Anticipating the locations in Europe of high‐risk areas for West Nile virus outbreaks in 2021

West Nile virus infections in humans are continuously increasing, and the virus has spread considerably in Europe over the past decade. The incidence of the disease was unusually high between 2018 and 2020. The resulting model identifies the West Nile virus outbreak‐prone areas during 2021, even in regions where the virus has not yet been discovered. It is remarkable that in Central Europe, new favourable areas are emerging, where early actions could lessen the impact of the disease.

| 983 GARCÍA-CARRASCO et Al. favour WNV occurrence and the distribution of cases throughout the study area in 2020 (Table 1).
The risk model was elaborated following several steps. First, through univariate logistic regression we assessed the environmental power of each environmental variable. Multicollinearity among the environmental variables was controlled by calculating pairwise Spearman correlation coefficients. If two variables, belonging to the same factor (Table 1), were correlated by more than 0.8, the least explanatory one was deleted. The false discovery rate was controlled to avoid an increase in type I errors due to the number of variables used in the analysis (Benjamini & Hochberg, 1995). Finally, we performed a multivariate forward-stepwise logistic regression in which a variable was added to the null model if the resulting regression was most significantly improved by the new variable. The result was a probability value (P) of WNV outbreak in each NUTS according to its environmental characteristics. The P value of each NUTS was transformed into a favourability value (F) using the favourability function (Real et al., 2006).
with n1 being the number of NUTS with reported WNV cases, and n0 the number of NUTS with no virus outbreak reported. The result was an F value (ranging from 0 to 1). This favourability model shows how the local probability of WNV outbreak differs from that expected by chance in Europe and thus identifies those localities with environmental conditions that favour outbreak occurrence. A detailed description of the methodological processes is available in the work of García-Carrasco et al. (2021).

| RE SULTS
Four variables were significantly associated with the areas of high environmental favourability for WNV outbreaks during the 2020 transmission season in Europe. Areas close to watercourses, with rice paddies and a high density of chicken farms, were the zones with the highest risk of occurrence of WNV cases in humans (see

| D ISCUSS I ON
Even though chickens are not competent species to reinfect biting mosquitoes with WNV, chicken farms constitute an ideal setting for WNV vector proliferation (Sowilem et al., 2019). A high density of chicken farms means a high quantity of blood meal available for ornithophilic mosquitoes, the same that transmit the virus from bird to bird, and eventually from bird to humans. Watercourses and paddy fields provide ideal breeding areas for mosquitoes and attract migratory birds (Hardy et al., 1983;Liang et al., 2015), which may carry the virus from other areas. Environmental temperature is widely known as an important driver of WNV transmission by increasing the replication rate of the virus (Hardy et al., 1983), shortening the gonotrophic cycle of mosquitoes and favouring the early transmission of the virus (Ciota et al., 2014;Hartley et al., 2012).
As temperature is involved in the distribution of human WNV cases in Europe, change in environmental temperatures might influence the spread of the virus. Climate change could produce changes in the distribution of vectors at higher altitudes, but also at higher latitudes (Semenza & Suk, 2018). This could explain the cases in Central Europe, as an increase in temperature is occurring mainly in Central and Northern Europe (Andriamifidy et al., 2019; I.P.C.C., 2014).
As the spread of the disease has been proven to be facilitated within the same watershed , our model suggests that special attention should be paid to the basins of the rivers Danube, Po, Elba and Aegean and the southern watersheds of the Iberian Peninsula. In these highly favourable areas, the disease is more likely to occur, manifest earlier and appear more intensely during the next epidemic season .
Prevention plays a more important role than ever in the case of infectious diseases. As survival in mosquitos might facilitate the annual recurrence of the virus at the same sites where outbreaks occurred the previous year (Rappole et al., 2000), our modelling approach should be regularly updated to continue predicting the risk in subsequent years. Following the incidence of WNV in Southern Spain during the summer of 2020, health authorities carried out fumigation plans in sensitive areas as well as campaigns to increase citizen awareness to reduce mosquito bites (using mosquito nets, avoiding being outside at sunrise and sunset, etc.

TA B L E 1 (Continued)
Although the first known cases of WNV in Europe were located in a few southern and eastern countries, the number of affected countries is increasing, including Spain in the south and reaching areas in higher latitudes as the Netherlands in Central Europe. This type of risk model may contribute to identifying the disease-prone areas on a yearly basis and could be regularly updated to better anticipate WNV outbreaks. Early actions in risky areas could lessen the impact of the disease in the following transmission season, reducing the number of affected people. Our study highlights the usefulness of pathogeography as an emergent discipline for the study of infectious diseases in constantly changing societies and environments.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

E TH I C A L A PPROVA L
Not applicable.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing not applicable to this article as no datasets were generated or analysed during the current study.