By Head of Statistical Consulting, Sam Miller and Principal Statistician, Alecia Nickless
According to clinicaltrials.gov 18,670 interventional studies were initiated around the world in the last twelve months. In every one of these, the principal investigator will have justified to themselves and an ethics committee that administering a novel intervention to the study participants balances potential benefits and risks appropriately so that a clinical trial can start. But as the many thousands of participants continue through their treatments, we all need reassurance that the balance is maintained as data emerges.
Independent Data Monitoring Committees (DMCs), or Data Safety Monitoring Boards (DSMBs), provide critical support to medical research by periodically reviewing clinical trial data without jeopardising the scientific integrity of the studies, which will often rely on blinding of patients and investigators to treatment. However, being independent means that the activities related to DMC meetings are often somewhat removed from day-to-day operations and the core study team, and as a result may not be so well understood by those not directly involved. Having contributed as statisticians to various DMC activities over many years, we thought it would be insightful to reflect on this important aspect of drug development.
DMC membership typically includes two or more clinicians with relevant specialisms for the trial, plus an experienced statistician. It is a weighty task to accept the invitation to join a DMC. The remit of this role is to ensure that it remains ethical to continue the study and to protect the ongoing scientific validity of the trial. Individual members spend a lot of time carefully scrutinising tables, figures and patient-level data looking for potential signals – both anticipated and unanticipated. DMC members have a duty to be curious and robustly discuss anything in the emerging data that might be of concern. Ultimately, they must decide whether to recommend to the sponsor for the trial to continue unchanged, continue with modifications, or terminate. Having observed these discussions, it is clear how seriously committee members take this important responsibility.
The remit of Exploristics in the DMC process (Figure 1) is typically to provide an independent, unblinded statistical and programming team to generate the DMC reports separately from the core study team. This is a big responsibility, requiring organisation and urgency to deliver on time, attention to detail to provide DMC members with accurate and specific data, and an appreciation of the wider context and concerns of the sponsor, regulators and patients. It is sobering when in a DMC meeting to think of the crucial part we play in the drug development process.
Delivery of all that is needed for a DMC to operate is complex and requires timely input from multiple groups to succeed. Figure 1 shows the interactions and flow of information between the various teams. The independent stats and programming team are an integral part of this process and must collaborate and build rapport with each group involved to understand the issues each face. For example, we need to recognise the challenge faced by the data management team in cleaning data “live” throughout the trial and taking cuts of the data to support DMC decisions. This means that the Exploristics team must be prepared to work with imperfect data and accommodate reduced reporting times (e.g. 2 or 3 days) in a way which would be unfamiliar to teams more used to reporting at the end of a study.
DMC members are typically eminent experts in their fields, and as such have multiple demands on their time. One way we can support them in fulfilling their duties is by ensuring the secure systems from which they access reports are simple to use, making the reports themselves easy to navigate and understand, and, when necessary, dedicate one-to-one time to provide further assistance. This might seem trivial, given the weighty nature of the DMC activities, but in practice we’ve seen what a difference it makes for the committee members when they have 1000’s of pages of output to review.
Because they are unblinded, the statistics and programming team may also be able to support the study in other project-specific ways. One example of this is the provision of patient profiles for subjects experiencing certain clinical events, so that that these cases can be adjudicated for evidence of drug-induced injury as soon as possible after an event has occurred. Another situation for a proof-of-concept study involved the provision of predictive probability of trial success to specific senior managers within the sponsor. This enabled them to make high-level plans for the program without providing information to the study team and bringing study integrity into question.
Given the DMC are on the lookout for unexpected safety signals, they and the statisticians and programmers supporting them must be flexible where required. When a possible trend is noticed the DMC may request additional tables or figures. For example, we have created bespoke graphs to illustrate the frequency of certain adverse events when subjects take concomitant medication in addition to an investigational treatment. This allowed the DMC members to assess consistency of timing and chronology of the adverse events. This is an ideal opportunity to use our creativity and experience to come up with useful presentations of the data that can be generated rapidly, even if these may not have been included in the original DMC charter or analysis plans.
The work of Data Monitoring Committees can be complex and laborious but is critical to protecting patients in ongoing clinical trials during drug development. As statisticians we are proud to be able to contribute key insights and expertise to support the multiple teams involved in this important process. This allows us to play our part in ensuring not only that clinical trials remain robust but are as safe as possible for those undergoing treatment.