Air temperature exposure-lag response on the incidence of COVID-19 in twelve Italian cities: a meta-analysis
This article was originally published here
Approx Res. March 16, 2022: 113099. doi: 10.1016/j.envres.2022.113099. Online ahead of print.
The answer to air temperature exposure lag on daily incidence of COVID-19 is unclear and there have been concerns about the robustness of previous studies. Here we present a high spatial and temporal resolution analysis using the Distributed Lag Nonlinear Modeling (DLNM) framework. Using almost two years of data, we fitted statistical models to twelve Italian cities to quantify the delayed effect of air temperature on the daily incidence of COVID-19, taking into account several categories of factors. potential confusion (weather, air quality and non-pharmaceutical interventions) . Covariance coefficients and matrices for the temperature term were then synthesized using random-effects meta-analysis to produce pooled estimates of exposure-lag response with effects presented as relative risk (RR ) and the cumulative RR (RRsperm). The cumulative exposure response curve was non-linear, with maximum risk at 15.1°C and decreasing risk at progressively lower and higher temperatures. The lowest RRsperm at 0.2°C is 0.72 [0.56,0.91] times that of the highest risk. Because of this nonlinearity, the shape of the delayed response curve necessarily varied with temperature. This work suggests that on any given day, an air temperature of around 15°C maximizes the incidence of COVID-19, with the effects being spread over the following ten days or more.