Interview: Issues in data-monitoring for complex clinical trials

OSP: Can you tell us a bit more about the Sally Hollis Memorial Lecture?​

JV-R: ​The Sally Hollis Memorial Lecture was established in 2021 in memory of Professor Sally Hollis. Sally was a much-loved and highly respected member of not just the Phastar team but the wider medical statistics community. Sally was best known for innovation. She was always thinking about how to improve things and she really placed herself at the forefront of that interface between academia and industry. That is why the lecturer alternates each year between academia and industry – to represent the substantial cross-discipline impact Sally had.

DA: ​I knew Sally so I felt really honoured to be asked but I also wanted to carefully consider what I should talk about. I thought of trial complexity and data monitoring because we do not know how to overcome all these issues, which makes it academic, but it is very much relevant to industry because of the impact on clinical trials.

Pic: Deborah Ashby, interim Dean of the Faculty of Medicine, Imperial College

OSP: How have clinical trials changed in recent years and what has driven this complexity?​

DA:​ While two-group trials are the bedrock of the pharmaceutical industry, increasingly people are questioning whether they are the most efficient method. New trial designs are busting the relatively straightforward framework we have all learned.

JV-R: ​The pandemic certainly sped up adoption of these more complex innovations, for example, decentralised clinical trial (DCT) models and platform trials. It just accelerated the industry in so many ways. We also saw a much more collaborative approach – everybody just came together to make it work.

OSP: How do complex trial designs, such as platform trials or trials with multiple endpoints, challenge traditional approaches to data monitoring?​

DA: ​During a clinical trial, accruing data is often seen in confidence by a data monitoring committee (DMC) to evaluate whether early termination of the study or other modifications are needed in the light of emerging results. For a classical two-group trial with a single primary endpoint there are well-established statistical approaches. But when we are dealing with multiple endpoints or treatments, we start to see new issues arising, particularly for the data monitoring committee which has to make decisions in real time.

JV-R: ​Complex trial designs can be difficult. There are lots of statistical intricacies and there are operational aspects. But they are growing in popularity and importance because they are efficient ways of getting treatments to patients. This is why it is vital we look at how data monitoring can adapt to a new clinical trial environment.

OSP: How can the data monitoring challenges of complex trials be overcome?​

DA: ​We must go back to first principles and do some thinking. And this is a message for those sitting on DMCs and those conducting the trials. We need to think about the questions we need to ask. One thing that is happening in several trials is data scenario planning, so people are ready and have done some of their pre-thinking about what happens if a treatment goes awry before they see the trial data. There is also probably another barrage of methodology that needs to be done by the next generation of PhD students which then explores how we do this more elegantly.

JV-R: ​I am a big advocate for communication. As statisticians, we might be able to understand these complex trial designs, but we are not just data analysts because ultimately clinical decisions come out of what we do. We have a really important role as communicators to make sure everyone can understand and interpret what the data is saying. When you have these complex designs, you need to make sure you are talking in a language other people can comprehend so you can take them on that journey with you.

OSP: Are there any emerging challenges or opportunities on the horizon?​

DA: ​I think COVID-19 changed the mindset and made the industry realise we can do more complex trials. If you look at the speed of what came out, you think: why wouldn’t you use that learning. Especially in trials that are comparing multiple treatments. It is not right for everything, but I think it has given us a better understanding of the possibilities. I hope that will push regulators and others to consider how we can handle the difficulties.

JV-R: ​We have a lot of talented statisticians who are great communicators. We can handle these more complex trials. Yes, they are more difficult and take a little bit more thought. But the benefit to the patient is huge. COVID-19 really has shown us what we can achieve when we all work together. I think it would have happened anyway, but it would have been a lot slower. Now we can embrace these patient-centred trials much more quickly, but we need to be educating everyone on these innovative designs.

OSP: Are there any top takeaways people can take from the lecture?​

JV-R: ​I think one of the key takeaways is that we still have a lot of work to do. We know that we are going to be asked to run these studies more and more but there is still an awful lot of work that needs to go on behind the scenes to answer some of the questions that still exist. We need to work together, collaborate, and break down those historic silos.

DA: ​One of the practical things that organisations can do, if they are not already, is in-house training for those who have not had much experience of data monitoring. Just oil the wheels and get the conversations going so we can harness the full potential of new, more complex trial designs.