Association between meteorological variations and the superspreading potential of SARS-CoV-2 infections

Background: While many investigations examined the association between environmental covariates and COVID-19 incidence, none have examined their relationship with superspreading, a characteristic describing very few individuals disproportionally infecting a large number of people.

Methods: Contact tracing data of all the laboratory-confirmed COVID-19 cases in Hong Kong from February 16, 2020 to April 30, 2021 were used to form the infection clusters for estimating the time-varying dispersion parameter (k_t), a measure of superspreading potential. Generalized additive models with identity link function were used to examine the association between negative-log k_t (larger means higher superspreading potential) and the environmental covariates, adjusted with mobility metrics that account for the effect of social distancing measures.

Results: A total of 6,645 clusters covering 11,717 cases were reported over the study period. After centering at the median temperature, a lower ambient temperature at 10^th percentile (18.2 °C) was significantly associated with a lower estimate of negative-log k_t (adjusted expected change: -0.239 [95 % CI: -0.431 to -0.048]). While a U-shaped relationship between relative humidity and negative-log k_t was observed, an inverted U-shaped relationship with actual vapour pressure was found. A higher total rainfall was significantly associated with lower estimates of negative-log k_t.

Conclusions: This study demonstrated a link between meteorological factors and the superspreading potential of COVID-19. We speculated that cold weather and rainy days reduced the social activities of individuals minimizing the interaction with others and the risk of spreading the diseases in high-risk facilities or large clusters, while the extremities of relative humidity may favor the stability and survival of the SARS-CoV-2 virus.