Ethical Dilemmas in Medical Innovation and Research: Distinguishing Experimentation From Practice Norman Fost
The words research and experimentation continue to have the power to evoke fear in potential subjects. But much of standard practice, particularly in critical care settings, involves interventions of unknown efficacy and safety. Innovation also abotmds in practice settings, typically unchecked by prospective or retrospective review. Historical attention has focused on the conflict of interest of the physician/investigator, but contemporary safeguards have reduced the risks of research and increased the likelihood that the patient/subject will have the opportunity to make an informed choice. Innovation and untested interventions in practice, in contrast, are often unknown to the patient and lack institutional safeguards. Some common cliches that suggest that research is more to be feared than standard practice will be examined, leading to the following conclusions: defining an intervention as experimental may be less important in ethical terms than the quality of prospective and retrospective review and the standards for informed consent, and the concerns that led to regulation of research should now be directed toward unproven interventions and innovation in the practice setting. Copyright 9 1998 by W.B. Saunders Company
As long as you promise not to learn anything from what you're doing, you don't have to go through an institutional review board.
I need permission to give a new drug to half my pa- tients but not to give it to all of them.
"When I use a word, " Humpty Dumpty said, in rather a scornful tone, "it means just what I choose it to mean--neither more nor less." "The question is, "said Alice, "whether you can make words mean so many different things." "The question is," said Humpty Dumpty, "which is to be master--that's all."
--Lewis Carroll 3
l ' f all you knew about human experimentation was what you read or heard in the mass me- dia, you might think that clinical research in- volves ethical problems that are more common, more serious, or more worrisome than standard practice.
There are several possible sources of such a viewpoint. Much of the modern bioethics move- ment had its origins in concerns about unethical experimentation involving human beings, begin-
ning with the disclosures at the Nuremberg tri- als. 4 Beecher 5 and others increased awareness of egregiously unethical research involving patients in the United States in the 1950s and 1960s. In 1973, hearings by Senator Ted Kennedy brought the infamous Tuskegee study 6 to public attention as well as other upsetting, if not unethical, stud- ies, such as studies on midtrimester fetuses kept alive after elective abortions for the purpose of nontherapeutic research. 7 Many of these exam- ples involved interventions of no conceivable or intended benefit to the subjects, confirming popular impressions that to be a research subject means to be used as a guinea pig, or to use Kant's more elegant phrase, being treated solely as a means.
Although the present federal rules and insti- tutional review boards have greatly reduced the chance of seriously unethical research, occa- sional scandals and further exposition of old scandals keep fear alive. The activities and report of the "Radiation Commission ''s received exten-
From the Departments of Pediatrics and History of Medicine, and the Program in Medical Ethics, University of Wisconsin School of Medicine, Madison, WI. Address reprint requests to Norman Fost, MD, MPH, Department of Pediatrics, University of Wisconsin Hospital, 600 Highland Ave, Madison, WI 53792. Copyright 9 1998 by W.B. Saunders Company 0146-0005/98/2203-0007508.00/0
Seminars in Perinatology, Vol 22, No 3 (June), 1998: pp 223-232 223
224 Norman Fost
sive press coverage, although most of the activi- ties occurred decades ago, before the current system of protections was in place. Legitimate controversies about research design, such as the use of placebos, are compared with the Tuskegee study, 9 even though there are few morally rele- vant points of comparison, a~ A recent change in the federal rules allowing consent to be waived in special situations in which experimental treat- ment may benefit unconscious patients in emer- gency settings was compared with the Nazi exper- iments) 2
Whether or not there are more serious or more common ethical problems in practice or research settings is partly an empirical question, but it also involves questions of definition. How we define words such as experimentation would seem critical to assessing how common it is, as well as a prerequisite to controlling it. The pur- pose of this article is to review some of the diffi- culties in defining and distinguishing experi- mentation, research, and standard practice by identifying and responding to several common misconceptions on the topic. But first, a few gen- eral observations about the limits of definitions in bioethics.
The Limits o f Def init ions
There is a tradition in the brief history of bio- ethics to appeal to definitions to solve complex moral problems. For many years, debates about withholding or withdrawing life-sustaining treat- ment, for example, focussed on whether or not the treatment involved extraordinary means. If it did, the argument went, there was no moral duty to treat. The problem, of course, was that the definition of extraordinary was elusive and not amenable to consensus. It also became ap- parent that some were using their own defini- tions as a short cut to justify actions, avoiding the difficult work involved in finding justifica- tions. In the early years of the "Baby Doe" de- bate, physicians reported that they commonly withheld standard treatment, such as repair of duodenal atresia, from infants with Down syn- drome, claiming that such treatment involved using extraordinary means, la The same physi- cians would argue that the use of the same treat- ment was quite ordinary, and obligatory, if the child did not have Down syndrome and was oth- erwise normal. 14 It was apparent that the means was not the morally relevant factor, but some-
thing about the intelligence and presumed qual- ity of life of the patient.
The same problem surfaces recurrently in the reliance on fetal viability as a morally relevant variable in judgments about the justifications for abortion and other prenatal and neonatal inter- ventions. 15 The concept is central to the land- mark United States Supreme Court decision in Roe V Wade) 6 But the court did not give us a clear definition of viability, stating only that it referred to a point at which the fetus "has the capability of meaningful life outside the mother's womb." This definition, of course, avoids the serious dis- cussion regarding what is meant by "meaningful life," as well as the ambiguity in what is meant by " the" mother. New reproductive technologies allow fertilized eggs to survive outside of the mother 's womb, if "mother" refers to the woman whose egg became fertilized, suggesting that all fertilized eggs are viable and therefore protectable by the state's important interest in potential life. My point is only that the court's simple definition of viability neither settles the complex moral questions nor even provides clear legal guidance.
A more contemporary example is the interest in defining some medical interventions as futile as a way of avoiding the difficulty involved in deciding whether treatment is owed to a patient. Although physicians are competent and war- ranted in making judgments regarding whether a treatment would be physiologically futile, ie, has no reasonable prospect of achieving its physi- ological purpose, the word is more commonly used to justify withholding treatments that are likely to be effective but are considered inappro- priate for social, economic, or other reasons. This kind of justification is not a matter of medi- cal expertise, and therefore cannot be defined by a physician) 7
The longstanding debate about distinguish- ing experimental care from standard practice is in this tradition. Whether a proposed interven- tion is experimental or not has important impli- cations for several policy questions: Must it be reviewed by an institutional review board? Should third-party payers offer reimbursement? What are the standards for informed consent? Many commentators have wrestled with defini- tions of key words--exper imentat ion, research, practice, and so on- - in the hope that clear definitions would make a major contribution to resolving some of the moral, legal, and policy
Ethical Dilemmas in Medical Innovation 225
questions. Although such analysis can be helpful in understanding the issues at stake, many of the critical questions cannot be resolved by defini- tions. I will try to illustrate this by analyzing a series of common myths or misunderstandings.
Definitions cannot do moral work, but they are necessary for communication. The defini- tions proposed by Levine, TM drawing mainly from standard dictionaries, seem useful: To experi- ment means to test something or try something out. Paraphrasing Moore 19 and Levineff much of standard practice involves experimenting. For example, a physician who sees a patient with a sore threat conducts an experiment: He swabs the throat, transfers the material to an agar dish, incubates it, and sees what grows. Depending on the result of that experiment, he may conduct another, involving an antibiotic, to see if it im- proves the patient's condition, and so on. All of this experimentation is done primarily in the interests of enhancing the well-being of the pa- tient.
Research, in contrast, is done primarily for the purpose of developing generalizable knowl- edge for the benefit of others. Research usually involves experimentation, as when the investiga- tor does something to the patient, such as admin- istering a new test or treatment. It need not in- volve new treatments; it may involve comparing standard treatments. Or it may involve interven- tions with no expected benefit to the patient, as in so-called nontherapeutic research studies, designed to advance knowledge about normal or pathological states. And research need not involve doing anything to the patient, as in medi- cal record review or studies of unidentified sam- ples.
Although these definitions are sensible, many use the word experimentation when referring to research as defined above, and use the word practice to refer to interactions that are intended to benefit the patient with no regard to the ad- vancement of knowledge. Despite the impurity of these uses, I will generally use the words exper- imentation and research as they are commonly used, to refer to activities that are intended to advance knowledge, and practice and treatment to refer to diagnostic and therapeutic activities designed primarily for the benefit of the patient.
With that as background, let us now turn to some popular truisms about claimed distinctions between research and experimentation, on the one hand, and standard practice on the other.
Alleged Truisms About the Distinction Between Experimentation and Treatment
1. Research involves trying new interventions of unknown benefit and risk; standard practice involves tested procedures with known benefits and risks.
Much of standard practice involves interventions that have never been adequately tested for safety or efficacy. It has been estimated that 80% of drugs prescribed by pediatricians have never been systematically studied in children. It has been a longstanding concern that clinical trials of new drugs generally exclude women, appar- ently because of concerns of pharmaceutical companies over possible fetal effects. The result is that drugs are approved and then prescribed and become part of standard practice without adequate information about appropriate doses or adverse effects on women or fetuses. Not only may drugs be used on populations in whom they have never been studied; they can be used for clinical purposes that have never been studied on any population. The Food and Drug Adminis- tration (FDA) does not prohibit or discourage physicians from using drugs for unapproved pur- pose. The FDA regulates the suppliers, not the prescribers, and accordingly restricts marketing practices, not use.
The use of unproven interventions is partic- ularly problematic in settings where the most seriously ill patients are treated; namely, emer- gency rooms and intensive care units. Some systematic reasons why clinical research has historically been less likely to occur in these settings are the pressure of clinical work leaves less t ime for clinicians to pursue scholarly in- terest, physicians attracted to these clinical spe- cialties may be more incl ined to action than study, and the imminence of death and the urgent nature of decisions makes it more diffi- cult to contemplate unfamil iar interventions or placebo-control led trials. Until recently, federal rules allowed little leeway for research in emergency settings when informed consent could not be obtained f rom the pat ient or his or her representatives.
Whatever the reasons, the history of neona- tology is replete with interventions that be- came standard practice over many years with- out adequate testing of safety or efficacy and were later found to cause harm without com-
2 2 6 Norman Fost
pensating benefits. 2~ A few examples will make the point.
A. Sulfonamides were given to many newborns for treatment or prevention of serious infec- tions before studies showed that the drug in- creased the risk of free bilirubin diffusing into the brain, causing kernicterus. 22
B. Chloramphenicol was also widely used for treatment of serious infection before studies showed that babies were dying of the "gray baby syndrome," a consequence of the in- ability of the immature liver to metabolize the drug. 2a
C. Exchange transfusion was performed on tens of thousands of otherwise healthy full-term newborns over a period of 30 years because of the unproven assumption that a serum bili- rubin above 20 mg% was associated with the risk of kernicterus. Many infants died of un- necessary procedures, and untold numbers developed complications, including transfu- sion-acquired infections. It is now clear that most of these infants were at no risk for devel- oping kernicterus. 24
D. Concentrated bicarbonate solutions were given intravenously to premature infants af- flicted with hyaline membrane disease to counter the respiratory acidosis brought on by carbon dioxide retention. Odell argued in vain that such therapy made no physiological sense, because the bicarbonate would bind with free hydrogen ions, and produce H2coa, which would dissociate into H~O and CO2, which could not be dissipated through the lungs. Despite animal studies showing the fu- tility of such treatment, the "Usher regimen" of administering concentrated solutions of sodium bicarbonate was standard practice for 15 years or more until Simmons showed that concentrated bicarbonate caused fluid shifts resulting in intracerebral hemorrhage. 25 The practice has been largely abandoned.
E. Oxygen inhalation for hypoxic newborns was used for decades without regard to dose be- cause it was thought unnecessary to assess toxicity for a naturally occurring substance, until Patz established the association between oxygen therapy and retrolental dysplasia.
F. Newborn screening for phenylketonuria (PKU) was mandated by legislation in most states without adequate studies assessing the
toxicity and efficacy of the screening test and the phenylalanine-restricted diet. The result was failure to appreciate that many infants falsely identified as having PKU in fact had a benign form of hyperphenylalaninemia and were exposed to toxic, and sometimes lethal, restrictions of an essential amino acid. The result was general protein malnutrition, caus- ing severe brain damage in otherwise normal infants, and occasional deaths due to a kwashiorkor-like s ta te . 26
In summary, many widely accepted standard practices involve interventions associated with considerable harm and little benefit. Although research interventions commonly involve new and completely untested practices, that is not always the case. Phase III randomized controlled trials may involve comparing two standard treat- ments to determine which, if either, is superior. The patient may not be exposed to any greater risk than would be the case if he or she were receiving standard care. In some settings, being a research subject may involve less risk than would be the case if the same treatment were being administered under the rubric of standard practice, because the monitoring for adverse ef- fects and the oversight from the review process may lead to more careful patient follow-up.
It should be noted that clinical trials involving unapproved drugs will generally present the pa- tient with much higher prospects for risk than benefit. The great majority of drugs that enter phase I testing never become approved.
2. Whether or not an intervention is research or standard practice is an objective question. The correct classification can be determined by answering a series of questions such as whether or not there is an intent to advance knowledge, whether patients will be randomized, whether there is an intent to disseminate the results, or whether the intervention deviates from standard practice.
As chair of an Institutional Review Board (IRB), I am frequently asked whether a proposed inno- vative procedure requires review and approval, ie, whether it is research. It would be helpful to have a litmus test that could resolve the question, but the answer is often elusive. One common approach relies on the intent of the physician. If his or her purpose is to help the patient, even
Ethical Dilemmas in Medical Innovation 227
if using an innovative, untested, or unproven technique to do so, then he or she is engaging in the practice of medicine. As stated most pun- gently by Lietman, 1 as long as the physician promises not to learn anything from what he is doing, he does not need to submit his plan to the IRB.
This reliance on intent is problematic because it relies on the integrity of the physician/investi- gator, whose potential conflicts at interest are at the root of the need for external review. Most physicians are presumably honest and want to do the right thing, and will also generally defer to IRB review for their own legal protection. But the occasional outlyer can avoid prospective peer review by claiming that he seeks only the welfare of his patients, acknowledging that he may not be able to avoid learning from his experiences. He may acknowledge that if a review of his records later uncovers some pattern or broader interest, he may disseminate such results through a presentation at a scientific meeting or a scholarly publication. These activities, particu- larly review of the medical records, might consti- tute research, but not of the sort typically requir- ing IRB review or informed consent from the patients. The would-be investigator could also argue that the future possibility of chart research does not change the nature of the innovative intervention, which was intended for the benefit of the patient and not for the creation of new knowledge.
To avoid these ambiguities, some prefer a more objective test of whether an intervention is research, such as whether or not the interven- tion deviates from standard practice or whether patients are being randomized. Deviation from standard practice is obviously not a sufficient rea- son to classify an intervention as research, be- cause much of standard practice is innovative. The Food and Drug Administration (FDA) regu- lations explicitly allow approved drugs to be used for unapproved purposes if the intent is to bene- fit the patient. I f the defining feature of research is the search for new knowledge, knowing that a physician is deviating from standard practice does not tell us whether he is doing so for the sole benefit of the patient before him, or for the benefit of future patients.
Using randomization as a basis for deciding what treatment will be offered or given will gen- erally be a valid indicator that there is a major research motive, but in rare instances a physician
may justifiably claim that he is randomizing treat- ments for some other reason. Such an occasion arose when the director of an infertility clinic was concerned about fairness in allocating a scarce resource--namely, fresh semen- - to women seeking artificial insemination from a donor other than their partners. At the time, fresh se- men was considered to be more effective than frozen semen, but there was no scientific evi- dence for this. To avoid bias and unfairness in distributing what he thought was a valuable ser- vice, the physician proposed randomizing fresh versus frozen semen, with each woman (with ap- propriate consent) given the opportunity to re- ceive fresh semen in alternate cycles. The physi- cian realized that randomization would open the possibility for medical record review and the pos- sible discovery of relative benefit in a way that would be publishable and of benefit to future patients, but he claimed that his purpose was not research but only to treat his patients in a fair and equitable way.
He eventually published a landmark paper on the relative efficacy of fresh versus frozen se- men. 27 He was later sued by a woman who was exposed to semen from a donor who became HIV positive some months after he had donated semen. He was HIV negative on the day of dona- tion, but in retrospect was possibly infected at the time. The suit was not based on negligence with regard to HIV testing, because the physician was following published standards of practice at the time, but was based on the claim that she was a research subject without having been in- formed of that, and that the physician had failed to comply with laws requiring institutional review of clinical research. He argued at his trial that there was no intent to do research, and that ran- domization did not establish such an intent, and the jury supported his view.
There is one other criterion sometimes used to determine whether an intervention is experi- mental, namely, whether it is reimbursed by third-party payors. This is problematic in two re- gards. First, it is tautological, because insurers need not, and usually do not, provide a coherent explanation of how they determine that some- thing is experimental. Second, to the degree that one can discern the underlying basis, it does not seem to be coherent. Sometimes the insurer seems to base the definition on the likelihood of a good outcome; eg, if a treatment cured a given cancer 5% of the time, it would be consid-
228 Norman Fost
ered experimental, even if it had been proven in a well-designed trial to have that benefit, and even if it had been standard practice for an ex- tended time. In these cases, the insurer appears to be making a judgment on some other grounds, such as cost versus effectiveness, and using the word experimental as a cover. The ab- sence of a coherent definition may account for the failure of insurance companies to withstand legal attacks on their policies. 2s
My own view is that there is no correct defini- tion, but that the essential element, the factor that best captures most of the ethical concerns, has to do with the intention of collecting infor- mation for the benefit of others. It is therefore not helpful to refer to highly innovative, un- tested changes in health care delivery, such as managed care organizations, as "massive experi- ments." They may be unwise or immoral for a variety of reasons, but it is not because human beings are being used to advance knowledge. As I will conclude, there are good arguments for studying innovations like this in formal ways. Such research projects are less likely to be mor- ally problematic than unstudied innovations.
3. Research is more tightly regulated than practice because experimental interventions raise more serious ethical issues than standard practice; being a research subject is generally more risky than being a patient.
Academicians are sometimes understandably perplexed that clinical research is so tightly regu- lated as compared with standard practice. They point out that the research subject is protected by multiple layers of oversight, including the competit ion and review process for funding of clinical studies; federal regulations, with their high standards for institutional review and in- formed consent; the scrutiny of the local IRB; the review of submitted manuscripts; and the open discussion of results when they are pub- lished. In addition, the investigator is likely to work in an academic center, surrounded by knowledgeable and critical colleagues, and in- quisitive residents and students, and is more likely to be aware of current developments than his counterpart engaged in pure clinical prac- tice. The practitioner, in contrast, may have little oversight of his activities, whether they be un- tested standard practices or adventures in inno- vative treatment. The major protection for the patient in the practice setting seems to be the
tort liability system, which most injured patients do not use because they are unaware that they may have been a victim of negligence. Why then is clinical research so much more tightly regu- lated than standard practice?
One reason is the historical discovery of egre- giously unethical conduct by physician-investiga- tors in the Nazi camps during World War I I, 4'29 and later studies of alarming misconduct by American researchers. 5 The latter disclosures in particular suggested that physicians who took ex- emplary care of their patients when acting in a clinical role lost their moral compass when they took on the role of researchers. Examples in- clude the studies of Dr Chester Southam in 1963 at the Jewish Chronic Disease Hospital in New York, in which he injected cancer cells into chronically ill patients, without their knowledge or consent, to learn about the role of the im- mune system in the development of clinical can- cer. ~~ Other studies, revealed by Beecher and others, include the dispensing of placebos to women who thought they were receiving birth control pills.
These examples suggested that otherwise hu- mane, sensitive, caring physicians ceased to act in a fiduciary role when they took on the role of researcher. The conflicting interests--serving the needs of the present patient as well as those of future possible patients and the wider social good- -apparent ly distracted the physician from his otherwise single-minded attention to the in- terests of the patient before him.
Another reason for the relatively greater regu- lation of research is because it is a discretionary activity. Although tight regulation of research may inhibit or slow down the rate of progress, with the consequence of suffering and death that would otherwise not occur, progress, it was ar- gued, was optional. 31 Causing harm is worse than failing to do good; virtually all societies have strict prohibitions against causing harm, whereas promotion of the good or the advancement of knowledge are seen as optional activities.
Finally, research is tightly regulated because it can be. The requirements for funding, as well as the need for FDA approval of new drugs and devices, presents the would-be researcher with a toll gate through which he or she must pass be- fore he or she can use a patient as a subject. IRBs can review all proposals for clinical re- search. Clinical care, in contrast, is so extensive that such oversight is logistically impractical. Pro-
Ethical Dilemmas in Medical Innovation 229
spective review and approval for all clinical inter- ventions is impossible. Thus, the need to rely on more general forms of oversight, such as peri- odic reviews of privileges and recertification, and reliance on posthoc corrections, such as mal- practice litigation and suspension of licensure.
These are historical, practical, and partly philosophical reasons for the tight regulation of research. But the protest of the academician is probably correct, ie, there are almost certainly more frequent and more serious ethical trans- gressions in the practice setting than in the re- search setting. This is partly because of the suc- cess of the existing research regulations, and also because of the profoundly greater volume of clinical practice.
Nonetheless, public fears of being a research subject seem to exceed the fears of being a pa- tient. Transgressions in the research arena con- tinue to attract the interest of the mass media in a way that seems disproportional to the magni- tude of the relative problems. Thus, a Presiden- tial Commission 8 exploring injuries to patients exposed to radiation mostly several decades ago, before the establishment of the current research regulations, evoked widespread press coverage. An apology to the victims of the admittedly un- ethical Tuskegee study also attracted major press attention. In contrast, there has been no compa- rable attention to the presumably much larger number of people who have been unknowingly injured by malpractice, misconduct, or other transgressions in the practice setting.
4. Research differs from standard practice in that the physician has serious conflicts of interest.
One of the more remarkable observations in the history of research involving human subjects is how sensitive, caring physicians can apparently undergo a Jekyll-Hyde transformation when act- ing as a researcher. Beecher's classic report was remarkable in the number as well as the outra- geousness of some of the studies. 5 One of the early questions was how could a physician who was otherwise widely acknowledged to embody the best qualities of respect and caring for his patients become transformed when he assumed the role of researcher. One explanation is that the physician/researcher is no longer just a phy- sician, whose sole purpose is to help the patient before him, but also a scientist with another larger constituency, namely, the larger group of
future patients who can benefit from his work as a scientist. The physician/researcher is a double agent, and is at risk of seeing the potential bene- ficiaries of his research as a more important cli- ent, in part because there are so many more of them.
There are personal benefits, as well, for the successful researcher. These include material gains, including academic promotion, lecture fees, and invitations to travel to meetings in ap- pealing locales. For some, career survival may be at stake, because funding for one's basic salary may depend on obtaining grants, which depends on showing success as an investigator. Success does not necessarily mean showing that a new drug is more effective than the old, but it does mean recruiting adequate numbers of patients into the study to allow for meaningful conclu- sions. It is this pressure to enroll subjects that may tempt the physician to understate the risks, exaggerate the benefits, or attempt interventions beyond those he would consider if his economic welfare were not at stake.
These potential conflicts of interest do not always materialize in ways that are adverse to the present patient's interests. Indeed, these other interests of the investigator are often congruent with, not in conf ict with the patient's interests. Many patients, particularly those with untreat- able conditions, want to be research subjects and are pleased that there are rewards and incentives for doctors to be researchers.
Conflicts of interests are not unique to re- searchers, of course. The physician involved in standard practice may also be distracted from his patient's welfare by a variety of potential con- flicts of interests, including incentives to provide unnecessary and dangerous care in fee-for-ser- vice systems, or providing too little care in capi- tared systems. These pressures are increasing in both the private sector and in academia, as man- aged care organizations bring more "bottom line" thinking to clinical practice, and as aca- demic departments become more dependent on clinical income for their survival. "You eat what you kill" is becoming a familiar expression among physicians who once relied on guaran- teed salaries to protect them from financial con- flicts in making clinical decisions.
Whether or not the research subject or the patient is in a more precarious position because of the potential conflicts of interest of the physi- cian is not self-evident. It depends on the facts
230 Norman Fost
of the case. That is why many patients, women and members of ethnic minorities as an exam- ple, may complain as bitterly about being ex- cluded from some research projects as they do about being included in others. It depends on their clinical situation, the benefits and risks of standard treatment, the particularities of the physicians involved, their financial resources, and many other variables.
5. Barriers to experimentation protect patients from harm.
The ability of the lay press and politicians to generate high levels of anxiety by exposing al- leged misconduct in experimentation suggests that the public is generally more fearful of exper- imentation than of standard practice. This may be partly because of the failure to distinguish so-called therapeutic research, studies that offer some possibility of benefit to the subject, from nontherapeutic research, studies that offer no conceivable medical benefit to the subject. The latter, particularly when strict standards of con- sent are not followed, feed the image of the re- search subject as guinea pig, exposed to discom- fort and risk with no possible benefit.
There is a more common misconception about the relationship between experimentation and harm, namely, that a typical patient has more to fear from experimental than standard treatment. As noted above (Truism No. 1), stan- dard practice, particularly in settings where the sickest patients present, may be dominated by untested, unproven interventions with risks that may greatly exceed benefits. The practicing phy- sician is not only free to use age-old unproven treatments, he or she is also free to innovate, using approved drugs for unapproved purposes, with none of the checks and balances that sur- round the explicit researcher.
The assumption that patients exposed to ex- perimentation were more likely to suffer harm seemed to be at the root of much of the objec- tion to recent changes in the federal rules involv- ing a waiver of consent for research studies in emergency settings. In 1996 the FDA and DHHS regulations were modified to allow research stud- ies to include patients unable to consent, and without relatives able to consent, under specified conditions; namely, a life-threatening condition for which existing treatments are unsatisfactory, and for which there is a promising experimental treatment. ~2
One of the more common examples of such a condition is severe head injury, affecting hun- dreds of thousands of people each year, with a high incidence of poor outcomes. Although FDA rules allowed unapproved drugs and devices to be used without consent in such a situation, the justification had to be based on the need of the patient for such experimental intervention. This principle precluded placebo-controlled studies, because no patient needs a placebo. The conse- quence was a succession of experimental inter- ventions, some, such as hypothermia, crani- otomy, or high dose barbiturates, involving considerable risk. The prohibition of placebo controls precluded prospective studies designed in a way that could adequately assess benefit and risk. Instead, patients were compared with non- comparable groups, either historical controls or patients treated in other settings, with unknown variables in their treatment or unknown differ- ences in severity of injury and other relevant vari- ables.
The new rule allows for such patients to be randomized into placebo-controlled studies in which all patients would receive standard treat- ment, but half would receive the promising new treatment. This was in contrast to the prior rules, under which all patients could receive the new treatment, also without consent, but with fewer safeguards; namely, no requirements for review by an IRB, and less chance for the results to be interpretable in a way that would allow for meaningful conclusions. Confused investigators have repeated Smithell's complaint, " I need per- mission to give a new drug to half my patients but not to give it to all of them."2
Critics seemed to assume that giving the ex- perimental treatment to all patients, without their consent, under the rubric of innovative therapy, exposed them to less harm than giving it to half the patients, without consent, with the protections of the many layers of review that nec- essarily accompany research. 33 The effect of this review is that every component of the experi- mental intervent ion- - the qualifications of the clinician, the inclusion criteria, the dose, the monitor ing for adverse effects, and the assess- ment for benef i t - - is assessed. In the absence of the research protocol, under the more restrictive rules, the identical treatment could be given by the most junior clinician, at any dose, for any indication, with no requirements for assessing benefits or risks. Under either system, uncon-
Ethical Dilemmas in Medical Innovation 231
senting patients receive innovative treatments. In the research context, the physician is con- strained by procedures designed to minimize the risks and maximize the benefits. 34 In the practice setting, the physician is constrained only by his or her imagination and the low risk of a malprac- tice suit.
Conc lus ion
Human experimentation has a troubled history, justifying the attention it has received. In the past 25 years, broad public disclosure of prior transgressions, coupled with increasingly de- tailed regulation and increasing oversight by lo- cal IRBs and federal agencies, has been associ- ated with a virtual disappearance of the egregiously unethical research that was once common. Twenty-five years ago patients had rea- son to fear being victimized by overzealous un- checked researchers. Today, patients are as likely to complain about exclusion from research pro- tocols as they are about inclusion.
Experimentation in the form of innovative therapy continues to be widespread. But public and political attention and the tightening of reg- ulations are addressed primarily to the physician whose innovations are most carefully considered, most subject to review, and most likely to respect the highest standards of informed consent.
All innovative treatment involves uncertain risks and benefits. Because of this, the standards for consent should be generally be high, regard- less of whether the intervention is part of a well- designed, well-reviewed protocol. High stan- dards for consent are separate from the question of whether the intervention is research or investi- gational. ~5
If the purposes of moral discourse and regula- tions are to protect patients from harm, max- imize their interests, and give them or their rep- resentatives the greatest opportunity to make informed choices about their medical care, then regulation of research involving human subjects may have come close to a point of diminishing returns. If further gains are to be made in achiev- ing these goals, attention should turn to the kind of experimentation and innovation that occurs outside of careful scrutiny, essentially unknown to scientists or patients.
Ryan Westergard and Leslie Klein provided valuable research assistance. Dr Alan Fleischman made helpful comments.
1. Lietman P: Personal communication, 1972. 2. Smithells R: Iatrogenic hazards and their effects. Post-
grad MedJ 51:39, 1975 (suppl 2) 3. Lewis Carroll: Alice's Adventures in Wonderland and
Through the Looking Glass. New York, NY, Book Crafts- men Associates, 1960
4. Annas GJ, Grodin MA: The Nazi Doctors and the Nu- remberg Code: Human Rights in Human Experimenta- tion. New York, NY, Oxford, 1992
5. Beecher Hgz Ethics and clinical research. N EnglJ Med 274:1354-1360, 1966
6. JonesJH: Bad Blood: The Tuskegee Syphilis Experiment, London, Free Press (MacMillan), 1981
7. Hearings Before the Subcommittee on Health of the Senate Committee on Labor and Public Welfare. 93rd Congress, 1st Session, 1973
8. Advisory Committee on Human Radiation Experiments: Research ethics and the medical profession. JAMA 276:403-409, 1996
9. Angell M: The ethics of clinical research in the third world [editorial]. N EnglJ Med 337:847-849, 1997
10. Varmus H, Satcher D: Ethical complexities of conduct- ing research in developing countries. N Engl J Med 337:1003-1005, 1997
11. Ho D: It's AIDS, Not Tuskegee. Time Magazine, Sep 29, 1997, p 83
12. Shuster E: Letter. New York Times, Nov 10, 1996 13. Shaw A, Randolph J, Manard B: Ethical issues in pediat-
ric surgery: a national survey of pediatricians and pediat- ric surgeons. Pediatrics 60:588-599, 1977
14. Fost N: Treatment of seriously ill and handicapped new- borns. Crit Care Clin 2:145-159, 1986
15. Fost N, Chudwin D, Wikler D: The limited moral signifi- cance of 'fetal viability.' Hastings Cent Rep 10:10-13, 1980
16. Roe v Wade 410 US 113 (1973) 17. Trnog RD, Brett AS, Frader J: The problem with futility.
N EnglJ Med 326:1560-1564, 1992 18. Levine RJ: The boundaries between biomedical or be-
havioral research and the accepted and routine practice of medicine (Jul 14, 1975); Addendum (Sep 24, 1975). The Belmont Report: Ethical Principles and Guidelines for the Protection of Human Subjects of Research. Ap- pendix 1:1-44, 1978
19. Moore F: Therapeutic innovation: Ethical bounc~aries in the initial clinical trials of new drugs and surgical procedures, in Freund P (ed): Experimentation With Human Subjects, New York, G Braziller, 1969
20. Odell GB: Therapeutic misadventures in neonatal care, in Gluck L (ed): Modern Perinatal Medicine, Chicago, IL, Year Book Medical, 1974
21. Silverman W: Overtreatment of neonates? A personal retrospective. Pediatrics 90:971-976, 1992
22. Silverman WA, Anderson DH, Blanc WA, et al: Differ- ence in mortality rate and incidence of kernicterns among premature infants allotted to two prophylactic antibacterial regimens. Pediatrics 18:614-624, 1956
23. Weiss CF, Glazko AJ, WestonJ~ Chloramphenicol in the newborn infant: A physiologic explanation of its toxicity when given in excessive doses. N Engl J Med 262:787- 794, 1960
232 Norman Fost
24. Newman TB, Maisels MJ: Does hyperbilirubinemia dam- age the brain of healthy full-term infants? Clin Perinatol 17:331-358, 1990
25. Simmons MA, Adcock EW 3d, Bard H, et al: Hyperna- tremia and intracranial hemorrhage in neonates. N Engl J Med 291:6-10, 1974
26. Holtzman N: Dietary treatment of inborn errors of me- tabolism. Ann Rev Med 21:335-356, 1970
27. Richter MA, Haning RVJr, Shapiro SS: Artificial donor insemination: Fresh versus frozen semen; The patient as her own control. Fertil Steril 41:277-280, 1984
28. Matthieu D: Ending the experiment: dilemmas of the ex- perimental/therapeutic distinction, in Blank RH, Bonnick- sen AL (eds): Emerging Issues in Biomedical Policy. New York, NY, Columbia University, 1993, pp 247-263
29. Lifton RJ: The Nazi Doctors: Medical Killing and Psy- chology of Genocide. New York, NY, Basic Books, 1986
30. KatzJ: Experimentation With Human Beings. NewYork, NY, Russell Sage Foundation, 1972
31. Jonas H: Philosophical reflections on experimenting with human subjects, in Freund P (ed): Experimentation With Human Subjects, New York, NY, George Braziller, 1970
32. Department of Health and Human Services, Food and Drug Administration: Protection of Human Subjects; In- formed Consent and Waiver of Informed Consent Re- quirements in Certain Emergency Research; Final Rules. 21 CFR Oarts 50, 56, 312, 314, 601,812 and 814
33. Kolata G: Ban on medical experiments without consent is relaxed. New York Times Nov 5, 1996, p 1
34. Fost N: When patients can't provide informed consent. Chronicle of Higher Education Jan 25, 1997
35. Tyson J: Evidence-Based Ethics and the Care of Prema- ture Infants. Future Child 5:197-213, 1995