A statistical framework for tracking the time-varying superspreading potential of COVID-19 epidemic
Zihao Guo,
Shi Zhao,
Shui Shan Lee,
Chi Tim Hung,
Ngai Sze Wong,
Tsz Yu Chow,
Carrie Ho Kwan Yam,
Maggie Haitian Wang,
Jingxuan Wang,
Ka Chun Chong,
Eng Kiong Yeoh
Affiliations
Zihao Guo
Jockey Club School of Public Health and Primary Care, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China
Shi Zhao
Jockey Club School of Public Health and Primary Care, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China; Centre for Health Systems and Policy Research, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China
Shui Shan Lee
Jockey Club School of Public Health and Primary Care, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China; Stanley Ho Centre for Emerging Infectious Diseases, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China
Chi Tim Hung
Jockey Club School of Public Health and Primary Care, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China; Centre for Health Systems and Policy Research, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China
Ngai Sze Wong
Jockey Club School of Public Health and Primary Care, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China; Stanley Ho Centre for Emerging Infectious Diseases, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China
Tsz Yu Chow
Jockey Club School of Public Health and Primary Care, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China; Centre for Health Systems and Policy Research, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China
Carrie Ho Kwan Yam
Jockey Club School of Public Health and Primary Care, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China; Centre for Health Systems and Policy Research, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China
Maggie Haitian Wang
Jockey Club School of Public Health and Primary Care, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China
Jingxuan Wang
Jockey Club School of Public Health and Primary Care, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China
Ka Chun Chong
Jockey Club School of Public Health and Primary Care, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China; Centre for Health Systems and Policy Research, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China; Corresponding author at: Jockey Club School of Public Health and Primary Care, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China.
Eng Kiong Yeoh
Jockey Club School of Public Health and Primary Care, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China; Centre for Health Systems and Policy Research, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong, China
Timely detection of an evolving event of an infectious disease with superspreading potential is imperative for territory-wide disease control as well as preventing future outbreaks. While the reproduction number (R) is a commonly-adopted metric for disease transmissibility, the transmission heterogeneity quantified by dispersion parameter k, a metric for superspreading potential is seldom tracked. In this study, we developed an estimation framework to track the time-varying risk of superspreading events (SSEs) and demonstrated the method using the three epidemic waves of COVID-19 in Hong Kong. Epidemiological contact tracing data of the confirmed COVID-19 cases from 23 January 2020 to 30 September 2021 were obtained. By applying branching process models, we jointly estimated the time-varying R and k. Individual-based outbreak simulations were conducted to compare the time-varying assessment of the superspreading potential with the typical non-time-varying estimate of k over a period of time. We found that the COVID-19 transmission in Hong Kong exhibited substantial superspreading during the initial phase of the epidemics, with only 1 % (95 % Credible interval [CrI]: 0.6–2 %), 5 % (95 % CrI: 3–7 %) and 10 % (95 % CrI: 8–14 %) of the most infectious cases generated 80 % of all transmission for the first, second and third epidemic waves, respectively. After implementing local public health interventions, R estimates dropped gradually and k estimates increased thereby reducing the risk of SSEs to approaching zero. Outbreak simulations indicated that the non-time-varying estimate of k may overlook the possibility of large outbreaks. Hence, an estimation of the time-varying k as a compliment of R as a monitoring of both disease transmissibility and superspreading potential, particularly when public health interventions were relaxed is crucial for minimizing the risk of future outbreaks.
