On Thursday August 23, three international rising stars in biostatistical research – Valentijn de Jong (Netherlands), Jennifer Thompson (UK) and Simon White (UK) – visited us in Melbourne to showcase their work at the monthly Victorian Branch meeting. Held at the University of Melbourne, this was a joint event with the SSA Biostatistics and Bioinformatics Section.
Benefitting from the fact that the International Society for Clinical Biostatistics (ISCB) conference was taking place in Australia for the first time (held jointly with the SSA’s Australian Statistical Conference), we had the opportunity to invite three fantastic early to mid-career biostatisticians from overseas to come and present their research in Melbourne.
The Biostatistics Bonanza was designed to feature presenters from a range of biostatistical discipline areas to enable us to highlight the diversity in biostatistics. Each presenter was given a 20 minute slot to provide a snapshot of their research. First up was Valentijn de Jong from the UMC Utrecht in the Netherlands. Valentijn is currently undertaking a PhD in Biostatistics focussed on individual patient data (IPD) meta-analysis and presented findings from his research on intervention studies with time-to-event outcomes. Known sources of heterogeneity in these studies include different follow-up times and time-varying treatment effects. Illustrated using five studies in epilepsy, Valentijn introduced a heterogeneity measure for time-to-event data and presented several analysis models to account for heterogeneity in IPD meta-analysis.
Dr Jennifer Thompson is a trial statistician and methodologist from the London School of Hygiene and Tropical Medicine in the UK. Her research interest is developing statistical methods for use in cluster-randomised trials, with a particular focus on stepped wedge trials. Jennifer provided advice on using generalised estimating equations (GEE) in a stepped wedge trial during her presentation. Using simulations, she evaluated the performance of GEE models with different working correlation matrices both with and without small sample correction and a commonly used mixed model (Hussey and Hughes) in stepped wedge trials. Based on the observed bias in the estimate and standard error and confidence interval coverage, the GEE model using the small sample correction by Fay and Graubard seemed to perform best with little difference between the different correlation structures.
Dr Simon White is a senior investigator statistician at the Medical Research Council Biostatistics Unit at the University of Cambridge in the UK. Simon has diverse research interests in Bayesian statistics, longitudinal analysis and dealing with neuroimaging data. His presentation was focussed on the particular challenges of dealing with neuroimaging data. He introduced us to the world of neuroimaging, which involves high dimensional (many variables, few participants) data. He illustrated the process of dimension reduction and signal detection in two brain studies: 1) a study testing the hypothesis that neurocognitive decline due to age could be counteracted by increased bilateral use of the brain in the older adult (compensation hypothesis) and 2) a study looking at the processes of the brain while watching Alfred Hitchcock’s (Bang! You’re Dead, 1961).
Thanks to all three of our speakers. We look forward to following their brilliant careers in biostatistics!
Sabine Braat, Jessica Kasza, Karen Lamb