The rapid global spread of COVID-19, caused by SARS-CoV-2, highlighted the need to understand its transmission dynamics. Super-spreading events (SSEs), where a few individuals infect many others, are known to significantly impact outbreaks of various infectious diseases, including SARS and MERS. While anecdotal evidence suggested SSEs might have occurred in the early COVID-19 outbreak, direct confirmation was lacking. This research aimed to fill this gap by reconstructing transmission chains using genomic data and Bayesian inference to quantify the role of SSEs in driving the initial spread of the virus in Wuhan, China. Understanding the occurrence and characteristics of SSEs is crucial for developing effective prevention and control strategies.

Previous studies have documented the role of SSEs in other viral outbreaks, such as SARS and MERS. These events significantly amplify disease transmission, contributing to the severity and rapid spread of outbreaks. Traditional methods for identifying SSEs often rely on epidemiological tracing data and statistical models, but these can lead to inaccuracies and false negatives. The concept of the basic reproductive number (R₀) and its distribution (y), proposed by Lloyd-Smith et al. (2005), provides a framework for understanding the heterogeneity of transmission. SSEs are considered events in the right tail of the distribution of y, and their propensity can be identified by estimating the skewness of this distribution. While some reports suggested the potential for SSEs in the early COVID-19 pandemic, a quantitative analysis based on robust genomic data was lacking.

The researchers utilized 208 publicly available SARS-CoV-2 genome sequences collected during the early stages of the outbreak in China. They first constructed a dated phylogeny using these sequences, incorporating temporal information to resolve the evolutionary relationships between the viral strains. This phylogeny provided the basis for reconstructing the transmission tree, mapping the flow of infections among individuals. Bayesian inference under an epidemiological model was used to estimate the parameters of the offspring distribution (the number of secondary infections caused by each infected individual). This distribution was assumed to follow a negative binomial distribution, allowing for overdispersion (variance exceeding the mean), which is a signature of SSEs. The researchers estimated the mean (R₀) and variance of the offspring distribution, as well as the dispersion parameter, a key indicator of transmission heterogeneity. To assess the robustness of their findings, the researchers conducted sensitivity analyses, including the examination of the impact of phylogenetic uncertainty. They re-analyzed the data after removing some individuals, and also analyzed randomly selected trees from the MCMC chain to evaluate the influence of phylogenetic uncertainty on the estimates of the offspring distribution parameters.

The study's key finding was the identification of super-spreading events (SSEs) during the early phase of the COVID-19 outbreak. The analysis revealed a significant overdispersion in the offspring distribution, meaning that the variance was substantially larger than the mean. The estimated dispersion parameter was 0.23 (95% CI: 0.13–0.39), which is considerably less than 1 and further supports the presence of overdispersion. The mean of the offspring distribution (R₀) was 1.23 (95% CI: 1.09–1.39), indicating that, on average, each infected person infected 1.23 others. However, the large variance highlights the heterogeneity in transmission, with some individuals causing far more infections than others. The transmission tree reconstructed from the phylogenetic analysis identified 18 pairs of patients with a high probability of direct transmission (bidirectional probability > 0.5). Sensitivity analysis, by removing the top three transmission pairs, showed that the results were robust to the uncertainty in the phylogeny. The study also found that uncertainty in phylogeny might lead to overestimation of the dispersion parameter. This suggests that the true extent of SSEs may be even greater than initially estimated. Comparison of the dispersion parameter with estimates for other diseases such as Ebola and SARS revealed similar values, indicating the considerable heterogeneity of COVID-19 transmission during its initial spread.

The findings demonstrate that SSEs played a significant role in the initial spread of COVID-19, emphasizing the importance of considering this heterogeneity in developing control strategies. The low dispersion parameter (0.23) indicates highly skewed offspring distribution, which is consistent with the presence of super spreaders. The genomic approach employed in this study offers a powerful alternative to relying solely on epidemiological data for identifying SSEs, which can be susceptible to biases and limitations. The early implementation of interventions by the Chinese government might have suppressed further SSEs after January 1st, 2020. Factors like higher binding affinity of SARS-CoV-2 to human ACE2 receptors, aerosol transmission, and surface stability likely contributed to the ease of transmission compared to SARS-CoV. The variability in the incubation period also poses a challenge to epidemiological tracing efforts, making genomic approaches particularly valuable in revealing the true nature of transmission patterns.

This study provides strong evidence of super-spreading events during the early stages of the COVID-19 pandemic using a robust genomic approach. The identification of SSEs emphasizes the importance of considering transmission heterogeneity in public health strategies. Future research should investigate the underlying factors contributing to super-spreading events, including individual characteristics and environmental conditions, to better inform targeted interventions and prevent future outbreaks.

The study is limited by the sampling frequency of SARS-CoV-2 genomes during the early outbreak phase. The relatively small number of complete genomes available might have resulted in some limitations to the accuracy of the reconstructed phylogeny and transmission tree. The study focused on the initial two months of the outbreak, which may not fully represent the transmission dynamics throughout the entire pandemic. The uncertainty in phylogeny could lead to slight overestimation of the dispersion parameter, potentially underestimating the importance of super-spreading events.