COVID-19: Climate and early global patterns of the outbreak
COVID-19: Climate and early global patterns of the outbreak. Jefferson T, Heneghan C.
Published on June 3, 2020 | Last modified on June 17, 2020
Transmission Dynamics of COVID-19
||Ficetola GF, Rubolini D. Climate affects global patterns of COVID-19 early outbreak dynamics. medRxiv 2020:2020.03.23.20040501 2020
Temperature and humidity are strongly associated with the variation of the growth rate of Covid-19 cases across the globe in the early phase of the outbreak.
The relationships between the incidence of Covid-19 and climate was insensitive to the potential confounding effects of air pollution and socioeconomic variables (population size, density and health expenditure). There is a strong relationship between local climate and Covid-19 growth rates, suggesting the possibility of seasonal variation in the spatial pattern of outbreaks, with temperate regions of the Southern Hemisphere becoming at particular risk of severe outbreaks during the austral autumn-winter.
The figure below is reproduced from the Supplementary material.
Confirmed Covid-19 cases cumulative growth curve (Thailand).
The first two confirmed cases were reported on Jan 22nd 2020. The 25 case threshold was reached on day 35 (Feb. 4th, the blue dots). The lag phase is followed by an exponential phase of confirmed cases (exponential phase, orange and red dots).
Covid-19 growth rates peaked in the Northern Hemisphere with a mean temperature of ~5°C, and specific humidity of 4-6 g/m3. This equates to a relative humidity of 17 to 26% (conversions are done at https://planetcalc.com/2167/)
Growth rates were lower both in warmer/wetter and colder/dryer regions. The relationships were robust to the potential confounding effects of air pollution and socioeconomic variables, including population size, density and health expenditure
What did they do?
This is an ecological study correlating specific humidity and mean air temperature with early phase Covid 19 case growth. The outbreak event was defined as the day when at least 25 confirmed cases were reported in a given country or region. The authors constructed growth curves at 5 days intervals for the initial phase of the outbreak. Environmental variables were calculated as mean monthly values for temperature (°C) and specific humidity (g/m3) during a 30-days interval, including the day of the end of the exponential incidence phase and the preceding 29 days. The authors analysed levels of PM2.5 (as at 2016) by region, health expenditure per capita and demographic make-up in the over 65s, density and population size. The authors excluded Hubei province because the initial phase was over and carried out a sensitivity analysis by countries with and without control measure in place.
As population density was strongly related to PM2.5, and health expenditure was positively related to population 65+ the authors did not include these variables in the main analysis.
Analyses can be heavily affected when independent variables are highly correlated. For example, the population density was strongly positively related to PM2.5.
|Clearly defined setting
||Demographic characteristics described
||Follow-up length was sufficient
||Transmission outcomes assessed
||Main biases are taken into consideration
What else should I consider?
The lag time and exponential growth phase demonstrated in this paper need replication, which will then allow further analysis of what triggers the exponential phase.
About the authors