Estimating the epidemic size of super-spreading coronavirus outbreaks in real-time

Abstract

In the past two decades, novel coronaviruses such as severe acute respiratory syndrome coronavirus (SARS-CoV-1), Middle East respiratory syndrome coronavirus (MERS-CoV), and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) have caused significant epidemics and the recent COVID-19 pandemic. Super-spreading events (SSEs) occur when a small number of individuals infecting a disproportionately large number of people, leading to rapid and extensive outbreaks. Characterizing the heterogeneity in transmission and accurately estimating SSE sizes in real-time are crucial for effective surveillance and timely implementation of control measures. Relying on daily reports of confirmed cases and the population measures of average transmissibility, such as basic reproductive number R_0 and effective reproductive number R_t, are insufficient to reflect the explosive growth of ongoing SSEs. As a complement, a statistical framework based on back calculation was developed. It leverages data on the time of exposure, time of symptom onset, time of confirmation of the reported cases, and prior information about the epidemiological characteristics of SARS, MERS, and COVID-19. The framework underwent validation using simulated scenarios resembling SSEs of SARS, MERS, and COVID-19. Subsequently, it was retrospectively applied to real-world case studies, including the Amoy Gardens SARS SSE in Hong Kong in 2003, three nosocomial MERS SSEs in South Korea in 2015, and two COVID-19 SSEs in Hong Kong restaurants in 2020. The accuracy and precision of the estimation improve with longer observation time, larger SSE sizes, reduced uncertainty in the time from symptom onset to confirmation, and more precise prior information on the incubation period distribution and the mode of transmission. The framework can produce robust epidemic size estimation with relative error within 1 before 50% of the cases are reported when there is no secondary transmission in SARS and MERS simulated SSEs or when it is adjusted for as in COVID-19 simulated SSEs. In real-world cases, the true epidemic size was captured within the 95% CrI of the estimated sizes after 37% of cases were reported in the Amoy Gardens SARS SSE. Similarly, the range was 41-62% for the three nosocomial MERS SSEs in South Korea and 76-86% for the two COVID-19 SSEs in Hong Kong. By identifying and projecting SSEs, public health authorities gain insights into the underlying factors and mechanisms driving SARS, MERS, and COVID-19 outbreaks. Our framework can be integrated into existing surveillance systems. Ultimately, it becomes possible to monitor SSEs closely and enhance situational awareness.