On Tuesday 6th
December 2022 the Medical Section
and the Leeds–Bradford local group
of the RSS held a joint afternoon meeting entitled “Medial Decision Making,” chaired by Dr Ciarán McInerney from the University of Leeds.
The first talk was by Dr Maria Ibrahim, a nephrology trainee currently working at Manchester Royal Infirmary but who recently completed a post as a Clinical Research Fellow with NHS Blood and Transplant where the main part of her work was to use UK Transplant Registry Data to help answer clinical questions relating to organ transplantation. She described her work in risk communication and how to improve the way we communicate statistics to both clinicians and patients. There are difficult questions patients have to address (with help from their doctors) including how long one might expect to wait for a transplant, what is the expected survival following transplant, “should I take this organ, or wait and hope for a better match to come along,” etc. Different patients have different priorities, and different levels of numeracy. The project Maria worked on developed simple graphical tools for communication, including the challenging problem of communicating competing risks (death, transplanted organ failure, and so on).
The second talk was from Professor Deborah Ashby, Director of the School of Public Health at Imperial College London where she holds the Chair in Medical Statistics and Clinical Trials. Not least, she is also well known as a former President of the Society! Deborah talked about challenges of making decisions in clinical trials based on incomplete data – primarily from the perspective of members of a data monitoring committee. She contrasted the (relatively) simple problems of single endpoints and standard group-sequential stopping rules, with the far more common problem of trials with multiple important endpoints – either multiple efficacy endpoints of interest, or even just the everyday problem that efficacy needs to be balanced against safety and this led into a discussion of choosing relative weights of benefits and harms. Problems become even more complex with, for example, platform trials and adaptive designs for trials. One of Deborah’s key take-home messages was the importance of truly understanding what the purpose of the trial is.
The third speaker was Andrew Vickers, biostatistician at Memorial Sloan Kettering Cancer Center with interests in prostate cancer, risk prediction, decision-making. He was one of the originators of Decision Curve Analysis and this contributed to his talk. He kept us all awake with a lively presentation, encouraging us to take an active part by choosing between different projection models to answer the question of “which model is best to use?” He rightly pointed out that a typical statistician’s answer of “well, it depends on what’s most important” (e.g. sensitivity, specificity, etc.) really wasn’t good enough and we needed better ways to truly answer the question – not just bat it back to the clinician who might have asked us. He used this to introduce the notion of a “threshold probability” below which it would be appropriate to take no action and above it we might perform further tests or begin treatment, as appropriate.
The meeting generated much discussion and questions to all the speakers from the “in-house” audience, as well as those joining remotely.
Written by Simon Day,