Ethics in Research

Proposed Guidelines

Jessica WIIen Berg

When science takes man as its subject, tensions arise between two values basic to Western society: freedom of scientific inquiry and protection of individual inviolability.... At the heart of this conflict lies an ageold question: When may a society, actively or by acquiescence, expose some of its members to harm in order to seek benefits for them, for others, or for society as a whole?]

he term human research may connote a number of negative images: the Nazi experimentation during World War Il, the Tuskegee syphilis study, or the Willowbrook experiments. All of these abuses of medical experimentation share one thing in common—they did not employ satisfactory informed consent procedures, either because the subjects were not given adequate information, because they simply were not given a choice whether to participate or not, or because the subjects suffered from cognitive impairments that rendered them unable to consent. One positive outcome of these and other abuses in biomedical research has been the promulgation of codes of ethics that restrict human experimentation. Many of these codes have focused on informed consent procedures, reasoning that only when truly informed consent is obtained is experimentation justified. The present regulations, however, are not sufficient to protect all vulnerable2 groups, in particular, cognitively impaired subjects.

Cognitively impaired subjects present difficult issues in relation to informed consent. Because people suffer different degrees of cognitive impairment, problems may range from difficulties in understanding consent documents to

Jowmal ofLaw, Medicine & Ethics, 24 (1996): 18-35.

0 1996 by the American Society of Law, Medicine & Ethics.

severe incapacity rendering the subject incompetent to make any research decision about involvement in research. In the latter situation, third-party consent is necessary to protect individual inviolability. Thus research in this context raises a number of questions: (1) What experiments involving cognitively impaired subjects are permissible?; (2) How does one distinguish among levels of cognitive impairment?; (3) What protections should apply in various cases to ensure adequate informed consent?; and (4) Who should have the authority to consent to participation for an incompetent subject? To answer these questions, I initially examine the historical evolution of constraints on human experimentation, focusing on informed consent. Then I explore the contours of surrogate decision making. Third, I identify a set of proposed guidelines for research involving cognitively impaired subjects. In so doing, I evaluate some alternative proposed guidelines in this area. Although each one entails a number of useful suggestions, all fall short of encompassing the necessary elements needed for national guidelines. Finally, I take the position that local institutional review boards (IRBs) should decide, using clearly articulated national guidelines, which experiments are permissible. I recognize, however, that additional safeguards to ensure adequate informed consent may be necessary in certain situations, and identify some of the relevant issues that arise in research involving particular groups of cognitively impaired subjects.

Background

It has been said that "all contemporary debate on human experimentation is grounded in Nuremberg. "3 Thus the Nuremberg Code provides an appropriate starting point for evaluating the historical context of human experimentation in the United States. The code begins with the assertion that the "voluntary consent of the human subject is absolutely essential. This means that the person involved should have legal capacity to give consent. "4 Thus stated, the code would not allow for research involving those cognitively impaired subjects who lack legal capacity to consent. However, the strict prohibitions of the code have not been followed, prompting Jay Katz to suggest that its "language was too uncompromising and too inhospitable to the advancement of science. "5 On reflection, it has become clear that, in some cases, the only way to advance scientific knowledge with regard to a particular disease process is to study subjects who have the specific illness. Thus, in situations where the illness results in cognitive impairment, it necessarily follows that some experimentation will take place with incompetent6 subjects.

The 1964 Helsinki Declaration responded to the medical profession's objections to an unqualified consent requirement and allowed for proxy consent or even a complete waiver of consent if in the subject's best interests. 7 Unlike the Nuremberg Code, which was proposed by a "judicial" tribunal, the Helsinki Declaration was drafted by the World Medical Association. The two documents illustrate the competing interests of protecting individual rights and accommodating the need for scientific experimentation. Groups concerned with protecting individual rights, particularly judges and lawyers who have focused on victims of abuses, tend emphasize the former, often to the complete exclusion of the latter principle. Scientists and doctors, on the other hand, clearly have an interest in preserving the latter principle and may be less protective of individual autonomy. Although such characterizations overgeneralize the differences between legal and medical professionals (most people on either side recognize the need for both protection of subjects and protection of research), a comparison of the Nuremberg and Helsinki documents indicates the inappropriateness of using either the courts or scientists as the sole decision makers regarding the acceptability of a particular experimental proposal or in shaping the requirements for informed consent.

In 1978, the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research published the Belmont Report.8 Among other things, the commission addressed the issue of informed consent, noting the possibility of research involving "subjects that one might consider as incompetent. "9 Without specifying particular guidelines for proxy consent, the report stressed that each group should be evaluated separately, and that a subject's wishes should be taken into account to the greatest extent possible. In addition, a third party should be appointed to act on behalf of the patient and thus to substitute his/her consent for that of the subject.

The notion of substitute, or proxy, consent was further acknowledged in the Department of Health and Human Services's (DHHS) Rules and Regulations for the Protection of Human Research Subjects. 10 Unlike the Nuremberg Code, DHHS regulations allow for the "legally effective consent of the subject or the subject's legally authorized representative. "11 Although the regulations set forth specific elements required for informed consent, they do not address issues of substitute consent. Moreover, the Food and Drug Administration (FDA) guideline "Informed Consent of Human Subjects" utilizes the same language, and also fails to deal with the role of the legally authorized representative. 12

The end result seems to be an acceptance of the possibility of conducting research with cognitively impaired subjects without any clear guidelines regarding limitations or procedures. In 1978, the Department of Health, Education and Welfare had proposed rules that would deal with "individuals institutionalized as mentally disabled. "13 The purpose of the proposed rules was to set guidelines that would protect incompetent subjects. The proposal was not accepted for a number of reasons, most importantly because it singled out institutionalized populations for unwarranted special treatment. 14 First, vulnerability to coercive pressures is shared by a number of groups, not limited to institutionalized persons. Second, although the guidelines were intended to cover various groups of people, 15 use of the term mentally disabled seemed to target unfairly persons suffering from mental illness. As I shall stress later, the problems related to informed consent are shared by all cognitively impaired individuals, regardless of diagnosis. Furthermore, not all individuals within a diagnostic category are cognitively impaired, nor all to the same degree.

The failure of DHHS to promulgate appropriate guidelines has not gone unnoticed. The President's Commission for the Study of Ethical Problems in Medicine and Biomedical and Behavioral Research repeatedly urged the secretary of DHHS to adopt regulations governing research in this context. However, after the President's Commission concluded its work in 1983, the Reagan administration failed to recharter an Ethics Advisory Board, thus the issue has lain dormant. Neither of the two federal agencies that have promulgated rules in this area, DHHS and FDA, have the broader authority to regulate all research or to control policy in biomedical ethics. Although a few organizations have proposed guidelines for research involving cognitively impaired subjects,16 no authoritative national body has been involved in biomedical ethics issues in the past decade.

Recent cases, such as the University of California/Los Angeles schizophrenia study, have raised concerns that local IRBs are inadequately prepared to protect the interests of cognitively impaired subjects and that national guidelines are needed. 17 Many proposals have been made for a new national bioethics commission, although no consensus exists regarding the scope of responsibility and power of such a national board. 18 On December 2, 1994, President Clinton stated that, "[iln order to ensure that advice on complex bioethical issues that affect our society can continue to be developed," he is "planning to move forward with the establishment of a National Bioethics Advisory Commission in the next year. "19 In the following section, I stress the need for a national body to establish guidelines for research20 with cognitively impaired subjects and to ensure their implementation. Subsequently, I identify relevant considerations that should be used to shape the guidelines, including issues surrounding surrogate decision making. Finally, I take the position that local IRBs should be responsible for evaluating specific research protocols involving cognitively impaired subjects and for instituting additional safeguards when necessary.

National guidelines and local review

In a 1993 article, Katz argues that, in general, IRBs fail adequately to protect the rights of patient-subjects and that a national commission should be established "to administer and review, the human experimentation process. "21 Although this has been true to some extent, I disagree with his conclusion that "it is unlikely that members of IRBs will hold investigators to a standard of disclosure and consent that would protect subjects of research if doing so would place impediments on the conduct of research. "22 IRBs can do a better job at protecting patient-subject rights if given better directives. Presently, however, DHHS and FDA regulations are inconsistent with regard to issues of informed consent and fail to guide adequately local discretion.

Even with the creation of a national commission, local IRBs will still have discretion in deciding which protocols to approve.23 A national board cannot determine, ahead of time, which specific proposals will be acceptable, only what types of proposals. Terms like minimal risk or therapeutic benefit must be interpreted in the context of a particular experimental protocol. 24 "Compared to ... a national review process, local committees have the advantage of greater familiarity with the actual conditions surrounding the conduct of research. "25 Experimental proposals should be carefully reviewed, but long delays in processing can be detrimental to scientific progress. A national review board will be unable to review efficiently all experimental proposals involving cognitively impaired subjects. Local IRBs are adequate forums to address these issues provided that they are given clear guidelines.

Here, I primarily focus on three proposals as a starting point for shaping such guidelines: (1) a 1989 American College of Physicians (ACP) position paper on cognitively impaired subjects; 26 (2) the 1986 National Institutes of Health (NIH) Clinical Center Policy guidelines for research with cognitively impaired subjects;27 and (3) Mark WICClair's recommendations for research with cognitively impaired elderly persons.28 Although a number of other proposals have been made, these three are discussed because they set forth the specific issues that need to be addressed in developing guidelines. The ACP paper sets forth four positions designed to "allow progress in research without violating society's obligation to uphold the rights and protect the welfare of potential experimental subjects 2 advance consent to participation in research is proper; (2) in cases with no advance directive, a legally authorized surrogate can consent using a mixed-substituted judgment and best interests standard; (3) special protections should be established by IRBs for institutionalized patients; and (4) a national review board should review protocols that would not be allowed under the guidelines set forth.30 The NIH policy is designed complement NIH's counterpart to IRBs, the Institute Clinical Review Subpanels (ICRSs). The first part of the policy focuses on appointing a surrogate decision maker; the second part establishes guidelines for ICRS review in eight specific cases. Wicclair identifies seven conditions that should be applied to research involving elderly subjects; he feels that they best reflect the substituted judgment standard and therefore are most protective of individual autonomy.

In 1994, John Fletcher and Franklin Miller proposed the creation of a national commission for bioethics, preferably established by an executive order with a congressional mandate. 31 They set out a national bioethics agenda, stressing the need to reexamine all research regulations and noting that "the ethics of research with cognitively impaired [subjects] ... is long overdue for national review. "32 Gerald Dworkin also saw the need for legislation in this area.33 He felt that such legislation should: (1) indicate whether research is permissible with certain groups of incapacitated subjects; (2) define the limits and requirements of proxy consent; (3) ensure review of proposals by ethics committees; and (4) provide for a national ethics committee to oversee local committees.34 In the following sections, I first identify the concerns surrounding obtaining informed consent from cognitively impaired subjects and the limitations of proxy consent. Later, I explore the implications of the forgoing analysis for what types of research are permissible. Finally, I identify issues IRBs should be aware of when evaluating protocols involving specific groups of cognitively impaired subjects.

Surrogate consent

The informed consent of a research subject is necessary to protect the individual's autonomy. Autonomy can be defined a number of different ways. For example, Bruce Miller distinguishes among four senses of autonomy: free action, authenticity, effective deliberation, and moral reflection. 35 When cognitive impairment renders a subject incompetent to give consent, additional protections of autonomy are necessary.36 One possibility is the use of a proxy consent mechanism. There are a number of methods of proxy consent.37 To the degree that a particular model of proxy consent better protects a subject's autonomy than an alternative, the more protective model will be preferred. In addition, models of proxy consent that are less protective of autonomy should not be used to authorize certain types of research, for example, that which holds no potential for direct therapeutic benefit and involves greater than minimal risk. In the following section, I examine the role of advance care documents in a research context. Then, I address general issues of proxy consent. Thereafter, I focus on the limits of proxy decision making in specific research scenarios.

Advance care documents

Both the ACP paper and the NIH policy recognize the potential usefulness of advance care documents and surrogate decision makers.38 Position 1 of the ACP paper states that "researchers should develop and use a mechanism that allows a competent subject to consent in advance to a research protocol and designate a proxy to supervise the subject's later participation in the research project after he or she becomes incompetent. "39 The NIH policy is more restrictive in that it advocates the use of a Durable Power of Attorney (DPA) alone, without reference to other types of advance care documents.

Advance care documents, which include advance directives, living wills, and DPA, have recently become popular tools that enable patients to exercise greater control over treatment decisions. A recent survey of state law reveals that every state has one or more advance directive statutes.41 It is unclear, however, whether state advance directive laws apply to research because they were originally drafted to address treatment concerns.42 Those that use broad language43 may be interpreted to apply to research that has a potential for direct therapeutic benefit, but amendments would likely be necessary to adapt the laws to nontherapeutic research. Although I leave it to others to define the form and scope of advance care legislation, I identify several specific suggestions. First, although the Patient Self-Determination Act of 1990 provides for increased education of patients regarding advance directives, it does not indicate whether health care providers have any obligations to inform research subjects about state laws related to advance directives." Inveétigators could include, in all consent forms for all subjects, not merely those who are cognitively impaired, an advance care document that would allow a person to indicate his/her wishes regarding research, as well as to appoint a proxy. The form could contain space for the subject to describe any specific instructions and to list the name of a proxy and substitute proxy. Additionally, regular (treatment) advance care documents could be modified to include questions about an individual's inclinations toward research.45 Finally, it is important to note that many states still have variants of living wills, which are only effective when the subject is suffering from a life-threatening or terminal illness. Such documents would fail to cover numerous research situations involving cognitively impaired subjects. Thus, in drafting legislation in this area, it may be useful to keep in mind the limitations of various advance care documents in the research context.

Surrogates

Many advance care documents involve proxy decision makers. And regardless of the content of their advance directive legislation, all states permit some form of proxy decision making in the treatment context. A number of issues arise in the area of surrogate" decision making. First, who should designate a surrogate? Advance directives that allow an individual to designate his/her own surrogate decision maker are the most protective of autonomy and thus are preferable. However, because many people fail to implement such authorization, a secondary mechanism for appointing a surrogate is necessary.47 The NIH policy recommends that where the subject is unable to execute a DPA, the family seek a court-appointed guardian.48 This is often time-consuming and expensive.49 Furthermore, it may require greater deprivation of rights than necessary because the person may be capable of managing his/her financial affairs, but incapable of making a research or medical de- cision. A few states have statutes that provide for a hierarchy of decision makers, ranging from a previously appointed proxy to a spouse or other close family to, in some cases, even a close friend. 50 Such statutes are extremely useful because they automatically designate a surrogate, but avoid the need to get a judicial determination of incompetence.

Another issue is who the surrogate decision maker should be. An easy place to begin is with who is least suited for this role. Courts, for example, should not have the responsibility for making surrogate decisions. The judicial system is adversarial, based on the premise that if both sides vigorously argue opposite viewpoints, the "truth" will come out. Decisions involving participation in research, like most treatment decisions, are not necessarily adversarial. One author noted that "[j]udges are so accustomed to the adversary nature of most legal proceedings that they sometimes search to find a party with a contrary view even when the party is a stranger to the litigation and is using the litigation to forward its own agenda. "51 Only in those cases where the surrogate's decision is challenged by a party with a personal interest in the outcome is recourse to the courts appropriate. In these situations, the court's role is to determine whether the decision maker is applying the correct standard, and, if not, to appoint another decision maker.

Additionally, clinicians are generally not the proper decision makers, because research decisions create a conflict of interest between the role of the physician as a clinician and as an investigator.S2

Families are usually the best suited to make treatment decisions53 and, likewise, to make research decisions. In states with statutes that specify a hierarchy of decision makers, the legislation stresses the role of the family as the preferred surrogate when no other has been specified. At present, however, these statutes are not the norm, and many states appear to require court authorization of a surrogate. In the treatment context, common law authority supports the notion that an informal next-of-kin surrogate can make decisions. 54 The NIH policy recognizes this and proposes that in situations where the subject has not executed a DPA, and cannot presently do so, a next-of-kin surrogate should be appointed. State laws that codify the common law authority specify surrogates based on a policy judgment regarding which persons are generally the most appropriate decision makers. The underlying supposition is that such persons are generally in the best position to know the patient's values and preferences and thus are able to make the best substituted judgment. Moreover, because of the close relationship between the surrogate and the patient, it is also thought these individuals will make a decision that is in the patient's best interests when such a standard is appropriate. Ideally, states should codify the informal nextof-kin hierarchy of decision makers for both treatment and research contexts.55 If a physician or another interested party believes that the surrogate is not acting appropriately, they should be able to challenge the appoinment in court. Courts would thus step in at the end of the process to resolve disputes, a role for which they are much more suited.

Making decisions

Once an advance care document has been executed and/or a surrogate has been designated, the issue remains whether to authorize participation in light of a specific protocol. The surrogate must be educated both as to what standard to apply and as to how to apply the appropriate standard. The ACP paper, NIH policy, and Wicclair all emphasize the need to promote the best substituted judgment. 56 Whether and to what extent this is feasible depends, in part, on the individual involved.57 The substituted judgment model should be applied to formerly competent subjects who have left indications of their wishes. 'IWo situations may arise, and the evidence in each should be given different weight. First, the subject may have executed an explicit advance directive either authorizing or refusing research. Alternatively, the individual may not have executed an explicit advance directive, but he/she may have left other evidence of his/her wishes that can be used by the surrogate.

When an explicit advance directive authorizes participation in a particular research protocol, the proxy should consent. In these cases, the subject's autonomy is protected in the sense that the surrogate is merely implementing the subject's already expressed consent. S8 Thus with an advance directive, the proxy does not decide whether to consent to research, rather he/she decides whether to effectuate the advance care document. The surrogate, however, retains a continuing responsibility to ensure that the research in question has not changed substantially since the subject consented, and that the subject's condition has not changed so to alter the risk-benefit ratio." I take the position that there are no limits on the types of research (for example, high risk, little or no benefit) to which a subject can consent in an advance directive. 60 Because competent subjects do not have limitations placed on the types of research in which they can participate while they remain competent (as long as the protocol is approved by an appropriate review board), they should not have limits placed on the types of research in which they can consent, in advance, to participate should they become incompetent. The surrogate in these situations, however, has a greater responsibility to monitor the research and to ensure that continued participation is appropriate. As noted above, if for any reason the surrogate believes that the subject would have withdrawn from the protocol were he/she still competent, then the surrogate is under an obligation to withdraw the incompetent subject.

Where the subject has left an explicit advance directive refusing to participate in any research, the analysis is slightly different. In these cases, the proxy should honor the subject's refusal in all situations except where the research holds a possibility of therapeutic benefit and presents the only access to a particular treatment. In these circumstances, the proxy should use a best interests analysis to decide whether to override the proxy's previous competent refusal. Furthermore, court approval would be necessary in the same circumstances as would be required to override a regular treatment refusal." As a result, when an individual has left an explicit advance directive, that directive should be given a great deal of weight. Only in unique situations where in the experiment is in the subject's best interests may a proxy override a specific advance directive.

When a subject has not left an explicit advance directive but has expressed some beliefs or attitudes bearing on research participation, the proxy must evaluate the evidence62 in order to reach the best substituted judgment. Evidence that the subject would have refused to participate should be respected under all circumstances, except those outlined above, which allow a proxy to override an explicit advance directive refusing participation. Evidence that the subject would have consented is more difficult to evaluate. The ACP proposal is unnecessarily limited in this respect because it only permits implementation of a subject's wishes to participate that are documented in an advance directive. When there is no advance directive to authorize research, the surrogate is only permitted to consent to research that is in the subject's best interests. 63 Wicclair recognizes this shortcoming in the ACP suggestions and allows a proxy to consent to research that, "considering the nature of the proposed study, the potential benefits and harms to subject, and what is known about the cognitively impaired person ... [a surrogate] can conclude that the person would have consented."" Moreover, Wlcclair notes that "as the seriousness of possible harm or discomfort to the cognitively impaired person increases, the strength of the evidence supporting the conclusion that the person would have consented should also increase. "65

As a starting point for establishing the level of certainty needed, one might consider the standards and type of evidentiary proof courts require in surrogate treatment decisions." Both verbal and written evidence may be used to meet these standards.67 Surrogate decision makers can use varying judicial standards of proof to aid their understanding of the level of certainty needed for consent in particular situations. Admittedly, these degrees of proof are somewhat vague, but they provide a usable framework. For example, cases involving low risk and high potential for therapeutic benefit might require a preponderance of evidence-—greater than 50 percent probability—indicating that the subject would have consented. By contrast, research involving low risk, but little or no potential for therapeutic benefit may require clear and convincing evidence— greater than 75 percent probability—of the subject's wishes. Evidence of earlier discussions regarding research, when the subject was competent, should be given significant weight. An explicit advance directive would likely meet even a clear and convincing evidentiary standard. On the other hand, mere altruistic tendencies throughout life would probably not suffice under any standard. Under this framework, it is the responsibility of the surrogate to ascertain and implement the subject's wishes to the best of his/her ability.

Finally, the best interests model should be applied to formerly competent subjects who have left no specific evidence of their inclinations regarding participation in research or to subjects who have always been incompetent. As a result, the types of experiments in which these subjects can participate will be more restricted. Participation in protocols involving greater than minimal risk without the possibility of direct therapeUtic benefit would be impermissible because it will never be in the best interests of the subject. Even though the research may hold an intangible benefit, such as the altruistic nature of participating in research for the good of the community or the possible long-term benefits of finding a cure for the subject's disease, such considerations are generally inappropriate for a best interests analysis. The ACP paper is even more protective of subjects; it advocates a mixed best interests and substituted judgment standard when there is no advance directive. 68 Thus the surrogate should only consent to research that is in the best interests of the subject and that the individual would not have refused.69 It is therefore posSible that a surrogate might refuse to consent to research that involves low risk and some potential for therapeutic benefit and that is in the subject's best interests because he/ she determines that the subject would have refused to participate had he/she been competent. Wicclair presents analogous conditions, allowing for proxy consent when the surrogate does not believe the person would have refused and the research is either in the person's best interests if therapeutic or the risk of harm or discomfort is minimal if nontherapeutic.70

Incompetent revocation of previous consent

One final problem arises in the case of a subject's incompetent refusal after either an initial competent consent or a competently executed advance directive authorizing participation. The ACP paper notes that the proxy has the responsibility to withdraw a subject if continued participation would cause "substantial distress (for example, if the subject becomes uncooperative). "71 Although this position seem obvious at first glance, it is not so simple in cases where the experiment involves research entailing a "drug wash-out" period.n In these cases, it is likely that the subject's capacity will become progressively more impaired as he/she is taken off any previous stabilizing medications. These considerations also apply to subjects who have illnesses that result in progressive dementia, because they will suffer greater impairment as time passes. Many experiments involving elderly dementia subjects are designed as longitudinal studies, thus ensuring that the subjects will not be capable decision makers at a point when the research is still ongoing. Concurrent with a decline in capacity may be an increase in uncooperativeness as the illness limits more and more the person's ability to make decisions. It would be extremely dangerous simply to allow the subject to drop out of the experiment without providing any medication or monitoring of the subject. Moreover, it is unclear whether to follow the person's previous competent consent or present incompetent refusal.

These difficulties have prompted some to suggest that when a subject gives consent at the outset of a study, this should constitute a continuing authorization while incompetent. 73. This may not be legally or ethically acceptable because the subject's cooperation may be a necessary prerequisite to continuation of the study.74 Thus it has been suggested that the assent of the subject should be sought even when consent is unobtainable. This assent does not take the place of consent, rather it reflects a recognition that even subjects who are incapable of giving informed consent should be afforded some input in decisions concerning their participation in research. Even when surrogate consent is obtained, the subject may still be approached and asked whether he/she wishes to participate. In one article explaining the NIH policy, John Fletcher and Alison Wichman note that the guidelines were based on the principle, embodied in the existing DHHS regulations, that the assent of subjects is a necessary but not sufficient prerequisite to all research.7S Wicclair takes a less restrictive stance and requires subject cooperation only when the research is without therapeutic benefit. 76 This is compatible with laws that allow for treatment over an incompetent patient's refusal.n The assent, or at least cooperation, requirement should apply to all subjects except if the research has high therapeutic potential, is in the subject's best interests, and presents the only access to a particular treatment. 78 In these circumstances, the decision is more akin to a treatment decision, and thus incompetent refusals may be overridden in some cases.79

A more difficult situation arises when a subject, who has previously given consent, is neither assenting nor objecting, or when the subject's capacity to assent is suspect. Some authors have proposed the use of a research auditor80 to ensure that continued participation is appropriate. Instead of a research auditor, a proxy decision maker might be identified at the outset of the study, even when the individual in question is competent.81 This surrogate should have an opportunity to investigate the subject's specific wishes and would then have to monitor whether the continuing participation of the subject, after he/she is no longer capable of making decisions, is warranted. Using a surrogate under these circumstances, rather than a research auditor, would enable the substituted judgment standard to be used. Because a research auditor would likely be a stranger, he/she might have difficulty ascertaining the subject's preferences and, as a result, need to use a best interests standard. A surrogate, on the other hand, would have had more contact with the person while competent, and thus would be in a position to implement the subject's wishes.

Proposed research guidelines

The extent of protection afforded [cognitively impaired persons] should depend upon the risk of harm and the likelihood of benefit. The judgment that any individual lacks autonomy should be periodically reevaluated and will vary in different situations.82

To identify to what research a proxy can consent, I will focus here on two variables: the extent to which the research has the potential to provide direct therapeutic benefits to the subject,83 and the degree of risk entailed." Both risks and benefits will be treated along a continuum—thus greater protections are necessary as the risk increases and the potential for therapeutic benefit decreases. Initially, it is important to acknowledge that the term minimal risk, found in the federal regulations and utilized by a number of commentators, is somewhat vague.8S The regulations define it as meaning that "the probability and magnitude of harm or discomfort anticipated in the research are not greater in and of themselves than those ordinarily encountered in daily life."" This leaves much room for interpretation—What risks are ordinarily encountered?87 Merely riding in an automobile could be said to present a high probability and magnitude of harm. Whose life do we look at? Is the standard that of the ordinary healthy person, or does it vary according to the individual subject? For example, a cancer patient is exposed to a number of treatments that involve a high probability and magnitude of harm: Should these be used as the baseline for evaluating research proposals? Undoubtedly, the meaning varies in different circumstances, and the result is that minimal risk has become a term of art. It clearly does not mean only insignificant risk, but its exact scope is unclear. Although the term will be used in subsequent discussion, more guidance is needed at the national level as to what considerations are applicable. In the end, however, it is the responsibility of local IRBs to interpret minimal risk in light of specific proposals.

Research risk is a matter of potentialities. Thus it is misleading to speak of an experiment as being risky or benign. The quantification of risk involves an examination of both the degree or magnitude of harm that may occur and the possibility that such harm will occur. Thus a highrisk experiment could mean either a protocol that presents a high probability of causing a small amount of harm or one that presents a low probability of causing a high magnitude of harm. Risk calculations contain additional ambiguities. For example, it may be unclear what side-effects a drug may have, or how likely those side-effects are to occur. To the extent that the risks (both magnitude and probability) of a particular protocol are unknown, one may categorize it as entailing a higher risk than a protocol that involves known effects.

Furthermore, because risk is categorized along a continuum, the use of minimal risk in the federal regulations has led to such anomalous results as creating exceptions for minor increases over minimal risk. 88 This is a failure to define minimal risk and the vast category of "greater than minimal risk." Therefore, better guidance on the meaning of minimal risk is warranted, rather than on defining exceptions for minor increases over minimal risk. Once better defined, however, minimal risk may help to identify those types of research proposals that do not need additional protections.

As with research risk, the potential for direct therapeutic benefit is a product of a number of factors. First, there is the question of how much benefit is entailed. Although this generally involves a separate calculus in each case," an IRB will need to be as aware of the degree of benefit as it is aware of the degree of risk. Moreover, like risk calculations, benefits involve potentialities. A given protocol may only offer a 50 percent chance of directly benefiting any one individual. That is, if 100 persons were given the experimental medium, only fifty would directly benefit. 90 In many studies, however, not all participants receive the experimental treatment; some simply get a placebo, or another drug perhaps already deemed ineffective for them. Thus the benefit calculation must take into account how probable it is that a particular subject will get the experimental medium as well as the probability, that, once received, the intervention will help. In a simple experiment, 50 percent of the subjects would get the experimental treatment and 50 percent would get something else. If we use the example above, then the potential for direct therapeutic benefit would only be 25 percent (50 percent of 50 percent). It is significant to note here that this aspect of potential benefit is completely within the investigator's control—the potential could easily be brought back up to SO percent if the researcher utilizes a cross-over design. Thus, after a period of time, the group receiving the experimental treatment would be given the placebo and the group receiving a placebo would be given the treatment. To the extent that IRBs seek to maximize therapeutic potentials, they may be able to encourage investigators to alter the designs of their experiments.

Having established that both risk and benefit calculations are fraught with uncertainty, we are left asking what role should the IRB take to protect cognitively impaired subjects. Here, I discuss eight cases, differing in the level of research risk and the potential for therapeutic benefit, and thus requiring various levels of monitoring to protect the interests of the individual involved. In cases involving subjects who are competent to consent, the protections should focus on ensuring that the individual truly has the requisite capacity and that he/she actually understands the research. In cases involving incapacitated subjects and therefore proxy decision makers, considerations should focus on whether the proxy understands his/her role and whether he/she actually understands the research. The following framework is similar to the one identified in the NIH policy. However, while the NIH policy focuses on the degree of impairment and on the level of risk to decide whether to appoint a surrogate decision maker, this proposal looks at possible therapeutic benefits and degree of risk to decide what additional protections should be imposed. Level of impairment remains important because it will determine the extent of the additional protections. This framework allows for the separate consideration of risk-benefit calculations and surrogate decision-making limitations.

Research involving minimal risk and a potential for direct therapeutic benefit

Case 1 involves minimal risk to the subject and a potential for direct therapeutic benefit. 91 The subject is presumed to be a competent decision maker. No additional protections regarding the capacity of the subject are necessary beyond those normally in place in the treatment arena." Thus it remains the responsibility of the medical team to monitor the subject's capacity to consent and to take appropriate measures when capacity is impaired. As mentioned earlier, such measures may include seeking a proxy decision maker under the applicable state law.

In case 2 (also involving minimal risk and potential benefit), a surrogate decision maker is necessary, and the normal rules for proxy consent, outlined in the previous section, apply. The NIH policy allows a surrogate to consent to research involving minimal risk and a potential for direct therapeutic benefit when the subject is capable of executing a DPA. Under my framework, if the subject has, or can presently, execute any of the other advance care documents, research should be permitted. This policy recognizes that a subject can be incapable of consenting to participation in research, but be capable of understanding and executing an advance care document." When the subject is incapable of executing a DPA, the NIH policy authorizes the appointment a next-of-kin surrogate. It also requires review of the surrogate appointment through an ethics consultation. As stated previously, if a hierarchy of preferred decision makers is specified in advance, no additional protections beyond those available in the regular treatment context would be needed. In other words, when a choice between members of the same class is still necessary, further protections may be warranted. While review of the matter can be done by an ethics consultation service, it may be more appropriate to have a neutral third party, who is familiar with the research, to mediate among family members. In cases of clear conflict, recourse to the courts may be unavoidable.

Research involving greater than minimal risk and a potential for therapeutic benefit

Case 3 also involves a potential for direct therapeutic benefit. The risk involved in the study, however, is greater than minimal, and, as it increases, IRBs should take a more active role in overseeing the consent process. Where the subject is capable, no alternative treatment is available of at least equal value, and the experimental treatment is not available through any other source, no additional protections are necessary. This is similar to the pure treatment context, and should be handled accordingly. The capacity of the subject should remain a clinical judgment. This analysis changes slightly when alternative treatments that hold at least equal possibility for therapeutic benefit are available. In this case, protections are necessary. These might include further evaluation of the subject's capacity, and employment of mechanisms that can increase the subject's actual understanding. IRBs can focus on perfecting consent forms, on observing a random sample of consent interactions, or perhaps on audio- or even videotaping the consent process. 94 The mechanism employed should be commensurate with the risk involved—as the risk increases, more protective measures are necessary. Thus IRBs will need to use discretion in determining the appropriate degree of oversight.

One concern in this area is the prevalence of the therapeutic misconception'S A study of psychiatric patients found that many of them misconstrued the nature of the research in which they were involved and erroneously believed that they would be receiving the best for their condition. 96 More specifically, this study found that schizophrenic and depressed subjects involved in a doubleblind experiment were unaware that they would be randomly assigned different drug regimens and that the researchers themselves would not know which drug the subjects were taking.97 Understanding of the depressed patients was significantly improved when a videotape was utilized or a neutral educator disclosed information.98 Thus studies involving subjects with schizophrenia may require greater safeguards than those involving depressed subjects." IRBs and investigators can minimize the therapeutic misconception by focusing on the content of the disclosure, emphasizing the methodology of science, and treating consent as an ongoing process. 100

Case 4 is identical with case 3, except that the subject does not have the capacity to consent to participation. Under these circumstances, if the subject has not been adjudicated incompetent and can execute a DPA, the NIH policy calls for review of the surrogate appointment by an ethics consultation service to ensure that the "person appointed by the subject is capable of understanding the purpose and risks of the study." This raises, for the first time, the adequacy of protections regarding the surrogate's, as opposed to the subject's, consent. The real issue is not whether the surrogate's decision is the correct one, but whether the surrogate understands the appropriate standard to be used and applies it. This issue also arises in treatment decisions. In those cases, the capacity of surrogate decision makers is not evaluated. By the same token, where research preseno a potential for direct therapeutic benefit, a requirement that the ethics committee evaluate the surrogate's capacity or knowledge is unduly burdensome. This problem would be better handled by additional education of surrogates, a responsibility that could be the province of either the investigator or the IRB. An IRB might draft a basic document explaining what is meant by substituted judgment or best interests and give examples of how to apply the standards in specific cases. Furthermore, it could specify the level of certainty needed to make a decision. In cases where there is a concern that proxies will not understand the experiment, because of the complicated nature of the research, and there are high risks involved, the IRB might require investigators to take additional steps to ensure comprehension, such as employing a neutral educator. The extent of the protections required would depend on particulars of the case.101

Research not involving direct therapeutic benefit

The remaining cases do not involve either any potential for direct therapeutic benefit or the potential is so low as to be negligible. 102 In these situations, society is justified in taking a more paternalistic stance. Although people who have no decisional impairment can volunteer for such experimeno, protecting the autonomy of cognitively impaired subjects does not require that we allow them to consent to any experiments. Experimentation is permitted only within certain risk-benefit boundaries, and even competent subjects are not permitted to consent to all research. Research with incapacitated subjects may be even more limited. Consent by an incompetent person is not an autonomous decision. Proxy consent is an imperfect approximation of individual consent, and may therefore also be limited. To the extent that a person has previously indicated a desire to in research, especially through a specific advance care document, surrogate consent is permissible. However, when there is little or no evidence of a person's wishes on this matter, the only way to protect individual inviolability is to prohibit participation. 103 Furthermore, in cases with no or a very small potential for direct therapeutic benefit, subject cooperation is mandatory. When the subject has previously consented to research, even through an advance care document, but is presently refusing to participate, research may not continue. This highlights the importance of appointing a surrogate, at the outset of the study, who can monitor the subject's continued participation. It would then be the surrogate's responsibility to withdraw the subject from the experiment. 104 Finally, when no potential direct therapeutic benefit exists, only research that is relevant to the subject's condition is permissible.

Case 5 involves minimal risk to the subject and a low or nonexistent potential for direct therapeutic benefit. Protections are warranted to ensure both that the subject has the capacity to consent and that he/she actually understands the risks involved in the research and its lack of direct therapeutic benefit. This case should be treated similarly to case 3. Unlike case 3, höwever, this situation involves little or no possibility of direct therapeutic benefit. Thus smaller increments of risk may trigger greater protections. As the degree of risk increases,10S investigators may consider screening subjects' capacity or employing consent monitors. 106

Case 6 is identical with case 5, but surrogate consent

is necessary from the outset. Here, too, additional protections are appropriate. The concern in this context should not be so much whether the correct person was appointed as proxy, as whether the proxy understands the relevant facts and applies the correct standard of decision making. Wicclair suggests that the reasoning of the surrogate be scrutinized by "persons who do not have a vested interest in recruiting and that the surrogate demonstrates that it would have been reasonable for the subject to have consented had he/she still had the capacity. 108 When a subject has not left explicit instructions, this additional reasonableness requirement may be warranted. The standard should be objective—considering all of the factors, Would a reasonable subject consent to participation? When the subject has left clear instructions, a reasonableness requirement should not be used to negate an advance directive. The addition of a reasonableness standard when the subject has not left explicit instructions agreeing to participate in the research would force surrogates to evaluate objectively the risks and benefits of the study as well as to consider carefully whether the subject would have consented if still competent.

Case 7 also has little or no potential for direct therapeutic benefit, but involves a significant degree of risk. This situation requires the greatest safeguards. In addition to mechanisms described in earlier cases, investigators should employ a capacity screening device to identify those subjects whose capacity is impaired to the degree that they are unable to consent, those who are slightly impaired but able to consent, and those who are competent. The latter group should be treated as persons in case 5 are. Due to the high risk, however, even competent subjects should be encouraged to discuss the research with family or close friends and perhaps to appoint a proxy who can monitor the research should the subject become incapacitated. The first and second groups should both have a surrogate appointed. The first group, individuals who are incompetent to consent, are in case 8. The second group, including people whose capacity is impaired but they are still capable of consenting, needs the greatest attention. This group needs both protections during the initial consent process and some mechanism for monitoring ongoing participation in the research. As suggested earlier, rather than having an IRB appoint a separate consent or research auditor, a surrogate may be appointed at the outset of the study. The surrogate should be someone of the subject's choosing, and he/she would be responsible during the consent process for helping the subject understand the research. Moreover, if the subject's capacity becomes more impaired during the course of the study, the surrogate can evaluate the appropriateness of the subject's continued participation. Furthermore, IRBs should require that investigators identify a neutral third-party educator who can explain the high level of risk involved and the lack of any direct therapeutic benefit.

Case 8 is similar to 7 except that surrogate consent is needed at the outset of the study. This type of research is least justified in the absence of the subject's consent, and proxy consent under these circumstances can be severely limited. Where the subject has previously indicated that he/she would like to take part in the specific study, a surrogate may authorize experimentation. 109 The NIH policy allows research in this context when the family is intact, but requires an ethics consultation for family members to ensure that each understands the risks and benefits and to appoint a legal guardian. This may not be the best alternative. First, a hospital ethicist or ethics committee has no special ability to evaluate a surrogate's understanding of the risks and lack of benefit. Moreover, review by an ethics consultation service or a committee may suggest to investigators that the consultation is for the purpose of identifying unethical research, and thus trigger greater resistance to such interventions. 110 Additional scrutiny in this situation is more appropriately the role of the IRB. It may be necessary in this situation for the IRB to appoint an independent consent/research auditor. Ili One author has proposed that an administrative program run by DHHS arrange for patient-advocates to assist families in making research decisions.

Although DHHS patient-advocates are unlikely to become commonplace, it would be possible for IRBs to take a more active role in monitoring the consent process in certain cases. First, it is important to note that only where the subject has left specific advance instructions can a surrogate consent to research that has no potential for therapeutic benefit and involves greater than minimal risk. 113 Thus the surrogate's role is to evaluate the research and to decide whether the subject would have consented if he/she were competent. This requires that surrogates completely understand all aspects of the experiment, and that they be kept up to date about changes both in the experiment and in the subject's condition throughout the study. Under these circumstances, it may be useful to have a third party act as a contact for the surrogates, to update them on the progress of the experiment, as well as to be available to answer questions. 114 This consent or research auditor does not monitor consent, as much as act as a resource to help surrogates discharge their duties. Thus, research with incompetent subjects would be allowed in this context, but carefully constrained.

This eight-case policy raises two situations in which research with a cognitively impaired subject could not take place. 115 The first is where the experiment involves high risk, little or no potential direct therapeutic benefit, and an incapacitated subject who has not previously executed an advance care document specifically consenting to the research protocol and who presently is incapable of doing so. 116 The second is the case where there is greater than minimal risk, little or no potential direct therapeutic benefit, and, although the subject has a designated surrogate, he/she either was never previously competent or left little evidence on which to base a proxy decision. In both, the best interests standard of surrogate decision making would apply and should bar research participation. 117 The proxy decision in these cases would not protect any aspect of the subject's autonomy, and thus should not be used as a substitute for an individual's informed consent. 118

Unlike therapeutic situations where we might be more inclined to allow individual choice, even if it is of questionable competency, the research situations in cases 7 and 8 present a different scenario. The benefits are no longer at the individual level and thus without adequate informed consent (including proxy substituted judgment), the subject merely becomes a means to an end. Hans Jonas writes that "[w]hat is wrong with making a person an experimental subject is not so much that we make him thereby a means ... as that we make him a thing—a passive thing merely to be acted upon.... Only genuine authenticity of volunteering can possibly redeem the condition of 'thinghood' to which the subject submits. "119 For people who are incapable of consenting, only proxy consent based on strong evidence of what the person would have wanted if he/ she was competent will suffice. When there is no way to effectuate this and a best interests standard is used, consent in these situations merely reduces the subject to a thing to be acted on; there is no recognition of the individual's authority as an agent with control over his/her life. For such research participation to be permissible when it does not offer any direct therapeutic potential, society must decide that it has a right to use people in this manner. This is not to imply that such a decision is right or wrong, rather it is to point out that once we move beyond the substituted judgrnent standard, we are no longer effectuating individual autonomy. Furthermore, without a therapeutic basis, society cannot claim the right to act for the person's individual benefit. Thus the only way research can occur in this context is if society decides that the benefits to the whole outweigh the harm to the individual. Our society places considerable importance on individual rights, hence the balance will never likely favor research. Notwithstanding, "[i]f there are circumstances in which an obligation to protect autonomy may be outweighed by an obligation to conduct research, there is a need for a discretionary body to consider various prima facie obligations, competing interests, and circumstances, and to arrive at an appropriate decision. "120 This discretion must be exercised on the national level, because it entails a policy decision unsuited for local IRBS. 121

Other guidance for local IRBs

In 1978, the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research recommended that "the rights of subjects should be protected by local review committees operating pursuant to federal regulations...." 122 Even with federal guidelines, issues remain that IRBs should consider in relation to specific groups of cognitively impaired subjects. Presently, the Institutional Review Board Guidebook for Protecting Human Research Subjects says little about cognitively impaired subjects. In general, it identifies the ethical issues involved and the lack of national consensus. 123 With respect to emergency situations involving traumatized and comatose sub* jects, it simply notes the conflict between DHHS regulations (which require informed consent unless the research involves only minimal risk and an IRB approves a waiver of consent) and FDA regulations (which allow for a waiver of IRB review in emergency situations when four conditions are met). 124 For research with elderly persons, the guidebook acknowledges the lack of specific federal regulations and states that "IRBs should treat cognitive impairment in elderly subjects as they would cognitive impairment in any prospective subject. "125 This lack of clear guidance is due to a complete paucity of leadership at the national level. IRBs are thus left with the confusing task of not only reconciling conflicting federal regulations, but also adhering to various state laws. 126

Although discrete standards should not be applied to different groups of cognitively impaired subjects based on diagnostic categories, particular groups of experimental subjects do present distinct issues for IRBs to consider. 127 For example, the standard of surrogate decision making applied in each case varies based on whether the individual in question was decisionally capable at some earlier time or was never capable. Many people with mental illness will have intermittent periods of capacity. As a result, although certain types of experimentation may be impermissible at one time, they may be possible at another, either when the individual is competent to give his/her own consent or through the use of advance directives or instructions to a proxy in anticipation of later relapse.128 Other people, such as those suffering from Alzheimer's disease, will become progressively more impaired as the research progresses. Therefore, it is important to discuss all aspects of research with the subject during the early stages of the illness. In addition to differing levels of capacity, issues arise regarding the appropriate standard to apply. Because the onset of schizophrenia typically occurs during young adulthood, many subjects may not have left adequate evidence of their wishes while competent. Even among people who are over eighteen, it is unlikely that incapacitation was ever considered, and they may leave no evidence on which to base a substituted judgment. On the other hand, elderly subjects may leave a great deal of evidence regarding their wishes. The earlier information is disseminated to patients and their families regarding involvement in research, the greater likelihood that a subject will have considered participation and indicated a preference.

Research involving emergency room patients presents some of the most difficult problems. People in these circumstances are unlikely ever to have considered participation in research protocols. Moreover, family members or other surrogate decision makers may be unavailable at the critical time during which research must be initiated. Separate guidelines are warranted here. 129 Many of these subjects are unable to give consent, and often the medical staff has no time to obtain proxy consent. In essence, then, the decision to involve the subject in a research protocol is left wholly to the attending physician. A 1994 a congressional subcommittee investigation report concluded that " [a]t the heart of the matter is the absence of clear and consistent, patient-sensitive guidelines from the Food and Drug Administration and the National Institutes of Health. "130 New regulations should focus on distinguishing emergency treatment from research with emergency room patients. Although the former is likely to be permitted under a wide range of circumstances (within the confines of acceptable medical practice), the latter may be highly restricted. It is in this context that the delicate balance between the need for new scientific knowledge in order to advance medical treatment and the protection of individual autonomy is hardest to achieve.

Conclusion

Because of historical abuses in human experimentation, the doctrine of informed consent has been developed to protect individual autonomy. This doctrine is constantly set against the need for new scientific information to help find new treatments and eventually cures for various diseases. Research involving cognitively impaired subjects is a reality, and clear guidelines must be established both for protection of individual informed consent and for limitations on surrogate consent. In general, research that involves high risk and little or no potential for direct therapeutic benefit should be subject to the greatest protections. Moreover, IRBs must be made aware of specific issues that arise in research involving particular groups of cognitively impaired subjects. Only through a combination of national guidance and increased local IRB scrutiny will adequate protection for research with cognitively impaired subjects be achieved.

Acknowledgments

The author thanks Professor Richard J. Bonnie and Dr. Paul S. Appelbaum for their extremely helpful comments on earlier drafts.

References

1. J. Katz, Experimentation with Human Beings (New York: Russell Sage Foundation, 1971): at 1.

2. The present federal regulations include pregnant women, fetuses, institutionalized persons, prisoners, and children as "vulnerable" populations that need additional protections. I do not deal with the issue of vulnerability in the sense of being susceptible to coercive pressures, rather I focus on the difficulties relating to informed consent in research with cognitively impaired subjects.

3. G. Annas and M. Grodin, "Introduction," in G.J. An-

nas and M.A. Grodin, eds., The Nazi Doctors and the Nuremberg Code: Human Rights in Human Experimentation (New York: Oxford University Press, 1992): at 3.

4. The Nuremberg Code (1947). See full text as reprinted in Katz, supra note 1, at 305—06.

5. J. Katz, "The Nuremberg Consent Principle: Then and Now," in G.J. Annas and M.A. Grodin, eds., The Nazi Doctors and the Nuremberg Code: Human Rights in Human Experimentation (New York: Oxford University Press, 1992): at 235 (Katz argues that the Nuremberg Military Tribunal's main concern was to prevent recurrence of the atrocities of Nazi experimentation, and thus they did not seek to balance the advancement of science and individual inviolability.).

6. "Incompetent" is used here to refer to incapacity to make medical decisions; it is not necessarily a legal determination of incompetence. Only a court can make a determination of "competence." "Capacity" determinations, on the other hand, are relegated to the medical or mental health professions.

7. World Medical Association, Declaration ofHelsinki: Recommendations Guiding Medical Doctors in Biomedical Research Involving Human Subjects (1964). Reprinted in T. Mappers and J. Zembaty, eds., Biomedical Ethics (New York: McGraw-Hill, 1981): at 146-47.

8. National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research, The Belmont Report: Ethical Principles and Guidelines for the Protection of Human Subjects in Research (Washington, D.C.: U.S. Government Printing Office, OS 78-0012, 1978).

9. Id. at 13 (the list included "infants and young children, mentally disabled patients, the terminally ill and the comatose").

10. Department of Health and Human Services, Rules and Regulations for the Protection of Human Research Subjects, 45 C.F.R. 46.101-.409 (1983).

11. 45 C.F.R. S 46.116 (1991) (emphasis added).

12. 21 C.F.R. 50.20-.48 (1993).

13. 43 Fed. Reg. 53,954 (Nov. 17, 1978).

14. See, for example, P Appelbaum, C. Lidz, and A. Meisel, Informed Consent: L.egal Theory and Clinical Practice (New York: Oxford University Press, 1987).

15. 43 Fed. Reg. 53,954 (Nov. 17, 1978) ("'mentally disabled' individuals includes those who are mentally ill, mentally retarded, emotionally disturbed, psychotic or senile, regardless or their legal status or the reason for being institutionalized.

16. See infra page 20.

17. See J. Horowitz, "For the Sake of Science: When Tony Lamadrid, a Schizophrenic Patient and Research Subject at UCLA, Committed Suicide, It Set Off a National Debate: What Is Acceptable in Human Experimentation and Who Decides? " Los Angeles limes Magazine, Sept. 11, 1994, at 16; and Office for

Protection from Research Risks, "Evaluation of Human Subject Protections in Schizophrenia Research Conducted by the University of California Los Angeles" (May 11, 1994).

18. See, for example, J.C. Fletcher and A. Wichman, "A New Consent Policy for Research with Impaired Human Subjects," Psychopharmacology Bulletin, 23 (1987): 382—85.

19. Statement on Federal Funding for Research on Human Embryos, 30 weekly comp. Pres. Doc. 2459 (Dec. 2, 1994). No action was taken in 1995. It remains to be seen whether a commission will be established in 1996.

20. I use the terms research and experimentation interchangeably, except where specially noted. However, the two have different meanings, and it is useful to clarify exactly what is intended. The Belmont Report points out that "[t]he fact that a procedure is 'experimental,' in the sense of new, untested, or different, does not automatically place it in the category of research" (National Commission for the Protection of Biomedical and Behavioral Research, supra note 8, at 3). Experimentation and research, however, both involve unknown consequences and thus both should be subject to increased scrutiny. P. McNeill, The Ethics and Politics of Human Experimentation (New York: Oxford University Press, 1993). Even so, it is not realistic to require that every doctor who tries a new or experimental treatment on one patient get IRB approval. Thus the guidelines proposed here should only apply when the physician seeks systematically to gather general information, that is, when the doctor seeks not only to treat the individual patient but also to obtain information that will help treat other patients. Thus many "experimental treatments" may become research as they are repeatedly tested on different patients. IRB approval would not be avoided by categorizing each of a number of trials of an experimental medium or method as "individual treatment.'

21. J. Katz, "Human Experimentation and Human Rights," St. Louis University Law Joumal, 38 (1993): at 7.

22. Id. at 41. see also P Williams, "Why IRBs Falter in Reviewing Risks and Benefits," IRB: A Review of Human Subjects Research, 6, no. 3 (1984): at 4 (listing possible reasons why IRBs fail to evaluate adequately risks and benefits, but noting that there is "no evidence that unduly dangerous protocols are actually passing committee review").

23. Katz, supra note 21, at 158—59 ("In human experimenresearch ethics committees are ideal candidates to exercise discretion in the application of principles and rules to particular research proposals.... The degree of discretion exercised by committees should not mean that some committees can approve research on human subjects that is below the minimum ethical standard for such research. This position ... requires an active national body to oversee the system of review and provide specificity in research guidelines.").

24. 45 C.F.R. 46.101-.409 (1983); and 21 C.F.R. 50.20.48 (1993). More guidance at the national level on the meaning of these terms is necessary.

25. National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research, U.S. Department of Health, Education and Welfare, Institutional Review Boards: Report and Recommendations (Washington, D.C.: U.S. Government Printing Office, OS 78-0008, 1978): at 1.

26. American College of Physicians, "Cognitively Impaired Subjects," Annals of Internal Medicine, 111 (1989): at 843.

27. National Institutes of Health, Clinical Center Policy, Research Involving Impaired Human Subjects Clinical Center Policy for the Consent Process (Bethesda: National Institutes of Health, 1986).

28. M. Wicclair, Ethics and the Elderly (New York: Oxford University Press, 1993).

29. American College of Physicians, supra note 26, at 883.

31. J.C. Fletcher and F.G. Miller, The Promise and Perils of Public Bioethics (Baltimore: University of Maryland, 1994).

32. Id. at 13.

33. G. Dworkin, "Law and Medical Experimentation: Of Embryos, Children and Others with Limited Legal Capacity," Monash University Law Review, 13 (1987): 189-208.

34. Id. at 208.

35. B. Miller, "Autonomy and Proxy Consent," IRB: A Review of Human Subjects Research, 4, no. 10 (1982): at 1, 2.

36. Additional protections may also be necessary when the research presents a high risk of harm to the subjects, or when the subject population is considered unusually vulnerable to coercive tactics designed to induce participation.

37. Miller, supra note 35, at 3 (specific authorization; gen-

eral authorization with instructions; general authorization without instructions; instructions without authorization; substituted judgment; and deputy judgment). He concludes that all but one of the models, deputy judgment, protects autonomy in one or more of the four aspects. I do not use Miller's categories; instead, I address the standards of proxy decision making (best interests and substituted judgment) allowed in different contexts.. Miller's analysis of autonomy, however, is useful in explaining why one model of proxy consent is preferable over another.

38. Although Wicclair recognizes the existence of these documents, he does not specifically allude to them in his seven conditions. In his comments, however, he notes that "speculative judgments are discouraged by requiring explicit instructions and criteria to be given to surrogates." W:cclair, supra note 28, at 183.

39. American College of Physicians, supra note 26, at 884. 40. See T. Sunderland and R. Dukoff, "Informed Consent with Cognitively Impaired Patients: An NIMH Perspective on the Durable Power of Attorney," Accountability in Research (forthcoming, 1996) (evaluating the NIH Clinical Center Policy).

41. C. Sabatino, "Legislative Trends in Health-Care Decisionmaking," ABA Bioethics Bulletin, 3, no. 2 (1994): 10—11 (Advance directive is used in the study as an all encompassing term that includes living wills and DPAs).

42. See E. DeRenzo, "Surrogate Decision Making for Severely Cognitively Impaired Research Subjects: The Continuing Debate," Cambridge Quarterly of Healthcare Ethics, 3 (1993): at 543.

43. A number of states either do not permit a guardian to consent to research or only allow consent with prior court approval. Many of these states make a distinction between lifesaving potentially therapeutic experimental treatments (which may be allowed) and nontherapeutic research. Alaska Stat. S 13.26.150 (1995); Fla. Stat. Ann. S 744.3215 (West 1995); Ill. Rev. Stat. ch. 405, para. 5/2-10 (1995); Minn. Stat. S 525.56 (Supp. 1995); Nev. Rev. Stat. S 159.0805 (1993); N.H. Rev. stat. Ann. S 464-A:25 (1994); N.D. Cent. Code S 30.1-28-12(5312) (1995); Okla. stat. Ann. tit. 30, S 3-119 (Supp. 1995); and 20 Pa. Cons. Stat. Ann. S 5521 (1995).

Other states either explicitly or implicitly allow guardians to consent to experimental treatment or research. Cal. Health & safety Code S 24173 (Supp. 1995); D.C. Code Ann. S 61969 (1994); Ga. Code Ann. S 31-8-108 (Michie supp. 1995); Stat. Ann. 59-3018 (Supp. 1995); Mo. Ann. Stat. S 431.064 (Vernon 1995); Mont. Code Ann. SS 53-20-147, 53-21-147 (1993); N.J. stat. Ann. S 30:6D-5 (West 1995); N.Y. Pub. Health Law S 2442 (McKinney 1995); Wis. Stat. Ann. S 51.61 (1995); P.R. Laws Ann. tit. 24, S 4020 (1990); S.D. Codified Laws Ann. S 27B-8-20 (Supp. 1995); and Va. Code Ann. S 32.1-162.16 (Supp. 1995) (if nontherapeutic only if minor increase over minimal risk).

44. 42 U.S.C. SS 1395cc, 1396aa (1990).

45. See DeRenzo, supra note 42, at 546.

46. The term surrogate will be used rather than guardian. A guardian is a court-appointee who has much greater control over an incompetent individual's life than a surrogate decision maker. Casasanto, Simon, and Roman note a historical disagreement about whether a guardian should act in the ward's best interests or as a surrogate using a substituted judgment stan-

dard. M. Casasanto, M. Simon, and J. Roman, "A Model Code of Ethics for Guardians," Whittier Law Review, 11 (1989): at 545. Casasanto, Simon, and Roman suggest that a substituted judgment standard is preferable, and that a best interests standard only be used when there is no indication of the ward's preferences. Id. at 547. In addition, they state that the guardian should be careful not to act beyond the scope of his/her authority and should allow wards to make decisions on their own behalf when appropriate. Thus, in situations where the research subject has already had a guardian appointed, or in states where the only surrogate decision makers legally provided for are guardians, such guardians may function in essentially the same manner as any other surrogate decision maker. Their proposed model code's Rule 4 recognizes the guardian's responsibility to make medical decisions; it states, however, that the guardian cannot consent to "experimental treament ... without seeking review by the court or the ward's attorney or other representative." Id. at 561. This seems unnecessarily burdensome. The guardian in these situations should be as free to act as any other surrogate— being permitted to make decisions within the confines of the applicable guidelines.

47. A.A. Hamann, "Family Surrogate Laws: A Necessary Supplement to Living Will and Durable Powers of Attorney," Villanova Law Review, 38 (1993): at 106 ("All states have intestacy laws to provide for the of the decedent's property without a will. A comparable alternative is needed to provide for those individuals who fail to execute a living will or a durable power of attorney.").

48. NIH Clinical Center Policy, supra note 27, cases 5, 6.

49. N. Dubler, "Some Legal and Moral Issues Surrounding Informed Consent for Treatment and Research Involving Cognitively Impaired Elderly," in M.B. Kapp, H.E. Pies, and A.E. Doudera, eds., Legal and Ethical Aspects of Health Care for the Elderly (Ann Arbor: Health Administration Press, 1986): 247—57. Dubler argues that the courts are an improper forum to make health care decisions; however, she states that "there is a class of patients, primarily the long-term institutionalized, retarded, and mentally ill, who are demonstrably so vulnerable that public judicial process is required." Id. at 252. There is no reason why these groups should be singled out for differential treatment. All cognitively impaired subjects face the same questions of informed consent and substitute decision making, some, perhaps, to a greater degree than others. The courts are no better arbiters of the research process for the groups identified by Dubler than for others; recourse to the judicial system merely lends a false sense of legitimacy. Rather, clear guidelines should be set forth that limit what types of research can be done with cognitively impaired persons, regardless of the source of the impairment. Furthermore, local review boards should be responSible for taking into account additional considerations that are unique to particular groups, for example, coercion in an institutionalized setting.

50. Sabatino, supra note 41, at 10—11. These statutes do not presently apply to research decision making.

Missouri is the one exception as it specifies a hierarchy of proxy decision makers who can consent to research participation. Mo. Ann. Stat. S 431.064 (Vernon 1995).

51. Hamann, supra note 47, at 166. This has led to such anomalous scenarios as the Christian Fellowship of the Disabled being appointed as the guardian of a permanently unconscious individual who had no previous affiliation with the group, or the Americans United for Life Legal Defense Fund being allowed to argue against a family's decision to remove a feeding tube when the guardian ad litem was unable to find a physician to testify in favor of continuing treatment.

52. See infra notes 90—91 and accompanying text.

53. See Hamann, supra note 47 (arguing that the family is the most appropriate decision maker).

54. See, for example, Cruzan v. Diæctor, Missouri Dep't of Health, 497 U.S. 261, 286—87 (1990) (recognizing that a number of states permit close family members to make surrogate treatment decisions, but holding that the U.S. Constitution does not require that a state do so in all circumstances).

55. See, for example, Illinois Health Care Surrogate Act, Ill. Rev. Stat. ch. 755, S 40/25 (1993); Tex. code Ann. Health & safety S 313.002(6) (Supp. 1995); and Va. code 54.1-2986 (1994). Even under statutes that set out a hierarchy, it may be necessary to choose among people in a given class, for example, between a number of adult children or siblings. In such situations, some statutes authorize all the members of the class to make decisions and require either a majority or a consensus (Illinois). Others specify criteria, such as age, expressed interest, or availability of a particular member of the class (Virginia). Lastly, some statutes leave the ranking among class members unspecified, and thus open the possibility that the physician will have to exercise discretion (Texas). In situations where such a choice must be made, additional scrutiny may be appropriate.

56. American College of Physicians, supra note 26, at 845; NIH Clinical Center Policy, supra note 27, at 2; and Wrcclair, supra note 28, at 182.

57. See generally, S. Busby-Mott, "The Trend Towards Enlightenment: Health Care Decisionmaking in Lawrence and Doe," Connecticut Law Review, 25 (1993): 1159-225.

58. Miller, supra note 35.

59. See, for example, American College of Physicians, supra note 26, at 844; and E.W. Keyserlingk et al., "Proposed Guidelines for the Participation of Persons with Dementia as Research Subjects," Perspectives in Biology and Medicine, 38 (1995): 319— 62 (condition 10, at 352).

60. Compare Keyserlingk et al., supra note 59, at 351 (restricting the use of advance directives for research involving more than a minor increment over minimal risk except where the subject has experienced a similar level of pain or discomfort before consenting to the research).

61. States' laws vary on this point. For example, a judicial determination of incompetence may be necessary. Other smtes may require that a court the proxy decision. Compare, for example, Superintendent ofBelchertown State School v. Saikewicz, 370 N.W.2d 417 (Mass. 1977) (requiring court review) with Matter of Quinlan, 355 A.2d 647 (N.J. 1976) (noting that court review of all decisions would be unduly burdensome).

62. The "evidence" may include written preferences, such as those found in a general advance care document; practices throughout the subject's life (that is, previous research participation); and verbal communications regarding research.

63. American College of Physicians, supra note 26, at 845.

64. Wicclair, supra note 28, at 182 (conditions I(a), 2(a)).

65. Id. (condition 5).

66. See, for example, Cruzan u Director, Missouri Dep't of Health, 497 U.S. 261, 286-87 (1990) (holding that a state may require clear and convincing evidence of an incompetent patient's wishes before allowing the withdrawal of life support). It is not clear that all courts will apply the clear and convincing evidentiary standard, even to life-support withdrawal. However, in the research context, more so than the life-support situations, it is better to err on the side of caution. In those cases with no therapeutic benefit, no harm comes from not allowing participation except the generalized harm of not allowing implementation of an autonomous decision. This is the price society has to pay for protecting individual autonomy, because one cannot perfectly substitute one person's judgment for another's.

The suggestion to use evidentiary standards is not meant to imply that courts should be scrutinizing the evidence in each case; on the contrary, courts are the least appropriate forum for these types of decisions.

67. But see Keyserlingk et al., supra note 59, at 351 (allowing only the use of a written directive).

68. American College of Physicians, supra note 26, at 845. 69. Because the American College of Physicians paper only allows use of a subject's advance directive, this position does not allow a proxy decision maker to take into account other evidence of the subject's wishes. Thus a proxy could not consent for a subject who, although he/she had not executed an advance care document, has left other evidence that he/she would not want to participate in a research protocol, even if participation was in the subject's best interests. In this sense, the paper only permits a proxy to take into account evidence of a subject's wishes. By contrast, my position, outlined in the previous paragraphs, would allow a proxy also to consider evidence of a subject's positive attitude toward participation in research. Id. 70. Wrcclair, supra note 28, at 182 (conditions I(b), 2(b)). 71. American College of Physicians, supra note 26, at 844.

72. J. Lazarus, "Ethical Issues and Conflicts," Journal of the California Alliance for the Mentally Ill, S (1994) 20—21 (describing a patient who competently consented to research that involved taking him off medication, subsequently regressed and refused to take the experimental medication, and, later, after stabilizing again on the initial medication, "earnestly begged for another chance to try the experimental medication." Id. at 20.).

73. Id.; see also J. Warren et al., "Informed Consent by Proxy," N. Engl. J. Med., 315 (1986): 1124—28 (suggesting that investigators should reconsider when research begins, initiating contact while the subjects are still competent and thus can prospectively consent).

74. See, for example, Keyserlingk et al., supra note 59, at 349 (discussing the use of the "Ulysses Contract" (a self-binding psychiatric advance directive) in dementia research). 75. Fletcher and Wichman, supra note 18, at 383.

76. Wicclair, supra note 28, at 182 (condition 6).

77. If state law does not allow for treatment over an incompetent patient's refusal, research, even if it holds the potential for direct therapeutic benefit, would likewise be impermissible. 78. See infra pages 24—28.

79. State laws governing treatment refusals of incompetent patients would apply here. Thus it may be necessary to get a court order declaring the individual incompetent or to appoint a surrogate.

80. See, for example, Keyserlingk et al., supra note 59.

A consent auditor and the research auditor may be the same person. The different titles simply highlight the different roles— the consent auditor monitors the initial consent process to ensure that the subject has the ability to consent and actually understands the information. The research auditor would monitor the experiment to ensure that continued participation is appropriate given either changes in the protocol or changes in the subject's physical and mental state.

81. See Lazarus, supra note 72, at 21 (noting the "additional advocacy roles for family or friends of the patients").

82. National Commission for Protection of Human Subjects of Biomedical and Behavioral Research, supra note 8, at 5.

83. The phrase "research has the potential to produce direct therapeutic benefits" is used in place of "therapeutic research" to avoid confusion. The reason will become clear in the subsequent discussion of potential therapeutic benefits.

Indirect benefits are not discussed in this analysis. Although they are likely to be included in a competent subject's evaluation about whether to participate in a study, they are unlikely to be appropriate considerations for a surrogate decision maker. See Keyserlingk et al., supra note 59, at 327—28 (identifying direct versus indirect benefits).

84. See, for example, A.D. Milliken, "The Need for Research and Ethical Safeguards in Special Populations," Canadian Journal of Psychiatry, 38 (1993): at 683 (discussing the need to balance risk and potential benefit). In this section, I will not go into detail about the risks that IRBs should consider when deciding whether to accept a particular protocol. For my purposes, I assume that the protocol has already passed IRB review and that at issue is whether cognitively impaired subjects should be allowed to participate. Because some protocols are designed to include only cognitively impaired subjects, in some sense the protocol approval will overlap with the analysis presented here. For example, it would be useless to "approve" a protocol involving severely cognitively impaired subjects, but not to allow proxy consent to participation.

85. Agreement also has not been reached on the definitions of more fundamental terms such as harm and risk. See, for example, E. Meslin, "Protecting Human Subjects from Harm Through Improved Risk Judgments," IRB: A Review of Human Subjects Research, 12, no. 1 (1990): 7—10 (comparing various definitions of harm and risk and proposing a model for risk judgments); and T. Beauchamp and J. Childress, Principles of Biomedical Ethics (New York: Oxford University Press, 1994) (discussing risk assessment, uncertainty, and risk perception).

86. 45 C.F.R. S 46.102(1) (1991); and 21 C.F.R. S 50.3(1) (1993).

87. A 1995 consensus statement regarding informed consent in emergency research proposed a new risk category— "appropriate incremental risk." This is defined as "any potential risk associated with participating in the research protocol relative to the natural consequences of the medical condition, or any potential risk associated with receiving the experimental intervention relative to receiving the standard intervention for. the medical condition." M.H. Biros et al., "Informed Consent in Emergency Research: Consensus Statements from the Coalition Conference of Acute Resuscitation and Critical Care Researchers," JAMA, 273 (1995): at 1286. See also Keyserlingk et al., supra note 59, at 329 (noting that minimal risk must be evaluated in light of the subject's "everyday care and treatment." Thus the baseline of risk for a person with Alzheimer's would be higher than for a healthy person.).

88. Current DHHS regulations allow parents to consent to research involving their children that involves greater than minimal risk and no potential for direct therapeutic benefit when it is likely to yield generalizable information about the subject's condition. 45 C.F.R. S 46.406 (1991). The regulations state that the risk must "represent a minor increase over minimal risk." Id. Furthermore, the regulations require the approval of the secretary of DHHS. I suggest instead that when a research pro.tocol involves risk that is too high for IRB approval, review be the responsibility of a national review board. See infra pages 28.

89. It is hard to quantify benefits. For example, how do you compare potential life-spans—living pain-free and out of the hospital for two weeks versus confinement to a hospital bed for two months? Nonvithstanding, it is possible for IRBs to make general benefit calculations.

90. This simplifies the analysis greatly. The percentages, while being difficult to predict initially, may also not be this clear cut. For example, a drug could be likely to help 20 percent of the people 100 percent, 10 percent of the people 50 percent, and 2 percent of the people 30 percent. For the sake of argument, simple statistics are used.

91. As the potential for direct therapeutic benefit approaches zero, greater protections should apply.

92. I will not discuss protections in the treatment context.

If these were to be changed or augmented, similar protections should be applied to the research context.

93. If the subject in question has been adjudicated as incompetent, he/she can neither consent to research nor execute an advance care document.

94. See generally, J. Robertson, "Taking Consent Seriously: IRB Intervention in the Consent Process," IRB: A Review of Human Subjects Research, 4, no. 5 (1982): 1—5 (describing ways in which IRBs could improve consent forms and the consent process through the use of competency testing, monitoring, and restructuring).

95. Therapeutic misconception refers to a phenomenon that occurs when the investigator is viewed as both a researcher and an attending physician. The result is a blurring of the two roles and a misconception that the research is primarily conducted for the direct therapeutic benefit of the patient-subject. Both physicians and patients are subject to this phenomenon.

96. P.S. Appelbaum, L. Roth, and C. Lidz, "The Therapeutic Misconception: Informed Consent in Psychiatric Research," International Journal of Law and Psychiatry, 5 (1982): 319—29.

97. P Benson, L. Roth, and W Winslade, "Informed Consent in Psychiatric Research: Preliminary Findings from an Ongoing Social Science and Medicine, 20 (1985):

98. Id. at 1337.

99. B. Stanley, M. Stanley, and A. Lautin, "Informed Consent and Competency in Psychiatric Research," (unpublished manuscript, on file with the author, 1985) (medication use did not affect capacity, but diagnosis correlated with poorer decision making, that is, persons with schizophrenia were more likely chose to participate in high risk, low benefit studies).

100. Appelbaum, Lidz, and Meisel, supra note 14, at 248— 58. See also F. Grinnell, "Endings of Clinical Research Protocols: Distinguishing Therapy from Research," IRB: A Review of Human Subjects Research, 12, no. 4 (1990): 1—3 (noting that many subjects at the end of a research protocol are unaware that a potentially beneficial treatment may be discontinued).

101. See, for example, Robertson, supra note 94.

102. A low potential could mean either that the magnitude of benefit is low or that the percentage calculation approaches zero. The former should be treated cautiously because, as previously mentioned, it primarily involves an individual assessment. The latter could occur either because of the low probability that the treatment will help or because of the experimental design.

103. See National Commission for Protection of Human Subjects of Biomedical and Behavioral Research, supra note 8, at 5: "Respect for the immature or incapacitated may require protecting them as they mature or while they are incapacitated.... The extent of protection afforded should depend upon the risk of harm and the likelihood of benefit."

104. American College of Physicians, supra note 26, at 844. 105. To some extent, guidelines can carve out extremes along the continuum but not the middle. Thus one can specify restrictions on minimal and high-risk research, but appropriate protections for research that falls into the category of greater than minimal risk but lower than high risk should be determined by local IRBs. Factors to consider include the degree of risk (higher = greater protections) and the potential for direct therapeutic benefit (lower = greater protections).

106. See infra case 7.

107. Wlcclair, supra note 28, at 182 (condition 108. Id. at 182 (condition 3).

109. See, for example, American College of Physicians, supra note 26, at 844. This, in effect, is a situation where advance consent is sought from the subject in anticipation of later incompetence. For example, Alzheimer's patients in the first stage of the illness may want to consent to participation in high-risk research to be conducted during the later stages of the disease (for example, research designed to investigate the disease process in hopes of eventually finding a cure, but yielding no present direct therapeutic benefit to the subject). If a proxy is appointed to ensure that participation is still appropriate at a later date, such research should be allowed. As mentioned earlier, however, subject cooperation remains essential.

In setting up these guidelines, I want to distinguish the court cases that have allowed a parent or guardian to consent to the donation of an organ of an incompetent person to a sibling. See, for example, J.B. Korins, "Curran v. Bosze: Toward a Clear Standard for Authorizing Kidney and Bone Marrow Transplants Between Siblings," Vermont Law Review, 16 (1992): 499—539. These holdings have been based on a stretching of the best interests analysis to include the emotional harm to the incompetent person that would result from the loss of a sibling. Although this is a dubious interpretation of best interests, it is not improbable that, in these extreme situations, a court may authorize such measures. Even so, I choose not to base my general recommendations on this practice.

110. Compare the NIH Clinical Center Policy of having a bioethics consultant as a nonvoting member of an IRB. E. DeRenzo and A. Wichman, "A Pilot Project: Bioethics Consultants as Non-Voting Members of IRBs at the National Institutes of Health," IRB: A Review of Human Subjects Research, 12, no. 6 (1990): 6—8. In general, the use of such consultants was con-

sidered helpful (id.; and E. DeRenzo and F. Bonkovsky, "Bioethics Consultants to the National Institutes of Health's Intramural IRB System: The Continuing Evolution," IRB: A Review of Human Subjects Research, 15, no. 3 (1993): 9—10). However, most bioethics consultants will need additional training to serve in this capacity. DeRenzo and Wichman, id. at 8 ("individuals who were more familiar with the performance of human subjects research were appropriately perceived as better able to assist the ICRSs").

111. see 43 Fed. Reg. 53 (Mar. 17, 1978). The proposed regulations for research involving those "institutionalized as mentally infirm" stated that, in certain cases, the IRB should appoint a consent auditor to monitor consent.

Current regulations refer to the obligation of the IRB or subject-advocates to oversee the consent process in connection with activities involving fetuses, pregnant women, or human in vitro fertilization. See 45 C.F.R. S 46.205(a)(2) (1991).

112. P.M. Bein, "Surrogate Consent and the Incompetent Experimental Subject," Food and Drug Cosmetic Journal, 46 (1991): 739-71.

113. Where there is no evidence of the subject's wishes, the best interests standard applies and participation would be barred. Likewise, where there is some but insufficient evidence, the high level of certainty needed would not be met and thus participation would be barred.

114. R. Delgado and H. Leskovac, "Informed Consent in Human Experimentation: Bridging the Gap Between Ethical Thought and Current Practice," UCLA Law Review, 34 (1986): at 116.

115. The above analysis assumes that the IRB has already approved the research. Thus, at issue here is whether the participation of an individual subject is permissible.

116. See NIH Clinical Center Policy, supra note 27, case 8.

117. Research that involves a significant degree of risk (that is, greater than minimal) and offers little or no therapeutic benefit would never be in a person's best interests. See American College of Physicians, supra note 26, at 845. Even if one assumes some altruistic gain from participating in experimentation, this would be outweighed by the high risk. Thus a surrogate decision maker using the best interests standard should never consent to such experiments.

118. See Miller, supra note 35, at 6 (arguing that "deputy judgment" where there is "no authorization of the agent by the principal, and the agent is not able to make a decision that is implied by the principal's values and attitudes ... does not respect the autonomy of the principal in any of the four aspects of autonomy.

119. H. Jonas, "Philosophical Reflections on Experimenting with Human Subjects," in P.A. Freund, ed., Experimentation with Human Subjects (New York: G. Braziller, 1970): at 3—4.

120. McNeill, supra note 20, at 155 (McNeill also notes that "[tlhere needs to be an authority to define the cases to which the rule applies because it is not possible to specify in advance all the situations to which the rule or principle applies." Id. at 156). But see Keyserlingk et al., supra note 59, at 334 (arguing that even a national review board should not have the authority to allow research under these circumstances).

121. See DeRenzo, supra note 42, at 545 (discussing the benefits of a public forum).

122. National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research, supra note 25, at 1. See also Biros et al., supra note 87, at 1286 ("Because the local IRB has good insight into local medical practice, the local patient population, and the capabilities of local researchers, institutions, and resources, the IRB should be the monitoring body primarily responsible for maintaining vigilant oversight of clinical trials of emergency research.").

123. National Institutes of Health Office for the Protection from Research Risks, Institutional Review Board Guidebook: Protecting Human Research Subjects (Washington: D.C.: U.S. Government Printing Office, 1993): 6-26 to 6-32.

124. Id. at 6-41.

125. Id. at 6-47.

126. Id. at 6-31 ("IRBs should be aware of any applicable law in their state particularly those relating to consent by family members on behalf of persons incapable of consenting on their own.").

127. See, for example, J. Kane, L. Robbins, and B. Stanley, "Psychiatric Research," in R. Greenwald, M.K. Ryan, and J. Mulvihill, eds., Human Subjects Research (New York: Plenum, 1982): 193—205 (proposing specific issues for IRBs to consider with regard to subjects who have a mental illness, but noting that "the population has much in common with other patient populations. Therefore issues such as risk-benefit ratio of the proposed research must be addressed as with nonpsychiatric research." Id. at 194.); and E.G. DeRenzo, "The Ethics of Involving Psychiatrically Impaired Persons in Research," IRB: A Review of Human Subjects Research, 16, no. 6 (1994): 7—9, 11.

128. See, for example, DeRenzo, supra note 127, at 9 (discussing the use of a "Ulysses self-binding psychiatric advance directive); and Casasanto, Simon, and Roman, supra note 46, at 549 (discussing the role of a guardian when the ward "has a cyclical impairment such as severe depression." Casasanto, Simon, and Roman conclude that "ethical principles favoring self-determination seem to dictate that the wishes expressed by the person be adhered to if a person is in a lucid state, despite the judicial determination that he is incompetent. [Thus the] Model Code recognizes ... that a guardian is obligated, in limited situations, to respect the wishes of the ward even if contrary to the guardian's notion of best interest. ").

One danger that should be acknowledged is the guardian's potential liability. If the guardianship order is not limited, the guardian may be liable for allowing the ward to consent to a procedure that is not in his/her best interests. This can be avoided either by limiting the scope of the guardianship order or by expressly permitting the guardian to use a substituted judgment standard rather than a best interests analysis.

129. See, for example, Letter to Ron Wyden from Subcommittee Staff (Mar. 11, 1994) (highlighting abuses in this con-

text); Biros et al., supra note 87; and Protection of Human Subjects, 60 Fed. Reg. 49,086 (proposed Sept. 21, 1995).

130. Letter to Ron Wyden from Subcommittee Staff (Mar. 11, 1994). The present federal regulations appear to be inconsistent. Compare 45 C.F.R. S 46.116(d) (1991) with 21 C.F.R. s 50.236) (1993).

Copyright of Journal of Law. Medicine & Ethics is the property of Blackwell Publishing Limited and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However. users may print. download. or email articles for individual use.