The non-linear and interactive effects of meteorological factors on the transmission of COVID-19: A panel smooth transition regression model for cities across the globe

Abstract

The ongoing pandemic created by COVID-19 has co-existed with humans for some time now, thus resulting in unprecedented disease burden. Previous studies have demonstrated the non-linear and single effects of meteorological factors on viral transmission and have a question of how to exclude the influence of unrelated confounding factors on the relationship. However, the interactions involved in such relationships remain unclear under complex weather conditions. Here, we used a panel smooth transition regression (PSTR) model to investigate the non-linear interactive impact of meteorological factors on daily new cases of COVID-19 based on a panel dataset of 58 global cities observed between Jul 1, 2020 and Jan 13, 2022. This new approach offers a possibility of assessing interactive effects of meteorological factors on daily new cases and uses fixed effects to control other unrelated confounding factors in a panel of cities. Our findings revealed that an optimal temperature range (0°C–20 °C) for the spread of COVID-19. The effect of RH (relative humidity) and DTR (diurnal temperature range) on infection became less positive (coefficient: 0.0427 to −0.0142; p < 0.05) and negative (coefficient: −0.0496 to −0.0248; p < 0.05) with increasing average temperature(T). The highest risk of infection occurred when the temperature was −10 °C and RH was >80% or when the temperature was 10 °C and DTR was 1 °C. Our findings highlight useful implications for policymakers and the general public.