Drivers and epidemiological patterns of West Nile virus in Serbia

Abstract

Background: West Nile virus (WNV) is an emerging mosquito-borne pathogen in Serbia, where it has been detected as a cause of infection in humans since 2012. We analyzed and modelled WNV transmission patterns in the country between 2012 and 2023.

Methods: We applied a previously developed modelling approach to quantify epidemiological parameters of interest and to identify the most important environmental drivers of the force of infection (FOI) by means of statistical analysis in the human population in the country.

Results: During the study period, 1,387 human cases were recorded, with substantial heterogeneity across years. We found that spring temperature is of paramount importance for WNV transmission, as FOI magnitude and peak timing are positively associated with it. Furthermore, FOI is also estimated to be greater in regions with a larger fraction of older adult people, who are at higher risk to develop severe infections.

Conclusion: Our results highlight that temperature plays a key role in shaping WNV outbreak magnitude in Serbia, confirming the association between spring climatic conditions and WNV human transmission risk and thus pointing out the importance of this factor as a potential early warning predictor for timely application of preventive and control measures.

Keywords: Culex; West Nile virus; mathematical model; mosquito; vector-borne.

Figure 1

WNV cases recorded in Serbia between 2012 and 2023. Total number of cases by administrative area (NUTS3 level) (A, with inset map highlighting Serbia), by year (B), by week of symptoms onset (C) and by age group (D).

Figure 2

FOI model. (A) Frequencies of the stochastically predicted and observed Σ _y,i (total number of WNV human cases recorded during year y in region i ). Bars: observed Σ _y,i . Points and lines represent average and 95% quantiles of the frequencies of the stochastically predicted values, respectively. Values are shown aggregated by group. (B) Predicted  $\left(\overline{{\Sigma }_{y,i}}\right)$  and observed  $\left({\Sigma }_{y,i}\right)$  total number of WNV human cases for each region and year.

Figure 3

Estimated average (μ), standard deviation (σ) and magnitude (c, log-transformed) distributions (violin plots, A–C respectively) for each year. Dots represent estimated values for each epidemiological curve. Dashed horizontal lines are average values computed overall years.

Figure 4

Model predictions. (A) Model predictions (lines) for c (FOI magnitude) conditional to T_spring (average spring temperature) and ε (fraction of inhabitants older than 65 years). Green (orange) dots represent FOI-model values with associated ε belonging to the 0.15–0.2 (0.2–0.25) interval. (B,C) Model predictions for μ (day of the year when the FOI reaches its maximum) and σ (length of the epidemiological season) respectively; continuous and dashed lines provide average and CIs respectively, dots represent FOI-model values.