Modelling the Impacts of Non-pharmaceutical Interventions, Vaccination, and Weather Conditions on COVID-19 Transmission and Severity


Student thesis: Doctoral Thesis

View graph of relations


Related Research Unit(s)


Awarding Institution
Award date25 Aug 2023


The Coronavirus Disease 2019 (COVID-19) has caused significant economic and health burdens throughout the world after its emergence. The implementation of a recent global vaccination program can assist in controlling the pandemic, but uncertainties remain regarding the transmission dynamics and severity of infection. For instance, the emergence of new variants of the virus has raised concerns about the effectiveness of vaccines and the potential long-term effect of continued epidemics. It is crucial to continue research efforts to better understand the dynamic of COVID-19 and develop effective strategies for mitigating its impact. This thesis is divided into two main parts: the first part examines the effect of non-pharmaceutical interventions (NPIs) and vaccines on reducing the transmission of COVID-19. The second part investigates the effects of vaccination and weather conditions on the severity of COVID-19 infection, including hospitalization rates and case fatalities. How vaccine-induced immunity can provide transmission advantages on immune escape strains was also investigated.

During the initial outbreak, NPIs such as transportation restrictions, lockdowns, and quarantine measures were effective in reducing the COVID-19 outbreak size and keeping it under control. In Chapter 2, I developed a stochastic disease transmission model to estimate the impact of interventions on the transmission dynamics of COVID-19 in Wuhan, China, during the initial outbreak. In order to estimate the total number of infections more accurately based on the reported cases, I incorporated changes in the reporting rate, taking into account the limited detection ability at the initial stage. Our model captured the local Wuhan epidemic pattern as two-peak transmission dynamics, with one peak on February 4 and the other on February 12, 2020. The impact of intervention measures determined the timing of the first peak, leading to an 86% drop in the Re from 3.23 to 0.45. The improved diagnostic capability led to the second peak and a higher proportion of documented infections. Our estimated proportion of new documented infections out of the total new infections increased from 11% to 28% after January 26 when more detection kits were released. After the introduction of a new diagnostic criterion (case definition) on February 12, a higher proportion of daily infected cases were documented (49%).

In early 2021, vaccinations were introduced in collaboration with NPIs to reduce COVID-19 transmission. However, strict NPIs can have a serious impact on daily life and economic development. To cope with this, many governments make strategies for vaccine rollout. Increasing population immunity allows social distancing to be relaxed. This can be achieved through high vaccination coverage, which may allow for a lower strength of NPIs while keeping outbreak sizes at a manageable level. In Chapter 3, I developed an age-stratified compartmental model to optimize relaxation levels of social distancing measures (SDMs) during outbreaks triggered by different COVID-19 variants. The goal was to strike a balance between resuming economic activities and preventing a collapse of the hospital system in Hong Kong. My study was conducted prior to the Omicron outbreak, and its critical significance lies in serving as an early warning for both the public and government at a time when the Omicron variant was largely unknown. This study highlights the age-stratified mathematical transmission model can reproduce the incidence of previous wave of epidemic in Hong Kong. It quantifies how much optimal SDMs level could be lifted at given population vaccination and boosting coverages. My study results suggest that populations with less community-level immunity should ensure speedy vaccine and booster distribution to the majority across age groups and vulnerable subpopulations. Complete SDMs lift could be considered if the hybrid immunity could be achieved or the COVID-19 disease severity could be lowered as seasonal influenza.

As of 2022, the COVID-19 epidemic has continued for more than two years, and its severity was gradually reduced. Decision-makers have increasingly focused on COVID-19 hospital admissions as they provide a more accurate and immediate indication of recent infection severity and healthcare resource utilization. It is widely believed that vaccinations can increase population immunity, thus reducing infection severity and the risk of hospitalization. However, the impact of vaccination on the evolution of viruses and the subsequent risk of hospitalization remains unclear. In Chapter 4, I quantified vaccine protection and indirect effects on the case hospitalization rate (CHR) via affecting the transmission of new variants. I used data on vaccination coverage, COVID-19 cases, and CHR in 50 states of the United States and the District of Columbia during the first Omicron wave. My findings show that high vaccination coverage can increase population immunity against severe outcomes, directly reducing the COVID-19 CHR within a short time. However, high vaccination coverage may also increase the transmission (proportion) of Omicron immune-escape mutations, which in turn reduces CHRs indirectly because most of these mutations are associated with reduced severity of COVID-19 infections.

There have been studies suggesting that meteorological conditions are associated with the severity of COVID-19 infections. While some studies have reported associations between key weather indicators such as temperature and humidity, and COVID-19 mortality, the relationship between these exposures at different timings in the early stages of infection (from virus exposure up to a few days after symptom onset) and the probability of death after infection (also known as case fatality rate or CFR) is still unclear. In Chapter 5, I used Bayesian inference in conjunction with stochastic transmission models to estimate the instantaneous CFR of eight European countries. I also used distributed lag nonlinear models coupled with mixed-effect models to determine the exposure-lag-response associations between fatality rates and weather conditions to which patients were exposed at different timings. My findings suggest that environmental conditions may affect not only the initial viral load when patients are exposed to the virus but also their immune response around symptom onset. Specifically, warmer temperatures and higher humidity after symptom onset were found to be associated with lower fatality rates.