IRB: Ethics & Human Research
Increasing Common Rule Protections: IRB Consensus, Black Box Warnings, and Risk in Equipoise
On July 26, 2011, the U.S. Department of Health and Human Services (HHS) published proposed changes to the Common Rule, the federal regulation that governs much of the federally funded and supported research with human subjects.1The proposed changes are intended to enhance protections for research participants by streamlining and clarifying the process of institutional review board (IRB) review of research studies. Although the proposed changes sparked considerable debate, many commentators believe that research participants should receive additional protection through further changes to the Common Rule.
In this paper, we argue that the proposed changes to the Common Rule will not substantially improve the IRB review process or strengthen protections for research participants. We offer three changes to the Common Rule and use the controversy over the clinical trial Action to Control Cardiovascular Risk in Diabetes (ACCORD) to argue for these changes. Specifically, changes should be made to further unify review efforts across institutions in multisite trials, to develop ways of highlighting the most serious research risks, and to clarify when and how controversial risks need to be disclosed to research participants. To accomplish this, we argue for three things: a consensus IRB review process for multisite trials, using black box warnings of serious research risks, and application of arisk in equipoisecriterion.
The Relevance of the ACCORD Study
Three clinical trials made up the primary focus of the ACCORD study. In these, various treatment approaches were tested to identify the best ways to decrease the high rate of heart attack, stroke, or death from cardiovascular disease among people with type 2 diabetes who were at especially high risk of experiencing one of these events.2Despite prior evidence that intensive treatment may lead to excess death rates for research participants, the ACCORD model consent form did not mention a risk of excessive death found in the earlier studies in participants intensively treated for diabetes.3The evidence from these prior studies was considered by the ACCORD researchers to be “uncertain” or “uncorroborated,” as stated in their own protocol.4We argued previously that prior studies’ findings of risk of excessive death, even if uncorroborated, should be mentioned in a model informed consent document.5The ACCORD clinical trial was ended in February 2008 based on a recommendation of the study’s data and safety monitoring board because there was a higher overall death rate in the arms in which participants underwent intensive treatment. The ACCORD website states that “257 participants in the intensive group died, compared to 203 in the standard group—a difference of 54 deaths.”6Although the ACCORD study provides an excellent demonstration of the value of data and safety monitoring boards, it also gives an opportunity to reflect on whether different guidelines and frameworks should have been in place. The changes we recommend to the Common Rule are evaluated in light of the ACCORD trial.
IRB Consensus Process for Multisite Trials: IRB Reconciliation or a Central IRB
For a clinical trial to be methodologically sound, all prospective participants should be provided substantially similar consent information. In addition, the expectation is that sufficient information is provided so that potential participants can make an informed decision about whether to enroll in the study. Yet, in a multisite clinical trial, if the IRB at one or more sites makes substantive changes to a consent form and process, there currently is no formal process to ensure that the changes will be shared with the IRBs at the other sites.7The result is that potential research participants may make a decision to enroll based on consent information that varies across sites, and as a consequence, there may be important differences in the characteristics of enrollees across the sites. Moreover, the current IRB process fails to ensure a consistent and rigorous approach in the review of multisite trials despite dedicated and well-meaning local IRB personnel.
Another deficiency in the current system is that IRBs tend to be overworked, with limited access to specific expertise for every protocol.8This lack of adequate attention to some protocols by an IRB could occur even in a single-site research project. However, at least each participant in that case would be under the same consent process and review. The cost of having multiple IRBs review the same study for multisite trials is high in terms of resources, both time and money,9and in terms of delays in research.10Furthermore, variations in research that result from reviews of the same study by multiple IRBs may adversely affect the scientific quality of the study.11At the same time, experimental therapies often involve risk to research participants, and extra time and money spent on ethics review may actually be worth the cost if this provides extra safeguards.
The evidence suggests that having multiple IRBs review the same study generally does not provide additional protection to participants and, paradoxically, may have a negative impact on protections.12One negative result of having multiple IRBs review the same study is an apparent decrease in the readability of consent forms.13Furthermore, IRBs are often poorly situated to provide subspecialty expert review of very large and complex trials. Empirical studies indicate that most IRBs do not make significant substantive changes to centrally produced consent documents.14Finally, if important changes are made at one IRB, there is significant potential that such changes will not be propagated throughout the system.15
Our proposal is a mechanism that can facilitate fulfillment of the requirement in the Common Rule that “[w]hen any one of these [local] IRBs requires changes to the research protocol that are adopted for the entire study, investigators must re-submit the revised protocol to all of the reviewing IRBs.”16This requirement may be underutilized, particularly since protocols often are vague about the consent process and allow IRBs great latitude in imposing requirements without changing the actual protocol. Creating either a central IRB system or a mandatory consensus and reconciliation process for IRBs at institutions involved in multisite trials would provide an opportunity for more consistent and thorough review of a multisite study. A centralized IRB system could include a greater number of specialized experts on the roster of IRB members than the number currently available to IRBs at their specific institutions.
A central IRB for multisite studies can be consistent with local IRB review and can address concerns about local issues. For instance, concerns that the consent form approved by a central IRB will not include issues relevant to the local context of each research site can be addressed by including a customizable section that addresses state laws or institutional concerns.17The National Cancer Institute’s Central Institutional Review Board (CIRB), which was recently accredited by the Association of Human Research Protection Programs,18involves an expedited review for local IRBs that is designed to address concerns about local issues.19
If multiple local IRBs are used, there should be a process in place to mandate consensus or reconciliation for any changes proposed in trial protocols. The changes proposed in the protocols may include those related to the consent process. There should also be a central repository of issues and concerns that local IRBs raise because the current system has no formal mechanism for ensuring that an issue one local IRB raises results in notification about the issue to all local IRBs. Implementing these measures may reduce variation in review across sites and yield better data for analysis of IRB decisions.
Mandating consensus or reconciliation yields an IRB process similar to a legislative process where two houses of a congress might draft similar bills that contain significant differences. To become law, a uniform bill is required. In the legislative process the differences need to be mediated by a special committee and resubmitted to each house. In the same way, in a multisite trial a special committee could mediate differences in substantive changes not related to institutional general “boilerplate” requirements. Although this process runs the risk of additional bureaucracy, with careful structuring, the process could be undertaken in a substantive and efficient manner that would provide important protections to research participants.
As mentioned earlier, the ACCORD model consent document, contained as an appendix in the ACCORD protocol,20did not include a warning based on data from three previous studies of the risk of excess death for participants who would be in the intensively treated arm. The study protocol did, however, include a discussion of the three studies showing excess death from intensive treatment.21Because we do not have access to the consent documents approved by the IRBs at each study site, we do not know whether any of the IRBs required the ACCORD consent documents to include the warning of this specific risk. If some local IRBs did require the warning, then in our proposed review system that requirement would have triggered a mediation process across local IRBs. Alternatively, a central IRB, with experts in the field serving as members, would have provided a greater opportunity to include information about prior risks in the consent document. These experts who understood the risks associated with intensive treatment and understood the evidence of risk that was not taken seriously in the ACCORD protocol could have insisted that the protocol be changed.22At a minimum, our proposal would have increased the odds that prior evidence of risk from intensive treatment would have been included in the consent document used at all ACCORD sites.
The ACCORD trial is not an isolated case in this matter, as is demonstrated in the current debate about a research trial in neonates. There may have been prior evidence of serious adverse outcomes from treatment used in one of the arms in the Surfactant, Positive Pressure, and Pulse Oximetry Randomized Trial (SUPPORT). A determination letter from the Office for Human Research Protections (OHRP) issued to one of the sites provides a convincing and detailed timeline of ongoing controversy and concern in the field about the risk of low oxygenation in neonates.23In the 1950s and 1960s, concerns were raised about the use of lower oxygen levels in premature infants. However, it became more common to use such levels to avoid retinopathy, which is associated with eventual blindness. SUPPORT tested lower levels of oxygen against higher levels, with the result that the in-hospital death rate among those infants at the lower level was significantly higher.24Some critics of the study have specifically mentioned the variation across consent forms regarding potential research risks. Macklin and colleagues note, “About half the forms indicated that because all the treatments proposed in the study were ‘standard of care,’ there was no expected or predictable increase in risk.”25The OHRP determination letter does not prohibit the type of study that was conducted, but it does criticize the failure of the study to provide important information to the neonates’ parents. We should design consent processes so that important information is conveyed during the process. A revision in current practices in multisite trials is needed to achieve this result.
Emphasize Serious Risks: Black Box Warnings
Our second concern with current approaches used to develop consent processes is that a clinical trial may pose serious risks to research participants and that the consent document may fail to call adequate attention to especially serious risks. This concern about consent documents gives rise to our second proposed change.
In many trials clinically relevant risks are not life threatening and do not involve serious disabilities. Yet this is not the case for some trials. A consent document should not simply list serious risks along with other, less serious risks. Instead, serious risks should be set apart and given special emphasis, such as the Food and Drug Administration (FDA) does in its “black box warning” for some approved products.26IRBs could require a black box warning in consent documents to emphasize very important or serious risks that research participants might experience while participating in the trial. A black box warning is intended to create another tool within the consent process. We do not specify which risks deserve a black box warning because the determination of such risks will vary on a case-by-case basis, and the process should continue to rely upon careful deliberation of IRB members. The standard they should use is to include risks that a reasonable participant would want emphasized.
Whether a black box warning or some other method of emphasis is used, the placement and the style of emphasis are important. Methods of emphasis should be carefully studied and should comport with best practices gleaned from the field of risk communication. Regardless of the form the warning takes, it should be supplied in a way to ensure that a reasonably careful reader will see it in the consent form. If a black box warning had been used in the consent document for the ACCORD study, prospective participants recruited at each site would have seen the warning about the seriousness of the risks involved in that study.
Expanding the Guidelines for Disclosure: Risk in Equipoise
Our third proposed change to the Common Rule is designed to ensure that sufficient information is provided in consent processes for potential research participants to make informed decisions about whether to enroll in a study. In particular, there is no good guidance in the Common Rule about how to handle information in the consent process about serious research risks that have some data to support their existence, but for which there is dispute about whether the risk is real.27For this type of risk we propose adoption of a criterion for inclusion of a risk that we label “risk in equipoise”: “A serious risk of harm from a medical intervention is arisk in equipoisewhen disputed evidence of risk has not been accepted as disproven by a consensus among specialists in the appropriate medical field.”28Without this criterion for determining if an uncertain risk has adequate evidentiary support, the researcher may apply a personal standard by simply deciding the evidence for or against the risk as believable. However, when the stakes are high with respect to a serious risk that is in equipoise, the researcher should be required to use the more rigorous criterion related to whether the community of researchers has disproven the risk as real. In fact, when the risk is serious enough, this type of risk in equipoise may warrant emphasis in the consent document with the sort of black box warning proposed above.
The burden of proof to show that a disputed risk isnota risk in equipoise should rest on researchers, such as those in the ACCORD study, who did not mention a possible high risk in the model consent form. Given a strong burden of proof, the ACCORD researchers should have included the risk of death from intensive glycemic therapy in the model consent document. Since they did not, the local IRBs should have insisted on the inclusion of this information.
While it may appear from the ACCORD example that risk in equipoise is simply the precautionary principle in camouflage, it is not. In our context, the precautionary principle would be used to determine if a trial should be conducted; risk in equipoise, however, is a tool to decide what types of information should be included in the consent process.29In the ACCORD protocol, the death rate in the United Kingdom Prospective Diabetes Study (UKPDS) intensive treatment arm was compared with an epidemiological study; the ACCORD researchers argued that the comparison did not corroborate the UKPDS finding.30Also, trial data were combined, and the ACCORD researchers claimed that an excess death rate did not appear with the combination.31This additional information about UKPDS did not disprove the evidence, but the ACCORD protocol characterized the excess death as uncertain. Uncertainty about unexplained and uncorroborated evidence does not negate the fact that prior evidence, at the least, warns of a possible, maybe even likely, serious risk. Despite the fact that prior evidence of risk came in a controlled trial and was not disproven or debunked, the ACCORD protocol cited the evidence as “unexplained” and “uncorroborated” and did not cite it in the model consent form.32
The ACCORD researchers apparently believed that by arguing that prior evidence of risk was uncorroborated or unexplained, they had no obligation to include mention of the risk of excess death in the model consent form. But experts in the field did not reject the concerns, and so they counted as risks in equipoise. As Byron J. Hoogwerf states succinctly, “In earlier trials in type 2 diabetes, concerns had been raised about an increased risk of cardiovascular events and possibly death associated with glucose-lowering drugs, hypoglycemia itself, or both, and these were well known when ACCORD was convened.”33In our view, the risk of excess death as a risk in equipoise should have been mentioned in the ACCORD consent documents.34
What Is New about Our Proposed Changes?
Unlike the HHS’s proposed changes to the Common Rule, we offer an alternative method to mediate and reconcile concerns and disagreements among the local IRBs reviewing the same protocol for a multisite study. We offer this not as a way to save on resources, but as a way to help ensure that the consent process conveys adequate information to prospective research participants about the study and its potential risks. Our call for mediation on proposed changes to the consent document is unique. We offer a distinctive way to further the protection of human subjects if a central IRB is not adopted. Because the HHS proposal does not include attention to emphasizing serious research risks, we propose the use of a black box warning in consent forms. Finally, none of the proposed HHS changes or public comments about them deals with the issues we raise with a guideline involving risk in equipoise for including warnings in consent documents about disputed evidence of risk.
Our suggestions could be implemented in future clinical trials without any change to the Common Rule. However, we believe it would be best that the changes we recommend be included in a revised Common Rule. Although our proposals cannot guarantee that omissions such as those that occurred in the ACCORD model consent document do not happen in the future, there should be continued attempts to minimize the likelihood of such failures to provide potential research participants with relevant information about serious risk. Our proposals, if implemented by IRBs, would minimize the likelihood of such failures whether or not a revised Common Rule includes such changes.
Paul J. Ford, PhD,is the director of the NeuroEthics Program in the Department of Bioethics at the Cleveland Clinic Foundation;Douglas O. Stewart, PhD, is associate professor emeritus in the Department of Economics at Cleveland State University; andJoseph P. DeMarco,PhD,is professor emeritus in the Department of Philosophy at Cleveland State University.
1. U.S. Department of Health and Human Services, Food and Drug Administration. Human subjects research protections: Enhancing protections for research subjects and reducing burden, delay, and ambiguity for investigators. (Advance notice of proposed rule making.)Federal Register2011;76(143):44512-44531. https://www.hhs.gov/ohrp/humansubjects/anprm2011page.html.
2. National Heart, Lung, and Blood Institute. Questions and answers: Action to Control Cardiovascular Risk in Diabetes (ACCORD Study). March 15, 2010. https://www.nhlbi.nih.gov/health-pro/resources/heart/accord-trial/questions-answers#trial.
3. DeMarco JP, Stewart DO, Ford PJ, Patton DJ. Is there an ethical obligation to disclose controversial risk? A question from the ACCORD trial.American Journal of Bioethics2014;14(4):1-7.
4. ACCORD Study Group 2009, 8 ACCORD Study Group. Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial protocol. January 5, 2009. https://biolincc.nhlbi.nih.gov/static/studies/accord/Protocol.pdf.
5. See ref. 3, DeMarco et al. 2014.
6. See ref. 2, National Heart, Lung, and Blood Institute 2010.
7. Menikoff J. The paradoxical problem with multiple-IRB review.NEJM2010;363(17):1591-1593, p. 1592; Stair TO, Reed CR, Radeos MS, et al. Variation in institutional review board responses to a standard protocol for a multicenter clinical trial.Academic Emergency Medicine2001;8(6):636-641, p. 639.
8. Burman WJ, Reves RR, Cohn DL, Schooley RT. Breaking the camel’s back: Multicenter clinical trials and local institutional review boards.Annals of Internal Medicine2001;134(2):152-157.
9. Petersen LA, Simpson K, SoRelle Rd, et al. How variability in the institutional review board process affects minimal-risk multisite health services research.Annals of Internal Medicine2012;156(10):728-735.
10. Infectious Diseases Society of America. Grinding to a halt: The effects of the increasing regulatory burden on research and quality improvement efforts.Clinical Infectious Diseases2009;49(3):328-335.
11. Greene SM, Geiger AM, Harris EL, et al. Impact of IRB requirements on a multicenter survey of prophylactic mastectomy outcomes.Annals of Epidemiology2006;16(4):275-278.
12. See ref. 7, Menikoff 2010.
13. Burman W, Breese P, Weis S, et al. The effects of local review on informed consent documents from a multicenter clinical trials consortium.Controlled Clinical Trials2003;24(3):245-255.
14. McWilliams R, Hoover-Fong J, Hamosh A, et al. Problematic variations in local institutional review of a multicenter genetic epidemiology study.JAMA2003;90(3):360-366; see ref. 7, Stair 2001.
15. See ref. 7, Menikoff 2010.
16. U.S. Department of Health and Human Services. 45 CFR Part 46, Protection of Human Subjects, Subpart A, Basic HHS Policy for Protection of Human Research Subjects.
17. Flynn KE, Hahn CL, Kramer JM, et al. Using central IRBs for multicenter clinical trials in the United States.PLoS One2013;8(1):1- 4.
18. AAHRPP (Association for the Accreditation of Human Research Protection Programs). Latest AAHRPP accreditations include first NIH entity and first organization in Taiwan. December 13, 2012. https://www.aahrpp.org/apply/web-document-library/latest-aahrpp-accreditations-include-first-nih-entity-and-first-organization-in-taiwan.
19. Christian MC, Goldberg JL, Killen J, et al. A central institutional review board for multi-institutional trials.NEJM2002;346(18):1405-1408.
20. See ref. 4, ACCORD Study Group 2009.
21. See ref. 4, ACCORD Study Group 2009.
22. See ref. 4, ACCORD Study Group 2009.
23. Office for Human Research Protections. Determination letter to Richard B. Marchese. March 7, 2013. www.hhs.gov/ohrp/compliance/letters/2013.htm.
24. SUPPORT Study Group of the Eunice Kennedy Shriver NICHD Neonatal Research Network. Target ranges of oxygen saturation in extremely preterm infants.NEJM2010; 362(21):1959-1969.
25. Macklin R, Shepherd L, Dreger A, et al. The OHRP and SUPPORT—another view.NEJM2013;369(2):e3.
26. American Society of Consultant Pharmacists. https://www.ascp.com/articles/black-box-warning-resources.
27. Silverman H, Hull SC, Sugarman J. Variability among institutional review boards’ decisions within the context of a multicenter trial.Critical Care Medicine2001;29(2):235-241.
28. See ref. 3, DeMarco et al. 2014.
29. See ref. 3, DeMarco et al. 2014.
30. UK Prospective Diabetes Study Group. Effect of intensive blood-glucose control with metformin on complications in overweight patients with type 2 diabetes (UKPDS 34).Lancet1998;352(9131):854-865.
31. See ref. 30, UK Prospective Diabetes Study Group 1998.
32. See ref. 4, ACCORD Study Group 2009.
33. Hoogwerf BJ. Does intensive therapy of type 2 diabetes help or harm? Seeking accord on ACCORD.Cleveland Clinic Journal of Medicine2008;75(10):729-737.
34. See ref. 3, DeMarco et al. 2014.