ORIGINAL CONTRIBUTION Problematic Variation in Local Institutional Review of a Multicenter Genetic Epidemiology Study Rita McWilliams, MPH Context Sequencing of the human genome provides an immense resource for stud- Julie Hoover-Fong, MD ies correlating DNA variation and epidemiology. However, appropriately powered ge- Ada Hamosh, MD, MPH netic epidemiology studies often require recruitment from multiple sites. Objectives To document the burden imposed by review of multicenter studies and Suzanne Beck, MD to determine the variability among local institutional review boards (IRBs) in the ap- Terri Beaty, PhD proval of a multicenter genetic epidemiology study. Garry Cutting, MD Design A PubMed search was performed to determine the frequency of citations of multicenter studies by 5-year intervals from 1974 through 2002. A 7-question survey ROTECTION OF HUMAN SUB- was sent to all participating study centers to obtain information on frequency of IRB jects in research is an evolv- meetings, dates for submission and approval, use/nonuse of a specific consent form, ing process. The current sys- type of review performed, types of consent forms required, preparation time, and num- tem of institutional review ber of changes requested by the IRB at each center. Centers also provided a copy of Pboard (IRB) assessment of human sub- all consent forms they generated and IRB correspondence regarding the study. jects protection was established in 1974 Setting and Participants Thirty-one of 42 cystic fibrosis care centers in this single in response to highly publicized hu- US multicenter genetic epidemiology study of cystic fibrosis replied, yielding a 74% man research scandals in the 1960s and response rate. early 1970s.1,2 Federal regulation of re- Main Outcome Measures Frequency of published research studies and consis- search conduct and IRB function was tency among IRBs. implemented in 1979. When IRBs were Results The number of all published single-center studies has increased 1.3-fold since created, the common paradigm for hu- 1985, while the number of published epidemiology and genetic epidemiology multi- man subjects research consisted of a center studies increased by 8- and 9-fold, respectively, during this same period. Evalu- single investigator at one institution en- ation of the risk of the same genetic epidemiology study by 31 IRBs ranged from mini- rolling local participants, with the ma- mal to high, resulting in 7 expedited reviews (23%) and 24 full reviews (77%). The number of consents required by the IRBs ranged from 1 to 4; 15 IRBs (48%) required jor emphasis of regulation on the re- 2 or more consents, while 10 (32%) did not require assent for children. The most com- view of clinical trials. mon concern (52%) of IRBs pertained to the genetic aspects of the study. Over the past 25 years, research strat- egies and technologies have changed, Conclusions Review of a protocol for a multicenter genetic epidemiology study by local IRBs was highly variable. Lack of uniformity in the review process creates uneven often bringing together investigators human subjects protection and incurs considerable inefficiency. The need for reform, from multiple institutions to enroll geo- such as the proposed centralized review, is underscored by the ever increasing rate of graphically diverse pools of partici- genetic discoveries facilitated by the Human Genome Project and the unprecedented pants into epidemiological studies. opportunity to assess the relevance of genetic variation to public health. However, IRB procedures and their fed- JAMA. 2003;290:360-366 www.jama.com eral underpinnings have not corre- 2,3 spondingly kept pace. Because of the Author Affiliations: Bloomberg School of Public Health Johnson and Johnson, SciClone Pharmaceuticals, de focus of IRBs on clinical trials, others (Ms McWilliams and Dr Beaty) and McKusick-Nathans Code, Discovery Labs, and Osteotech. Dr Cutting owns Institute of Genetic Medicine (Drs Hoover-Fong, Ha- stock in Invitrogen and Life Technologies and has con- have asserted that IRBs “often have little mosh, and Cutting), Johns Hopkins Medical Institu- sulted for Roche Molecular Systems. insight into the needs of epidemiol- tions, Baltimore, Md; Department of Pediatrics, Drexel Corresponding Author and Reprints: Garry Cutting, 4 University College of Medicine, and St Christopher’s MD, McKusick-Nathans Institute of Genetic Medi- ogy.” Indeed, it is worth noting that Hospital for Children, Philadelphia, Pa (Dr Beck). cine, CMSC 1004, 600 N Wolfe St, Baltimore, MD one infamous human subjects re- Financial Disclosure: Ms McWilliams owns stock in 21287 (e-mail: [email protected]). 360 JAMA, July 16, 2003—Vol 290, No. 3 (Reprinted) ©2003 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/26/2021 VARIATION IN INSTITUTIONAL STUDY REVIEW search scandal—the Tuskegee Syphi- ent in the interpretation of regulations and consent form was provided as a lis Study—involved an epidemiology by the estimated 3000 to 5000 IRBs in template for the IRB application at each study rather than a clinical trial.1 the United States.2 To assess the bur- center. Each center was provided with The incorporation of genetic infor- den imposed by review of all types of additional information that included the mation into clinical and epidemiologi- multicenter studies, we determined the guidelines for genetic banking pro- cal studies has raised additional prob- yearly volume of single-center and mul- vided by the American Society of Hu- lems for the current IRB system. There ticenter studies published in the litera- man Genetics.10 are few standards set by the Office for ture since 1974. One method of evalu- To document the process of IRB Human Research Protections for ge- ating the impact on the current IRB approval, a 7-question survey (avail- netic studies, and substantial disagree- multicenter process is to submit a com- able on request from the authors) was ment exists within the research com- mon protocol to multiple IRBs.7 sent to all participating CF centers ask- munity about what constitutes minimal We are conducting a multicenter ge- ing the study staff to provide informa- risk in studies that are not clinical trials.5 netic epidemiology study to identify tion on the following: frequency of IRB According to Greely,6 “Research into hu- modifiers of cystic fibrosis (CF). Imple- meetings, dates for submission and man genetics has stretched current regu- mentation of this study required the approval, use/nonuse of the Johns Hop- lations of human subjects research be- participation of CF care centers across kins University consent form, type of yond the breaking point.” The inherent the United States. Institutional review review performed, types of consent rarity of the outcome and the large num- board review of the same study proto- forms required, preparation time, and ber of subjects needed to unravel com- col varied considerably. Here, we pre- number of changes requested by the IRB plex gene-gene and gene-environment sent results of a survey of participat- at each CF center. Each center was also relationships often require a multi- ing CF centers to document current IRB asked to provide a copy of all consent center study design to attain sufficient issues in conducting a multicenter ge- forms generated at their center and all statistical power to generate meaning- netic epidemiology study. IRB correspondence regarding the CF ful results. study. Variability among IRBs regard- Although Silverman et al7 have METHODS ing approval of this study was derived reported variability in the review of mul- A PubMed search was performed to from review of IRB correspondence and ticenter clinical trials, there have been no assess the number of English-language approved consent forms. Issues raised published reports examining IRB human multicenter studies in the litera- by centers and differences among con- approval for sites involved in multi- ture since 1974.9 Abstracts, letters to the sent forms were categorized and tabu- center genetic epidemiological studies. editor, review articles, and publication lated. In addition, current regulations are not types not containing original results were A matrix of consensus statements well suited to the complex issues raised excluded. In addition, collaborative stud- published from 1987 through 2001 was by genetic studies. According to Jam- ies of disease mechanism, disease treat- created to assess the most frequently rozik, “The current systems of ethical ment, or health care delivery were cited guidelines for genetic studies and oversight designed primarily to regu- excluded. The frequencies of citations to illustrate variability in use of these late intervention studies involving indi- of published, multicenter, English- statements in the consent forms. Data vidual patients associated with single language human studies in PubMed from on number of beds, obtained from the institutions have been completely over- 1974 through 2002 were tabulated in American Hospital Association, were taken by developments in clinical, 5-year increments. (The algorithm used used as a proxy for the size of the in- molecular, and epidemiological in the PubMed search is available on stitution.11 Extramural research rev- research.”4 When IRB committees do not request from the authors). enues obtained from the National In- allow a consistent method of consent The CF Twin and Sibling Study is a stitutes of Health (NIH) were used as among the participants, “selection bias multicenter genetic epidemiology study a surrogate for volume of research per- may be introduced and statistical power that was used as a case study to illus- formed at each institution.12 Number of is certainly decreased.”4 Therefore, IRBs trate variability in IRB review. The study beds and level of NIH extramural fund- are largely without guidance in the review involved collection of medical record ing in centers that did and did not re- of studies that incorporate genetics. data along with a blood sample from CF spond to the survey were compared by The current method of multicenter re- patients who attended CF care centers t test.