Seminar Series: Kevin McIntyre and Emma Smith
PhD Thesis Proposal Defense Public Lectures
An Examination of Global Perioperative Causes of Death
Kevin McIntyre, PhD Candidate
Program: Epidemiology and Biostatistics
Supervisor(s): Dr. Janet Martin and Dr. Saverio Stranges
Department of Epidemiology and Biostatistics
Schulich School of Medicine & Dentistry
Western University
Short Biography:
Kevin McIntyre is a 3rd year PhD candidate in the epidemiology stream of the Epidemiology and Biostatistics program at Western University. Kevin graduated with his first undergraduate degree in 2016 from the University of Ottawa in the Health Sciences program and his second degree in 2018 from the Psychology program at Lakehead University. He entered the MSc program in Epidemiology and Biostatistics at Western University in 2018 before transferring to the PhD program in April 2020. His current research interests focus on examining perioperative cause of death trends globally through meta-analytic techniques, COVID-19 epidemiology, Bayesian statistics, knowledge translation and causal inference.
Abstract:
The overall objective of this research is to understand the drivers of Post Operative Mortality Rates (POMR) so that interventions can be implemented to reduce it. This research aims to: 1. To estimate POMR adjusted and unadjusted relative odds of perioperative mortality (PM) over time overall and within each cause of death (CoD) from caesarean section, laparotomy, and treatment of open fractures; 2. To determine overall and CoD-specific POMR estimates across countries included in the CovidSurg dataset using Bayesian hierarchical modeling; and 3. To investigate the failure to rescue (FTR) pathway to PM using propensity score modeling. This project consists of a systematic review of studies reporting on specific CoD during the perioperative period among patients who underwent a bellwether procedure since 2015. A series of multilevel meta-regressions will be constructed to assess the relative odds of PM for each CoD by year for each country. Following the meta-regressions, a Bayesian hierarchical model using data from the CovidSurg cohort study will assess the current risk of CoD-specific PM in individual countries. This will provide countries with relevant estimates that can be used to adapt their national surgical obstetric and anaesthesia plans to improve surgical safety. Furthermore, we will investigate the role of FTR through an IPTW analysis of death following a complication occurring after surgery. This research aims to find the driving causes of perioperative mortality globally. We aim to establish a foundation from which future research can specifically address major causes of death to intervene, thereby potentially leading to the prevention of countless perioperative deaths worldwide.
Keywords: Global Health, Global Surgery, Cause of Death, Post Operative Mortality Rate, Failure to Rescue, Multilevel Meta-regression, Bayesian modeling, Inverse Probability of Treatment Weighting
Nonparametric analysis of cluster randomization trials with multiple endpoints
Emma Smith, PhD Candidate
Program: Epidemiology and Biostatistics, Biostatistics Collaborative Specialization
Supervisor(s): Dr. Vipul Jairath and Dr. Guangyong Zou
Department of Epidemiology and Biostatistics
Schulich School of Medicine & Dentistry
Western University
Short Biography:
Emma Davies Smith is a fourth-year Ph.D. candidate in Biostatistics at Western University. She previously completed her BSc in Statistics and MSc in Applied Statistics at the University of Guelph, where her research focused on quantifying the impact of C difficile on health-related quality of life. Before pursuing her doctorate, Emma worked as a statistical analyst and data scientist in the insurance and ad tech industries. Her current research interests include the design, analysis, and interpretation of clinical trials, with focus on nonparametric treatment effects and cluster randomized trials.
Abstract:
Cluster randomization trials in which intact social units are randomly allocated to different groups have become the gold standard to evaluate the effectiveness of non-therapeutic interventions, including algorithms for the management of diseases and innovations in the of provision of health care. There is a large body of literature on methodological developments and applications in the field of cluster randomization trials. However, there exists a paucity of methods for such trials with subjective endpoints that are ordinal in nature. Zou (2021) proposed nonparametric methods for cluster randomization trials, focusing on estimation of the probability that a randomly selected patient from the treatment arm does better than a randomly selected patient from the control arm in terms of an endpoint, termed the win probability. Many cluster randomization trials use multiple endpoints to capture disease complexities and improve study power. These endpoints may be of equal or hierarchical clinical importance. The proposed thesis will extend methods developed by Zou (2021) to cluster randomization trials with multiple endpoints, accounting for both intracluster correlation and correlation among endpoints. Analytic methods for point estimation and confidence intervals will be developed for both the global win probability and hierarchical win probability, and simulation studies will be used to assess the performance of the developed methods.
Keywords: clinical trials, cluster randomization, treatment effects, nonparametric statistics, multivariate outcomes, global methods, composite outcomes, interval estimation
Date: Friday, October 7th
Time: 1:30 pm - 2:30 pm
Location: PHFM 3015 (Western Centre for Public Health and Family Medicine)
