Bayesian inference of transmission chains: opportunities and challenges for characterising SARS-CoV-2 transmission dynamics
File(s)
Author(s)
Geismar, Cyril
Type
Thesis
Abstract
Assessing the transmissibility of a pathogen primarily involves quantifying the amount, speed, and patterns of transmission. During the COVID-19 pandemic, these dynamics varied by time and location, driven in part by the emergence of variants of concern (VOCs). Successive waves of infection prompted the continuous reassessment of transmission dynamics and public health policies. However, large-scale epidemic models faced challenges in rapidly evaluating changes in transmissibility and capturing transmission heterogeneities in high-risk settings, such as in hospitals and care homes.
The unprecedented availability of epidemic surveillance data enabled statistical inference of transmission chains, providing opportunities to reveal detailed insights into local transmission, but also to characterise broader epidemiological properties of SARS-CoV-2. Nevertheless, limited SARS-CoV-2 genetic diversity hindered the resolution of transmission trees. This thesis investigates the extent to which outbreak reconstruction tools can enhance our understanding of SARS-CoV-2 transmission dynamics and how they can be integrated into routine surveillance to improve public health response.
In Chapter 2, the reconstruction of household transmissions at the national level revealed that 20% of transmission events had negative serial intervals across VOCs. In Chapter 3, we developed a method to quantify group-level transmission assortativity and established guidelines on the minimum data required for transmission chains to elucidate transmission patterns. In Chapter 4, we assessed the contribution of healthcare workers (HCWs) and patients in nosocomial transmission by integrating epidemiological, genetic, and contact data under simulated real-time conditions. We found early on that HCWs were more likely to transmit to other colleagues than to patients.
Integrating outbreak reconstruction tools into surveillance systems, even in the absence of genetic data, holds significant potential to support real-time modelling and public health response. Leveraging diverse data streams, at various scales, to address different needs, our work demonstrated that Bayesian inference of transmission chains provides valuable insights into SARS-CoV-2 dynamics.
The unprecedented availability of epidemic surveillance data enabled statistical inference of transmission chains, providing opportunities to reveal detailed insights into local transmission, but also to characterise broader epidemiological properties of SARS-CoV-2. Nevertheless, limited SARS-CoV-2 genetic diversity hindered the resolution of transmission trees. This thesis investigates the extent to which outbreak reconstruction tools can enhance our understanding of SARS-CoV-2 transmission dynamics and how they can be integrated into routine surveillance to improve public health response.
In Chapter 2, the reconstruction of household transmissions at the national level revealed that 20% of transmission events had negative serial intervals across VOCs. In Chapter 3, we developed a method to quantify group-level transmission assortativity and established guidelines on the minimum data required for transmission chains to elucidate transmission patterns. In Chapter 4, we assessed the contribution of healthcare workers (HCWs) and patients in nosocomial transmission by integrating epidemiological, genetic, and contact data under simulated real-time conditions. We found early on that HCWs were more likely to transmit to other colleagues than to patients.
Integrating outbreak reconstruction tools into surveillance systems, even in the absence of genetic data, holds significant potential to support real-time modelling and public health response. Leveraging diverse data streams, at various scales, to address different needs, our work demonstrated that Bayesian inference of transmission chains provides valuable insights into SARS-CoV-2 dynamics.
Version
Open Access
Date Issued
2024-11-01
Date Awarded
01/06/2025
License URL
Advisor
Cori, Anne
Jombart, Thibaut
White, Peter
Sponsor
National Institute for Health Research (Great Britain)
Publisher Department
School of Public Health
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)