COMMENTARY “ ” POLICY consist of dozens of pages, oft en written in legalese that obscures rather than clari Institutional Review Boards, fi es the nature of the experiment. Senior researchers say morale is low among clini Professionalism, and the Internet cal and translational scientists, especially those early in their careers (5). Robert Martensen Whether cumbersome or not, IRBs tend Published 5 May 2010; Volume 2 Issue 30 30cm15 to vary considerably in their responses to a given protocol, a variation that the Inter Even as the Internet generates pressures that erode professional authorities of all kinds, net easily exposes. Marked variation in IRB it also provides opportunities for researchers and their institutional review boards to responses challenges their professionalism bolster their status as trusted sources. To this end, we must work to improve clinical pro on at least two counts: (i) fairness—that tocol design and approval procedures and maintain the integrity of the study participant like should be treated alike by deliberative recruitment process in clinical trials. bodies operating in the public sphere and (ii) capacity for sound judgments. A study of three Baltimore-area IRB responses to Th e translational enterprise by defi nition re ticipants by the most expedient means pos a single protocol concludes the follow quires collaboration among multiple compo sible that are consistent with the safety of the ing: “Inconsistencies in these reviews raise nents within organizations and between them participants. Clinical principal investigators questions as to the validity and effi ciency and the public. From research design onward (PIs) come from diverse disciplines and en of the IRB process … Validity can be de through study participant recruitment and gage in clinical research with wide ranges of fi ned as the ‘extent to which any measuring physician/patient treatment decisions, trans risk and potential benefi t. Clinical trial pro instrument measures what it is intended lational medicine depends on a modicum of tocols may be devised for a single study or to measure.’ It is important that the IRB trust among the various parties. If an ability may consist of prepackaged documents for process reliably measure with adequate to provide sound judgments in the face of the participation in multicenter trials. Regard validity the degree of safety of scientifi c uncertainties of disease is the sine qua non of less of the risk-to-benefi t ratio or protocol experiments in order to preclude harm to being a professional in health care (1), what type, PIs typically submit their protocols, subjects.” (6) challenges and opportunities does the Inter wait weeks for review by an IRB, receive sug IRB administrators tend to assume that net pose to translational medicine researchers gestions and corrections from the IRB, revise IRB members deliberate within a frame and clinical trial managers and sponsors be and resubmit, wait for another IRB review, work based on logical positivism. Th at is, ing perceived as trusted sources (Fig. 1)? Th is and perhaps revise and resubmit and wait for they believe that members apply fi xed reg Commentary focuses on two: (i) communi an additional review. Most IRB requests for ulations more or less accurately, resulting cation and workfl ow between researchers and revision take one of three forms: (i) changes in objectively right and wrong decisions. institutional review boards (IRBs) and (ii) the in the logistics and supervision of the re Th e prevalence of variation troubles these fruits of their collaborations—clinical trials search, as in requesting that someone with administrators, as it fl ies in the face of and study participant recruitment. In this con an MD degree be present while participants their positivistic worldview. When looking text, the term IRB represents IRBs as entities undergo a physical test; (ii) alterations in the at examples of variation, physicians and in themselves and also as vital components of research design, such as suggesting that re researchers who are running audits of IRB human research protection programs. searchers collect diff erent data; or (iii) revi studies conclude that variations come from Professionalism in IRB review and re sions to patient consent forms. two sources: dubious application of federal cruitment of clinical trial participants is chal What do researchers say about this pro regulations and hasty judgments made by Downloaded from stm.sciencemag.org on May 11, 2010 lenged by the Internet’s ability to expose and cess? Th e phrase “death by a thousand duck overburdened board members and admin create variation, a seemingly innocuous word bites” seems to describe the experience of istrators. In short, they tend to assume that that lies at the heart of science and its man many (3). Writing in the Annals of Inter the variation results from members’ lack agement. According to W. Edwards Deming, nal Medicine, a Stanford team estimated of time and knowledge. Almost regardless a statistician who wrote extensively on statis that it took $56,000 in administrative costs of the nature of the critique, IRB leaders tics and management: “Th e central problem and over 15,000 pages of paper to tweak tend to propose the same remedy, which is of management in all its aspects, including an already-approved research protocol greater fi nancial support for staff training planning, procurement, manufacturing, re that simply compared the progress made and staff expansion. search, sales, personnel, accounting, and law, by patients who had attended two diff erent But what if the standard administra is to understand better the meaning of varia types of addiction treatment programs (4). tive account of the IRB process is wrong? tion. …” (2) And a 2008 report in Science found that Evidence from interviews with IRB chairs paperwork is among the biggest threats across the United States suggests that dif RESEARCHERS AND IRBS to pricing large clinical trials out of reach ferences in decisions from one board to Most readers would agree, I expect, that re (5). Researchers complain that obtaining the next are products of how boards de searchers want to select clinical research par- participant consent has become onerous liberate, not of board members’ mistaken as well. In cancer drug trials, for exam judgments (7). What matters, it turns out, Offi ce of NIH History, Bethesda, MD, 20892, USA. E-mail: ple, prospective participants are routinely is an IRB’s pattern of “local precedents,” [email protected] asked to sign complicated forms that can which are previous decisions made by the
www.ScienceTranslationalMedicine.org 5 May 2010 Vol 2 Issue 30 30cm15 1 COMMENTARY “ ”
Fig. 1. Gutenberg to Google. For its potential to transform authority and trust relations of all kinds, nothing like the Internet in scale has occurred since the invention of mass printing by Gutenberg around the year 1440.
IRB that board members use to guide their modifi cations to previously approved pro existence of a fi nancial COI is disclosed, evaluations of subsequent protocols. By tocols can diff er substantially across IRBs its amount is not. Th erefore, no one out drawing on them, members tend to read in multicenter trials and can be contradic side the sponsors has the means to assess new protocols as permutations of stud tory. Without disputing the importance of a COI’s potential “oomph” to distort sound ies that they have previously debated and adequate material resources, IRBs seeking judgment in conveying risks and benefi ts settled. Instead of working from general less variability might seek to bolster their to prospective trial participants. According rules to specifi c cases, these IRB mem conceptual resources (7). to the U.S. Food and Drug Administration bers tend to work from case to case. IRBs (FDA), “Direct advertising for study sub may vary not because their judgments are SPONSOR COMMUNICATION IN jects [is] the start of the informed consent mistaken but because they strive to make PARTICIPANT RECRUITMENT process” (8), and the Offi ce for Human Re locally consistent decisions over time. Lo Once an IRB has approved a protocol, it search Protections, which is part of the U.S. cal precedents tend to be idiosyncratic to makes its way onto the Internet in the form Department of Health and Human Services an individual IRB but stable within each of participant recruitment ads. Operation (DHHS), has stated that “the information board. Th us, when a local precedent has ally, the IRB may believe its work is done, provided on these [patient recruitment] Downloaded from stm.sciencemag.org on May 11, 2010 not been established, a PI may help shape but that does not mean the world is done Web sites may constitute the earliest com the solution. Once precedents are set and with the trial or the IRB that approved the ponents of the informed consent process” decision-making inertia is established, fu trial protocol. Although sponsors take over (9). Not surprisingly, recruitment practices ture PIs lose infl uence over decisions made public communication, the Internet pub on the Web vary considerably by sponsor. later on the same recurring issue. lic has little means to make distinctions In a 2002 DHHS study of 22 Web sites for From a local-precedents perspective, between a trial sponsor and the IRB that patient recruitment for 110 clinical trials, two common IRB behaviors may account approved the trial. How does the Internet only a quarter mentioned potential ben for both variability and lengthy review aff ect the perception of sponsors and, by ex efi ts and none mentioned potential risks timelines. First, IRBs disagree as to the tension, the IRB approval process as trusted of the protocol (10). More recent studies amount of scrutiny a given protocol will sources during participant recruitment? (11) suggest that there have been no sig require. Whether protocols are expedited Everyone who uses the Internet gets nifi cant changes in sponsor behavior since depends on how risky IRB leaders consider information from multiple sources with that time. Furthermore, euphemisms are the protocol to be for participants’ bodies disparate interests, values, and voices. common, such as when recruitment Web and minds. Second, in full-board reviews Whether a translational entity is offi sites refer to “new drug treatments” rather (as compared to initial reviews, which cially nonprofi t or not, fi nancial confl icts than using the phrase “experimental or un don’t require a quorum) IRBs can arrive at of interest (COIs) suff use all aspects of proven drug.” diff erent overall decisions concerning the U.S. health care. Sometimes these are dis Another problem with patient recruit
SCIENCE TRANSLATIONAL MEDICINE C. BICKEL/ SCIENCE TRANSLATIONAL CREDIT: approval of similar protocols. Requested closed, sometimes not. Even when the ment Web sites is that balanced presen
www.ScienceTranslationalMedicine.org 5 May 2010 Vol 2 Issue 30 30cm15 2 COMMENTARY “ ” tations—those that offer complete and preparatory steps and provide acceptable the moment, “selling” trial participation noneuphemistic information of clini precedents organized by type of research to participants may not be ubiquitous, but cal significance—are not the norm, even and typical protocols. An alternative might it is common, from trial recruitments on though unbiased information is crucial to be to establish a Central Protocol Service Craigslist to slick Web presentations by a potential participant’s decision-making Center that includes live advisers and In large pharmaceutical concerns, such as process. A recent report by the Hastings ternet interaction. Detailed pre-review Wyeth (12). Center (an independent bioethics research within the PI’s home department, which is Vibrant social contracts of small and institute) that analyzed 171 diabetes and routine in some organizations and not in large scope depend on trust between the 184 depression clinical trials noted that others, makes sense, especially concern parties. When researchers don’t trust the “38% of the [enrollment Web] sites … did ing research design and trial logistics. IRB process, and when IRBs respond rou not appear to provide balanced descrip At the initial review, instead of an ex tinely to researchers with unreasonable tions of the studies.” Moreover, the report tended back-and-forth exchange of drafts demands, translational throughput suf stressed that “nearly 75% provided some and edits, why not provide an option to fers. When research overseers routinely description of incentives … yet roughly PIs that authorizes IRB staff to insert de turn a blind eye to the proliferation of half of these failed to mention risks or fault edits—acceptable precedents—as corrupted speech on Internet trial recruit what the study involved.” No sites used part of the initial review process? Appli ment sites, the public has less of a reason the term “risk(s),” and only one used the cants could bring their computers to the to trust professional authorities. term “side effect(s)” (and then only to initial meeting with IRB staff and edit In the broadest terms, the Internet note their absence). This study also found their applications right then and there. multiplies the facts, interests, values, and that a “more than minimal risk” occurred Researchers could choose or decline this voices at play in many domains, including in 67% of the diabetes trials and in 77% of service. Those who opted in would receive translational medicine. Instead of profes depression studies, most of which had for- an initial review report immediately after sionally controlled speech, heteroglossia profit sponsors. Recruitment Web sites the IRB meeting that focused only on out prevails. Dissidents can and do organize also often omit basic study criteria, such standing issues, if any. themselves easily. Learned knowledge is as the length of the study, the source(s) of At the formal review, why not observe widely available, but medical illiteracy is funding, and the specifics of subject par limits on inquiry? Given that the primary prevalent. Private interests and govern ticipation. According to Hastings Center purpose of the IRB review is human sub ments tend to cherish secrets, but the researchers, “Many online recruiting sites ject protection, what is the logic for hav Internet promotes transparency. As for include some—but not full—descriptions ing IRB members discuss a protocol’s re its potential to transform authority and of what will be involved. … Thus the in search design, as when IRBs suggest that trust relations of all kinds, nothing like formation provided appears inconsistent the PI gather additional data? Especially the Internet in scale has occurred since with federal guidance and weighted to when the protocol has been pre-reviewed the invention of mass printing. Societies ward encouraging research participation” by the PI’s home department, should not changed more slowly then, but not long (11). Concerning the genuine uncertainty questions of research design be moot for after the mass circulation of vernacular and clinical risk faced by trial partici the IRB unless they undermine safety or Bibles, books, and pamphlets took off in pants, we must ask whether trial sponsors’ consent standards? the 16th and early 17th centuries, estab blatant use of euphemism and elision is To maintain the integrity of the par lished authorities in religion, politics, and professional and ethical. ticipant recruitment process, IRB leaders medicine across Europe wobbled or top might insist as part of protocol approval pled. Those at the top then felt that their SOME MODEST RECOMMENDATIONS that sponsors provide balanced presen world had been turned upside down, and In terms of conceptual resources, suppose tations to be given to prospective par they were right (13). Downloaded from stm.sciencemag.org on May 11, 2010 all IRBs had access to a Web resource that ticipants and then follow through with Even as the Internet generates pres contained exemplar protocols and problems compliance monitoring and enforcement sures that erode professional authorities as sources of model decisions that allowed mechanisms. Small research centers may of all kinds, it also provides opportuni IRB members to address ethical concerns find this difficult to accomplish on their ties for researchers to bolster their status effi ciently: to identify a subsequent proto own. The Internet, however, makes vir as trusted sources. Whether local and na col’s essential problem amid all of its par tual consortiums straightforward, just as tional leaders of translational medicine ticulars, to clarify its resonance with a prior it facilitates ongoing surveillance. Large are willing to accept that challenge as a case, and to render a consistent decision. regional players with hundreds or thou genuine calling remains to be seen. At the Embodying this wisdom on the Internet sands of trials, such as the Partners system moment—and regardless of institutional might eliminate administrative bottlenecks at Harvard, control access to participants rhetoric to the contrary—translational at the pre-review, initial review, and full and researchers alike. If translational throughput stalls, atomized scramble review levels. Th e proposed Web resource leaders have the will, then ensuring pro and hustle seem ascendant, dissidents could be established and maintained at an fessional standards in communication proliferate, and trust seems in short sup IRB or consortium level as part of the re should be doable. Being known as the ply. Systematic attention to variations in search infrastructure. trusted source in participant recruitment translational processes worldwide would At the pre-review level, it would not be in a clinical research organization’s geo ameliorate important problems that now difficult to provide researchers with In graphical region might even enhance its arise routinely at each stage in the transla ternet “protocol navigators” that outline marketing to prospective participants. At tional enterprise.
www.ScienceTranslationalMedicine.org 5 May 2010 Vol 2 Issue 30 30cm15 3 COMMENTARY “ ” REFERENCES AND NOTES 6. J. M. Hirshon, S. D. Krugman, M. D. Witting, J. P. Furuno, Health and Human Services, Washington, DC, 2002). 1. Royal College of Physicians, Doctors in Society: Medical M. R. Limcangco, A. R. Perisse, E. K. Rasch, Variability in 11. R. Klitzman, I. Albala, J. Siragusa, J. Patel, P. S. Appelbaum, Professionalism in a Changing World (Royal College of institutional review board assessment of minimal-risk Disclosure of information to potential subjects on re Physicians, London, 2005). Note: The document credits research. Acad. Emerg. Med. 9, 1417–1420 (2002). search recruitment web sites. IRB 30, 15–20 (2008). Richard Horton, editor of The Lancet, for drafting most 7. L. Stark, Behind Closed Doors: Deciding How to Treat 12. Alzheimer’s disease clinical trial recruitment Web site, of the article. Human Subjects from the Clinic to the Committee Room http://www.icarastudy.com/. 2. W. E. Deming, Out of the Crisis (MIT Press, Cambridge, (Univ. of Chicago Press, Chicago, in press). 13. R. L. Martensen, The Brain Takes Shape: An Early History MA, 2000), p. 20. 8. U.S. FDA clinical trials guidelines, http://www.fda.gov/ (Oxford Univ. Press, New York, 2004). 3. N. Fost, R. J. Levine, The dysregulation of human sub ScienceResearch/SpecialTopics/RunningClinicalTrials/ 14. The author’s views are personal and do not necessarily jects research. JAMA 298, 2196–2198 (2007). GuidancesInformationSheetsandNotices/ucm116329. represent those of the U.S. National Institutes of Health 4. K. Humphreys, J. Trafton, T. H. Wagner, The cost of insti htm. or the U.S. government. tutional review board procedures in multicenter obser 9. U.S. Offi ce for Human Research Protections, http://www. 10.1126/scitranslmed.3000884 vational research. Ann. Intern. Med. 139, 77 (2003). hhs.gov/ohrp/policy/clinicaltrials.html. 5. D. Malakoff , Clinical trials and tribulations. Spiraling 10. Clinical Trial Websites, A Promising Tool to Foster Informed Citation: R. Martensen, Institutional review boards, profes costs threaten gridlock. Science 322, 210–213 (2008). Consent (Offi ce of Inspector General, Department of sionalism, and the Internet. Sci. Transl. Med. 2, 30cm15 (2010). Downloaded from stm.sciencemag.org on May 11, 2010
www.ScienceTranslationalMedicine.org 5 May 2010 Vol 2 Issue 30 30cm15 4