ICTU and NIHR RSS Imperial and partners Hubs’ 1st Contemporary Clinical Trials Methodology Meeting

Victoria Cornelius

Chair

Leila Janani

‘Exploring Group-Sequential Design: An Introduction and its Application in the BACHb Trial’

Stuart Pocock

‘Stopping guidelines for interim analyses’

Statistical stopping guidelines provide a necessary objectivity to aid the decision-making of Data Monitoring Committees. A major clinical trial may stop early for either overwhelming efficacy, harm or futility. The speaker will present examples from real cardiology trials to illustrate the variety of challenging scenarios that can occur. Group sequential designs provide a valuable theoretical basis for interim analyses, but wise DMC recommendations rely on a totality of evidence that is often not readily captured by formal stopping boundaries.

Laurent Billot

‘Discussant on the potential benefits of group sequential designs’

Methods for group sequential designs are well established and have been around for a while; however, they tend to be underutilised. Most of the time, investigators only consider a limited number of interim analyses with conservative stopping rules. But is this the best use of group sequential designs? With the rise of adaptive trials, Laurent will reflect on the potential for group sequential designs to allow simple trial adaptations while bringing substantial efficiency benefits.

Suzie Cro

Chair

Lukas Pin

‘A gentle introduction to Response-Adaptive-Randomisation’

In this talk, we illustrate the diversity of Response-Adaptive Randomisation (RAR) designs, tracing its evolution since 1933. Our journey will illuminate the practical and statistical challenges encountered in implementing RAR in clinical trials, including balancing possible design’s objectives (patient-oriented vs. efficacy), endpoint types, valid hypothesis tests for the final analysis, burn-in periods, sample sizes, and the algorithm’s aggressiveness. Through concrete examples, we aim to furnish attendees with a foundational understanding of RAR, highlighting its significance and the considerations necessary for its effective application. This presentation is designed as a primer, ideal for those new to the field, and promises to both introduce and shed light on the complexities and nuances of RAR in clinical trials. 

Sofia Villar

‘Beyond the basics, how does response-adaptive randomisation look in practice?’

In my talk I will present some of the more practical challenges of using response-adaptive randomisation  (RAR) as an adaptive element of a trial design. Through some specific case studies I will try to illustrate how to compare designs using a RAR assignment mechanism and to present some of the less well known practical challenges implementation can bring. 

John Norrie

‘Discussant on Response adaptive randomisation’

Response Adaptive Randomisation (RAR) has a long history as a statistically elegant way of learning from within the study, and so sending proportionately fewer participants onto a randomised allocation that will give lesser benefit. However, the uptake of RAR has been slow, and there have been controversial applications. There are many different realisations of RAR, and these designs do generate quite strong opinions in both directions. I will focus more on some practical issues around implementation, and likely benefits and possible disbenefits, also from the patient perspective, and how clinicians regard RAR. 

Rachel Phillips

Chair

Alan Montgomery

‘Sample size re-estimation in the HI-Light vitiligo trial’

The HI-Light trial tested topical corticosteroid (TCS) and narrowband ultraviolet B (NB-UVB) for localised vitiligo in a three-arm, blinded, randomised trial. The primary aim was to compare each of TCS+NB-UVB combined and NB-UVB alone with TCS alone. In this talk I will describe changes to the study aims made after the study was funded, and to the target sample size based on checking study design assumptions prior to completion of recruitment.

Dominique-Laurent Couturier

‘Blinded and Unblinded Size Re-estimation for DAH, a Complex Endpoint’

“Days alive and at home” (DAH) is a recent patient-centred outcome measure for perioperative trials, defined as the total days spent at home during the follow-up period. DAH is typically left skewed, zero-inflated due to death and censoring, and bimodal due to re-admission times. In this work, we propose a novel modelling for this complex endpoint and revisit the sample size re-assessment of the NOTACS trial performed by assuming a zero-inflated log-normal distribution for the hospital stay duration.

Ruth Goodall

‘Sample size re-estimation in the STREAM trial’

ABSTRACT COMING SOON…