We are pleased to announce that UCLan and Lancashire Clinical Trials Unit will be hosting a Royal Statistical Society seminar (Lancashire and Cumbria local group) on Wednesday 25th of May, 3 - 5 pm. The topic of the seminar is “Assurance and Sample Size Determination for Clinical Trials”. 

We welcome Dr Kevin Wilson (School of Mathematics, Statistics & Physics, Newcastle University) and Dr S. Faye Williamson (Biostatistics Research Group, Population Health Sciences Institute, Newcastle University) to talk about assurance as a Bayesian alternative to power and sample size calculation, with application to clinical trials. The details of the talks are below:

Dr Kevin Wilson (School of Mathematics, Statistics & Physics, Newcastle University)
Title: "Assurance: a Bayesian alternative to power – an illustration in planning a diagnostic study for ventilator associated pneumonia"
Abstract: When planning medical studies such as randomised clinical trials, power calculations are performed to choose a sample size. The idea is that in the analysis following the study we should have a good chance of detecting that the new treatment is superior to the old, for example, if this is true. This power calculation is based on a chosen value for the difference in effect between treatments and, if this is far from the true difference, the ability of the study to detect a difference may be much lower than anticipated. The assurance includes our uncertainty on this difference and produces a more robust sample size to mis-specification. In this talk I will give an intuitive explanation of assurance and power, and illustrate how both can be used to choose a sample size for a diagnostic study into ventilator associated pneumonia, emphasising the advantages (and disadvantages) of assurance.

Dr Faye Williamson (Biostatistics Research Group, Population Health Sciences Institute, Newcastle University)
Title: "Sample size determination for rapid COVID-19 diagnostic tests using assurance"
Abstract: In response to the ongoing and continually evolving COVID-19 pandemic, early and accurate detection of infectious individuals is critical to outbreak containment. This relies heavily on the sample size chosen to evaluate the accuracy (e.g. sensitivity and specificity) of the diagnostic test. Too small a sample size will lead to imprecise estimates of the accuracy measures, whereas too large a sample size may delay the development process unnecessarily.  We apply the Bayesian assurance-based approach, introduced in the previous talk, to guide sample size determination for diagnostic accuracy studies of COVID-19 rapid viral detection tests. By utilising information from the preceding laboratory study, this approach has the potential to reduce required sample sizes compared to traditional power-based approaches. This increased efficiency would allow for improved evidence development, which is particularly important when accurate diagnostic tests need to be deployed rapidly.

The programme of the seminar is as follows:
3 pm Tea and Coffee
3.30 pm Dr Kevin Wilson
4.15 pm Dr Faye Williamson
Followed by Questions and Discussion

Location: Brook Building room BB312
University of Central Lancashire (UCLan)
Preston
PR1 2HE
 

Dr Kevin Wilson (School of Mathematics, Statistics & Physics, Newcastle University)
Dr S. Faye Williamson (Biostatistics Research Group, Population Health Sciences Institute, Newcastle University)

RSS Lancashire and Cumbria Local Group