Updating Protections for Human Subjects Involved in Research

RESEARCH involving human subjects funded by the federal government has been subject to policies and regulations requiring local institutional review and oversight for more than 30 years.^{1 - 2} Nonfederally funded research also largely follows these rules. Backed by federal law, these rules evidence a societal commitment to the advancement of scientific knowledge provided that the advances occur in accord with ethically sound principles and practices.

The combination of federal and local control embodied in the current regulations constitutes a sound regulatory and philosophical approach, but the extent to which the current system protects the rights and welfare of research subjects while furthering important research goals is less clear. The last major revision of the Federal Policy for the Protection of Human Subjects, a joint effort of the Department of Health and Human Services and the Food and Drug Administration (FDA), was in 1981. The policy was reissued without change in 1991 and endorsed by 15 departments and agencies of the federal government (hereinafter referred to as the Common Rule³ ). There has been no major reexamination of the regulation of human subjects research for more than 17 years.

Regulatory changes cannot be everyday occurrences, but regulations must evolve along with changes in the practices to which they are to apply.⁴ The largely self-policing nature of human subjects research oversight leaves it open to charges of institutional inertia in failing to keep pace with new developments.^{4 - 5} Furthermore, since the last major modifications in the federal rules in 1981, there have been significant changes in the nature of clinical science, in the way that research is funded, and in the institutional environment of research.

The model research activity for which the current regulations were designed was that of a solo investigator, working in a single institution, with a discrete cohort of subjects. Typically, that investigator submitted a protocol to the department chair and a local institutional review board (IRB) for approval prior to submission of a grant application to the National Institutes of Health (NIH) for funding.

This model now accounts for only a minority of investigator-initiated individual project (RO-1) awards each year. Many RO-1s do not involve human subjects but are preclinical or "basic" research.^{5 - 6} Today, many research subjects participate in multisite studies, conducted simultaneously at several institutions in the United States and, often, in other countries as well.⁷ Multisite trials require many research investigators and a large number of subjects. Multisite research is scarcely addressed in the current regulations, although it creates unique problems for IRBs, the informed consent process, and the monitoring of research activity.⁴

Problems with specific subject populations that many consider particularly vulnerable have also arisen, reflecting issues that were unresolved when the regulations were promulgated. Data are lacking on the exact number of human research subjects and what proportion are at risk, but there have been a sufficient number of cases of dubious research practices with vulnerable populations to raise concern about the adequacy of the existing regulations.^{8 - 11}

In addition, the FDA has promulgated a "narrow exception" to federal informed consent requirements for some research in emergency settings.¹² Other complicating elements in the research environment include the changing conception of informed consent requirements,^{13 - 15} the desire by some who are ill to have greater access to research drugs,^{16 - 17} and the view that women and children have been underrepresented in clinical studies.¹⁸

Another significant development since the last revision of the regulations is the greater presence of privately sponsored research.^{19 - 20} Concerns about investigators' and institutions' actual or potential conflicts of interest have intensified as the proportion of privately sponsored research (which may not be peer reviewed) has increased relative to federally sponsored research. Financial pressures on researchers may influence the conduct of clinical trials, and there is concern that, in some settings, scientific investigators are compensated according to the number of subjects who complete protocols. Although professional pressures such as the need to conduct research have long been a part of the scientific enterprise, these factors are augmented by direct or potential financial interests on the part of investigators.^{20 - 21} Aside from modest guidance aimed at IRB members, the current regulatory system was not designed to give guidance about conflicts of interest.

As a result of the new developments in research design and methods, problems with research conducted with members of vulnerable populations, changes in the financing and organization of research, and calls for greater scrutiny of the adequacy of IRBs, it is an opportune time to assess and reevaluate the current protections for human subjects involved in research.

The Human Research Ethics Group is part of the Project on Informed Consent funded by the Annenberg Public Policy Center and administered by the Center for Bioethics at the University of Pennsylvania Health System. The group includes representatives of the disciplines of medicine, nursing, law, philosophy, political science, psychology, and sociology. About half of the group members are faculty of the University of Pennsylvania, Philadelphia, with the rest drawn from other institutions. Members have long-standing interests in issues concerning the use of human research subjects, and some bring perspectives from professional experience in federal regulatory agencies. The Human Research Ethics Group's goals were (1) to conduct a systematic and comprehensive review of the current state of the procedures and regulations concerning the use of human subjects in biomedical and behavioral research in the United States and (2) to offer recommendations for the improvement and reform of these procedures and regulations.

The Human Research Ethics Group benefited from the work of 2 important governmental studies relating to the current state of human subjects regulations. The studies were produced by the President's Advisory Committee on Human Radiation Experiments (ACHRE)²² and the General Accounting Office (GAO).⁶ Some of the observations in these reports relate to our recommendations.

The ACHRE issued its final report in October 1995. The committee found that at least some patients are confused about the distinction between research and therapy, a confusion that is especially troubling in procedures that expose individual subjects to greater than minimal risk and do not hold out the prospect of direct benefit to them. ACHRE recommended that IRBs allocate more time to reviews of such studies, that information provided to potential subjects explain risks and benefits more clearly, that federal agencies sponsoring research be identified to subjects, and that the financial implications of participating (or not) be clarified for potential subjects. Expressing doubts about the adequacy of current oversight mechanisms, the committee stated that "the adequate protection of human subjects requires that the system be subjected to regular, periodic evaluations that are based on an examination of outcomes and performance and that include the perspective and experiences of subjects of research as well as the research community."^{22 (pp527-528)}

In its March 1996 report, Scientific Research: Continued Vigilance Critical to Protecting Human Subjects, the GAO noted the heavy workloads faced by IRBs, registered concerns about the thoroughness of IRB reviews, and identified several challenges to the independence of local evaluation in light of collegial and funding pressures. In its conclusion, the GAO report notes that "federal and IRB reviewers rarely observe the interaction between researchers and subjects during the informed consent process or throughout the course of the study."^{6 (p24)}

Not only is there a need to update federal regulations to cover certain subject populations and to ensure the thoroughness of IRB review, but the responsibilities of the IRB system need to be better matched by the resources at its disposal. Unless these problems are addressed, the review process will be unable to achieve its ultimate goal of ensuring that research subjects are protected, truly informed, and voluntary participants in the scientific enterprise.

The following recommendations for reform of the current system for the protection of human research subjects constitute a consensus of the members of the Human Research Ethics Group. These judgments are especially commended to appropriate federal and state regulatory agencies, to the National Bioethics Advisory Commission, and to professional organizations of scientific investigators who work with human subjects.

The recommendations are divided into 3 sets: recommendations for populations in need of special protections, recommendations for IRBs, and a list of policy issues that require further study. Within each set, the recommendations are ordered from the most important general changes to more specific suggestions.

A. For individuals planning for subsequent possible loss of capacity, state statutes for durable power of attorney for health care should be amended to permit surrogate consent for research involving interventions that hold out the prospect of direct personal benefit to individual subjects or that entail no more than a minor increase over minimal risk to the subject.— Statutes governing the durable power of attorney for health care operationalize an ethical consensus that competent adults should be able to determine in advance who they would want to represent them in making their health care decisions, should they become incapacitated. These surrogate decision-making arrangements have been favorably received by patients, families, and health care professionals.

In practice, decision makers appointed by patients (usually family members or close friends) are typically viewed by most in our society as acceptable surrogate decision makers if certain conditions are met.²³ These conditions include a presumption that the agent is acting according to the wishes of the patient, so far as they are known, and that a proposed intervention holds some potential for benefit to the patient or no more than a minor increase over minimal risk. However, the appointed agent may not have the legal authority to permit the patient to be enrolled in research, even that which involves procedures that may be beneficial to the individual subject. Statutes governing the durable power of attorney for health care can be read as allowing the patient's representative to authorize only standard therapy on behalf of the patient.

The possibility of enrollment of an incompetent subject in research involving interventions that are potentially beneficial to the individual patient-subject and in research that involves minimal incremental risk should be part of the durable power of attorney for health care authority for several reasons: it is an expression of patient autonomy, it is an opportunity for the surrogate to act for the potential benefit of the now incompetent subject, and it might benefit future patients and therefore society in general. But such a "research advance directive" should only be permitted if the individual has explicitly recorded his or her wishes concerning a surrogate decision maker for research participation.

Advance directives for clinical decision making are not used as frequently as had been hoped. However, clinical investigators seeking to recruit subjects who may become incompetent will have a strong incentive to call potential subjects' attention to research advance directives. Advance consent planning should include the identification of a surrogate who can make decisions about continuing or withdrawing a subject from a study in light of the subject's known preferences and the actual evolving net benefit of study participation as the study proceeds.

Another important consideration in research with those who are cognitively impaired stems from an analogy with the regulations governing pediatric research. Under certain conditions, investigators are required to obtain assent from children who are potential subjects. Arguably such an assent provision should also apply to cognitively impaired patients, including those who have completed advance directives of the kind contemplated in this recommendation.

Finally, statutes authorizing the creation of a research advance directive should not be interpreted as excluding the authority of surrogates to permit participation of individual patients in clinical studies from which there is a reasonable expectation of direct benefit to the patient.

B. Special provisions concerning cognitively impaired persons should be added to the federal regulations.— The number of persons with dementia is expected to increase dramatically in our aging population. Should these neurodegenerative illnesses result in serious impairment in decision-making capacity, there may be substantial barriers to important research.²⁴ Yet much can be gained from research with the cognitively impaired that will benefit them or, more likely, future patients. Accordingly, we propose that specific provisions be added to federal regulations to deal with cognitively impaired persons.²⁵

New regulations should include a requirement that principal investigators proposing research with human subjects who are cognitively impaired or who are at foreseeable risk of becoming cognitively impaired during the research process include a written section in the protocol that addresses the importance of the research and an assessment of the risks and benefits for the subjects. Material such as this is already generally included in protocols for research with cognitively impaired persons, and current regulations may be read as requiring it, but these conditions should be explicit.

A written description concerning the expected degree of impairment of the subjects, as well as the means by which it will be determined that they have lost the ability to make decisions regarding research participation, should also be specifically required by federal regulations. An example of the kind of issue that should be anticipated is how it will be determined that an impaired patient-subject should continue in or be withdrawn from the study and who will make such a determination.

Institutional review boards and investigators should be sensitive to the modifications that may be needed in informed consent procedures for subjects with varying degrees of cognitive impairment. For protocols involving incompetent subjects, IRBs should be specifically required to determine that risks are justified and that alternative decision-making arrangements are both ethically and legally adequate. For example, IRBs should determine that alternative decision-making measures are in place that will be used for any subject who is or becomes incompetent during the course of the study, including the appointment of surrogate decision makers, in accordance with pertinent state law.

Current regulations authorize IRBs to require consent procedures be monitored, a particularly appropriate measure for research involving those with cognitive impairments. However, this requirement is rarely mandated by IRBs, a deficit we address in later recommendations.

C. The recruitment and retention of women, minorities, and children in biomedical and behavioral research is an important national goal and should be vigorously pursued.— Federal policies and initiatives in this area have been in place for several years, representing agreement that the potential benefits as well as burdens of research participation should be genuinely available to all societal groups and that the information generated by research should be of benefit to all. But the pace and scope of improvement with regard to fairness in subject selection are unclear. Increased access to and the opportunity to participate in clinical trials are important national goals in order for science to function ethically and thereby to be worthy of the public trust.

Minority groups have been subject to well-known cases of exploitation in medical research.^{26 - 27} Research involving minority participation engages issues of cultural and linguistic differences and the potential for increased risk of coercion and exploitation. The resulting fragility of trust must be faced by the medical community even as it strives for greater equity in research participation among all segments of society.

Although there appears to be growing support in the scientific community for inclusion of women of reproductive age in clinical research, debate continues about the proper stance toward pregnant women as research subjects.¹⁸ Pregnant women should not be arbitrarily excluded from studies, including trials of drugs, that present a favorable balance of risks and benefits for them and their fetuses. The question is not that of the legal relationship between a mother and her fetus, but rather who, as a matter of policy, should be considered to be in the best position to make decisions on behalf of the fetus. The presumption should be that pregnant women are.

Like women, children can also be the recipients of medication that has not undergone thorough scrutiny for this patient population. In spite of formidable methodological challenges, drugs and devices that are likely to be used in children should be subject to systematic investigation to establish dosage rates, contraindications, and secondary effects in this population.

D. Consent forms should include clear language that discourages unrealistic expectations in research.— Some subjects in phase 1 studies assessing toxic effects are seriously or terminally ill. Compared with the attention paid to other "vulnerable" populations, the extent to which patients like these may be motivated to consent to participate in research by unrealistic expectations of benefit has not been adequately studied. Yet, phase 1 studies are very important to the overall progress of science and they must be conducted.

Ample evidence, both direct and indirect, indicates that many research subjects in these trials do not fully appreciate that they are unlikely to benefit from participation.^{28 - 30} Admittedly, many of these patients would consent to participate even knowing of the remote chance of benefit, but the research community is not thereby relieved of the obligation to take all reasonable measures to promote their understanding.

The language of many consent forms encourages an illusion of therapeutic benefit, whether intentionally or not.²² For example, many phase 1 trial consent forms refer to "treatments" that will be given to subjects.

All phase 1 consent forms should include the phrase, prominently displayed in bold type on the first page, "This medical research project is not expected to benefit you," and the face sheet of all consent forms should include an abstract of the study goals written in readily understandable language. Every effort should be made to avoid ambiguity about personal benefit in consent forms for phase 1 research.

E. Human subjects protections should be extended to cover research that is not federally funded and not regulated by the FDA.— Although the Common Rule might be argued to represent a "standard of practice" in the protection of human subjects, the fact remains that research funded by states or by private sources may be conducted outside the Common Rule's requirements or those of the FDA.³¹ It is striking and unacceptable that there is no universal statutory requirement for informed consent to research irrespective of funding source. Steps must be taken, perhaps including federal legislation, that would ensure that all human research subjects are afforded the protections the Common Rule provides.²²

F. Research in emergency settings should be conducted under a waiver of informed consent only when there is specific evidence that advance consent procedures are impracticable and when there are additional protections for incapacitated patient-subjects whose relatives or friends cannot be located within a reasonable period.— The FDA's new rule on emergency research provides a narrow exception to the requirement that researchers obtain and document informed consent from each human subject or from his or her legally authorized representative prior to initiation of an experimental intervention.¹² The rule applies to studies of life-threatening conditions for which available treatments are unproved or unsatisfactory. These rules also include requirements for consultation with representatives of the community from which the subjects will be drawn, requirements for public disclosure, the establishment of an independent data monitoring committee to exercise oversight of the research, and an assurance by the investigator that the research could not be carried out without the waiver. Studies to be done under the waiver cannot include subjects who could have given advance consent to participate or those who were capable of doing so at the time the study was actually initiated.

Research in the emergency setting conducted under a waiver of informed consent presents special ethical issues.¹¹ In this setting it is sometimes necessary to initiate interventions that offer the prospect of direct benefit to individual subjects within a brief "therapeutic window." Subjects in these circumstances are often incapacitated, and no legally authorized representative may be available to give consent to study participation.

Although the FDA regulation holds such studies to a higher level of scrutiny than other research involving human subjects, the waiver of consent provision calls for still further safeguards on the research. These further conditions are absolutely essential not only because of the exception to consent, but also because the research design may include standard treatment, standard treatment in combination with a placebo, or a placebo alone. The new rule specifies that risks are to be reasonable in light of the subject population's medical condition, the risks and benefits of the standard therapy, and the proposed new intervention. In some study designs, not all subjects will be guaranteed treatment, and some may not even receive the standard treatment. In fact, the condition of those in the treatment arm may actually be worsened by the experimental intervention.

To gain a waiver from informed consent, investigators should give an IRB specific assurance that the study cannot be conducted using subjects who have given advance consent to participate. The current regulations do not specify that such an assurance must be provided. While there are obviously many emergency studies for which advance consent is clearly impractical, as in the case of traumatic motor vehicle crashes, others, such as repeat myocardial infarctions, lend themselves to recruitment through advance consent.

Safeguards in emergency situations must be stringent. If, after a reasonable period of time (eg, 12 hours), a relative or friend willing to speak for the patient who has been entered into a study has not been located, and if the patient is still incapacitated, an independent physician, not connected to the study or emergency medicine, should assess whether continued participation in the study is in the patient's medical interest. This assessment should involve a determination that the treatment afforded the patient under the study is at least as medically beneficial as any that would otherwise be provided. Such an arrangement, involving a physician who is not associated with the research, would provide a further layer of accountability. It may also help to encourage the study team members to make their best efforts to locate a friend or relative of the patient within a reasonable time rather than involve another physician. It is also of crucial importance that IRBs receive timely reports on compliance with all aspects of the protections that are triggered when it is not possible to obtain informed consent and a waiver is granted to conduct emergency research.

G. As individuals in situations of unequal power, sometimes called "status relationships," students should not be required by their instructors to participate in research without being given reasonable alternatives to participation.— Although students are not categorized as a vulnerable population in current federal regulations, under certain circumstances they clearly qualify for this category of protection and thus should be included. For example, medical students may be asked to volunteer to participate as subjects in studies of basic physiology as part of laboratory course work. Students are in an inherently subordinate position to their instructors and are therefore vulnerable to coercion. Obviously these considerations apply to others in status relationships, such as laboratory workers.

Although students may be assured that their research participation is voluntary, they may feel unable to decline to participate for fear of retaliation by the instructor. It is also possible that the true risks of participation cannot be properly assessed, partly because of the student's reluctance to disclose personal information to their instructors (eg, drugs they are taking or a pregnancy), thereby exposing them (or their fetuses) to possible risks. Therefore, we urge a generic exclusion of students from recruitment as normal volunteers while enrolled in the researcher's course.

A. Local review should be continued.— The IRB is and should remain an essential element of the current regulatory system, which relies on local, institution-based review of proposed research. The primary virtue of local review is that, in principle, it makes oversight and monitoring more feasible and effective. Among the more important functions that a system of local review permits is the opportunity to monitor consent processes and study outcomes.

For various reasons, however, IRBs do not seem to have fulfilled this aspect of their potential.⁴ For example, it appears that IRBs do not typically conduct postapproval review beyond the mechanical processing of paperwork submitted by the investigator. One of the reasons for the absence of more active monitoring is that IRBs are increasingly overburdened with the need to adhere to paperwork requirements, many of which contribute little to actual subject protection.

The need to engage IRBs more actively with the actual conduct of research at their institutions stems from the fact that paper compliance with regulations in a proposal does not imply sound execution. Even well-designed protocols that appear to present little risk can pose significant dangers to subjects' well-being when investigators are inexperienced or careless, staff are untrained, or equipment is inadequate. A local IRB, which has the responsibility to protect the rights and welfare of subjects and also has an incentive to reduce the institution's risk of liability, is uniquely suited to oversee research and to take necessary action to minimize its risks, provided mechanisms are developed to ensure that IRBs can and do fulfill their larger obligations.

Local IRBs are also well positioned to ensure that state and local laws and institutional policies are followed. They can also play an important educational role for an institution's research staff, especially by providing information and guidance on such matters as consent requirements and subject selection. It will be easier for IRBs to meet these goals if they are liberated from much of the bureaucratic trivia with which they are now burdened, provided, of course, that mechanisms are developed to ensure that IRBs can and do fulfill their larger obligations.

B. Redundancy in the work of IRBs should be reduced.— Institutional review boards should be given more discretion in the appropriate distribution of their energies. For instance, IRBs should be enabled to designate which projects need the most extensive continuous review. The primary motivation for this recommendation is that, as the ACHRE and GAO reports note, the main function of IRBs has been to monitor the documentation associated with studies.^{4 ,6 ,22} Remarkably little is known about what actually transpires in the course of research, partly because IRBs do not currently have the practical ability to invest more of their resources in monitoring riskier studies.

To achieve a more appropriate distribution of IRBs' efforts, some proposals should be exempt from continuing review while others should be subjected to expanded and even continuous review. For example, not all annual project updates need to be reviewed by every IRB member but may be screened for substantive problems or potential inquiries by the chairperson or by a subcommittee of the IRB. If study data are reviewed by a data safety monitoring board, then they may not need to be reviewed by the IRB.³² It may also make sense to have the same primary reviewer take responsibility for following a study as new reports are submitted. Similar arrangements might be applied to minimal risk studies and to studies where the risks and potential benefits have not changed since the last review. In short, an emphasis on paper compliance is only appropriate insofar as it tends to support the IRB system's true goal: the protection of human subjects.

C. The educational functions of IRBs, for investigators and for their own members, should be expanded.— Most protocols are approved by IRBs with changes of varying degrees of significance. Institutional review boards have traditionally modified around 80% to 90% of all proposals that they review.² Institutional review boards endeavor to work with researchers, and as a result, very few proposals must finally be rejected.³³ By returning protocols for revisions to investigators, IRBs perform a significant educational function. The educational function of IRBs should be recognized and reinforced by providing for better training for IRB members and by offering training programs for investigators conducted by veteran IRB members and others knowledgeable about the ethical issues of research with human subjects. For example, institutions holding multiple project assurances might be required to conduct or sponsor at least 1 educational session per year for IRB members, local investigators, and other interested persons. Admittedly, educational programs are easy to mandate but difficult to implement in a meaningful fashion. Nevertheless, more effort should be given to devising accessible venues for training in research ethics.

In addition, a regular forum for communication among IRB chairs from major research institutions would be of great benefit. Any number of professional societies with an interest in promoting professionalism in health care and research could take the lead to form such an organization.

D. Local IRBs must have the practical ability to revise protocols in multisite studies.— In multisite research activities, a number of institutions are required to follow the same protocol that has been reviewed by many IRBs. Many such projects come before the local IRB on a "take it or leave it" basis. In such cases the IRB has little leeway to make the protocol less risky or more beneficial for subjects. In addition, IRBs are under pressure to approve multisite research projects that often bring funding to participating institutions because of their lucrative sponsorship or because IRBs at reputable institutions have already approved them. As a result, IRBs may be reluctant to invest a great deal of time reviewing these protocols, even though they may be able to contribute substantially to their successful execution. For example, local IRBs may know best what sort of supplemental techniques might be used to improve recruitment and consent processes in their community (eg, videotapes, discussion groups). Although in theory local sites can vary recruitment and consent processes, IRBs might not find it practicable to recommend or require these techniques under the conditions specified for a multisite trial that is being considered.

The regulations governing local IRBs should be adjusted to accommodate the realities of multisite trials. One option would provide for national or regional reviews of the scientific aspects of such trials by representatives from potential participating institutions, coupled with local review for issues of subject selection and recruitment. A program along these lines is under way in the field of human immunodeficiency virus infection research funded by NIH, the AIDS Clinical Trials Group. While this program has produced a number of administrative problems, this model, with appropriate improvements, could be applied to other multi-institutional studies.

E. Institutional review boards should be informed of potential financial conflicts of interest on the part of investigators.— With important exceptions, financial conflicts of interest have not received sufficient attention in the research community.^{20 - 21 ,34 - 35} Conflicts of interest are created by situations that threaten to undermine the researcher's objectivity and truthfulness due the scientific community and the trust or loyalty owed to subjects.

Potential conflicts may come in several forms. For example, when private industry negotiates contracts with investigators to conduct research on a per-subject basis or when NIH contracts require a minimum number of subjects, there may be, or there may be an appearance of, conflict of interest between the requirements of the contract and the best interests of subjects.

Incentives to recruit or retain subjects in order to ensure successful completion of a project could expose them to significant inconvenience, if not risk. This possibility manifests itself when physicians are offered "finder's fees" for each of their patients who enrolls in a study. Institutional review boards must be alert to the possibility that those fees or other incentives (such as authorship or committee assignments) could distort the physician's judgment in counseling a patient about a study.

Another version of this problem that stems from the close relationship between the academic and business worlds occurs when an investigator is a director or functions as a consultant for a company that has also awarded him or her a research grant. Institutional review boards must be informed of any potential conflicts of interest so that they may elect to review relevant aspects of the study and evaluate the propriety of the reimbursement or other incentive. In some instances they may decide that patients or subjects in the study must be informed of such relationships. In general, IRBs must be provided with information that enables them to minimize actual and apparent conflicts of interest.

F. The ways that IRBs relate to their communities need to be reexamined, and new initiatives considered.— There is ample room for misinformation and misunderstanding by the general public about the nature of participation in research. Institutions conducting human subjects research need to make their policies and practices more accessible to members of the community. New ways of informing the public about research and eliciting public approval of research activities, such as the use of the Internet and public-access cable television channels, should be explored. Special programs might be conducted in which IRB members discuss questions submitted by members of the general public about the nature of research and the rules governing research. Activities such as these would benefit IRBs, investigators, and the public and further participation in research and, thus, the growth of scientific knowledge.

G. Institutional review boards should receive significantly increased support from their institutions and from research sponsors.— Many IRBs are overburdened by the quantity and complexity of proposals they review and by the oversight demands of federal agencies. Some IRBs report growing difficulty in attracting dedicated faculty to what is too often a largely thankless job. Administrative support is often greatly strained by record-keeping requirements. If current trends continue, there is a distinct danger that IRBs will be considered as little more than protection against legal liability.

Research institutions and especially the sponsors of research must accept the expenses of human subjects review as part of the cost of doing business. The burden on IRBs could be reduced if sufficient staff could be recruited. A system of recognition and rewards for IRB service should be implemented by local institutions. For example, IRB members may be allowed a course load reduction, their service recognized in a dean's letter that is included in tenure and promotion decisions, or the time of service credited toward administrative leave to pursue their own research. Work on an IRB should be recognized as the invaluable professional activity it is.

H. The National Bioethics Advisory Commission should sponsor a study of the management and costs of the IRB system.— Our recommendations for improvements in the IRB system, like those made by other groups, proceed with mostly anecdotal information about the economy and effectiveness of local review. Little is known about the actual costs of IRBs, the size of their staffs, the budgets they have been allocated by their institutions, and the relationship of these costs to the indirect costs accrued as part of research today. Further, IRBs and the institutions that they serve are very diverse. Little is known about how IRBs actually function, how they are administered, how they are structured, and what the characteristics are of the most effective IRBs. Information of this kind would indicate how much of an investment has truly been made in the human subjects review process at the local level and whether that investment has paid off and might suggest (as we expect) that a higher level of support is both necessary and feasible, considering the magnitude of the medical research enterprise.

I. Institutional review boards that review a significant number of protocols involving people with severe disabilities should be required to include, as members or as consultants, persons knowledgeable about disabled persons' needs.— Institutional review boards need to ensure the protection of those with significant disabilities being recruited into research who might be vulnerable to undue influence.^{25 ,29} Persons with severe disabilities may feel socially isolated and might display compliant behavior to gain acceptance and a sense of participation. External positive inducements should be minimized. Professionals who are trained to work with persons with disabilities or those who have physical disabilities themselves are more likely to be aware of and sensitive to these considerations.

Current regulations provide that IRBs include persons knowledgeable in the areas in which an IRB reviews proposals involving vulnerable subjects. A more specific indication that this regulation applies to research involving subjects with disabilities would help ensure that it is honored and would be consistent with other public policies with respect to persons with disabilities.

A. The placement and role of the Office for Protection From Research Risks (OPRR) in the regulatory system should be reassessed.— The primary mission of the federal regulations is to protect research subjects. One important obstacle to reform in this area is structural: The agency charged with enforcing and interpreting the regulations, the OPRR, is part of the NIH, which is its major client and one of the nation's leading sources of research funding. As a matter of principle, the OPRR should not be located within the structure of any government funding agency, and its charter should specify that it is independent. As an independent agency, it would still be accountable to the professional and lay constituencies it serves, and a suitable reporting structure would need to be devised.

With more authority and control over its own processes, the OPRR could consider ways that the major goals of the human subjects protection system could be better served, in the spirit of our recommendations in this report. For example, the current system of audits of IRBs is inadequate to ensure the goals of the review system are realized. Inspection of paper records cannot substitute for personal contact with investigators, staff, and randomly selected subjects themselves (with appropriate privacy protections), to assess how well official research standards reflect reality. But under its current structure, the OPRR is hamstrung in its ability to experiment with different procedures and advance new regulations by the laws that govern IRBs, since the regulations cannot be changed without the approval of more than a dozen government departments and agencies and the NIH.

The OPRR should provide guidance on emerging policy issues in the use of human subjects, but it should also be able to grant local institutions more flexibility to better serve the spirit of the regulations. Its ability to meet these goals would be much improved if it were independent of any federal departments that sponsor or conduct research with human subjects.

B. Financial risks to which subjects are exposed, owing either to the costs of experimental drugs or devices or to harm that results directly from participation in research, must be minimized.— Not all studies protect subjects from financial hazards of research participation. These hazards may be of 2 kinds: research-related injuries or the cost of investigational drugs and devices. Either may arguably present an unfair burden on those who elect to participate in research that will finally benefit our entire society. The vast majority of such harms are anticipated and are the responsibility of the patient and third-party payers. Debate persists over responsibility for unanticipated harms.

Currently, compensation for harms associated with study participation is theoretically dealt with by the malpractice system in the case of negligence or through the normal mechanisms of compensating for clinical care. However, blatant deception (which depends on the cause of action and the injury alleged) must usually be proven for a subject to win a "research malpractice" suit involving deficiencies in informed consent. In reality those who are sick, aged, disabled, or dying are unlikely to recover for any losses. A number of previous task forces have considered a no-fault insurance system or pool for research-induced injury. This could be established by the federal government, funded by a mandatory assessment on federal grants, along with participation by private funding sources.

Similarly, asking research subjects to pay the cost of investigational drugs and devices may be unfair. Sponsors or institutions might be required to bear the costs of study drugs.

Our group did not reach agreement on the manner in which the current system should be reformed, concerning either research-related harms or the costs of investigational drugs and devices. One problem lies in the difficulty of assessing the cost of such a system because of the elastic nature of potential claims that an insurance scheme might generate. However, the group believes the questions are important enough to deserve further discussion.

C. The National Bioethics Advisory Commission should consider whether some form of national review ought to be required for particularly sensitive research.— In the final analysis, science is a social enterprise. Like all science, biomedical and behavioral research with human subjects can only fulfill its great promise if it is worthy of the wider society's trust. Some kinds of research, especially those that involve special populations and novel interventions (eg, human cloning), are so sensitive that some form of national review, perhaps conducted by a special panel, might be in order. The NIH Recombinant DNA Advisory Committee is an example of this special oversight.

Research with fetal tissue is an example of a research effort that has been vexed by controversy and may have benefited from national review early in its history. Fetal tissue implants have shown promise in the treatment of neurologic and other disorders, but the research remains highly controversial because the tissue derives from elective abortions. In the absence of a widely acceptable compromise, Congress amended the Public Health Service Act in 1993. The amendment permitted federal funding, but without the guidance and oversight that federal funding usually requires. A standing national body with a history of probity on ethical issues might have been able to develop a solution that would have met with acceptance among the various stakeholders in this problem and not have caused the research now being done to move into the private sector and outside the scope of existing human subjects protections.

Another category of research that could be added to the agenda of a standing national body is experimentation that might affect the human germline, a field in which there is likely a dearth of local expertise outside the investigative team. Experience with the Ethics Advisory Board (1978-1980) of the former US Department of Health, Education, and Welfare, which considered several sensitive research issues (among them in vitro fertilization, pediatric research, and psychiatric research), provides some reason for optimism about the effectiveness of a national review body.

D. The National Bioethics Advisory Commission and other national bodies and professional organizations should study the conditions under which research results are to be considered proprietary.— Commercial entities have a well-recognized right to protect the fruits of their investment from competitors, and society gains from the protection of proprietary interests. But in the case of health care, these protections compete with the potential benefit that those who are currently ill might realize if health professionals had access to new information. The tensions between these principles and the ways proprietary interests can be balanced against subjects' protection and their best interests should receive more intensive consideration than they have received thus far.^{21 ,36}

All people deserve the impressive benefits that research involving human subjects creates. But they also deserve the assurance that their rights and welfare will be protected at all times if they are participants in biomedical research. Reexamination and reform of existing human subjects regulations are in order so that both goals can be achieved.