Statistics, communication, impact: the Florence Nightingales of today

Date: Monday 13 June 2022, 9.00AM
Location: Lyon
Lyon, France
RSS Event

Share this event

Session as part of the Society de Francaise de Statistique 53rd annual conference in Lyon:
Dr Catey Bunce, Royal Marsden. 
Tips for Applied Medical Statisticians from Florence Nightingale - Data Visualisation

12th May 2020 was the bicentenary of the birth of Florence Nightingale. Whilst many are aware of Florence Nightingale as a pioneer of modern nursing, less may be aware that she is an inspirational female leader of applied medical statistics. I believe that Florence Nightingale was inspirational in her ability to communicate statistical concepts to those without statistical training and she was perhaps ahead of her peers in using data visualisation in this. There is today evidence of poor statistics within medical research and there is evidence also that people can suffer harm because of statistical misunderstandings. Better communication between statistician and non-statistician has been encouraged for many years. Data visualisation is a vital tool for communication and one that statisticians need to engage with in order to improve understanding. Ironically 2020 was the year that the COVID-19 pandemic struck the UK and other countries. It was arguably the year that data visualisation came into its element with animations and interactive tools. Transparency in relation to the robustness of data was at times lacking and this was not conducive to maintaining confidence in findings. My talk will simply reflect upon the use of data visualisation over time and the challenges facing applied medical statisticians in providing messages that are clear to non-statisticians but do not compromise on statistical integrity.

Keywords: Statistical Communication; medical statistics.

Dr Roger Beecham, Leeds
Scientific Reform and Visual Data Science: Retiring the EDA/CDA dichotomy

Concerns around the replicability of published scientific findings has prompted much introspection into the way in which scientific knowledge is produced. To address issues of data fishing, searching exhaustively for discriminating patterns in a dataset, picking and then publishing those that are statistically significant, an argument is made that research findings should only be claimed through pre-registered confirmatory data analyses. Pre-registration studies are, though, somewhat inimical to the more informal research environments typical of modern applied data analysis (‘Data Science’). In this talk I enumerate some of these challenges and demonstrate, through an analysis of road crash data in the UK, how nascent visualization techniques can be used to navigate and inject statistical rigour into contemporary data analysis environments.

Keywords. Exploratory data analysis; confirmatory data analysis; data visualization; replicability crisis; graphical inference, uncertainty visualization.

Acknowledgement: This work is supported by Wave 1 of The UKRI Strategic Priorities Fund under the
EPSRC Grant EP/T001569/1, particularly the ‘Digital Twins: Urban Analytics’ theme within that
grant & The Alan Turing Institute.
Dr Victoria Cornelius, Imperial
Data visualisation of adverse events in randomised clinical trials

Assessing the benefit-harm balance is a critical part of a randomized controlled trial as this informs prescribing decisions and the design of future studies. Despite this, harm outcomes in trial publications are poorly presented. Poor presentation can obscure an informative assessment of the true harm profile. Methods to analyse efficacy outcomes in randomised controlled trials (RCTs) have made substantial progress over the last 50 years, but the analysis and presentation of adverse events (AEs) have not. AE data is particularly difficult to analyse due to its multi-faceted nature. There is a lack of guidance on what and how to visually display complex AE data. We previously undertook a review to identify methods specifically developed to analyze AE data. In this talk we look at three visual analysis methods that can aid detecting signals or adverse drug reactions.  We explore their value using data from a COVID-19 treatment trial and a COVID-19 vaccination trial. We describe the trials, the role they play in assessing harm and then demonstrate the impact and value of the visualisations, as well as discussing their differenced and limitations.  
Keywords: Adverse events; Harms; Randomised Controlled Trials; Data Visualisation.
Victoria Cornelius (Imperial College London)
Roger Beecham (University of Leeds)
Catey Bunce (The Royal Marsden)
Nicola Fitz-Simon and Catey Bunce on behalf of RSS Medical Section