Risk of rabies reintroduction into the European Union as a result of the Russo‐Ukrainian war: A quantitative disease risk analysis

The importation of rabid animals poses a continual threat to rabies freedom in the European Union (EU). Following the Russian invasion of Ukraine in 2022, the European Commission eased the rules on companion animal importations for Ukrainian refugees through derogations to the EU Pet Travel Scheme (EU PETS). As Ukraine remains endemic for canine rabies, this paper aimed to quantitatively assess whether the change in regulations affected the risk of rabies introduction to the EU.


| INTRODUC TI ON
Rabies is a major pathogen of concern to both human and animal health.The disease is caused by the classical rabies virus (RABV) which is transmitted to people following a bite, usually from a rabid dog.In almost all cases, fatality is certain once clinical signs occur (WHO, 2023).As such, numerous public health efforts have been made to eradicate the disease in Europe (WHO, 2018).Both parenteral vaccination of companion animals and the development of oral rabies vaccination for wildlife, have resulted in eradication from most of the European Union (EU), being restricted to the eastern member states with cases reported in Hungary, Lithuania, Poland and Romania since 2016 (Lojkić et al., 2021;Robardet et al., 2019;WHO, 2023).In order to maintain rabies freedom, strict measures are in place regarding the importation of companion animals via the EU Pet Travel Scheme (EU PETS), to minimize the continual threat of reintroduction from rabies-endemic countries (Vega et al., 2021).One of the main threats concerns Eastern European countries like Turkey, Russia and Ukraine, where canine rabies is endemic (Riccardi et al., 2021;WHO, 2023).In Ukraine, 1600 annual cases were reported in animals between 1996 and 2020 (Makovska et al., 2021).
On 24 February 2022, Ukraine was invaded by Russian troops, leading to the ongoing conflict in the country.To date, nearly 5 million refugees have fled to the EU from Ukraine (UNHCR, 2023) with many bringing their pets with them (Sandvik, 2022).In response to this crisis, the European Commission (EC), on 25 February 2022, declared a relaxation to the rules on importation under the Article 32 derogation within Regulation (EU) No. 576/2013 on the noncommercial movement of pet animals (Fortuna, 2022).The period of derogations continued until 30 June 2023.The usual EU PETS rules for a non-listed non-EU country, require that animals must first be identified, usually via a microchip, before receiving rabies vaccination.A 30-day period must then elapse before a serological test can be performed to identify whether the animal has mounted an adequate immune response (tests positive).Following a positive result, they must undergo a 3-month waiting period in the country of origin before being permitted entry into the EU (DEFRA, 2014) (see Figure 1).However, in exceptional circumstances where people may need to suddenly depart from a country, the regulation allows for entry into the EU without normal compliance with EU PETS as long as animals are isolated for a required necessary period and the owner has applied for a permit (EUR-Lex, 2013).
It is unlikely that such wide-scale use, necessary to accommodate the influx of refugees and their pets, was considered when the regulation was written.To the authors' knowledge, no previous quantitative risk assessments (QRAs) have been published to assess how the implementation of derogations to EU PETS could impact the risk of rabies entry into the EU.Within the literature, only a handful of opinion pieces were identified, which subjectively hypothesize an increase in risk as a result of the change (Silverwood, 2022a;WVA, 2023).
Therefore, the aim of this work was to answer the question 'Has the risk of rabies entry into the EU increased with the easing of importation requirements for the pets of Ukrainian refugees?'.In order to assess this, a QRA was performed with the aims of addressing: (1) whether there is an increase in risk with implementation of derogations to EU PETS and (2) how the level of risk is affected with a reduced compliance to these schemes.The risk is expressed as the annual probability of at least one rabies introduction into the EU from Ukraine and as the number of years between introductions.

| MATERIAL S AND ME THODS
Ethical approval was granted on 2 May 2023 by the University of Bristol's Animal Welfare and Ethical Review Body (AWERB).The study was given the unique reference number UIN/23/019.
A 3-month desk study was undertaken from 1 June 2023 to 5 September 2023.In constructing the stochastic models, information was obtained from scientific papers and from official government and international organization websites.

| Transmission pathways
Transmission scenarios were considered for both EU PETS (Figure 2) and derogations to this scheme (Figure 3).Both pathways outline all possible routes of entry of an infected companion animal (dog or cat) from Ukraine into the EU assuming clinical signs are not displayed prior to entry.
The EU PETS pathway was adapted from previous quantitative risk assessments (QRAs) for non-listed non-EU countries (Goddard et al., 2012;Jones et al., 2005;Ramnial et al., 2010).The derogation scenario assumes all EU countries implement immediate home quarantine upon entry which spans a period of 4 months.Within this time period, it was assumed that animals would receive vaccination upon entry and undergo the normal 30-day period before

Impacts
• The derogations to the European Union Pet Travel Scheme (EU PETS) pose a significantly lower risk of annual rabies introduction into the EU from Ukraine under the scenario of 100% compliance according to this model.
• Reduction in compliance within both EU PETS and derogation schemes had a large effect on the annual risk of rabies entry with a 74-fold increase in risk under the derogations scheme and a 10-fold increase under EU PETS.
• A 4-month period of home isolation in the derogation model eliminates contact with other potentially rabid animals during the waiting period and may explain the significantly lower risk compared to EU PETS.serological testing.Given an adequate immune response, the animal would then undergo a 3-month waiting period before official entry into the EU.If an animal had previously received a valid vaccination, that is, an approved vaccine 30 days prior to entry, they would not require vaccination upon entry but would undergo the other requirements (Figure 1).This was based upon the more detailed information provided by the Irish and Finnish governments (GOV.IE, 2022;Ruokavirasto, 2022).This decision was made due to the lack of clear guidance from other EU countries, where official veterinarians would mainly determine the implementation of the derogations on a case-by-case basis (IFAW, 2022).
For both transmission pathways, two compliance scenarios were considered: (a) 100% compliance and (b) reduced compliance to the schemes.To model reduced compliance, the values for the probability of receiving vaccination, serological testing and a border check were reduced from 100%, where p = 1, to lower values obtained from the literature and modelled by pert probability distributions.
These values are described in the following text.

| Probability pet is infected (P I )
The total recorded number of infected Ukrainian dogs and cats for 2022 was 366 animals (WHO, 2023).It was assumed that there was no underreporting of cases as the figure is in line with a decrease across previous years.The mean incubation period of rabies was given as 35 days (Crozet et al., 2023).Both values were used in the following equation to calculate the mean number of animals infected with rabies but not yet observed (Jones et al., 2005): To model uncertainty around this value, this equation was described by a gamma distribution (Jones et al., 2005): In order to then calculate the probability of a random animal incubating rabies (P I ), this value was divided by the total pet population, estimated as 6.25 million animals based on a canine population of 750,000 and a feline population of 5.5 million (PetFoodIndustry, 2014):

| Probability pet is vaccinated (P V )
With 100% compliance P V = 1.Where compliance was reduced, the values 0.56, 0. entering the UK via EU PETS obtained from Jones et al. (2005) as no specific data was available regarding Ukraine.Where an animal had been previously vaccinated within the derogation scenario, P V = 0 as it was assumed the first vaccination would be accepted and no further vaccinations would occur.

| Probability pet was previously vaccinated (P PV )
The uncertainty surrounding this value was described via a pert distribution with 0.09, 0.467 and 1 as the minimum, most likely and maximum values for the level of vaccination of animals in Ukraine (MEEREB, 2018;Nychyk et al., 2013;Picot et al., 2017).

| Probability pet is not protected (P NP )
Where infection occurs prior to vaccination or an animal remains unvaccinated, P NP = 1.Even following vaccination, failure can occur and an animal may remain unprotected.The vaccine efficiency was estimated from studies (Bahloul et al., 2006;Kallel et al., 2006;Minke et al., 2009;Sihvonen et al., 1995) based on the commercially available vaccines Rabisin, Nobivac and Madivak used in EU PETS as described by other QRAs (Goddard et al., 2012;Jones et al., 2005).As specific data was unavailable for Ukraine, it was assumed the same vaccines would be used to maintain compliance with the scheme.To estimate uncertainty, a beta distribution was used and uncertainty around vaccine protection (P v+ ) for each vaccine (Ra+, N+, M+) was given by the equation (Jones et al., 2005): where n v is the number of protected animals and N v is the number of vaccinated animals.
Vaccine protection from each vaccine (Ra+, N+, M+) was weighted equally to give a final estimate of protection, which was then subtracted from 1 to give the probability that a vaccinated animal is not protected: 2.2.5 | Probability pet is serologically tested (P T ) With 100% compliance P T = 1.Where there was reduced compliance, a pert distribution was used with the minimum, most likely and maximum values of 0.8, 0.98 and 0.998, respectively, based on information provided by Jones et al. (2005), as no information could be found relating to non-compliance of serological testing from Ukraine.
In pathways when there was non-compliance and vaccination did not occur, it was also assumed that the probability of being tested would equal 0.

| Probability of a false positive serological test (P TP )
Two serological tests, the fluorescent antibody virus neutralization test (FAVNT) and the rapid fluorescent focus inhibition test (RFFIT), are used to assess the immune response to rabies vaccination via EU PETS (Crozet et al., 2023).
To assess the number of false positives the specificity of the tests must be subtracted from one (1-Sp).Test specificity was determined from data provided by Cliquet et al. (1998) on the efficacy of the two tests.To estimate the level of uncertainty, a beta distribution was used for the following equation (Jones et al., 2005): where s indicates the number of true negatives and n indicates the number of unprotected animals.Both test specificities were then weighted equally and subtracted from 1 to give the probability of a false positive result.

| Probability pet becomes infected (P BI )
Becoming infected in the waiting period depends on the prevalence of rabies in the country where the waiting period is being undertaken.For this reason, P BI was not included in the derogation model as the level of rabies is negligible within the EU (GOV.UK, 2023;Ribadeau-Dumas et al., 2016).The annual pet cases (366) for 2022 (i) were used to determine the prevalence in Ukraine (WHO, 2023), and uncertainty was modelled via a gamma distribution: To calculate the probability that a given animal was clinically infected on a particular day (P I ), this value was then divided by the total number of companion animals in Ukraine (6.25 million) (PetFoodIndustry, 2014) multiplied by 365 in the following equation: The probability of becoming infected in the period between vaccination and entry into the EU was given by where T was 121 days for full compliance to EU PETS and where P BI = P I with non-compliance to model immediate entry into the EU.The above equations were based on those by Goddard et al. (2012).

| Probability pet does not show clinical signs (P NC )
Animals with clinical signs would be denied entry; however, some animals may incubate disease without displaying signs in the period T before entry and be allowed into the EU.Where animals were infected before vaccination, the period to entry equalled the time period from vaccination to official entry into the EU.This period was 121 days under EU PETS and under the derogation model was 151 days if the pet had been previously vaccinated or 121 days where they had not been previously vaccinated (Figure 1).Previously vaccinated animals would have undergone an additional 30-day period for the vaccination in Ukraine to have been considered valid prior to entry into home isolation.Therefore, in this scenario, the probability of an animal incubating rabies was the probability that the incubation period was greater than the waiting period (Goddard et al., 2012): where IP indicates the incubation period and T indicates the length of the period to entry.
Within EU PETS, it is also possible for animals to become infected during the waiting period.In this case, the time period is dependent on the date they are infected.Infection was equally likely on any day between day 1 and day t (last day of the waiting period), therefore the probabilities associated with each day of infection were summed and divided by the total waiting period to give the average probability for infection on any given day (Goddard et al., 2012) Where there was non-compliance t was equal to 1, indicating immediate entry.

| Probability checked (P C )
With full compliance to EU PETS, P C = 1.With reduced compliance, this value was placed at 0.9 as no specific information was available in the literature concerning this.The value was chosen

| Probability passes import check (P PC )
This measure is separate from disease status and based solely on compliance with the required documentation for entry.As data was unavailable for Ukraine or Eastern European countries, this value was estimated from information provided by Goddard et al. (2012).
The values for each country were summed to estimate average entry via EU PETS and a beta distribution was used to calculate the level of uncertainty (Jones et al., 2005): where n PC is the number of animals passing the checks and N C is the number of animals checked from each country.It was assumed that even though this was for the UK, the values would be similar across countries via EU PETS.

| Risk estimation
The nodes were multiplied across each pathway in the transmission pathways and they were summed to give a risk value (R) for each scenario (EU PETS and Derogation).The annual probability of importing at least one rabid animal into the EU (P R ) was then calculated from (Jones et al., 2005): where N is the total number of imported domestic animals.This value was calculated using data on the number of Ukrainian refugees who applied for asylum or a similar protection scheme within the EU countries since 2022 to give a worst-case scenario This was multiplied by the number of refugees to give 683,246 animals.
To identify the number of years between entries, the following calculation (Goddard et al., 2012) was used: Using open source software XL Risk for Microsoft Excel (Pyscripter., 2021), stochastic modelling was performed through 10,000 iterations via a Monte Carlo simulation and the median, 2.5th percentile and 97.5th percentile were calculated.If the 95% confidence interval (CI 95%) did not overlap then the results were considered significantly different.

| RE SULTS
The median annual probability of at least one rabid pet being imported into the EU was calculated as 4.25 × 10 −2 (CI 95% 1.44 × 10 −2 -9.81 × 10 −2 ) via EU PETS and 3.63 × 10 −3 (CI 95% 1.18 × 10 −3 -9.34 × 10 −3 ) via the derogation scheme when 100% compliance was modelled (Table 1).In this scenario, there was no overlap between the 95% confidence intervals of both schemes (EU PETS and derogation) therefore, the values can be considered significantly different.Additionally, there is approximately an 11-fold difference between both median values, with no crossover of the interquartile ranges in the boxplot (Figure 4) indicating that the distributions are distinct.The 97.5th percentile values indicate worst-case scenarios and the risk could be as high as 9.81 × 10 −2 with 100% compliance to EU PETS and 9.34 × 10 −3 in the derogation situation.The worst-case scenario in the derogation scheme (97.5th percentile, 9.34 × 10 −3 ) remains at an approximately 1.5 times lower risk than the best case scenario (2.5th percentile, 1.44 × 10 −2 ) at 100% compliance to EU PETS.
The number of years between rabies entries ranged from 10.19 to 69.33 with a median value of 23.54 for the EU PETS scenario and from 107.12 to 850.2 with a median of 275.28 years in the derogation scenario (Table 2).
When compliance to both schemes was reduced, the risk level increased, as expected.There was also a smaller difference between the risk distributions.Each of the median values sit outside the interquartile range of the other boxplot (Figure 5) indicating it is likely the distributions are different; however, as the 95% confidence intervals overlap, it is not considered significant.The median annual probability of importing a rabid animal was 4.46 × 10 −1 (CI 95% 1.60 × 10 −1 -7.31 × 10 −1 ) when reduction in compliance was modelled for EU PETS, a value approximately twice the risk level of the derogations at 2.22 × 10 −1 (CI 95% 3.86 × 10 −2 -5.47 × 10 −1 ).In a worst-case scenario, the risk increased to 7.31 × 10 −1 via EU PETS and 5.47 × 10 −1 with the derogations (Table 1).The number of years between entries varied from 1.38 to 6.27 years with a median of 2.24 for EU PETS and 1.83 to 25.93 with a median of 4.5 years for the derogation scheme (Table 2).
Risk via the derogation scheme was more sensitive to reductions in compliance.The median value with 100% compliance with the scheme (3.63 × 10 −3 ) was 74-fold lower than reduced compliance (2.22 × 10 −1 ).There was only approximately a 10-fold difference between both scenarios within EU PETS (Table 1).Overall, 100% compliance to the derogation scheme carried the lowest level of risk, whilst reduced compliance to EU PETS was associated with the greatest risk.

| DISCUSS ION
Stochastic modelling of two possible canine rabies transmission pathways based on EU PETS and on derogations to that scheme in response to the war in Ukraine found that the estimated implementation of the derogations at 100% compliance carried the lowest level of annual risk (3.63 × 10 −3 ), whilst reduced compliance with EU PETS carried the greatest risk level (4.46 × 10 −1 ).There are currently concerns that there would be an increased risk of rabies introduction via derogations to EU PETS for Ukrainian pets (Silverwood, 2022a;WVA, 2023).In contrast, the results of this study indicate that implementation of derogations, as estimated in this model, significantly reduced the annual risk of rabies introduction with an 11-fold decrease in risk following 100% compliance.Even when compliance was reduced, the risk level remained at approximately half that of entry via EU PETS.Model design is known to have an impact on risk outcomes and stochastic models are sensitive to assumptions made in their development (Berriman et al., 2018;Goddard et al., 2012;Høgåsen, 2005).
The assumption that animals would undergo an immediate home isolation upon entry into the EU destination country may have influenced risk in this study as a result.
Additionally, the elimination of the waiting period in the country of origin, in which animals may become infected due to contact with rabid animals, reduced the number of pathways of rabies entry in the derogations scheme, reducing the risk.
Compliance level had a large impact on disease risk, especially in the derogation model where a 74-fold increase in risk occurred with uncertain compliance.As expected, previous studies on rabies risk have also noted this effect where, for example, the risk was seen to increase by approximately two orders of magnitude in studies by Goddard et al. (2012) and Jones et al. (2005).Additionally, Napp et al. (2010) showed that by implementing stricter controls, there was a 270-fold decrease in disease risk.This highlights the  Sandvik, 2022;Silverwood, 2022b).In order to better understand the risk via this scheme, future research should focus on surveying official veterinarians on the guidance issued as well as questioning Ukrainian refugees on the information they received and their levels of compliance with the guidance.
In this model, the risk associated with EU PETS was higher than that of other QRAs.At 100% compliance to this scheme, 4.25 × 10 −2 annual importations would occur.In comparison to Goddard et al. (2012), where the risk level was 4.79 × 10 −3 from movements into the UK from all countries via EU PETS, this result was almost an order of magnitude greater.This may be explained by a higher prevalence of rabies in Ukraine as it is the only importing country considered in this paper.Additionally, an increase in importations may have increased the risk.Høgåsen (2005) found that as imports were increased with 100, 1000 and 10,000 animals, the risk level also increased proportionally.As only data from 2022 was considered in this study, where there is an influx of Ukrainian refugees, it would be useful to assess the risk levels  option for rabies management in similar crisis situations in the future.
The 2022 European Union One Health Zoonoses report indicated that there was only a single imported rabies case in the EU that year, which came from Morocco (EFSA and ECDC, 2023).It is important that full compliance is maintained to minimize the risks of rabies entry and this could be achieved through effective public education, which has been shown to be a key preventative measure in reducing disease risk (Riccardi et al., 2021;Vega et al., 2021).Managing risk while providing a compassionate solution for refugees is crucial and efforts should be made to provide a compromise which balances the biosecurity of receiving countries with humanitarian aid.Therefore, further research should focus on finding solutions and developing regulations to effectively deal with this emerging issue, especially in light of the growing number of climate refugees across the globe (Vince, 2022) as well as other circumstances such as international conflict.

| CON CLUS ION
To conclude, the suggested model for derogations to EU PETS, which includes a 4-month quarantine rather than a waiting period in Ukraine, has a significantly reduced annual risk of rabies introduction following 100% compliance than EU PETS.Even in the scenario of reduced compliance this risk remained reduced under the derogation scheme, although this was not statistically significant.
Therefore, the suggested scheme could be a viable solution to accommodating the pets of refugees whilst managing risk to diseasefree areas in times of crisis.

ACK N OWLED G EM ENTS
The authors would like to thank Dr Irene Bueno Padilla for her guidance and input in the initial conception and design of the project.
We are also grateful to the University of Bristol for allowing the undertaking of this study, and in particular Dr Nicola Rooney, Dr Andrew Carr and Dr Laszlo Talas for discussions and support during study planning.Additionally, the authors would like to thank Sophia McDonald for her peer review of the paper, and Dawn Roberts for the discussion which provided inspiration for the project.

CO N FLI C T O F I NTER E S T S TATEM ENT
We have no conflicts of interest to disclose.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing is not applicable to this article as no new data were created or analysed in this study.

Tirion
89 and 1 to describe the minimum, most likely and maximum values of a pert distribution were used to model uncertainty.The values were based on the compliance level of animals Comparison of the requirements of the European Union Pet Travel Scheme (EU PETS) and the estimated implementation of derogations under Regulation (EU) No. 576/2013 on the non-commercial movement of pet animals to EU PETS.The border with the European Union (EU) is indicated by the bold dashed blue line.The 3-month waiting period is outlined with blue dotted lines.Entry indicates official entry into the EU.Under the derogation scenario, although entry into the country has occurred immediately, official entry does not occur until after the 3-month waiting period.A 4-month home isolation period is outlined by a light blue box in the derogation scenario.
, calculated as A pert distribution used to estimate uncertainty around the period, with the following values 1, 35 and 365 for the minimum, most likely and maximum values, respectively, (Abdulmajid & Hassan, 2021), and 1000 samples were taken from this distribution via a Monte Carlo simulation in order to calculate the above probabilities.
Transmission pathway indicating all possible routes of disease entry into the EU through the European Union Pet Travel Scheme (EU PETS) for Ukraine (a non-listed non-EU country).The dotted line on the right indicates the border of the EU.WP denotes the waiting period.Figure adapted from Jones et al. (2005).as it has been used in previous QRA's(Goddard et al., 2012;Kwan et al., 2017) and does not vary significantly from the values for other markers of reduced compliance (vaccination and serological testing) in this model.The probability of being checked was not included in the derogation model as all animals are allowed immediate entry into the EU.

E 3
Transmission pathway indicating all possible routes of disease entry into the EU through an estimated pathway of implementation of the derogations to the European Union Pet Travel Scheme (EU PETS).The dotted lines outline a period of home isolation with the left line denoting the time of entry into the EU country and the right line denoting official entry into the EU country following home isolation.WP denotes the waiting period.

from
Ukraine to the EU in previous years to assess the effect this has on the risk level.Additionally, entry of unvaccinated companion animals from Ukraine could potentially influence the transmission dynamics of rabies within eastern, rabies-endemic areas of the EU and the risk of rabies reintroduction may have a geographical gradient across the EU based on the distance to the Ukrainian border.Both of these considerations were beyond the scope of this current study but should be explored in further studies.The findings of this paper indicate that compliance to a scheme such as suggested by the model for derogations, may be a viable F I G U R E 4 Box plot comparing the annual risk of rabies introduction between the European Union Pet Travel Scheme (EU PETS) and the scenario of derogations to this scheme assuming 100% compliance to the rules on vaccination, serological testing and border checks.Boxes indicate interquartile range and horizontal lines represent the median value.TA B L E 2 Estimated years between rabies entries into the European Union (EU) via the EU Pet Travel Scheme (EU PETS) and the derogation scheme under the scenarios of 100% and reduced compliance.
Rebecca Cobby https://orcid.org/0009-0008-6669-1373F I G U R E 5 Box plot comparing the annual risk of rabies introduction between the European Union Pet Travel Scheme (EU PETS) and the scenario of derogations to this scheme with reduced compliance to the rules on vaccination, serological testing and border checks.Boxes indicate interquartile range and horizontal lines represent the median value.