Weather Patterns as Predictors of West Nile Virus Infection Risk in Greece ^†

Abstract

West Nile Virus (WNV) poses a recurring public health threat in Greece, with outbreaks influenced by meteorological conditions. This study examines associations between key weather variables (e.g., temperature and total precipitation) and WNV incidence from 2010 to 2024 using ERA5 environmental data and available epidemiological data. Distributed lag nonlinear models were applied to identify time-lagged impacts of climate on WNV transmission. The results aim to reveal critical meteorological thresholds that could improve early warning systems and vector control. Understanding these climate–infection links can support predictive modeling and enhance public health preparedness amid growing climate variability and global vector-borne disease risks.

1. Introduction

West Nile virus (WNV) is a neurotropic, mosquito-borne virus belonging to the Flavivirus genus [1]. In nature, it is sustained through a mosquito–bird–mosquito transmission cycle, with Culex species widely recognized as the primary vectors. Wild birds are the predominant reservoir hosts of the virus [2]. Humans are incidental hosts, while vulnerable population groups at risk are the elderly and people with an underlying disease or with immunosuppression. Still no vaccine is available for humans [3,4].

Several studies across Europe and beyond have identified a range of environmental factors associated with the prevalence of WNV-infected mosquitoes and human cases [5,6,7]. Temperature has consistently emerged as a key driver, with higher temperatures linked to earlier onset and prolonged mosquito seasons, increased infection rates, and, in some cases, reduced adult mosquito survival during peak summer periods [8,9,10,11,12,13]. Precipitation has shown a more complex relationship: reduced rainfall has been associated with higher mosquito infection rates and increased WNV circulation in some areas, while early-season precipitation has been linked to delayed mosquito emergence but greater overall abundance later in the season [8,9,10,11,14]. Ecological and landscape variables, such as proximity to breeding habitats, elevation, vegetation indices, land use, and habitat fragmentation, have also been found to influence mosquito abundance and infection dynamics [11,15]. Additional climatic variables, including wind speed and overwintering conditions, have been shown to affect mosquito survival and WNV transmission [16,17,18]. These findings reflect the multifactorial nature of WNV transmission dynamics and highlight the increasing influence of climate change. Ongoing shifts in temperature, precipitation, and the frequency of extreme weather events are altering the ecological balance in many regions, creating more favorable conditions for mosquito proliferation and virus amplification. As a result, the interplay between changing climate conditions and the geographic expansion of WNV has become a growing concern in public health research [19,20].

In this study, we investigated the short-term effects of 2 m air temperature and total precipitation on the occurrence of WNV human cases in Greece from 2010 to 2024. Employing a space-time-stratified case-crossover design and distributed lag nonlinear models, we aimed to capture the delayed and nonlinear associations between environmental conditions and disease incidence at the provincial (NUTS3) level. By quantifying these associations through Odds Ratios (OR), our goal was to understand the spatiotemporal dynamics of WNV transmission and to identify critical environmental thresholds that may support the development of targeted public health interventions and early warning systems.

2. Materials and Methods

2.1. Investigation Area and Data

Our study focused on the provinces of Greece (NUTS3 level). Reported WNV human case data spanning the period 2010–2024 were obtained from the National Public Health Organization (EODY) and included essential information such as symptom onset and clinical classification (e.g., asymptomatic, WNF (West Nile Fever), or WNND (West Nile Neuroinvasive Disease)), all georeferenced at the NUTS3 level. This study included 41 out of the 52 provinces of Greece (Figure 1 (left)), selected based on the presence of a non-zero number of reported human cases through the study period. The multiannual variability of the total number of observed human cases each year is presented in Figure 1 (right).

Meteorological variables, specifically daily mean 2 m temperature and total daily precipitation, were retrieved from the ERA5 dataset provided by the European Centre for Medium-Range Weather Forecasts (ECMWF).

2.2. Methodology

Using a space-time-stratified case–crossover design, we investigated the short-term effects of meteorological variables on WNV human cases in Greece. Based on the methodology of [6], distributed lag nonlinear models were applied within a conditional logistic regression framework to capture delayed and nonlinear associations. The analysis covered human cases reported between 2010 and 2024. We examined the influence of several environmental factors—2 m air temperature and total precipitation—on the occurrence of human infections. In this study, we used OR to quantify the strength and direction of the association between environmental variables and the occurrence of WNV human cases. The OR is defined as the ratio of the odds of the event occurring in the exposed group to the odds in the non-exposed group: OR = (odds of the event in the exposed group)/(odds of the event in the non-exposed group). Specifically, we compared the odds of WNV occurrence under specific environmental conditions—defined as values above the 75th percentile of 2 m temperature and precipitation distributions—across progressively higher levels. An OR greater than 1 indicates increased odds of WNV cases with exposure to the corresponding elevated environmental conditions.

3. Results

We identified a significant, positive, and delayed effect of 2 m temperature and total precipitation on the risk of WNV human cases in Greece. Heatmaps (Figure 2 (left)) present the overall trends in the association between weekly average temperature and total precipitation with WNV incidence across different lag periods. These visualizations highlight general positive associations suggesting that warmer and wetter conditions may promote environments conducive to virus transmission. In addition, we simulated an increase in temperature of 1, 2 and 3 °C and in total precipitation of 10, 20 and 30 mm, both relative to the 75th percentile of their respective distributions (Figure 2 (right)). This approach allowed us to identify the time windows during which these environmental changes had the greatest impact on disease risk. The analysis revealed that increased temperature was associated with higher incidence at a lag of 1.5 weeks, while elevated total precipitation showed a significant effect at a lag of around 3.75 weeks. These findings underscore the relevance of environmental factors in shaping the spatiotemporal dynamics of WNV transmission and may inform the design of early warning systems based on environmental monitoring.

Table 1. Differences in Odds Ratios (OR) for weekly 2 m temperature and cumulative precipitation exposures, compared to reference values (26 °C for temperature and 30 mm for precipitation), at lag 1.5 and 3.75 weeks, respectively.

2 m Temperature		Total Precipitation
Value (°C)	Difference (ORvalue-OR26 °C)	Value (mm)	Difference (ORvalue-OR30 mm)
26	Ref	30	Ref
27	0.11	40	0.03
28	0.23	50	0.07
29	0.34	60	0.10

presents the OR at different temperature and total precipitation levels, relative to the reference values (26 °C and 30 mm, respectively). For temperature, compared to the reference value of 26 °C, the OR increases progressively with higher temperatures. Specifically, at a lag of 1.5 weeks, at 27 and 28 °C, the OR is 0.11 and 0.23 higher, respectively, while at 29 °C the OR is 0.34 higher, indicating the probability of occurrence is 34% higher at 29 °C than at 26 °C. For total precipitation, at 40, 50, and 60 mm of precipitation, the OR is 0.03, 0.07, and 0.1 higher at a lag of 3.75 weeks. These findings confirm that both higher temperatures and increased precipitation are associated with an elevated risk of the health outcome, with the effect being more pronounced for higher temperature values.

4. Discussion

Understanding the environmental drivers of WNV transmission is crucial for anticipating outbreaks and ensuring public health. In Greece, where WNV has become endemic, identifying the climatic conditions that favor transmission can support the development of early warning systems and targeted interventions. This study aimed to clarify the short-term associations between temperature, precipitation, and the incidence of human WNV cases by applying advanced statistical modeling capable of capturing delayed and nonlinear effects and using high-resolution epidemiological and meteorological data spanning a 15-year period.

Our findings reveal a significant, positive, and temporally lagged association between 2 m air temperature, total precipitation, and the occurrence of human WNV cases. Specifically, higher temperatures were linked to increased WNV risk with a lag of 1.5 weeks, while precipitation showed the strongest effect about 3.75 weeks later. These results suggest that warmer and wetter conditions may foster mosquito activity and enhance viral transmission.

Previous studies highlighted that seasonal climatic anomalies could elevate WNV transmission risk across different years [21,22]. Furthermore, numerous studies exploring short-term meteorological influences have underscored that increases in weekly mean temperatures are often followed by delayed increases in WNV cases, a pattern observed across diverse countries and climatic zones [6,16,23,24,25,26]. Our findings align with these results, reinforcing the role of temperature in shaping transmission dynamics. Similarly, several studies have reported positive associations between cumulative precipitation and WNV incidence [6,24,26], which is consistent with our observations and further highlights the dual impact of temperature and rainfall on disease risk.

Our research contributes to an understanding of how short-term weather variability influences WNV transmission risk, offering valuable evidence to support the development of early warning systems and targeted public health interventions in Greece. Limitations of this study include the reliance on reported human cases, which may underrepresent true incidence due to asymptomatic infections or underreporting. Moreover, other environmental or ecological factors—such as vector abundance, avian host dynamics, or land use changes—not considered in this study. Future research that integrates entomological and ecological data could further enhance predictive models and inform more comprehensive surveillance strategies.