Associations between meteorology and COVID-19 in early studies: Inconsistencies, uncertainties, and recommendations

Abstract

Meteorological variables, such as the ambient temperature and humidity, play a well-established role in the seasonal transmission of respiratory viruses and influenza in temperate climates. Since the onset of the novel coronavirus disease 2019 (COVID-19) pandemic, a growing body of literature has attempted to characterize the sensitivity of COVID-19 to meteorological factors and thus understand how changes in the weather and seasonality may impede COVID-19 transmission. Here we select a subset of this literature, summarize the diversity in these studies' scopes and methodologies, and show the lack of consensus in their conclusions on the roles of temperature, humidity, and other meteorological factors on COVID-19 transmission dynamics. We discuss how several aspects of studies' methodologies may challenge direct comparisons across studies and inflate the importance of meteorological factors on COVID-19 transmission. We further comment on outstanding challenges for this area of research and how future studies might overcome them by carefully considering robust modeling approaches, adjusting for mediating and covariate effects, and choosing appropriate scales of analysis.

Keywords: COVID-19; Humidity; Meteorology; SARS-CoV-2; Temperature; Virus transmission.

Fig. 1

(a) Countries from which data were drawn by the 43 studies synthesized in this review. If one city was the focus of a study, the entire country was shaded. (b) Meteorological predictor variables considered by studies.

Fig. 2

Studies focusing on the sensitivity of COVID-19 to (a) temperature and (b) humidity over the global domain versus their hindcast period. Studies are shaded by the sign or result of their key finding, and a brief summary of the study is provided. A positive relationship implies that increasing temperature or humidity is associated with additional COVID-19 cases or other COVID-19 transmission metrics, while the opposite is true for a negative relationship. Several studies do not provide directionality but rather report an optimal range.

Fig. 3

Mediating and covariate effects accounted for in individual studies. Effects are separated into environmental variables (e.g., meteorology, air quality) and non-environmental variables (e.g., demographics, mobility, non-pharmaceutical interventions, unobserved location-specific effects). The size of the scatter points is proportional to the number of effects considered.

Fig. 4

Same as Fig. 2 but for studies examining only China.