A cross-sectional analysis of meteorological factors and SARS-CoV-2 transmission in 409 cities across 26 countries

There is conflicting evidence on the influence of weather on COVID-19 transmission. Our aim is to estimate weather-dependent signatures in the early phase of the pandemic, while controlling for socio-economic factors and non-pharmaceutical interventions. We identify a modest non-linear association between mean temperature and the effective reproduction number (Re) in 409 cities in 26 countries, with a decrease of 0.087 (95% CI: 0.025; 0.148) for a 10 °C increase. Early interventions have a greater effect on Re with a decrease of 0.285 (95% CI 0.223; 0.347) for a 5th - 95th percentile increase in the government response index. The variation in the effective reproduction number explained by government interventions is 6 times greater than for mean temperature. We find little evidence of meteorological conditions having influenced the early stages of local epidemics and conclude that population behaviour and government interventions are more important drivers of transmission.


nature research | reporting summary
April 2020 Data Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A list of figures that have associated raw data -A description of any restrictions on data availability Field-specific reporting Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection. Meteorological variables (mean temperature, dew point temperature, solar radiation, wind components and precipitation) were derived from ERA5 reanalysis product "https://cds.climate.copernicus.eu/cdsapp#!/search?type=dataset".
Socio-economic and demographic characteristics were extracted from the OECD Regional and Metropolitan database "http://www.oecd.org/regional/regionalpolicy/regionalstatisticsandindicators.htm" and Worldcities database.
Data were processed and harmonized at city-level. The city-level data used in the main and supplementary analysis of the paper are available in the GitHub directory https://github.com/fsera/COVIDWeather/ We used a two-stage ecological approach to examine the impact of meteorological variables on SARS-CoV-2 transmission. in the first stage we estimated the effective reproduction number (Re) early in the epidemic in 409 locations (city or small region) within 26 countries. In the second stage, we estimated the association between city-level Re (allowing for standard errors) with meteorological variables (mean temperature, relative and absolute humidity, solar radiation, wind speed and precipitation), controlling for confounding by total population, population density, GDP per capita, percentage of population >65 years, PM2.5, and nonpharmaceutical interventions (OxCGRT Government Response Index). The analysis was performed considering the two-level (cities and countries) structure of the data using a multilevel meta-regression model Data in this study were obtained from the well-established MCC Collaborative Research Network. The current MCC network covers 750 cities in 43 countries/regions. For this study, 26 countries provided daily time-series of COVID-19 cases for a total of 502 locations (cities or small regions). COVID-19 data were downloaded from public available repository or obtained from health agencies. The MCC network provided access to COVID-19 data at the city (or small region) level. Analysis case data at this fine spatial scale reduces information bias and confounding compared to using large regions or country level data.
The MCC network provided access to COVID-19 data at the city (or small region) level. Despite the opportunistic nature of the sample, we achieved a reasonable global coverage. Overall, we collected 2,771,137 COVID-19 cases, representing 44.8% of the cumulative cases registered by 31 May 2020 in the Johns Hopkins database "https://coronavirus.jhu.edu/map.html".
The research data are 1) Time-series of COVID-19 confirmed cases collected in 502 cities within 26 countries. Whenever possible COVID-19 data were downloaded from existing public repository. For some countries data were obtained from health authorities. The downloaded data were processed and harmonized by FS.
2) time-series of meteorological variables (mean temperature, relative and absolute humidity, solar radiation, wind speed and precipitation) derived from ERA5 reanalysis product. The data were downloaded by RS in NetCDF format and processed with R version 4.0.3.
3) Time series of Pollution levels (PM2.5). The data were derived from CAMS near real time. The data were downloaded by FS in NetCDF format and processed with R version 4.0.3. 4) Time fixed city level covariates derived from the OECD regional and metropolitan and World Cities database. The data were downloaded and processed by FS.