First Expert Elicitation of Knowledge on Possible Drivers of Observed Increasing Human Cases of Tick-Borne Encephalitis in Europe

Abstract

Tick-borne encephalitis (TBE) is a viral disease endemic in Eurasia. The virus is mainly transmitted to humans via ticks and occasionally via the consumption of unpasteurized milk products. The European Centre for Disease Prevention and Control reported an increase in TBE incidence over the past years in Europe as well as the emergence of the disease in new areas. To better understand this phenomenon, we investigated the drivers of TBE emergence and increase in incidence in humans through an expert knowledge elicitation. We listed 59 possible drivers grouped in eight domains and elicited forty European experts to: (i) allocate a score per driver, (ii) weight this score within each domain, and (iii) weight the different domains and attribute an uncertainty level per domain. An overall weighted score per driver was calculated, and drivers with comparable scores were grouped into three terminal nodes using a regression tree analysis. The drivers with the highest scores were: (i) changes in human behavior/activities; (ii) changes in eating habits or consumer demand; (iii) changes in the landscape; (iv) influence of humidity on the survival and transmission of the pathogen; (v) difficulty to control reservoir(s) and/or vector(s); (vi) influence of temperature on virus survival and transmission; (vii) number of wildlife compartments/groups acting as reservoirs or amplifying hosts; (viii) increase of autochthonous wild mammals; and (ix) number of tick species vectors and their distribution. Our results support researchers in prioritizing studies targeting the most relevant drivers of emergence and increasing TBE incidence.

1. Introduction

In Eurasia, the most common viral tick-borne disease in humans is tick-borne encephalitis (TBE), caused by a flavivirus (TBEV), which is transmitted by the bite of infected hard ticks found in woodland habitats [1]. In Europe, the most important tick vector is Ixodes ricinus, whereas in Russia and Asia, it is Ixodes persulcatus. In addition, in Asia, Haemaphysalis concinna also seems to play a major role [2,3].

Less frequently, other routes of transmission occur [4,5,6], which involve mostly the consumption of unpasteurized milk/milk products originating from infected animals or, alternatively, potential pathways such as handling infected material, blood-borne infection, and solid organ transplantation [5].

Vector-borne transmission of TBEV involves hard ticks [1], which transmit TBEV to a variety of small and large mammals, birds, and humans [7,8]. Each tick life stage has a preference for, or most often uses, certain vertebrate groups (often dependent on the body size of these hosts) [1]. In addition, infected migratory birds are suspected of carrying the virus to new suitable areas (e.g., [8,9]). TBEV is transmitted within ticks mainly through trans-stadial [10] and occasionally trans-ovarial transmission [11]. The importance of trans-stadial transmission is related to the fact that hard ticks take only one blood meal per life stage. In addition, co-feeding transmission (i.e., a non-viremic transmission) is now widely accepted for TBEV and occurs when infected and uninfected vectors feed in spatiotemporal proximity to each other on the same host [12]. Indeed, efficient circulation of the TBEV in tick populations is achieved by naive ticks taking a blood meal from viraemic animal hosts (systematic transmission) as well as by co-feeding [13,14]. The TBE incubation period is known to depend in the laboratory on the virus inoculation doses, virus subtypes, and host innate and adaptive immunity. In the case of vector-borne transmission, the human incubation period of TBE is usually between 7 and 14 days (range: 2 to 28 days with a median of 8 days) [15], but in the case of foodborne TBE, the median incubation period is shorter (3.5 days), and neuro-invasive disease is more common compared to vector-borne transmission [15] (reported as 38.9% in [4]). When the human infection is symptomatic, the clinical course is frequently biphasic, with influenza-like symptoms (i.e., fever, fatigue, headache, muscle aches, and nausea) and signs that range from mild meningitis (inflammation of the membrane that surrounds the brain and spinal cord) to severe meningoencephalitis (inflammation involving the brain) with or without paralysis [16,17]. After acute TBE, in up to 50% of patients, a post-encephalitic syndrome can develop, causing long-lasting morbidity that often affects the quality of life [16]. The burden of human TBE can be assessed by the calculation of the DALYs (disability-adjusted life years, i.e., the loss of the equivalent of one year of full health). From an individual perspective, DALYs of 0.23 per TBE case (95% uncertainty interval: 0.22–0.24) were estimated in Slovenia in 2009–2013, where neurological sequelae of TBE had the largest impact on the overall DALY measurement [18]. In Sweden, the social burden of TBE (cost of illness and death) has been estimated at an annual cost of €24.5 million over the 2015–2019 period [19].

TBEV is tentatively grouped into seven subtypes, mostly according to their phylogenetic relationships. Indeed, virus strains that differ by less than 10% of nucleotides in the polyprotein-coding gene are provisionally proposed to belong to the same subtype [20]. Each subtype can be distinguished by its different phylogeography, ecology, virulence, and pathogenicity [21]. Within the TBEV viral species, three main subtypes are already defined by the International Committee on Taxonomy of Viruses: the European TBEV (TBEV-EU), the Siberian TBEV (TBEV-Sib), and the Far Eastern TBEV (TBEV-FE), of which the TBEV-EU is predominant in Europe [22]. The co-circulation of different subtypes was already demonstrated in a hyperendemic area [23].

There is no specific antiviral treatment for TBE; the human TBE risk can be reduced by using tick repellents on skin and clothing, personal protective clothing, pasteurization of milk, and vaccination of individuals [16]. The list of the available inactivated vaccines was recently reviewed by Kubinski et al. [24]. Increasing vaccination among all age groups can be the most effective and efficient strategy to reduce the burden of TBE on endemic areas and protect the health of the whole population [25]. TBE vaccines are effective, well tolerated, and cost-effective in endemic countries [26]. Currently, there are no vaccines available for veterinary use. However, proof of concept of the immunogenicity of a new adjuvanted TBEV vaccine was demonstrated in an in vivo mouse model [27]. If this low-cost vaccine could be validated, its use in animals, especially ruminants, may mitigate food-borne transmission through raw-milk consumption.

In Europe, TBE is endemic in several central, northern, and eastern countries. Some countries experienced novel autochthonous TBE in humans, like the Netherlands in 2016 [28], Belgium in 2018 and 2020 [29,30], and the United Kingdom in 2019 [31]. Since 2017, there has been a gradual increase in reported TBE cases in the European Union/European Economic Area (EU/EEA) (see also the atlas of diseases from ECDC at the following address: https://atlas.ecdc.europa.eu/public/; accessed on 15 March 2023). In 2020, 3817 cases of TBE were reported (97.8% confirmed mainly by the detection of specific IgM and IgG antibodies in serum and cerebrospinal fluid, usually by enzyme-linked immunosorbent assay, and some after which a TBEV-specific virus neutralization test was applied) in 24 countries of the EU/EEA, with a clear seasonal pattern (95% of TBE human cases occurring between May and November and 5% between December and April) [32]. This represents an incidence of 0.9 cases per 100,000 inhabitants, i.e., 50% more than the 2016–2018 baseline.

In addition, TBE cases were generally more frequently reported among men and in the age group of 45–64 years [32]. This increased notification rate should be interpreted with caution, since multiple factors may play a role in it, but the number of reported cases is likely to vastly underestimate the number of infections (as a baseline, a correction factor for underestimation of 4.5 was estimated by [33]), since the majority of infections by TBEV are asymptomatic (i.e., manifestations with mild clinical symptoms that may remain undiagnosed). Furthermore, there is no mandatory notification of TBE in many countries. Moreover, a recent 4-year prospective cohort study performed in France from 2016 to 2019 (N = 494 cases of encephalitis) indicates that TBEV makes up 8% of encephalitis cases with an identified cause and, compared to a cohort from 2007 (N = 222), a significant increase in TBEV occurred (i.e., the number of causes of TBEV amongst all causes of encephalitis in adult patients was 3/106 and 26/257 in 2007 and 2016–2019, respectively; Fisher’s exact test, p-value = 0.013) [34].

An emerging infectious disease is defined by the United States National Institute of Allergy and Infectious Diseases (NIAID; https://www.niaid.nih.gov/research/emerging-infectious-diseases-pathogens (accessed on 15 March 2023)) as “infectious diseases that have newly appeared in a population or have existed but are rapidly increasing in incidence or geographic range, or that are caused by one of the NIAID category A, B, or C priority pathogens”. TBEV is within category C (emerging pathogens that could be engineered for mass dissemination in the future because of their availability, ease of production and dissemination, potential for high morbidity and mortality rates, and major health impact). Tick-borne encephalitis is also included in the WHO group of vaccine-preventable diseases [35].

Studies on the initial emergence patterns of TBEV across Eastern Europe late last century highlighted the range of interacting social and environmental drivers potentially involved in increasing the risk for human populations (e.g., [36]) through mechanisms acting to increase tick abundance, hazards (e.g., scrub regeneration on an abandoned farmland), and human exposure in forests (e.g., increased unemployment of some groups and increased wealth and leisure for others).

The aim of this study is to investigate, for the first time, possible drivers for the current observed emergence in Europe, including the invasion of new countries and increased incidence of TBE in humans. A multi-criteria decision analysis (MCDA) method was chosen because it allows the systematic integration of information from a range of sources [37] and aims at improving repeatability and transparency [38].

2. Materials and Methods

The methodology followed in this expert elicitation of knowledge is the same as that previously published by the UREAR-ULiège for other emerging/zoonotic diseases [39,40,41], but adapted by a panel of experts for TBE. Briefly, we listed 59 possible drivers grouped in eight domains and elicited forty European experts to: (i) allocate a score per driver, (ii) weight this score within each domain, and (iii) weight the different domains and attribute an uncertainty level per domain. An overall weighted score per driver was calculated, and drivers with comparable scores were grouped in several terminal nodes using a regressive tree analysis.

2.1. Study Objective

The objective of the study was to prioritize TBE drivers in order to understand the factors that influence the observed emergence or increased incidence of TBE in humans across Europe. Using the following algorithms at December 31, 2022: (((tick-borne encephalitis virus[Title/Abstract]) OR (TBEV[Title/Abstract]) OR (tick-borne encephalitis disease[Title/Abstract])) AND (human[Title/Abstract]) AND (Europe[Title/Abstract])) search strings were conducted in PubMed (US National Library of Medicine, National Institutes of Health). The results of the search (N = 135 articles from 1994 through 2022) identified 29 review papers; other papers were related to field/epidemiological surveys (N = 45), biology studies (N = 20), diagnosis (N = 13), experimental studies (N = 1), treatments (N = 2), vaccination (N = 8), and vectors (N = 17). If we add AND (Driver[Title/Abstract]) in the algorithm, no occurrence is found.

2.2. Questionnaire Design

A questionnaire is used to determine the main drivers of the observed emergence or increasing incidence of TBE in humans. A driver was defined as a factor that has the potential to directly or indirectly precipitate (“drive”) or lead to the emergence or increasing incidence of TBE in humans. A former questionnaire created to rank drivers of the emergence of animal and zoonotic diseases [39,40,41] was adjusted to capture specific possible drivers for TBE. Overall, fifty-nine drivers were identified and classified in eight different domains (Table A1). The domains (D) were: (D1) disease/pathogen characteristics (N = 12 drivers); (D2) distance to Europe and the country of the expert (spatial-temporal scales) (N = 3 drivers); (D3) ability to monitor, treat, and control the disease (N = 11 drivers); (D4) European farm characteristics (N = 5 drivers); (D5) global change (N = 4 drivers); (D6) wildlife interface (N = 5 drivers); (D7) human activity (N = 8 drivers); and (D8) economic and trade activities (N = 11 drivers). These were formatted in an Excel^® (Microsoft, Redmond, WA, USA, 2016) file with one spreadsheet per domain, each domain harboring its respective drivers. Each driver had a score with its own definition, which could range from 0 to 4 (i.e., 5 modalities) or 1 to 4 (i.e., 4 modalities), and an intra-driver weight point. A spreadsheet was added, in which the 8 domains were listed with an inter-domain weight.

2.3. Expert Elicitation to Assess Drivers of Emergence or Increasing Incidence of TBE in Humans

An expert elicitation of knowledge was conducted, which consisted of gathering the opinions of people with recognized scientific expertise (indicated by at least one publication as first or co-author) and/or experience on TBE, TBEV, ticks, and/or tick-borne diseases (Table A2). Experts are from Europe with regional, national, or international activities, have different types of employment (government institution, university, research/scientific institution, diagnostic laboratory, or hospital), and cover different sectors relevant for TBE/TBEV (public health, animal health, environment, food safety, or laboratory diagnostic). In addition, specific attention was dedicated to capturing all disciplines related to the different domains of drivers. The number of years of professional expertise followed a normal distribution (Shapiro-Wilk test; p-value= 0.38), with an average of 19.9 years (standard error: 10.5).

For guidance purposes, an explanation letter accompanied the questionnaire (Appendix B). Each expert was contacted personally and responded in their own capacity (not on behalf of their institution). In order to capture the degree of variability in the experts’ knowledge, the data generated by elicitation were based on the scores provided by the experts. The elicitation was conducted over the course of two months (August–September 2022).

2.4. Scoring, Weighting System, and Level of Uncertainty

The elicited experts were asked to provide four types of information. First, they were asked to score the drivers (as established in Table A1). For each driver, the higher the score, the higher the driver’s chance to contribute to the emergence or increasing incidence of TBE in humans. Secondly, experts were requested to weigh each driver within a specific domain (intra-domain weight). This relative weight was determined using the Las Vegas technique [42]. Briefly, experts were given a number of points to be distributed among drivers according to their importance in the specific domain (proportional pilling). If all the drivers in a given domain had been considered equivalent by experts, each of them would have received the same score. Thirdly, the relative importance of each domain was subsequently weighted by experts (inter-domain weight). Finally, the level of uncertainty was asked at the domain level and not per driver (to reduce the length of the questionnaire to be completed by each expert), using a scale from 0 (minimal uncertainty in the scoring) to 100 (maximum uncertainty in the scoring).

2.5. Calculation of an Overall Weighted Score for Each Driver and the Ranking Process

To obtain the overall score per driver, an aggregation method that combined the two types of weighting (i.e., intra- and inter-domain) was used. First, the driver score (coefficients attributed by experts) was standardized by dividing it by the number of possibilities. Indeed, some drivers were allotted coefficients from 0 to 4 (5 possibilities) and others from 1 to 4 (4 possibilities). Afterwards, this standardized score was multiplied by the intra-domain weight and the inter-domain weight, as given by the expert. These results led to an overall weighted score for each driver and per expert:

$${\mathrm{OWS}}_{\mathrm{Dr}i}= {\mathrm{SDr}}_{\mathbf{i}} \times {\mathrm{WDr}}_{\mathbf{i}} \times {\mathrm{WDoj}}_{\mathbf{j}}$$

(1)

In this formula, OWS_Dri = overall weighted score for a specific driver_i; SDr_i = standardized score for a specific driver_i; WDr_i = intra-domain weight for a specific driver_i; WDo_j = inter-domain weight for a specific driver_i included in a specific domain_j. Furthermore, all drivers were ranked based on the median overall weighted score obtained for each driver, taking into account the answers of all the experts who answered the questionnaire. The statistical difference in the median, depending on the specific driver or the group of drivers considered, was assessed through a non-parametric Kruskal-Wallis equality-of-populations rank test and median regression analysis (Stata SE 14.2; StataCorp, College Station, TX, USA).

2.6. Cluster Analysis

A cluster analysis was carried out using a regression tree analysis (Salford Predictive Modeler^®, Version 8.2, Salford Systems, San Diego, CA, USA). Since the median overall weighted score (median OWS_Dri) is a continuous variable, the aim was to obtain groups of drivers with a minimal within-group variance and comparable likelihood to play a role in the emergence or increasing incidence of TBE in humans. In addition, the statistical difference between medians after grouping drivers in clusters was assessed using a non-parametric Kruskal-Wallis equality-of-populations rank test and a median regression analysis. Indeed, each driver is characterized by a median (based on all experts’ answers) and grouped. The test highlights potential significant differences between groups in terms of driver medians after clustering.

2.7. Sensitivity Analysis to Test the Robustness of the Expert Elicitation

In order to identify whether the ranking of drivers of the observed emergence or increasing incidence of TBE in humans was influenced by the choice of experts, three different sensitivity analyses were performed: comparison of 10 bootstraps each (random choice of 30 experts amongst 40) with the ranking of drivers obtained with all experts elicited as reference; comparison of French experts (N = 17) or other experts (N = 23) with the ranking of drivers obtained with all experts elicited as reference; and comparison of groups of experts coming from countries with a similar level of endemicity (increasing order level of endemicity from A to E). In order to have a sufficient number of experts per group, we aggregated the first two levels, group A-B, i.e., no autochthonous cases reported and sporadic cases or low endemicity, only in a few regions (N = 29), and the last three levels, group C-D-E, i.e., low to moderate, moderate to high, and low to high endemicity (N = 11) (for the details, see Table A3). The difference between the above ranking of drivers was tested using the Pearson coefficient of correlation test. If this coefficient was close to 1 and the p-value was less than 0.05, the correlation between the two rankings of drivers tested was considered significant.

3. Results

3.1. Response Rate and Field of Expertise Mobilized by the Experts

Forty professionals with scientific and/or field knowledge or experience regarding TBE and TBEV were contacted, and all agreed to participate. The fields and diversity of expertise are summarized in Table A2.

3.2. Estimating the Overall Weighted Score and Ranking of the Observed Emergence or Increasing Incidence of TBE in Humans

The medians of the weight between the eight domains of drivers as well as for the different 59 drivers were not equal according to the non-parametric Kruskal-Wallis test (Chi-squared test = 105.5 with 7 d.f. and α = 0.05, p-value = 0.0001; and Chi-squared test = 785.4 with 58 d.f. and α = 0.05, p-value = 0.0001, for the weights between domains and the weights of the different drivers, respectively) (Figure 1).

Figure 1. Boxplot of the relative importance of the eight domains of possible drivers of the observed emergence or increasing incidence of TBE in humans (N = 40 European experts). Legend: The bold line represents the median of the score distribution between the different experts attributed to each domain; the solid lines at the top and bottom of each rectangle represent, respectively, the first and third quartiles; adjacent lines to the whiskers represent the limits of the 95% confidence interval; small circles represent outside values. The eight domains of drivers are: D1, disease/pathogen characteristics; D2, distance to Europe and the country of the expert (spatial-temporal scales); D3, ability to monitor, treat, and control the disease; D4, European farm characteristics; D5, global change; D6, wildlife interface; D7, human activity; and D8, economic and trade activities.

The median of the weights of domains D2, D4, and D8 was significantly lower than the median of domain D1 as a reference (median regression; p-value = 0.019). The median of the weights of domains D5 and D7 was significantly higher than the median of domain D1 as a reference (median regression; p-value < 0.001).

Nine drivers out of 59 were ranked in decreasing order as having a very high importance (N = 1) or a high importance (N = 8) in the probability of playing a key role in the emergence or increasing incidence of the TBE in humans. The most important driver was related to changes in human behavior/activities leading to more contacts with the pathogen in high-risk areas (e.g., recreational activities, forest activities, and mushroom picking) (D7-1). The following eight drivers had a high importance in the probability of playing a key role: changes in eating habits or consumer demand leading to an increased exposure to the hazard through food (D7-2); changes of landscape, e.g., landscape fragmentation, creation of barriers, landfill sites (D5-4); the influence of humidity in the survival and transmission of the pathogen/disease (D5-2); the difficulty to control reservoir(s) and/or vector(s) (D3-3); the influence of temperature in the survival and transmission of the pathogen (D5-3); the number of wildlife compartments/groups (e.g., small mammals, birds, and ungulates) playing a role as reservoirs or amplifying hosts for the pathogen and potential spread (D1-6); the increase of autochthonous (indigenous) wild mammals in Europe and neighboring countries (D6-2); and the number of tick species vectors of the pathogen and their distribution within the country (D1-7) (Figure 2).

Figure 2. Ranking of the overall weighted score for each potential driver of the observed emergence or increasing incidence of TBE in humans (boxplot based on input from 40 European experts). Legend: The x-axis represents the drivers with the following codification: D1 to D8 refer to the eight domains of drivers, and D1_1 to D8_11 refer to a specific driver (for the codification, see Table A1). A relation to Figure 3 was provided by the group, which named “very high importance”, “high importance” and “less importance”, as possible drivers of the observed emergence or increasing incidence of TBE in humans.

3.3. Cluster Analysis

Three clusters were identified by regression tree analysis (Figure 3) that were significantly different (non-parametric Kruskal-Wallis equality-of-populations rank test; chi-squared test = 22.6 with two degrees of freedom (d.f.) and α = 0.05; p-value = 0.0001). These three clusters were classified as “very high importance” (N = 1 driver), “high importance” (N = 8 drivers), and “less importance” (N = 50 drivers).

Figure 3. Aggregation of drivers of the observed emergence or increasing incidence of TBE in humans into three homogenous groups using a regression tree analysis. Legend: N, number; SD, standard deviation. D1-01 to D8-11 refer to a specific driver (for the codification, see Table A1).

3.4. Sensitivity Analysis of the Impact of Experts on the Final Ranking of the Observed Emergence or Increasing Incidence of the TBE in Humans

The result of three different sensitivity analyses indicated that, irrespective of the experts excluded, excluding some experts only had very limited or no significant effects on the ranking compared to the reference (all experts elicited). Firstly, using 10 bootstraps of 30 experts amongst 40, the Pearson coefficient of correlation between each bootstrap against the ranking of 40 experts as a reference was very high (value between 0.97 and 0.99, with a p-value < 0.0001). Secondly, comparing the ranking of drivers either between French experts (N = 17) or other experts (N = 23) and all experts as a reference, the Pearson coefficient of correlation was of 0.92, with a p-value < 0.0001. Thirdly, comparing the ranking of drivers either between experts coming from countries with no autochthonous cases reported and sporadic cases or low endemicity (N = 29) or experts coming from countries with low to moderate, moderate to high, and low to high endemicity (N = 11), and all experts as a reference, the Pearson coefficient of correlation was respectively of 0.96 and 0.90, with a p-value < 0.0001.

3.5. Level of Uncertainty Per Domain of Drivers

The level of uncertainty per domain of drivers is depicted in Figure 4. The medians of the uncertainty between domains of drivers were not equal according to the non-parametric Kruskal-Wallis test (chi-squared test = 19.91 with 7 d.f. and α = 0.05, p-value = 0.018). The median of the uncertainty in the domain D2 (distance to Europe and the country of the expert) was significantly lower (around 10) than the median of other domains, which was between 20 and 30 (median regression; p-value < 0.02). In addition, there is no linear (Pearson coefficient correlation = −0.17; p-value = 0.69) or non-parametric (Spearman rank correlation = −0.39; p-value = 0.34) relationship between the median weight and the median uncertainty by domain.

Figure 4. Level of uncertainty per domain of drivers. Legend: The bold line represents the median of the level of uncertainty attributed by experts using a scale from 0 (minimal uncertainty in the scoring) to 100 (maximum uncertainty in the scoring); the solid lines at the top and bottom of each rectangle represent, respectively, the first and third quartiles; adjacent lines to the whiskers represent the limits of the 95% confidence interval; small circles represent outside values. The eight domains of drivers are: D1, disease/pathogen characteristics; D2, distance to Europe and the country of the expert (spatial-temporal scales); D3, ability to monitor, treat, and control the disease; D4, European farm characteristics; D5, global change; D6, wildlife interface; D7, human activity; and D8, economic and trade activities.

4. Discussion

Fifty-nine possible drivers of the emergence or increasing incidence of TBE in humans were ranked and aggregated into three homogenous groups according to the present expert elicitation of knowledge. Only the first nine most important ranked drivers will be further discussed, with a focus on the two categorized in the regression tree as “very high importance” nodes (N = 1) and “high importance” nodes (N = 8), respectively. Moreover, the sensitivity analysis showed limited and no significant effects of the experts involved, indicating an acceptable robustness of the elicitation for drivers included in the two first terminal nodes of the tree. In addition, the median level of uncertainty for all domains of drivers was moderate (between 20 and 30 on a scale from 0 to 100) except for the domain D2, which was lower (around 10).

The most important driver for the observed emergence or increasing incidence of TBE in humans recognized during this expert elicitation are changes in human behavior/activities leading to more contacts with the pathogen in high-risk areas (e.g., recreational activities, forest activities, and mushroom picking) (D7-1). This driver was first implicated in TBEV emergence by [36], who suggested that increased unemployment and marginal employment in forests due to the socio-economic transition were driving an increase in human TBEV exposure across eastern Europe. Other studies have linked changing human behavior to new autochthonous cases of TBE in infected areas [30,36,43]. Preventive actions are mostly related to tick bite avoidance (wearing protective clothing such as long sleeve tops and long trousers tucked into socks or boots and using tick repellents on the skin and/or clothing) [35]. More studies are needed to understand how the risk of tick bite exposure varies between different types of recreational and forest activities [44], to refine and tailor information and awareness campaigns for the general population or for particularly high-risk social groups (e.g., [45]).

The second important driver is the change in eating habits or consumer demands, leading to an increased exposure to hazards through food (D7-2). Although it is not well documented, there is a trend toward consuming minimally processed and locally produced foods, such as raw milk and certain raw milk cheeses. Slovaks consume traditional products made from raw sheep milk; however, sheep milk is often mixed with goat milk for higher yield, which increases the risk of TBEV milk infection. In Slovakia, the alimentary route of TBEV transmission is responsible for up to 20% of all reported human TBE cases [6]. Preferences for raw milk/milk products like cheese have changed over the last decade and vary between countries. In the past, consumers considered raw milk cheese to be less safe [46]. However, cultural determinants should also be considered, like pro-raw milk consumers in France who focus on the traditional and authentic character of the product [47] and in Switzerland who focus on positive health effects like allergy protection and support of the gut microbiome [48] or the international FACEnetwork (Farmhouse and Artisan Cheese and Dairy Producers European Network, www.face-network.eu (accessed on 15 March 2023)). There are also scientific data suggesting that raw milk consumption is associated with the prevention of atopy (i.e., an exaggerated IgE-mediated immune response) (e.g., see the PASTURE European Project [49]). Moreover, as explained by some stakeholders in a recent report [50], the consumption of raw milk products varies across cultures. Currently, raw milk is cheap compared to pasteurized milk and considered a standard product; the perception of healthy products is another aspect, and people increasingly look for these kinds of food products [50]. More studies are needed to estimate the consumption and safety (specifically for TBE) of raw milk and raw milk products by European consumers, as well as their contamination by the TBEV.

A third driver identified in this study is a landscape change, e.g., landscape fragmentation, the creation of barriers, and landfill sites (D5-4). The importance of landscape predictors of TBE was already identified within diverse frameworks like the link between land cover, land use, and land ownership and reported TBE incidence in rural parishes [43] or the hazard related to functional resources in the landscape determining where the reservoirs, vectors, and viruses are present, and exposure as the attractiveness and accessibility of people to hazardous areas and habitats [51]. Given the diverse and rapid land use change processes affecting the regions in which TBEV is circulating, it is critical to understand the joint socio-ecological mechanisms and the drivers of interactions between wildlife and domestic vertebrate reservoirs, ticks, and people that link hazards, exposure, and risk to landscape structure.

The fourth and sixth drivers, in order of importance, are the influence of humidity (D5-2) and temperature (D5-3), respectively, on the survival of the vector and consequently successful (or increased) transmission of TBEV. The main vectors of TBEV are hard ticks of the genus Ixodes (especially Ixodes ricinus across Europe and Ixodes persulcatus in north-eastern regions). The habitat suitability for the vector (suitable sites for breeding, questing, and resting) is influenced by the relative humidity and temperature, and these parameters contribute as predictor data for modeling the risk for tick-borne diseases like TBE [52,53]. Several laboratory experiments demonstrated the importance of humidity and temperature for tick survival (e.g., [54,55,56]). Both I. ricinus and I. persulcatus are vulnerable to desiccation. Therefore, these ticks require a relative humidity above 80% in their microhabitats to quest and survive [57,58]. The microclimatic temperature is another key driver of vector-borne disease transmission, as replication of arboviruses within the poikilothermic vectors is highly dependent on environmental temperature [59,60]. Indeed, the extrinsic incubation period and the blood meal digestion period are influenced by the temperature surrounding the vectors [60,61,62]. Additional microclimatic data are needed for a more accurate prediction of vector-borne diseases like TBE [61]. The impact of climate change on the risk of TBEV transmission is a question of interest. For example, a rise in temperature can increase the length of the tick questing season and, depending on the context, can also increase the numbers of susceptible ticks (larvae and nymphs) and the number of infected nymphs co-feeding on the same hosts, resulting in increased human infections [62]. Furthermore, a rise in temperature can increase the spread of ticks to areas with higher elevations (e.g., [63,64]) and may also drive changes in human behavior [65]. Theories explaining the consequences of climate change on tick-borne diseases in Europe, including TBE, are discussed in a recent review paper (and include effects on the intra-stage development of ticks, an extended transmission season, changes in the ecological balance of species abundance and interactions in the ecosystem, and climate-related range expansion of vectors, reservoir hosts, or human populations) [66].

However, more studies are needed to accurately predict the distribution of ticks and TBE. Tick-borne disease risk models need to be developed at a fine enough spatial resolution to capture the focal distribution of TBEV within the range of its widespread Ixodes vector and need to integrate climate and landscape drivers and the dynamics of wildlife and domestic hosts and people, aligning with a One Health perspective, and capitalize on advances in digital epidemiology and artificial intelligence (e.g., [67,68,69]) and citizen science approaches using, e.g., apps to collect data [70].

The fifth and seventh drivers were, respectively, the difficulty of controlling reservoir(s) and/or vector(s) (D3-3) and the number of wildlife compartments/groups (e.g., small mammals, birds, ungulates) playing a role as reservoirs or amplifying hosts for the pathogen and its potential spread (D1-6). Host vertebrates susceptible to TBEV (i.e., that can replicate TBEV with or without expression of clinical disease) with high viraemia (rodents (Apodemus, Myodes, Micrototus, Micromys, Pitimys, and Arvicola), insectivores like hedgehogs (Erinaceus europaeus) and moles (Talpa europea)) [71] or prolonged viraemia (some rodents like the red vole (Myodes rutilus) [72,73], and the bank vole (Myodes glareolus) [74,75]) play a key role in the transmission of TBEV to ticks [17] and co-feeding hosts [10]. Foxes (Vulpes vulpes L) can also be experimentally infected by infestation with TBE-infected ticks; these foxes became viraemic and suffered from encephalitis and paralysis [76]. Ticks (Ixodes ricinus, Ixodes canisuga, and Ixodes hexagonus) are able to infest red foxes [77], and TBEV was isolated from two adult I. ricinus and I. hexagonus ticks [78]. In addition, a first occurrence of clinical encephalitic manifestations due to TBEV in a roe deer (Capreolus capreolus L.) [79] and in a horse [80] was recently documented in Europe, suggesting that TBEV should be included in the differential diagnosis of roe deer and horses presenting neurologic disorders. It is critical to understand the relative role of these diverse hosts in transmission in different parts of the range of TBEV in order to predict further spread and impacts and design appropriate mitigation. Moreover, the control of reservoir hosts seems to be very difficult (e.g., [81]), and the control of ticks is currently impossible. In addition, seasonal bird migration has been identified as a risk factor for TBEV spread since birds are known to transport ticks and tick-borne pathogens such as TBEV [8,82].

The eighth most important driver identified is the increase of autochthonous (indigenous species) wild mammals in Europe and neighboring countries (D6-2). The wild mammal inventory was recently updated for some species and revealed an increase in abundance/range, driven by conservation efforts [83,84]. Such a trend should have some consequences for ticks and tick-borne diseases. As an example, the spatio-temporal increases in Western roe deer (Capreolus capreolus) in Europe should increase deer density and the density of bloodmeal sources for seeking ticks, which could potentially amplify tick populations [85,86]. A higher abundance of large mammals may lead to a greater tick population but also to other effects, such as dilution of the acarologic risk, predation on competent or non-competent hosts, or competition with specialized predators [87,88,89]. According to the country and scale of study, the abundance of ungulates or carnivores correlated either positively or negatively with the number of TBE cases (e.g., [87,89,90]). Moreover, few hard data on distribution, abundance, tick infestation rates, and TBEV infection rates are available for small rodent population dynamics, a species that is susceptible to TBEV with a high viremia [17]. More studies are needed to fill this gap or to investigate surrogate data (e.g., food supply during winter increases survival of small rodents [91] or breeding success of barn owls (Tyto alba) reflects the trends of small rodent populations [92]) to allow trend observation or trend analysis.

The ninth most important driver is the number of tick species that are competent vectors of the pathogen and their distribution in the country (D1-7). Although more than 14 species of ticks can be infected by TBEV [93], I. ricinus and I. persulcatus ticks are considered to be the main vectors of the European subtype and the Siberian and Far Eastern subtypes of TBEV [17,88,93]. However, the additional potential role of Dermacentor reticulatus was recently evidenced in experimental [94] and field [95] conditions. Dermacentor reticulatus is frequently found on pets, mainly on dogs (their movements without tick control can contribute to the spread of ticks) and is active in the winter or during the colder months [96,97,98]. Its range has also expanded in Europe [99,100,101,102]. Such a discovery deserves attention in order to better understand the eco-epidemiology of TBEV in relation to the trend of increasing human TBE cases, particularly since these ticks regularly feed on domestic ruminants, which may then transmit TBEV through their milk. The known European distribution of I. ricinus, I. persulcatus, and D. reticulatus ticks is available on a dedicated online platform [103,104].

Possible limitations exist with experts’ elicitation of knowledge (for a review, see [105]), like cognitive bias, overconfidence, or responses being conditioned by the recent high-profile findings (e.g., TBEV in unpasteurized milk) rather than taking a longer-term view. In this elicitation, the sensitivity analysis addresses these limitations to some degree. In addition, expert elicitation of knowledge is complementary and not in opposition to or a substitute for empirical One Health studies that address social and environmental drivers of risk in different geographical areas of emergence.

5. Conclusions

Scientific knowledge on possible drivers of the observed emergence or increasing incidence of TBE in humans is scant. In this context, expert elicitation of knowledge and multi-criteria decision analyses, in addition to clustering and sensitivity analyses, allowed the identification of nine drivers of either very high or high importance. These drivers should be the focus of future studies so we can increase and refine our understanding of the epidemiology of TBE in Europe and support decision-making to reduce TBEV exposure and its impacts.