which it was proposedas a solution, is false. Clinical equipoise and all other forms of equipoise make sense as a normativerequirement for clinical trialsonly on the assumptionthat investigators have a therapeutic obligation to the research participants. The "therapeuticobligation" of investigators, forming one horn of the RCT dilemma,constitutesa therapeutic misconception about the ethics of clinicaltrials.The presumption that RCTs must be compatible with the ethics of the physician-patient relationship assumeserroneousa form of therapy, the RCT is that ly thus inappropriately applying the principlesof therapeuticbeneficence that governclinand nonmaleficence ical medicine to the fundamentally

which aredisate to clinicalresearch, tinct from therapeutic beneficence and therapeuticnonmaleficence. Clinical equipoise is neither necessarynor sufficientfor ethicallyjustifiable RCTs. The use of placebo controlswhen proven effectivetreatment exists violates clinical equipoise; however, when methodologically indicated, their use is no different in principle from any researchinterventionthat poses risksto subjectswithout the prospectof benefiting them.37 In many cases, the risks of withholding effective treatment are excessive, and the use of placebo controls would thus be unethical. Nevertheless,it is the unacceptable level of risk, not the violation of investigators' alleged "theramakes these that peutic obligation,"

Even though from treatment

the patient

may derive benefit the therapy, but applicable

being evaluatef activity is not personal of generally

rather the acquisition

scientific knowledge.
differentpracticeof clinical research. It is impossibleto maintainfidelityto doing what is best medicallyfor patients in the context of RCTsbecause these are not designed for, and may conflict with, personalizedcare. Although ethically appealing,the project of bridgingthe gap betweentherapy and researchvia the doctrine of clinicalequipoiseis doomed to fail. The insight that the RCT contravenes the ethics of the physician-patient relationship led Samuel Hellman and Debra Hellman to argue that the RCT is unethical and that other methods of evaluating treatments should be employed.36This stance, however, would deprive patients and society of the benefitsthat flow from rigorousscientific evaluation of experimental and standard treatments. The more reasonable conclusion is that RCTs should be governedby ethical norms appropritrials unethical. In other cases, including the hypericum-sertraline trial, use of placebo controls when proven effective treatment exists is ethicallyjustifiable. By conflatingthe ethics of clinical trials with the ethics of therapeutic medicine,proponentsof the similarity position may also contributeto the lack of adequateinformedconsent. If view the ethics of cliniinvestigators cal trials through a therapeuticlens, they may explicitlyor implicitly foster the therapeutic misconception participants-that is, among research the tendency of participantsin trials to confuse clinicaltrialswith medical care. Researchparticipantsneed to know that the overall activity is aimed not at their own ultimatebenefit, but at discoveringnew knowledge to help future patients. If they think that clinical trial participation then they cannot is a form of therapy,

ungive informedconsent.Moreover, like the therapeuticcontext, the pacannotdelegatethe decitient-subject sion to the physician-researcher. In the therapeuticsetting, a patient can decide to trust the physician to choose the best treatmentbecausethe

alreadydo with increasingfrequency. Paymentswould add to the cost of conductingclinicaltrials,but it might help preventthe therapeuticmisconTo ception among trialparticipants.38 be paid signifiesthat the trialparticipant is not merely a patient seeking

To be trustworthy, investigators must themselves 1 c understand clearly the ways in whic research forthri from clinical practice and convey this ;esearch subjects.
therapy.If additional expenditureis necessary to motivate clinical trial participation, then this is a price worth paying for enhanced professional integrity and informed consent. An Alternative Ethical Framework n view of the theoretical and practical problems associated with the similarityposition and its logical offspring, clinical equipoise, an alternafor the ethics of clinitive framework cal trialsis needed.The most promisethics ing recenttreatmentof research has been developed by Ezekiel Emanuel,David Wendler,and Christine Grady.39 They proposeseven ethical requirementsfor all clinical research: (1) scientific or social value; (2) scientificvalidity;(3) fair subject selection; (4) favorable risk-benefit ratio; (5) independentreview;(6) informed consent; and (7) respect for enrolled research participants. This frameworkis built on the difference between research and therapyand on the core value of protectingresearch from exploitation. participants Yet even this formulation of an ethical framework appropriate to clinicalresearch testifiesto the hold of the similarityposition. The authors endorseclinicalequipoise,claimingit is implied by the requirements of value, validity,and risk-benefitratio. We contend, by contrast,that the en-

physicianhas the patient'sbest interests at heart.The investigator has the interests of future patients at heart, and so cannot decide for the subject whether or not to participatein the research. To be trustworthy, investigators must themselves understand clearlythe ways in which clinical research differs from clinical practice to potenand convey this forthrightly tial research subjects. It is worth pondering, however, the practicalconsequencesthat might ensue if physicians,investigators, patients, and ethicistsunderstoodclinical trialswithout distortionby therapeutic misconceptions. Would refor valuable cruitmentof participants clinical trials become substantially more difficult, slowing progress in medical care?The fact that clinical trialsare no longer seen as a mode of therapy leaves unchanged the real prospect of therapeutic benefits offered to patients from trial participation, includingthe opportunityto receive promising investigational agents, ancillarymedical care, expert diagnosticevaluations,and education about their disorder. Nonetheless, some patients might be less inclined to participatein clinical trials when they appreciate the differences between these scientific experiments and medicalcare. To attract enough subjects, researchersmight have to pay people for their participation,as researchers in industry-sponsoredclinical trials

dorsement of clinical equipoise rendersincoherentany accountthat arises from the differenceposition. The most importantnext step for research ethics is to develop this "non-exploitation"frameworksystematically in a way that avoidsany conflationof with medicalcare. clinicalresearch Those who agree that physicianwho conduct clinicaltriinvestigators als are not governed by therapeutic beneficence still might argue that clinical equipoise providesimportant methodologicalguidance for justifying clinical trials. Freedmanand his colleagues have argued that clinical equipoiseis both an ethicaland a scientific principle:"That principlecan be put into normative or scientific language. As a normative matter, it definesethicaltrialdesignas prohibiting any compromise of a patient's right to medical treatment by enrolling in a study.The same concern is often stated scientificallywhen we assertthat a study must startwith an honest null hypothesis,genuine medical uncertaintyconcerning the relative merits of the various treatment arms included in the trial'sdesign."40 Nevertheless, whatever is valid in clinicalequipoise methodologically -the honest null hypothesis-can be stated more clearlyand without confusion with the therapeutic obligaof tion, by appealto the requirement scientific value: no researchparticipants should be exposed to the risks of valueless research. Clinical trials must be designed to answervaluable scientific questions. If the answer is alreadyknown or the question is trivial, then there is no honest null hypothesis, and a clinical trial should not be conducted.But this is logically independent of whether all the patients enrolled in the trial would receive medical treatment that is believed by the expertmedicalcommunity to be at least as good as the standardof care. This alternative framework provides accurateethical guidance conwithout precerningclinical research that the ethics of therapeutic suming medicineshould governclinicaltrials.
2003 May-June

We illustrate this by applying the seven ethical requirements to the exof the ample hypericum-sertraline trial. Scientific or social value and scientific validity. The study has social value owing to the widespreaduse of herbalremedies.Since the efficacyof hypericumin treatingdepression(especiallymajordepression)was uncertain, therewas an honest null hypothesis that hypericumwould be no better than placebo. It would have been unreasonable to design the trialas an active-controlled superiority trial, since it is highly unlikely that hypericum could be shown to be more efAn active-confective than sertraline. trolled equivalence trial would lack becausethe finding sensitivity" "assay that the reduction in symptoms of depressionexperiencedby those trial participantsreceivinghypericumwas not significantly different for those receivingsertalinewould not validly supportthe inferencethat hypericum was effective.41 It would remainpossible that neither treatmentwas effective in the study sample-as was in fact shown. The study,therefore,was properly designed as a three-arm trial. placebo-controlled Fair subject selection.There is no evidence to suggest that particularly vulnerablepatientswere recruitedinfor this study,which inappropriately of decluded a sample representative pressedpatients. Favorable risk-benefit ratio. Risk-benefit assessment of research protocolsultimatelycomes down to a matter of judgment. With respectto the use of the placebo control-the aspectof the trialthatviolatedclinical equipoise- the risks to participants from an eight-weektrial,with careful criteriaand monitoring, exclusionary not and werejustifiable were excessive valueof the knowlby the anticipated edge to be gained from the research. Hence, the placebocomponentof the study had a favorable risk-benefit ratio. Eliminatingthe placebo would ratiounfahave made the risk-benefit vorableby virtue of underminingthe scientificvalidityof the research.
2003 May-June

Independent review, informed consent, and respectfor enrolled research participants. The report of the study assertedthat IRB approval was obtained at all sites and that all subjects gave informed consent. In for addition,the describedprocedures for risk monitoring subjects possible of harm indicatedan acceptablelevel of respect. In sum, this study was ethically justifiable despite violating clinical equipoise;moreover,had it been designed in accordance with clinical equipoise, it would have been deficientand theremethodologically fore ethicallyquestionable. CharlesWeijer,a leading advocate of clinicalequipoiseand the similarity position, has recentlyclaimed that trialsin the con"Placebo-controlled text of seriousillnessessuch as depression or schizophrenia are ethically egregiouspreciselybecause no competent physician would fail to offer therapyto a patient with the condition."42Although we agree that depressionis a seriousillness,the hypericum-sertraline trial demonstrates that there is nothing "ethicallyegregious"about the use of placebo controls in trialsof treatmentfor depression, as long as the ethical requirements for clinical researchare satisfied. Whether or not one agreesthat, all things considered, the placebo control was ethical in this trial, the ethical justification of placebo controlshas nothing to do with the therapeutic practiceof competent physiethcians. In any case, the alternative with its seven requireical framework ments providesadequateguidancefor clinical trials without appeal to the incoherent doctrine of clinical equipoiseand without conflatingthe ethics of researchwith the ethics of therapy. Disclaimer arethe views The opinionsexpressed reanddo not necessarily of the authors flect the policy of the National Institutesof Health,the PublicHealthSerof Health vice, or the U.S. Department and HumanServices.

References 1. Hypericum Depression Trial Study of Hypericum (St Group,"Effect Perforatum Disorder: John's Wort)in MajorDepressive a Randomized Controlled JAMA287 Trial," (2002):1807-1814. 2. Ibid., 1813. 3. S.M. Stahl,Essential Psychopharmacoloand Bipolar Disorder (New gy of Depression York: Press, 2000). University Cambridge 4. P.S. Appelbaum, L.H. Roth, C.W. Lidz, P. Benson, and W. Winslade,"False Hopes and BestData:Consentto Research Hastand the Therapeutic Misconception," 17, no. 2 (1987):20-24. ingsCenter Report the Ethics 5. P.S. "Clarifying Appelbaum, a PathToward Avoidof ClinicalResearch: AmerMisconception," ing the Therapeutic icanJournal 2, no. 2 (2002):22. of Bioethics Tri6. B. Freedman, "Placebo-Controlled IRB als and the Logicof ClinicalPurpose," 12, no. 6 (1990):1-6. the Concepts 7. R.J. Levine,"Clarifying of Research Center Ethics," Report Hastings 9, no. 3 (1979):21-26. 8. NationalCommissionfor the Protecand of Biomedical tion of HumanSubjects BehavioralResearch,The BelmontReport (Washington, D.C.: U.S. Government Office, 1979) p. 3. Printing 9. E.J. Emanuel, D.Wendler, and C. Grady, "What Makes Clinical Research Ethical?" JAMA283 (2000):2701-2711. 10. J. Sugarman, N.E.Kass,S.N. Goodman, P.Perentesis, P.Fernandes, and R.R.Faden, "What Patients Say About IRB20, no. 4 (1998):1MedicalResearch,"

11. EG. Miller, D.L.Rosenstein, and E.G. DeRenzo, "Professional Integrityin Clinical Research," JAMA 280 (1998):1449-54. A 12. R.R.Fadenand T.L.Beauchamp, Consent and of History Theory Informed (New York: Oxford University Press, 1986):200-232. and 13. D.A. Braunholtz, S.J.L.Edwards, Trials "Are Randomized Clinical R.J.Lilford, Evidence Good ForUs (in the Shortterm)? for a "Trial Effect," Journalof ClinicalEpi54 (2001):217-224. demiology 14. J.W. Berg, P.S. Appelbaum, ConC.W.Lidz,and L.S. Parker, Informed 2nd and Clinical sent:LegalTheory Practice, edition(NewYork: OxfordUniversity Press, 2001):280-283. and Regulation 15. R.. Levine,Ethics of Yale 2nd ed. (New Haven: Clinical Research, Press,1986):8-10. University and clinicalresearch" 16. J. Katz,"Ethics a tributeto Henry K. Beecher," revisited: Center Report 23, no. 5 (1993):36. Hastings


Ethics 17. C. Fried, MedicalExperimentation: Myths," Journalof Law,Medicine6& PersonalIntegrityand Social Policy (New 24 (1996):252-259. York: American Elsevier, 1974). 30. R. Temple and S. E. Ellenberg, Trials andActive-Con"Placebo-Controlled 18. Ibid., 5 of New Treattrol Trialsin the Evaluation 19. Ibid.,94. 1: and Scientific Part Ethical ments: Issues, 20. A.B.Hill, "Medicalethics and con133 (2000):455of nternalMedicine trolled trials,"British MedicalJournal 1 Annals 63. (1963):1043-1049. T.P. Laughren,"The Scientificand 21. C. Fried,MedicalExperimentation: 31. Trials Basis for Placebo-Controlled PersonalIntegrityand Social Policy (New Ethical an FDA and in Schizophrenia: Depression York: American Elsevier, 1974):8. 16 Psychiatry European Perspective, 22. D. Marquis, "Leavingtherapy to (2001):418-423. chance,"HastingsCenter Report13, no. 4 32. Departmentof Health and Human (1983):40-47. Services. Protection of Human Subjects. 23. Ibid.,40. Code of FederalRegulations.45CFR46, 24. F. Gifford, "The Conflict Between 1991. ClinicalTrialsand the TheraRandomized 33. K.J. Rothman and K.B. Michels, and "The Journalof Medicine peutic Obligation, ContinuingUnethicalUse of Placebo 11 (1986):347-366. Philosophy Controls," NEJM331 (1994):394-8. 25. B. Freedman,"Equipoiseand the 34. See B. Freedman,K.C. Glass, and Ethics of Clinical Research," NEJM 317 C.Weijer,"PlaceboOrthodoxyin Clinical (1987):141-145. and Methodological I: Empirical Research. 26. Ibid., 144. & Ethics 24 Journal Myths, ofLaw,Medicine 27. B. Freedman,"Placebo-Controlled (1996):243-51; and B. Freedman,K.C. in "Placebo Trialsand the Logicof ScientificPurpose," Glass,andC.Weijer, Orthodoxy Clinical Research.II: Ethical, Legal and IRB 12, no. 6 (1990):5. Myths,Journal of Law,Medicine and J.E Childress, Regulatory 28. T.L. Beauchamp, & Ethics 24 (1996):252-259. edition Ethics Biomedical 5th of Principles "PlaceandS.E.Ellenberg, 35. R. Temple (New York: Oxford University Press, bo-ControlledTrials and Active-Control 2001):323-327. in the Evaluation of New Treatments: 29. B. Freedman,K.C. Glass, and C. Trials Annals Scientific 1: Ethical and Part Issues," in ClinicalRe"Placebo Weijer, Orthodoxy search. II: Ethical, Legal and Regulatory of InternalMedicine 133 (2000):455-63;

and EG. Miller,"The Ethics E.J.Emanuel of Placebo-ControlledTrials-a Middle Ground,NEJM345 (2001):915-919. 36. S.Hellmanand D.S. Hellman, "Of of the RanMice But Not Men: Problems domized Controlled Trial, NEJM 324 (1991):1585-1589. 37. EG, Miller and H. Brody, "What Trials Unethical? MakesPlacebo-Controlled AmericanJournal of Bioethics2, no. 2 (2002):3-9. 38. N. Dickert and C. Grady,"What's the Priceof a Research Approaches Subject? New for Research to Payment Participation, England Journal of Medicine 341 (1999):198-203. andC. D. Wendler, 39. SeeE.J.Emanuel, EthMakesClinicalResearch "What Grady, ical? 283 (2000):2701-2711. JAMA 40. B. Freedman, K.C. Glass, and C.Weijer,"PlaceboOrthodoxyin Clinical II: Ethical,Legaland Regulatory Research. & Ethics 24 Journal ofLaw,Medicine Myths, (1996):253. 41. R. Temple and S.E, "Placebo-Contrialsin the Trials andActive-Control trolled Part1: Ethof New Treatments: Evaluation Annalsof Internal ical and ScientificIssues, Medicine 133 (2000):455-63. 42. C. Weijer,"WhenArgumentFails, AmericanJournal of Bioethics2, no. 2 (2002):10.



2003 May-June