Secondary Logo

Viewpoint: IRBs, Hospital Ethics Committees, and the Need for “Translational Informed Consent”

May, Thomas, PhD; Craig, J M., PhD; Spellecy, Ryan, PhD

doi: 10.1097/ACM.0b013e31806743ef
Research Issues

In this paper, the authors contrast the model of informed consent oversight employed by most IRBs with that most commonly employed by hospital ethics committees at academic medical centers. The challenge of developing adequate consent forms is complicated by competing aims that are difficult to reconcile. “Completeness” in conveying highly technical information is often at odds with comprehensibility for lay audiences. The authors argue that the problems posed by consent are tied to the fact that in clinical research, consent documents are perceived as the primary mechanism for securing informed consent, as opposed to viewing the consent process and conversation as the primary mechanism. Concerns for informed consent should be refocused to the consent process rather than IRB review of documents, using teams of consent monitors, and this new emphasis should be piloted through the National Institute of Health’s recent turn toward clinical and translational science awards.

Dr. May is associate professor of bioethics, Center for the Study of Bioethics, Medical College of Wisconsin, Milwaukee, Wisconsin.

Dr. Craig is director, Clinical Ethics Center, and adjunct assistant professor, Department of Medical Humanities, Memorial Health System and Southern Illinois University School of Medicine, Springfield, Illinois.

Dr. Spellecy is assistant professor of bioethics, Center for the Study of Bioethics, Medical College of Wisconsin, Milwaukee, Wisconsin.

Correspondence should be addressed to Dr. May, 8701 Watertown Plank Road, Milwaukee, WI 53226; telephone: (414) 456-8451; e-mail: (

Developing a consent form that passes institutional review board (IRB) scrutiny is one of the greatest challenges for gaining approval of clinical research. This is true both for researchers seeking IRB approval and for IRBs themselves. For researchers, revisions and modifications to consent forms are the most common reason for delay in IRB approval, with one study finding 91% of IRB requests for changes related to consent forms.1 Another study found that delays related to “picky changes” to consent forms were a leading motivator for “cheating” the IRB review system.2 The challenge of developing adequate consent forms is complicated by competing aims that are difficult to reconcile. “Completeness” in conveying highly technical information is often at odds with comprehensibility for lay audiences. The research literature itself has called attention to the problem that many informed consent documents are written in such detail, or at such a reading level, that many participants may not comprehend them.3

In this article, we argue that the problems posed by consent are tied to the fact that in clinical research, consent documents are perceived as the primary mechanism for securing informed consent. This, in turn, results in a need for detail and completeness, while at the same time comprehensibility requires language no higher than an eighth-grade reading level. Unfortunately, this approach to consent is both difficult to implement and less effective in securing actual informed consent than other models. For example, a U.S. General Accounting Office (GAO) study found that IRBs report that the majority of their time is spent reviewing consent documents.4 At the same time, the GAO study found that deficiencies in consent document review accounts for a very high percentage of violations found during Food and Drug Administration and Office for Protection from Research Risks inspections of clinical trials and IRB operations, and were among the most serious violations cited in warning letters.4 Most important, this approach focuses informed consent oversight on written documents: the GAO points out that “IRB reviews generally do not involve direct observation of the research study or of the process in which a subject’s consent is obtained.”4

Below, we contrast the model of informed consent oversight employed by most IRBs with that most commonly employed by hospital ethics committees (HECs) at academic medical centers. Whereas IRBs emphasize oversight of written documents, HEC oversight of informed consent emphasizes mechanisms for enhancing conversation between patients and health care providers. This is done through establishing mechanisms for addressing ethical issues (now a requirement for hospital accreditation by the Joint Commission on Accreditation of Health Care Organizations). At many academic medical centers, this mechanism is an ethics review or consultation service established as an arm of the ethics committee itself. Ethics consultation is designed to facilitate informed consent by clarifying values and their application to treatment plans.5 This difference in emphasis for oversight, we argue, leads to significant differences in how documents and conversations are viewed within the informed consent process by patient-care clinicians and clinical researchers. We examine how differences in origin, patient/subject population, and committee makeup may account for the dissimilar approaches taken to informed consent oversight by IRBs and HECs, and we discuss several problematic implications of this for consent in clinical research.

It is important to note at the outset that we do not mean to suggest that researchers fail to have conversations with potential research subjects; we recognize that they do have such conversations. Our concern is one of emphasis in securing consent, and the role envisioned for both conversation and documents in this process (an envisioned role perhaps best accounted for by IRB review expectations and processes). A common problem for many researchers is one of obtaining informed consent “merely to meet the letter of the law,”6 a problem we believe relates to the inordinate emphasis placed on consent documents by IRBs.7 We suggest that IRBs adopt an approach to informed consent oversight more like that employed by HECs, using the newly created National Institutes of Health (NIH) clinical and translational science awards (CTSA) to pilot this model. Significant attention has been directed of late to “translational research” from the “bench to the bedside.” Perhaps here, attention to translation from the bedside to the research setting would be valuable.

Back to Top | Article Outline

The Focus of HECs and IRBs

Although both IRBs and HECs are concerned with protecting rights and welfare (for HECs, of patients; for IRBs, of subjects), each takes a very different approach to this same general purpose. Historically, HECs arose as a response to difficulties in identifying what values a patient might wish to guide decision making (in cases like that of Karen Ann Quinlan), whereas IRBs arose as a result of scandals (like the Tuskegee syphilis study) in which informed consent was subverted or ignored altogether. Thus, HECs have focused largely on interpretation of values and their applicability to particular circumstances (a task that emphasizes conversation over documents), whereas IRBs have focused largely on ensuring that risks are known to subjects (a task that has emphasized written documentation that risks have been disclosed). The practical implications of these foci are perhaps most clear in each type of committee’s approach to informed consent: whereas IRBs spend significant time reviewing consent forms, HECs rarely spend significant time concerned with documents, instead emphasizing conversation that interprets patient wishes.

The different emphases of HECs and IRBs are likely also the result of different patient/subject populations and committee makeup. First, consider the population for whom each type of committee is concerned to protect rights. The focus of HECs is on patients who, in the vast majority of cases, enter their circumstances involuntarily; IRBs focus on subjects who, in general, voluntarily enter the circumstances of participating in research. In addition, hospital patients (especially those in the ICU, a common setting for ethics-related problems) are often unresponsive, and even those who are responsive often have limited or questionable decision-making capacity because of the severity of their illness or the administration of medications. Research subjects, on the other hand, normally posses full decision-making capacity, especially in phase I trials which seek healthy volunteers to test the safety and tolerability of a drug or device. In fact, decisional capacity is assumed for enrolling patients in most clinical trials, even phase II and phase III trials. Of course, there are exceptions to this (e.g., in research for mental illness, Alzheimer disease, or other vulnerable populations), but these exceptions themselves require additional safeguards for research to proceed (a requirement that implies decisional capacity where these additional safeguards are not necessary).8 As fully competent adults who voluntarily participate, most research subjects look more like parties to a legal contract than do most hospital patients. In general, then, approaching informed consent for research subjects seems naturally inclined toward a more “legalized” contract model that emphasizes written documents.

In addition, consider the makeup of each type of committee. HECs are staffed primarily by clinical care physicians, nurses, hospital administrators, clergy, social workers, etc.9 Although there is little empirical research on the composition of IRBs (a lack noted in the literature), both anecdotal evidence and those few studies which have been done suggest that IRBs have among their membership many more bench scientists and clinical researchers for whom at most only a portion of their time is devoted to patient care.10,11 For example, one survey of medical school faculty serving on IRBs found 94% reported that they engaged in research (74% engaged in clinical research).12 There is a significant difference in professional culture for these different types of biomedical professionals. The members of an HEC are likely to have far more extensive bedside contact with patients than the members of an IRB. Even clinical researchers conducting trials that involve extensive contact with subjects interact with these subjects in a very different way than, for example, the way in which an attending physician or nurse interacts with patients in a hospital setting. This is partly attributable to the differences between hospital patients and research subjects discussed above, and partly attributable to the different focus of each clinical professional: clinical researchers have a “dual loyalty” to the patient and the scientific study, whereas nonresearch professionals have the patient as a sole focus of their duties.13 Even for these researchers, then, there is a widely recognized problem of “distance” that can lead to an emphasis on formal documents. Given the more limited, “distant” interaction with patients/ subjects, it should come as no surprise that the membership of IRBs would be less inclined to emphasize conversations over documents.

Back to Top | Article Outline

The Role of Documents in Informed Consent

Perhaps because of the differences in origin, patient/subject population, and committee makeup, there exists a substantial difference in the approach to consent in each setting. In clinical medicine, informed consent has long emphasized the process over documents, with the idea of patient understanding of risks both morally and legally trumping a signed document. Accordingly, hospital consent forms are not recognized, by themselves, as sufficient for consent.14,15 Indeed, some institutions and clinicians have eschewed informed consent documents altogether in favor of documenting the informed consent conversation within the patient’s medical record.16 In our own experience with this approach at prominent medical centers, the rationale given is that the presence of a signed document can distract clinicians from the need for conversation. This, in turn, can curtail a clinician’s awareness of the need for conversation that might identify misunderstandings or clarify the meaning of risks outlined in the informed consent document. The legal system has affirmed this approach, with signed documents failing to serve as protection where patient understanding has not been achieved.17,18

One example of the HEC approach that directly relates to the role of documents versus conversation is apparent in the way advance directives, which are mechanisms designed to capture informed consent for incapacitated patients, are actually employed in a clinical setting. Although advance directives themselves are formal documents which are signed, dated, and witnessed, their use in a clinical setting is as “reference material.” In practice, clinicians do not simply apply the written document to treatment plans: treatment plans are determined through interpretation of the reference document, usually through discussion with family and/or surrogates identified within the document itself (in fact, assistance in interpretation of an advance directive is a common reason bedside clinicians request an ethics consultation through an HEC).5 The reason for this is related to uncertainty in the understanding of benefits and risks that a formal, static document simply cannot capture: few people can be fully aware of the complex circumstances that might affect the understanding of benefit in advance of future (unknown) illness.19 For example, if a patient indicates that she or he does not wish to be placed on mechanical ventilation, did the patient mean this to apply to temporary life support if some level of recovery might be expected (having perhaps only envisioned a persistent vegetative state when completing the document)? If not, what level of recovery would make temporary ventilator support worthwhile? How long would such support be acceptable? Thus, the details of the advance directive document are trumped by the understanding achieved through conversations that interpret the formal document in the context of uncertainties.

This latter point is important for the role of documents in securing informed consent in research. Like the uncertain circumstances for which advance directives are meant to apply, research by definition involves uncertainty concerning both potential risks and benefits. This fact is blurred by the formal declaration of risks in research consent documents, which might easily be taken by subjects to constitute greater certainty and completeness than such a declaration is meant to imply. Clinical trials are regularly stopped because of unanticipated adverse events. Relevant to this, Agrawal and Emanuel20 have distinguished a potential research subject’s “comprehension—understanding of the factual components of the information—and appreciation—what the information means to a particular person.” Indeed, many researchers have directly linked the problem of therapeutic misconception—a phenomenon in which one misunderstands research as therapy, normally attributable to a lack of appreciation of the uncertainties in research—in part to consent forms.21–23 The need for conversation that interprets patient/subject appreciation of relevant uncertainties, then, is similar. However, despite a shared circumstance of uncertainty in appreciation of benefits and risks, in research emphasis is placed on the informed consent document itself. This creates problems for informed consent that have long been recognized in clinical medicine, and which have led to the emphasis on conversation over documents.

Two examples serve to illustrate the need for revision in how informed consent is approached in biomedical research. The Kennedy Krieger Institute lead-abatement study sought to evaluate the long-term effectiveness of three different methods of lead abatement. Although the IRB-approved consent form stated that lead abatement may not fully prevent lead exposure in children,24 critics and the Maryland Court of Appeals raised concerns that the consent given by the parents in the study was not sufficiently informed.25 A similar question of consent marked the experience of Jesse Gelsinger, who suffered from ornithine transcarbamylase deficiency, a disease that usually proves fatal in infancy. Though Gelsinger survived to young adulthood, the disease would eventually claim his life, even with standard treatment. Days after beginning a novel, nontherapeutic gene-therapy trial, Gelsinger died at the University of Pennsylvania as a result of his participation in research.25 Although many ethical issues are raised by the Gelsinger case, informed consent (or lack thereof) was a key factor both in the case and in the subsequent lawsuit. Like the Kennedy Krieger Institute lead-abatement study, critics argued that Gelsinger and his family, though he signed a consent form approved by the IRB, did not actually consent to the study because they were not provided with information in a plain, usable format. Importantly, both studies employed IRB-approved consent forms, which were later criticized as inadequate in conveying key information.

Jay Katz, in a personal statement included in the final report of the Advisory Committee on Human Radiation Experiments,26 describes informed consent criteria in today’s world as often “obscuring rather than clarifying what participation in research entails,” a problem Paul Appelbaum relates to consent forms that simultaneously convey “too much information and yet failed to convey the right information.”6 In fact, lengthy consent forms may be counterproductive to actual informed consent by trying the focus and attention of subjects.27 Simply put, too much detail can lead subjects to dismiss potential risks as technicalities and fine print. For example, few people read carefully and take seriously the fine print of warnings on aspirin bottles. In addition, other problems, such as the long-recognized fact that the manner in which information is presented (“20% chance of failure” versus “80% chance of success”) will significantly influence the takeaway message for potential subjects, creating problems for ensuring that context does not undermine appreciation of risks. This is exacerbated by the static nature of documents. Studies of research consent practices have indicated that enhanced consent forms have only limited success in improving patient understanding, and that extended, face-to-face discussion is the most effective way to improve patient understanding and appreciation of risks.28 These problems, along with problems related to the appreciation of uncertainty discussed earlier, undermine the viability of reliance on forms and documents as the primary mechanism for securing informed consent.

We do not mean to imply that we advocate doing away with written informed consent documents in research; rather, we advocate a different perspective on the role of such documents. In fact, adopting a perspective in which the consent document is not viewed as the primary mechanism for conveying risks might allow for even more detail (and technical accuracy) in the documentation of risks. Consider the example of how we convey the risks of prescription medications. The manufacturers’ written description of risks included with the drug would never suffice for informed consent; but this is not its purpose. Rather, it is reference material, with patient understanding of the risks accomplished and documented through conversations with physicians, pharmacists, and other health care professionals, who interpret this reference material. Likewise, the research consent form should be viewed as reference material that serves to supplement the conversation outlining risks. Simply put, the perspective taken on the purpose of the informed consent document should not be that it serves as the primary mechanism for conveyance of information (a perspective that requires both completeness and comprehensibility for all potential participants). Instead, the informed consent document should be viewed as a reference with comprehension emphasized via an ongoing process of conversation between people.

Back to Top | Article Outline

The Viability of an Emphasis on Conversation over Documents

One reason for IRB emphasis on consent documents that we have not yet addressed is that documents provide a tangible focus for review. In addition, it is widely recognized that IRBs are overworked and understaffed. Refocusing attention to conversation rather than documents might be seen as nonviable in these circumstances. However, we believe there are viable, practical solutions that could actually reduce the workloads of many IRBs (especially at institutions with significant research programs). This would involve expanding the idea of consent monitors currently used primarily for vulnerable populations and by smaller research programs, in which all consent conversations are conducted in tandem with trained, specialized consent teams whose sole job responsibilities involve assisting researchers to obtain consent.

HECs have demonstrated that the type of focus on conversation over documents just described is viable. Like IRBs, HECs are staffed by busy professionals who volunteer their time. Nonetheless, most offer assistance in conversations meant to identify and apply patient values and wishes. For larger, tertiary care institutions, this has often required that professional, full-time ethics consultation teams be made available through the HEC, with individuals hired to perform this specific task. The ethics consultation model uses trained specialists to facilitate conversation that reduces problems related to uncertainty by promoting appreciation of risks and benefits, which is the same fundamental question facing research participants.5 To be certain, ethics consultants do not assist in the consent process in all cases (as we envision for research consent teams). Instead, they are used in problematic cases that involve particular uncertainty in the evaluation of risks, burdens, and benefits. For reasons discussed above, however, all clinical research (by definition) involves uncertainty in risks and benefits that is tied to recognized problems of subjects misunderstanding the nature of research. We suggest that although our focus in this article is on clinical research, our suggestions may be fruitful if modified for social and behavioral research. This calls for consent teams to be universally employed in the research setting.

Translating the HEC model of consent to the research setting in the form of consent teams would allow a redirection of attention to broader research ethics concerns. Ultimate responsibility for achieving informed consent would still lie with the researchers, and IRBs would still need to review the consent process, but this could be done via reports. For researchers, by facilitating consent through these teams rather than documents, protocol development could then emphasize one of the two competing goals of completeness and comprehensibility: Both researchers and IRBs could focus on the adequacy of documents to serve as reference materials, akin to the model used for prescription medications described above. This would reduce the challenge of achieving “complete” conveyance of highly technical information in writing at an eighth-grade reading level while simultaneously improving participants’ comprehension of risks and understanding of the uncertainty of research-related risks. This should also reduce delays in research approval by addressing the most common reason for requested changes (minor changes to consent documents). In turn, a common motivation for cheating the IRB review process would be addressed (as discussed at the outset of this paper).

In addition, the benefits of this model extend to improved oversight of research. Consent problems such as those described in the Gelsinger case above should be reduced, improving legal protection of the institution in addition to facilitating more ethical research. Ethics consultation at hospitals does not absolve attending physicians of ultimate responsibility for patient care (as our proposal should not absolve clinical researchers of responsibility), but it augments protections within the process to facilitate better consent.

The remaining obstacle to our recommended approach to consent becomes one of funding: using consent teams to facilitate consent in all clinical trials would require full-time personnel devoted to this purpose. Estimating the cost of such a program is difficult, given the well-documented range of overall IRB costs among small, medium, and large institutions.1–4 However, there has been some discussion of a formula of one FTE per 300 protocols per year as a minimum requirement for IRBs generally, though this formula has not been empirically investigated.3,4 At such a rate, administrative personnel costs for an entire IRB could range between approximately $30,000 to $460,000.3 Because these are the personnel costs for an entire IRB, the actual costs for adding consent monitors would be a fraction of this amount.

Perhaps a more informative approach is to examine the costs of hospital clinical ethics consultation services, because our proposed model is based on this approach. Several consultation services with which we are familiar suggest that the workload for full-time ethics consultants is approximately 60 to 80 case consultations per year, per full-time ethics consultant. An institution could use this to estimate staffing needs in general, although a higher workload might be expected for our proposal because consultations for the services we describe are often quite involved, lasting days or sometimes even weeks.

Estimated annual case loads for IRB protocols range from an average of 95 for low-volume IRBs to 2,782 for high-volume IRBs.29,30 Depending on the types of protocols in question and the number of continuing reviews that the above estimates reflect, the staffing needs for the types of service we propose could range anywhere from 1 to a high of 20 (although the high is, in our opinion, unlikely to be necessary). This represents a very broad potential range that would need to be refined through pilot program study.

The paucity of local institutional support for IRBs themselves suggests it is unlikely that institutions will fund the necessary additional personnel.31–33 The initiative to emphasize protection of research subjects began at the federal level; funding-enhanced mechanisms to shore up this deficient agenda must now occur. Funding for such specialists could initially be provided, in “pilot project” form focusing at first on high-risk clinical studies, through the NIH’s new CTSA initiative. Research ethics and the protection of human subjects is already a visible, stated focus of CTSAs as envisioned by the NIH. CTSAs would seem an ideal mechanism to implement this new approach to consent, both because of the new CTSA emphasis on translation (in this case, between bedside and research settings) and because of the general, institution-wide “infrastructure” concept these CTSA centers represent. Like ethics consultants in the patient-care setting, consent teams would represent an interdisciplinary team approach for enrolling subjects in clinical research throughout institutions that receive CTSAs.

In addition, because CTSA centers are located at institutions with significant research programs, piloting this idea through the CTSA initiative would allow some of the most overworked IRBs to streamline their focus. This should lessen the types of violations described by the GAO4 and allow IRBs to spend more time evaluating aspects of research protocols other than consent forms and documents, a charge envisioned for IRBs from the outset. Indeed, Wood et al34 argue that the Gelsinger case problems, as well as other recent research controversies, “would not have been rectified by a better informed consent form” and that more systematic changes in the focus of IRBs is needed. Thus, we believe that concerns for informed consent should be refocused to the consent process rather than IRB review of documents, and that this new emphasis should be piloted through the NIH’s recent turn toward clinical and translational science awards.

Back to Top | Article Outline


1 Stair T, Reed C, Radeos M, et al. Variation in institutional review board responses to a standard protocol for a multicenter clinical trial. Acad Emerg Med. 2001;8:636–641.
2 Giles J. Researchers break rules in frustration at review boards. Nature. 2005;438:136–137.
3 Sharp R. Consent documents for oncology trials: does anybody read these things? Am J Clin Oncol. 2004;27:570–575.
4 United States General Accounting Office. Scientific research: continued vigilance critical to protecting human subjects. Washington, DC: United States General Accounting Office; 1996. Publication GAO/HEHS 96-72.
5 Task Force on Standards for Bioethics Consultation. Core competencies for health care ethics consultation. Glenview, Ill: American Society for Bioethics and Humanities; 1998.
6 Titus S, Keane M. Do you understand? An ethical assessment of researchers’ description of the consenting process. J Clin Ethics. 1996;7:60–68.
7 Appelbaum P. Examining the ethics of human subjects. Kennedy Inst Ethics J. 1996;6:283–287.
8 45 Code of Federal Regulations 46.111(b).
9 McGee G, Spanogle J, Caplan A, Asch D. A national study of ethics committees. Am J Bioeth. 2001;1:60–64.
10 Francis L. IRBs and conflicts of interests. In: Spece R, Skimm D, Buchanan A, eds. Conflicts of Interest in Clinical Practice and Research. New York, NY: Oxford University Press; 1996:429.
11 Gray P, Cooke R, Tannenbaum A. Research involving human subjects. Science. 1978;201:1094–1101.
12 Campbell E, Weissman J, Clarridge B, et al. Characteristics of medical school faculty members serving on institutional review boards: results of a national survey. Acad Med. 2003;78:831–836.
13 Churchill L. Physician–investigator/patient–subject: exploring the logic and the tension. J Med Philos. 1980;5:215–224.
14 Bottrell M, Alpert H, Fischbach R, Emanuel L. Hospital informed consent procedure forms: facilitating quality patient–physician interaction. Arch Surg. 2000;135:26–33.
15 English D. Valid informed consent: a process, not a signature. Am Surg. 2002;68:45–49.
16 Moore R. Consent forms—how, and whether, they should be used. Mayo Clin Proc. 1978;53:393–396.
17 Sard v Hardy, 281 Md 432, 379 A 2d 1014 (1973).
18 Faden R, Beauchamp T. A History and Theory of Informed Consent. New York, NY: Oxford University Press; 1986.
19 May T. Bioethics in a Liberal Society. Baltimore, Md: Johns Hopkins University Press; 2002.
20 Agrawal M, Emanuel E. Ethics of phase 1 oncology studies: reexamining the arguments and data. JAMA. 2003;8:1075–1082.
21 Moreno J, Caplan A, Wolpe P. Updating protections for human subjects. JAMA. 1998;280:1951–1958.
22 King N, Henderson G, Churchill L. Consent forms and the therapeutic misconception. IRB. 2005;27:1–7.
23 Lidz C, Appelbaum P, Grisso T, Renaud M. Therapeutic misconception and the appreciation of risks in clinical research. Soc Sci Med. 2004;58:1689–1697.
24 Ross L. In defense of the Hopkins lead abatement studies. J Law Med Ethics. 2002;30:50–57.
25 Nelson R. Nontherapeutic research minimal risk, and the Kennedy Krieger lead abatement study. IRB. 2001;23:7–11.
26 Advisory Committee on Human Radiation Experiments. Final report. Washington, DC: U.S. Government Printing Office; 1995.
27 Epstein L, Lasagna L. Obtaining informed consent: form or substance. Arch Intern Med. 1969;123:682–688.
28 Flory J, Emanuel E. Interventions to improve research participants’ understanding in informed consent for research. JAMA. 2004;292:1593–1601.
29 Byrne M, Speckman J, Getz K, Sugarman J. Variability in the costs of institutional review board oversight. Acad Med. 2006;81:708–712.
30 Wagner T, Cruz A, Chadwick G. Economies of scale in institutional review boards. Med Care. 2004;42:817–823.
31 Wagner T, Bhandari A, Chadwick G, Nelson D. The cost of operating institutional review boards (IRBs). Acad Med. 2003;78:638–644.
32 Sugarman J. The role of institutional support in protecting human research subjects. Acad Med. 2000;75:687–692.
33 Ellis G. Keeping research subjects out of harm’s way. JAMA. 1999;282:1963–1965.
34 Wood A, Grady C, Emanuel E. The crisis in human participants research: identifying the problems and proposing solutions. Available at: ( Accessed March 16, 2007.
© 2007 Association of American Medical Colleges