placebos inclinical practice and research · virtually all patients accepted the placebo...

J7ournal ofMedical Ethics, 1996; 22: 140-146

Placebos in clinical practice and research

P P De Deyn and R D'Hooge Department ofNeurology, Middelheim General Hospital and Laboratory ofNeurochemistry and Behaviour, University ofAntwerp, Antwerp, Belgium

AbstractThe main current application ofplacebo is in clinicalresearch. The term placebo effect refers to diverse non-specific, desired or non-desired effects ofsubstances orprocedures and interactions between patient andtherapist. Unpredictability of the placebo effectnecessitates placebo-controlled designs for most trials.Therapeutic and diagnostic use ofplacebo is ethicallyacceptable only in few well-defined cases. While"therapeutic" application ofplacebo almost invariablyimplies deception, this is not the case for its use inresearch. Conflicts may exist between the therapist'sHippocratic and scientific obligations. The authorsprovide examples in neuropsychiatry, illustrating thatobjective scientific data and well-considered guidelinesmay solve the ethical dilemma. Placebo control mighteven be considered an ethical obligation but someprovisos should be kept in mind: (a) no adequatetherapy for the disease should exist and/or (presumed)active therapy should have serious side-effects; (b)placebo treatment should not last too long; (c) placebotreatment should not inflict unacceptable risks, and (d)the experimental subject should be adequately informedand informed consent given.

Placebo-controlled randomised clinicaltrialsShapiro' defined placebo as "any therapeutic proce-dure (or that component of any therapeutic proce-dure) which is given deliberately to have an effect, orunknowingly has an effect on a patient, symptom,syndrome, or disease, but which is objectivelywithout specific activity for the condition beingtreated". Since the earliest days of medical science,and certainly since the advent of chemotherapy,placebo effects have been considered more often anuisance than a therapeutic tool. Both inert andactive therapies were observed to produce effectsbeyond their predicted physiological properties.These rather surprising observations on the clinical

Key wordsMedical ethics; placebo; research; epilepsy; Alzheimer'sdisease; deontology.

consequences of the administration of placebosbecame known as "the placebo effect", comprisingthe total ofunexplained consequences of administer-ing placebos as well as active treatments. Due to,amongst other things, the unpredictability ofplaceboeffects in individuals and patient groups, placeboscame to play an irreplaceable role as "inactive"controls in randomised clinical trials (RCTs).The first rather isolated placebo-controlled

clinical trial took place in 1916 and was conductedby Macht who compared the analgesic action ofmorphine with that of physiological saline.2 It wasonly in the 1940s and 1950s that the large-scaleuse of placebos in clinical research emerged, simul-taneously with scientific knowledge pertaining to theplacebo effect. The unabated need for placebo-controlled clinical trials is illustrated by several anec-dotes in the development of presumedly therapeuticprocedures. Surgical ligation of the internalmammary artery, once proclaimed to be efficaciousfor treatment of angina pectoris, might serve as anexample. In the 1 950s, considerable relief ofsymptoms was reported for patients with anginapectoris subjected to bilateral ligation of the internalmammary artery. In the early '50s, Italian cliniciansBattezzati and colleagues3 were the first to apply thistechnique, and the Reader's Digest published anenthusiastic report.4 A year later promising resultswere also reported by American researchers whobased their enthusiasm upon data from non-con-trolled trials.5-7 It only took two double-blindplacebo-controlled studies, involving 35 subjects, todisprove the presumed efficacy of this putativetreatment. Fourteen subjects were placebo-treated(sham-operated) and 21 underwent a ligation of theinternal mammary arteries under local anaesthesia.89The placebo procedure consisted of parasternal skinincisions without ligation, while the "active" treat-ment consisted of parasternal skin incisions followedby ligation of the internal mammary artery. It wasdemonstrated that internal mammary artery ligationdid not increase cardiac muscle perfusion and hadno effect on the pathophysiology of coronary arterydisease. Although deception was obvious in thesetwo placebo-controlled trials (patients were notinformed about the possibility of sham operation),

copyright. on January 29, 2020 by guest. P

rotected byhttp://jm

e.bmj.com

/J M

ed Ethics: first published as 10.1136/jm

e.22.3.140 on 1 June 1996. Dow

nloaded from

http://jme.bmj.com/

P P De Deyn and R D'Hooge 141

the double-blind evaluation of the procedure's effectin 35 patients prevented wide-spread introduction ofthis non-efficacious surgical procedure.

Demonstration of safety and effectiveness of a

drug is a legal requirement for marketing drugs inmany countries. In the USA, evidence submitted tothe Food and Drug Administration (FDA) to meetthis requirement has to include results of "adequateand well-controlled investigations" capable of distin-guishing "the effect of a drug from other influences,such as spontaneous change in the course of thedisease, placebo effect, or biased observation".According to FDA regulation this usually impliesstandard clinical trial features such as (inactive)control groups, randomised assignment to treatment,and blinded outcome assessment. Difference inoutcome among patients, concurrently randomisedto therapy, should not be due to personal beliefs,secondary interests, and/or prejudices of patients,investigators or sponsors. FDA official Leber'"stresses the necessity of RCTs. Even thoughLeber states that "[t]here is no alternative to therandomised controlled design - no ifs, ands, or

buts!", he certainly does not take the RCT device asinfallible. In any case, open studies, case series andstudies relying on external or historical controls donot allow sound conclusions. Historical controls are

highly unreliable, especially in regard to diseaseswith poorly known or highly variable natural courses

such as insomnia, anxiety, depression and pain syn-dromes." '-3 As Leberl' argues, it is often impossibleto show improvement to be unrelated to pharmaco-logical effects of the administered agents without theinclusion of a placebo arm.

Deceptive use ofplaceboThe two major ethical problems in the use ofplacebo involve deception when used either thera-peutically or in research without adequate patientinformation and consent, and the potential conflictbetween the Hippocratic and scientific obligations ofthe therapist-researcher. Even though it remainsinadmissible to use placebo for therapeutic or diag-nostic purposes, most therapists do occasionallyapply placebo in a deceptive but rather benevolentway. Use of placebo for therapeutic and diagnosticuse is inadmissible since the patient may be withheldfrom an active treatment, may be denied his or herright to self-determination and finally, because thereis no scientific ground for the application of placeboin diagnosis. Deception is certainly unacceptablewhen applying placebo as an inactive control inclinical research.

Park and Covil4 demonstrated that deceipt maynot be necessary in therapeutic application. Theyshowed that in neurotic patients presenting signs ofanxiety, placebo could be used openly as a therapeu-tic agent. Virtually all patients accepted the placeboand improved over a one-week treatment. Park and

Covi'4 found that a number of patients even refusedto believe they had received placebo. Alternatively,in discussing placebo use with patients, one mightconsider another ethically acceptable or preferabledescription of placebo: "Placebo is a medicinewhich does not act directly through known bodilymechanisms but which may work through themind".

Although many authors'5 vividly defend con-sequentialist or utilitarian strategies and deceptive useof placebo, others point to the important unethicaland non-scientific use of placebo in so-called thera-peutic or diagnostic contexts. Goodwin et al16surveyed the knowledge of placebo action in 60 houseofficers and 39 registered nurses, and their patterns ofplacebo use. Most respondents underestimated orwere unaware of the power ofplacebo effects, and thefollowing unfortunate patterns of placebo useemerged: (a) to prove the patient "wrong" (ie, toexpose a patient thought to be exaggerating, imagin-ing or faking symptoms); (b) in disliked or "undeserv-ing" patients (for example, alcoholic, psychotic ormanipulative and demanding patients); (c) in situa-tions where standard treatments either fail or makepatients worse, (d) as a group activity to act out staffannoyance towards "difficult" patients.

Deception in the therapeutic use of placebo mightbe ethically acceptable in few well-defined cases, forexample, when the physician is dealing with subjectswith a history of substance abuse or when subjectshave to be withdrawn from certain addictive agents.Indeed, in these cases, giving placebo withoutobtaining consent of the patient, will almost invari-ably contribute to the patient's well-being, and inaddition does not imply that one withholds apossibly beneficial medical treatment. Other thera-peutic usage of placebo in non-informed patients isregrettable, ethically unacceptable and illustrative ofignorance and prejudice. It needs to be stressed,however, that malingerers or drug addicts are notrelieved more efficiently by placebo. Quite on thecontrary, most studies suggest that such patients areless likely placebo responders. Also, complaining or"manipulative" patients are no more likely torespond to placebo than patients who are well likedby the hospital staff.'7

In the scientific application of placebo, however,informed consent should be mandatory, implyingthat thorough information with regard to the ratio-nale, the design, the eventual standard therapeuticprocedures for the given disease, the randomisationprocedure as well as the chance of being randomisedto placebo should be communicated and discussed.Moreover, informed consent procedures requiresubjects to be informed about all risks and benefitsassociated with the trial and their right to withdrawat any time. Some authors disagree with this andoppose compulsory informed consent. Brewin'8argues that "too much information may be as bad astoo little". This author also suggests that some



e.bmj.com

/J M


e.22.3.140 on 1 June 1996. Dow

nloaded from

http://jme.bmj.com/

142 Placebos in clinical practice and research

investigators, once in possession of written informedconsent, might become less concerned than theyshould about their ethical responsibilities. Still otherauthors perceive an incompatibility betweeninformed consent and placebo control. Levine'9warns us about subjects who are thoroughly informedof the expected therapeutic and adverse effects ofpharmacologically active agents in placebo-con-trolled trials. In several clinical trials, subjectsunblinded the study by correctly guessing their treat-ment assignment. Levine'9 proposes a modificationin consent procedures that would eliminate or atleast reduce this possibility without defeating theethical purpose of informed consent. Levinesuggests adding irrelevant side-effects in theprovided information that are of the same order ofimportance as the actual expected side-effects. Thisis consistent with the ethical purposes of informedconsent in that it entails disclosure of materialrisks.

Another poignant issue concerns the patient'sability to understand placebo-controlled trials. Dothe patients really know what placebo means, dothey realise that they have a certain chance of receiv-ing placebo and why they will be on it? Stanley20reviewed the literature and concluded that irrespec-tive of their condition and whether or not they werepsychiatric patients, patients were fairly able tounderstand the risks and benefits of a proposedtreatment and the purpose of a particular treatmentor procedure.

Conflict between Hippocratic andscientific obligationsThe therapist-researcher is subject to two funda-mental ethical obligations, a Hippocratic one and ascientific one. The therapist's Hippocratic obligationobliges him/her to apply all existing knowledge for thebest possible treatment of individual patients. On theother hand, in agreement with the researcher's scien-tific obligations it is unethical to produce unsound sci-entific data. Thus, it is the duty of the researcher toacquire new knowledge so that future patients mightbenefit, and to communicate accurately this know-ledge to the scientific community in order to con-tribute to the collective benefit. However, these twoobligations may on certain occasions be in contradic-tion to the execution and design ofplacebo-controlledRCTs. According to certain people, randomisation byitself means that patients are no longer treated purelyfor their own good but are used at best for the benefitof future sufferers from their condition, at worstmerely to satisfy scientific curiosity, and that they riskbeing treated inappropriately. It is thought that somekind of sacrifice is being demanded of them and thatthey should either be given a full explanation or elsenot randomised.The primary question is whether it can be

ethically justified to deprive a certain percentage of

patients in placebo-controlled trials of their "right toreceive" treatment of acknowledged efficacy merelyin order to verify whether or not other active treat-ments exist. Withholding treatment of provenefficacy clearly violates the therapist's commitmentto individual patient welfare. However, the conflictbetween therapist and researcher more often is emo-tionally, rather than scientifically, based. Patientscan be allocated to standard therapy control groups,or to placebo groups when no standard therapyexists, without violation of Hippocratic commit-ment; developing new treatments of an already (par-tially) treatable disease need not necessarily bedangerous to the subject. Some therapists stress thepragmatic viewpoint of the RCT: only the applica-tion of an accepted treatment as control arm cananswer the question whether the new treatmentimproves on a standard method. The use of placebo-controlled RCT designs is the only way to minimisethe number of ineffective drugs and therapeutic pro-cedures. The conviction that ignorance may causepatient harm may not only be used as a justificationfor research but in addition renders research anethical obligation. The important ethical difficultiesassociated with the widespread application of anew treatment without a trial, and consequentlypotentially without specific effects but with varyingdegrees of side-effects, is far greater than thoseassociated with the trial itself. The optimal andtherefore often placebo-controlled and ethicallyfounded RCT meets the duties of benefiting societyand increasing knowledge without jeopardising thewell-being of the experimental subjects.

There are many examples of marketed com-pounds without scientifically demonstrated efficacyand, in addition, several companies market so-calledalternative medicines without needing to meetrigorous drug regulatory requirements for proof ofefficacy. Recently, Pope2' correctly stated that fortydifferent compounds for treatment of Alzheimer'sdisease are available in Europe without any provenefficacy. The Ginkgo biloba case is an examplewhere rigorous scientific standards of efficacy are notmet.2' 25 In 1988, 52224 million prescriptions forGinkgo biloba extract involved a cost of DM 370million to former West German health insurance. In1992, sales were comparable with total costs ofapproximately DM 368 million. The drug, licensedunder the 1986 drug law which does not requirescientific evidence from controlled trials but merelypositive therapeutic "experiences", is promoted fortreatment of disorders as diverse as peripheralarterial disease, memory impairment, vertigo andimpotence. Kimbel,25 writes that "[t]he obvious dis-crepancy that Ginkgo biloba extracts are among themost prescribed medicines in France and Germanybut not licensed in Anglo-American andScandinavian countries suggests widely differingstandards of acceptance. Would it not be advisableto base any therapeutic conclusions on the criteria of



e.bmj.com

/J M


e.22.3.140 on 1 June 1996. Dow

nloaded from

http://jme.bmj.com/


leading regulatory agencies - ie, evaluation of allpublished and unpublished studies and use of estab-lished standards for clinically relevant efficacy?"

Acceptability ofplacebo-controlledrandomised clinical trialsOne definitely needs to be fully informed aboutcurrent biomedical knowledge with regard to thedisease at hand in order to be able to make ethicaldecisions correctly. Some examples pertaining to thedevelopment of new agents against depression,schizophrenia, epilepsy and dementia will illustratethis necessity.

DEPRESSIONWith the advent of a variety of potent anti-depressiveagents, one might be tempted to conclude thatplacebo-controlled trials should be considered ethi-cally unacceptable. However, we believe that the dif-ference in improvement rates between drug-treatedand placebo-treated non-endogenous patients is notbig enough to make placebo control unethical.

Endogenous depressions have placebo responserates around 30 per cent, while neurotic or reactivedepressions have rates approaching 70 per cent.26The more severe depressions (Hamilton Depressionscore above 20) have placebo response between 30and 40 per cent, whereas those with less severeillness (Hamilton score below 14) have placeboresponse rates greater than 50 per cent. Brown et al26found no difference between baseline clinical anddemographic characteristics of placebo respondersand non-placebo responders in a large group ofpatients with major depression, illustrating thelimited predictability of placebo response. Quitkinet al 27 analysed the heterogeneity of placeboresponses in 144 depressive patients (DSM-IIIcriteria) randomly assigned to placebo medication infour double-blind antidepressant drug trials. Half ofthe patients were rated as improved for at least oneweek: 23 per cent with abrupt improvement; 27 percent with gradual improvement. Gradual improve-ment was observed later in the course of treatment,more resembled drug response, and was more likelyto persist than abrupt improvement. The authorsattributed gradual improvement to spontaneousremission.

Although placebo use may not always be advisable(for example, in depressed patients at significant riskof suicide, with psychotic features, or with severefunctional impairment), Brown's 28 suggestion toapply placebo in an open manner in mild tomoderate depression might hold some merit. Heproposes to inform the patients about the fact thattheir condition tends to respond to placebo (hedefines placebo adequately), and that in an attemptto treat them, they will receive placebo for a periodof six weeks after which the need for other treatmentwill be evaluated. The authors provide, of course,

possibilities for escape in case of deterioration of thesymptomatology and/or increased risk for complica-tions such as suicide.

SCHIZOPHRENIASeveral authors have questioned the ethicalacceptability of short-term placebo treatment ofchronic schizophrenics.29 Even though double-blindplacebo-controlled trials provide the most reliableevaluation of antipsychotic agents, some cliniciansfear that subjects exposed to placebo or inactive newagents might suffer lasting harm when the trial leadsto relapse. Sixty-six per cent of patients givenplacebo and only eight per cent of those givenfluphenzine decanoate relapsed during a trial thatshowed no differences between groups - neither inany clinical or social variable measured at the end ofseven years follow-up, nor in the number of relapsesafter the trial.29 Placebo-controlled clinical trials ofantipsychotic agents can still be acceptable on condi-tion that escape treatment is provided by applyingwell thought-through exit criteria within reasonabletime after relapse.

EPILEPSYThe first guidelines of the International LeagueAgainst Epilepsy on the development of anti-epilepticagents state that new agents cannot be introducedunless it has been proved that patients withintractable epilepsy become seizure-free or experi-ence an appreciable (for example, 25 or 50 per cent)reduction in seizure frequency.30 The placebo-controlled add-on design is the appropriate clinicaltrial to identify such agents. In this type of clinicaltrial, the patient's baseline medication is maintainedbut in addition, either placebo or the new investiga-tional drug is introduced. However, research on newanti-epileptic drugs with similar efficacy as drugsalready marketed but with fewer or no side-effectsshould also find a place. The placebo-controlledadd-on design is obviously not the best approach inthis latter case. Equally active anti-epileptic agentswith fewer side-effects might be missed since side-effects are difficult to assess in add-on designs due todrug interactions and difficulties in analysing indi-vidual drug actions. Determination of the activity ofa compound may also be impossible in add-ondesigns when the agent tested influences the bloodlevels of concomitant agents.3' Add-on trials mayoverestimate the toxicity of the test compound,especially if toxicity of the new compound is similarand additive to that of concomitant drugs. Finally,placebo-controlled add-on trials are often expectedto yield an unlikely 50 per cent reduction in subjectseizure frequency, whereas Schmidt32 found only 15per cent reduction with co-administration of amarketed drug.

Since epileptic seizures can result in serious dis-comfort or lasting disability, the sometimes moreappropriate placebo-controlled, single-drug designs



e.bmj.com

/J M


e.22.3.140 on 1 June 1996. Dow

nloaded from

http://jme.bmj.com/


face adverse ethical judgment. (Of course, certainseizure types such as primary generalised tonic-clonic seizures are imminently more dangerousthan, for example, simple focal seizures without sec-

ondary generalisation; also, patients who onlypresent seizures during sleep are less likely todevelop complications). We argue that placeboapplication in clinical epilepsy research can be con-

sidered ethically acceptable under certain condi-tions and in the following trial designs: (a)placebo-controlled add-on designs in intractableepilepsy; (b) developmental drug monotherapyversus placebo in pre-surgical video-EEG or video-invasive neuroelectrophysiological monitoring (inorder to be able optimally to observe epilepticseizures in this patient population, the baselinemedication is withdrawn in the pre-surgical work-up; following this short observation period, theefficacy of the new developmental drug can becompared against that of placebo); (c) active controldesigns with attenuated form of the active control(in this design, an active control group is used butthe administered dosage of the classical anti-epileptic agent is too low to be really effective);(d) placebo-controlled design in de novo patientspresenting with a first-ever epileptic seizure.We suggest that monotherapy designs b and c

should always be preceded by more than one

placebo-controlled add-on design indicatingprobable anti-epileptic efficacy of the test com-

pound. Monotherapy trials were proposed toovercome the deadlock of no-difference outcome,33and have actually been used in the recent develop-ment of the anti-epileptic agents vigabatrin and fel-bamate. The acceptability of the designs rests upon a

variety of factors such as type, frequency and onsettime of seizures, investigational circumstances (forexample, pre-surgery investigation of patients),preset safeguards such as escape criteria, duration oftreatment, and (tentative) anti-epileptic efficacy ofthe test drug. Preset escape criteria are mandatory indesign (b) trials. Bourgeois et al 34 used as primaryefficacy parameter the time required to reach either a

certain number of seizures or a fixed number ofseizure-free days, whichever first. After a predeter-mined number of seizures or, for example, one

(secondarily) generalised seizure, patients returnedto the preregistration treatment schedule. Thisprotocol was considered justified in the case of felba-mate because of its promising clinical profile andstrong interaction with other antiepileptics. In favourof a placebo-controlled design instead of an active-controlled design was the lack of sensitivity of thelatter.35

In design (c), the test compound is comparedagainst attenuated active control. The active control(a marketed anti-epileptic) can be administered indifferent doses (the lowest dose is not consideredefficacious and such doses should be seen as

pseudoplacebos) or at one fixed low dose. Patients

with sufficient seizure control but side-effects can beenrolled in this type of trial. Although the exact per-centage of seizure-free patients suffering side-effectsis unknown, one could still find it unethical to runthe risk of losing seizure control only in the hopethat an adverse effect would disappear. Therefore,inclusion criteria eliminating patients with severeepilepsy and escape criteria are again mandatory.Treatment failure, the primary efficacy variable inthese trials, is defined by specific exit criteriarelating to either seizure frequency or seizureseverity.

Placebo-controlled designs could be considered inde novo patients presenting with a first-ever seizure,provided that development of epilepsy is not tooimminent as it might be in patients with epileptiformelectroencephalographic elements or in those withpredisposing structural lesions. As early as 1881,Gowers noted that seizures apparently begetseizures, an observation that remains controversialup to the present.36 According to some authors butnot to others, one of the most decisive factors inlong-term seizure remission might be the number ofpretreatment seizures.3738 This consideration alsoholds for non-de novo patients in whom additionalseizures could change the prognosis of the disease.Only a trial with a placebo arm and an active anti-epileptic arm would be reliable in helping to resolvethe issue of the risk of epileptogenesis after a first-ever epileptic seizure.

ALZHEIMER'S DISEASEWith recent FDA approval of tacrine as a drug againstAlzheimer's disease, one might question the accept-ability of placebo-controlled clinical trials. For severalreasons we argue that placebo-controlled clinicaltrials in dementia are not only acceptable but ethicallyand scientifically preferable: (a) cholinesteraseinhibitors such as tacrine cannot nearly be calledstandard therapy as only 23 per cent of Alzheimerpatients benefit from tacrine treatment39; the clinicalrelevance of the therapeutic efficacy, at best onlysymptomatic and very limited, could even bedoubted; (b) cholinesterase inhibitors are not devoidof side-effects since some 30 per cent of patientsdevelop transaminitis, a reversible increase in livertransaminases, and some 15 per cent develop gas-trointestinal complaints39 40; (c) placebo treatmentdoes not inflict unacceptable risk on the patient suf-fering from this slowly progressing neurodegenerativedisorder; (d) investigating new anti-dementic agentsin cholinesterase inhibitor-controlled designs couldlead to false positive findings (type II errors) whenapplying equipotency or no-difference outcomecriteria; (e) cholinesterase inhibitor-controlleddesigns would require larger patient populations inorder to reach sound conclusions: (f) cholinesteraseinhibitor treatment as an active control arm woulddefinitely unblind the study because of the highfrequency of transaminitis.



e.bmj.com

/J M


e.22.3.140 on 1 June 1996. Dow

nloaded from

http://jme.bmj.com/


ConclusionSince it is inadmissible to perform ill-designedclinical trials and to market compounds or employtreatments without specific effect (efficacy notexceeding that of placebo) and/or with serious side-effects, properly controlled RCTs form the onlyscientifically valid tool. Nature of the diseaseprocess, duration of the study period, therapeutic-toxic ratio of the agent tested, availability and appro-priateness of alternative therapy, and many otherconsiderations all play a role in clinical trial design.Even though the placebo-controlled RCT remainsthe gold standard in therapy development, the needfor and acceptability of placebo control has to beevaluated case by case, considering and reconcilingboth scientific and ethical issues. Often, placebocontrol might even be considered an ethical obliga-tion but some provisos should be kept in mind: (a)no adequate therapy for the disease should existand/or the (presumed) active therapy should haveserious side-effects; (b) placebo treatment shouldnot last too long; (c) placebo treatment should notinflict unacceptable risks on the patients, and (d) theexperimental subject should be adequately informedand informed consent given.

Peter P De Deyn, MD, PhD, MMPR, is Professor ofNeurology in the Department of Neurology at theUniversity of Antwerp. Rudi D'Hooge, MA, MSc,PhD, is Research Fellow in the same department.Address for correspondence: Universiteitsplein 1, 2610Welregk-Antwerp, Belgium.

References1 Shapiro AK. Factors contributing to the placebo effect.Their implications for psychotherapy. American Journalof Psychotherapy 1961; 18: 73-88.

2 Delay J, Pichot P. Medizinische Psychologie. Stuttgart:Thieme, 1973.

3 Battezzati M, Tagliaferro A, de Marchi G. La legaturadelle due arterie mammarie interne nei disturbi di vas-colarizzazione del miocardio: nota preventiva relativade primi dati sperimentali e clinici. Minerva Medica1995; 46: 1178-88.

4 Ratcliff JD. New surgery for ailing hearts. Reader'sDigest 1957; 71: 70-3.

5 Brill IC, Rosenbaum WM, Rosenbaum EE, Flanery JR.Internal mammary artery ligation. Northwest Medicine1958; 57: 483-6.

6 Harken DE. Clinical conference: long-term manage-ment of patients with coronary artery disease.Circulation 1958; 17: 945-52.

7 Kitchell JR, Glover RP, Kyle RH. Bilateral internalmammary artery ligation for angina pectoris. AmericanJ7ournal of Cardiology 1958; 1: 46-50.

8 Cobb LA, Thomas GI, Dillard DH, et al. An evaluationof internal-mammary-artery ligation by a double-blindtechnique. New England Journal ofMedicine 1959; 260:1115-8.

9 Diamond EG, Kittle CF, Crockett JE. Evaluation ofinternal mammary artery ligation and sham procedurein angina pectoris. Circulation 1958; 18: 712-3.

10 Leber P. Is there an alternative to the randomized con-trolled trial? Psychopharmacology Bulletin 1991; 27: 3-8.

11 Lasagna L. Testimony given at Hearings before theSubcommittee on Antitrust and Monopoly of theCommittee on the Judiciary, United States Senate,87th Congress, 1st Session S. Res 52 on S. 1961; 1552:Part 1: 282-3.

12 Leber PD. The implicit assumptions of active controltrials (a critical examination). Controlled Clinical Trials1983; 14: 133.

13 Leber PD. Hazards of inference: the active controlinvestigation. Epilepsia 1989; 30: S1: 57-63.

14 Park LC, Covi L. Nonblind placebo trial. An explo-ration of neurotic patients' responses to placebo whenits inert content is disclosed. Archives of GeneralPsychiatry 1965; 12: 336-45.

15 Elander G. Ethical conflicts in placebo treatment.J7ournal ofAdvanced Nursing 199 1; 16: 947-5 1.

16 Goodwin JS, Goodwin JM, Vogel AV. Knowledge anduse of placebos by house officers and nurses. Annals ofInternal Medicine 1979; 91: 106-10.

17 Lasagna L, Masteller F, von Felsinger JM, BeecherHK. A study of the placebo response. American JournalofMedicine 1954; 16: 770-9.

18 Brewin TB. Consent to randomised treatment. Lancet1982; ii: 919-21.

19 Levine RJ. The apparent incompatibility betweeninformed consent and placebo-controlled clinical trials.Clinical Pharmacology and Therapy 1987; 42: 247-9.

20 Stanley B. Informed consent in treatment and research.In: Weiner JB, Hess AK, eds. Handbook of ForensicPsychology. New York: John Wiley & Sons, 1987:63-85.

21 Pope K. FDA chief answers critics. Academic Focus1992; 28: 28a-d.

22 Schonhofer PS, Schulte-Sasse H, Manhold C,Werner B. Sind Extracte aus den Blattern desGinkgobaumes bei peripheren Durchblutungs- undHirnleistungsstorungen im Alter wirksam? Beurteilungvon Tebonin und Rokan. Internal Praxis 1989; 29:585-601.

23 Schonhofer PS. Ginkgo biloba extracts. Lancet 1990; 1:788.

24 Kleijnen J, Knipschild P. Ginkgo biloba. Lancet 1992;340: 1136-9.

25 Kimbel KH. Ginkgo biloba. Lancet 1992; 340: 1474.26 Brown WA. Predictors of placebo response in depres-

sion. Psychopharmacology Bulletin 1988; 24: 14-7.27 Quitkin FM, Rabkin JG, Stewart JW, et al.

Heterogeneity of clinical response during placebo treat-rnent. American Jrournal of Psychiatry 1991; 148: 193-6.

28 Brown WA. Placebo as a treatment for depression.Neuropsychopharmacology 1994; 10: 265-9.

29 Curson DA, Hirsch SR, Platt SD, et al. Does short termplacebo treatment of chronic schizophrenia producelong term harm? British Medical J7ournal 1986; 293:726-8.

30 Meinardi H. Clinical evaluation of antiepileptic drugs:Guidelines of the International League AgainstEpilepsy. In: De Deyn PP, D'Hooge R, Schafer A, eds.The ethics ofanimal and human experimentation. London:John Libbey, 1994: 279-85.

31 Theodore WH, Rauberras RF, Porter RJ. Felbamate: aclinical trial for complex partial seizures. Epilepsia 1991;32: 392-7.

32 Schmidt D. Two antiepileptic drugs for intractableepilepsy with complex partial seizures. Jrournal of



e.bmj.com

/J M


e.22.3.140 on 1 June 1996. Dow

nloaded from

http://jme.bmj.com/


Neurology, Neurosurgery and Psychiatry 1982; 45:1119-24.

33 Gram L, Schmidt D. Innovative designs of controlledclinical trials in epilepsy. Epilepsia 1993; 34 S7: 1-6.

34 Bourgeois B, Leppik JF, Sackellares JC. Felbamate: adouble-blind controlled trial in patients undergoingpresurgical evaluation of partial seizures. Neurology1993; 43: 693-6.

35 Pledger GW, Kramer LD. Clinical trials of investiga-tional antiepileptic drugs: monotherapy designs.Epilepsia 1991; 32: 716-21.

36 Elwes RDC, Johnson AL, Reynolds EM. The course ofuntreatable epilepsy. British Medical Journal 1988; 297:948-50.

37 Beghi L, Tognoni G. Prognosis of epilepsy in newlyreferred patients: a multicentre prospective study.Epilepsia 1988; 29: 236-43.

38 Reynolds EH. The influence of antiepileptic drugs onthe natural history of epilepsy. Epilepsy Research 1991;S3: 15-20.

39 Knapp MJ, Knopman DS, Solomon PR, et al. A 30-week randomised controlled trial of high-dose tacrinein patients with Alzheimer's disease. J7ournal of theAmerican Medical Association 1994; 271: 985-91.

40 Watkins PB, Zimmerman HJ, Knapp MJ, et al.Hepatotoxic effects of tacrine administration in patientswith Alzheimer's disease. Jtournal of the AmericanMedical Association 1994; 271: 992-8.

News and notes

First International Conference on DNA Sampling

This first international conference on DNA samplingwill be held in Montreal, Quebec, Canada fromSeptember 6-8 this year.The conference will provide a forum for interdiscipli-

nary discussion on: DNA sampling and banking;patenting and commercialisation; legal status ofhumangenetic material and information; models of consentand confidentiality; policy and ethical concerns; andgenetic epidemiology and diversity.

The conference is being organised by the ResearchCenter in Public Law (CRDP), Faculty of Law,Universite de Montreal, in collaboration with theHealth Law Institute, University of Alberta, QuebecNetwork of Applied Genetic Medicine, Quebec HealthResearch Fund.For information contact: Ms Samaa Elibyari, Tel:

(514) 343-2142, Fax: (514) 343-7508, e-mail:genet(crdp.droit.umontreal.ca.



e.bmj.com

/J M


e.22.3.140 on 1 June 1996. Dow

nloaded from

http://jme.bmj.com/

placebos inclinical practice and research · virtually all patients accepted the placebo...

Documents