The risks of participating in a clinical trial cannot be predicted with certainty at the time of trial design, especially when evaluating novel therapies. Accordingly, once the trial is initiated, periodic and ongoing review of accumulating study data is necessary to ensure the continued appropriateness of enrolling and treating patients in the trial. This oversight activity is often conducted by a data monitoring committee (DMC), also known as a data and safety monitoring board, generally composed of scientific, medical, statistical, and other experts. Ideally, this oversight process ensures that a clinical trial is stopped if the benefit-risk balance for participants or the expected value to society no longer justifies continuing.
The clinical trial enterprise and all its stakeholders benefit from having qualified, independent DMCs with sufficient resources and flexibility to fulfill their critical and complex mission. Although all clinical trials have inherent risks, DMCs have the formidable task of protecting vulnerable study participants from unpredictable but preventable harm that may arise during the trial. Qualified DMC members with relevant expertise must have access to, and periodically review and analyze, study data to advise the trial sponsor as to whether a trial should be continued, modified, or terminated.
Other trial oversight bodies that do not review accumulating study data cannot fulfill this function. In contrast to DMCs, institutional review boards (IRBs) do not receive accumulating data, and the membership of IRBs is not tailored to any specific trial. The primary role of IRBs is to ensure that risks known at the time of trial initiation are minimized and clearly communicated to trial participants. Although IRBs also must conduct periodic review of ongoing trials, they do not review interim results. IRBs rely on information provided by DMCs as part of their shared responsibility for trial oversight and protection of human research participants. Similarly, study steering committees may review general metrics of study quality and progress or consider study protocol changes to enhance enrollment or to increase the number of study end points reached, but they do not review unblinded results.
Despite nearly 5 decades of experience with DMCs, there continues to be substantial variability in how these bodies are constituted and how their roles are defined and implemented. The Clinical Trials Transformation Initiative (CTTI) recently completed a project to better understand, and make recommendations to improve, the effectiveness of trial oversight by DMCs.1 CTTI is a public-private multisector partnership that identifies and promotes practices that enhance the quality and efficiency of clinical trials. To this end, CTTI investigates ineffective or inefficient practices related to clinical trials and initiates projects to address them. CTTI uses a systematic approach that includes literature reviews, surveys, qualitative interviews, focus groups, and multistakeholder meetings, with the goal of making recommendations that will improve the conduct of clinical trials. The full set of the resulting DMC recommendations is available.1
Investigators, trial participants, sponsors, and nearly all others involved in the conduct of a clinical trial should remain masked or blinded to treatment assignments when feasible, to protect the integrity of the trial and, specifically, to avoid bias in the assessment of outcomes or the attribution of adverse events. In contrast, DMC members, who generally do not assess individual outcomes or adverse events but consider accumulating results, should be unblinded to treatment assignment at all times, even from the earliest points in the trial. This approach is consistent with the 2006 US Food and Drug Administration Guidance.2
Access to accumulating study data by treatment group is necessary for DMC members to draw valid conclusions about the balance of harm and benefit in the trial. Without knowledge of treatment assignments, a masked DMC would lack the information necessary to identify safety or efficacy concerns, and it also may fail to recognize emerging patterns or irregularities in the conduct of the trial. Further, DMC actions based on differences in efficacy and safety are fundamentally asymmetric; a trend suggesting increased efficacy and safety of a new treatment supports continuation of the trial, whereas the opposite trend supports consideration of stopping for harm or futility. Masking of the DMC members to results by treatment would make them unable to carry out their obligation to appropriately interpret interim trial findings, thereby introducing undue risks to study participants and potentially compromising the trial.
With rare exception, DMCs should not review trial safety data in isolation but rather should consider the overall risks in the trial relative to the potential benefits to study participants. As such, DMCs must have full and simultaneous access to both safety and efficacy data to formulate informed risk-benefit assessments at various intervals in the trial. Without a definitive indication of the direction of the effect on the primary efficacy end point or the relative magnitude of a potential clinical benefit among study groups, the extent of risk that can be tolerated cannot be determined with any degree of confidence. In studies of serious conditions or those without any existing treatment options, higher levels of risk may be tolerated than in studies of less serious conditions or those with existing effective therapies. Valid judgment regarding trial risks and benefits cannot be made in the absence of either efficacy or safety data. Specifically, attempting to assess harm alone, without the context of clinical benefit, would, in most cases, be ill advised.
To minimize the risk of false-positive findings (ie, control for type I error), any interim analyses of efficacy data must be preplanned—particularly for confirmatory trials with regulatory implications—with precisely defined criteria for claiming success and adjustments for multiple looks at the data. However, when prompted by safety concerns (eg, emerging evidence of toxicity), previously unspecified analyses of efficacy data used by a DMC to assess ongoing risk-benefit in a trial do not necessarily carry the penalties or require the statistical adjustments typically associated with interim efficacy analyses, because these analyses by the DMC are not intended to lead to early stopping with the claim of trial success. Various statistical monitoring methods exist but were not discussed in the CTTI project and are beyond the scope of this Viewpoint.
The DMC must be able to understand, judge, and react to a wide range of expected and unexpected events and patterns in accumulating trial data. While this virtually always requires members to have expertise in the disease area under investigation, clinical trial conduct, and statistical methodologies, it may be beneficial—depending on the specifics of the trial and the level of risk—to also include bioethicists, patient representatives, or members with other relevant expertise.
Effectively serving as a DMC member requires expertise, experience, objectivity, independence, and the ability to work well in a multidisciplinary team. However, there are a relatively modest number of experienced personnel available, and assembling a well-qualified and effective group is often difficult. To address this challenge, inexperienced prospective DMC members should complete formal training (eg, online programs3) and review instructive case studies. Didactic training programs alone or unstructured and unsupervised “on-the-job” training are insufficient. Ideally, DMC members should receive practical experience under the mentorship of an expert in trial data monitoring, in the setting of actual open and closed DMC proceedings.
DMC members should not have substantial scientific, financial, or other conflicts of interest; even an appearance of conflict of interest may affect perceptions regarding the quality of trial oversight. Senior researchers with important expertise in the area under study will often have had prior interactions with the sponsor or trial investigators, leading to an appearance of conflict of interest. If these interactions have been minor (eg, prior DMC service for the same sponsor in the past), the conflict of interest can be addressed through disclosure to the sponsor and other DMC members. Such disclosures should be updated whenever a new relevant relationship or activity occurs, and review of conflicts of interest should be a recurring agenda item at each DMC meeting.
Trial sponsors typically compensate DMC members for their time and effort, similar to the compensation provided to other trial support personnel. The magnitude of the compensation should be commensurate to the effort and expertise of the DMC members, while avoiding any potential appearance of excess that could be viewed as coercive, biasing, or creating a conflict of interest.
DMC members are typically appointed by the trial sponsor based on recommendations from, or in consultation with, the trial steering committee or investigators and with the agreement of the DMC chair. Sponsors should select DMC members based only on their qualifications for this role and avoid, for example, appointing key opinion leaders to DMCs with the goal of rewarding past collaborations or of enhancing future publicity regarding the trial or therapy should the study results be positive. DMC work requires a deep understanding of and experience with the statistical challenges associated with the interpretation of early, sparse clinical trial data and the consequences associated with early termination of a trial or the modification of the trial design based on knowledge of interim data.
Corresponding Author: Roger J. Lewis, MD, PhD, Department of Emergency Medicine, Harbor-UCLA Medical Center, 1000 W Carson St, Bldg D9, Torrance, CA 90509 (firstname.lastname@example.org).
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Lewis reported that he currently participates and has participated as a member of multiple data monitoring committees (DMCs) for commercial, federal, and other sponsors and that both he and his institution receive financial compensation for these activities; Dr Lewis also reported that he is the Senior Medical Scientist for Berry Consultants LLC, a statistical consulting group that specializes in the design, implementation, and oversight of clinical trials. Both Dr Lewis and his institution are compensated for these activities. Dr DeMets reported serving as a DMC member for trials sponsored by AstraZeneca, Harvard Partners clinical trial group, Actelion, Duke Clinical Research Center, GlazoSmithKline, Public Health Research Institute, Dalcore, Amgen, Sanofi, Baxter, Boehringer Ingelheim, Roche, and Biotronik; conducting an independent analysis of the APPROVE trial under contract through the University of Wisconsin with funding from Merck; serving as a statistical advisor on issues related to clinical trials sponsored by AbbVie; consulting with Portola on the design and analysis of clinical trials; and receiving lecture fees from Sanofi. Dr Calis reported no disclosures.
Funding/Support: Funding for this article was made possible, in part, by the US Food and Drug Administration through grant R18FD005292 and cooperative agreement U19FD003800. Partial funding was also provided by pooled membership fees from Clinical Trials Transformation Initiative (CTTI) member organizations.
Role of the Funder/Sponsor: The funders/sponsors had no role in the preparation, review, or approval of the manuscript or the decision to submit the manuscript for publication.
Disclaimer: The views expressed in this Viewpoint do not necessarily reflect the official policies of the Department of Health and Human Services, nor does any mention of commercial practices or organization imply endorsement by the United States Government.
Additional Contributions: We thank all those who participated in the CTTI DMC project, whether as team members or as participants in surveys, focus groups, or additional discussions. We also thank Stephen Carlson, PhD, in affiliation with Whitsell Innovations Inc, for his editorial assistance with the manuscript.
Lewis RJ, Calis KA, DeMets DL. Enhancing the Scientific Integrity and Safety of Clinical TrialsRecommendations for Data Monitoring Committees. JAMA. 2016;316(22):2359-2360. doi:10.1001/jama.2016.16070