A quantitative risk assessment on the change in likelihood of rabies introduction into the United Kingdom as a consequence of adopting the existing harmonised Community rules for the non-commercial movement of pet animals

FINAL REPORT

31st August 2010

Ashley Goddard1, Neil Donaldson1, Rowena Kosmider1, Louise Kelly1,3, Amie Adkin1, Daniel Horton2, Tony Fooks2, Andrew Breed1, Conrad Freuling4, Thomas Müller4, Susan Shaw5, Gunilla Hallgren6 & Emma Snary1.

1

Centre for Epidemiology & Risk Analysis, Veterinary Laboratories Agency, New Haw, Addlestone, Surrey, KT15 3NB 2 Rabies and Wildlife Zoonoses Group, Veterinary Laboratories Agency, New Haw, Addlestone, Surrey, KT15 3NB 3 Department of Mathematics and Statistics, University of Strathclyde, Glasgow G1 1XH 4 FLI - Friedrich-Loeffler-Institut, Bundesforschungsinstitut für Tiergesundheit, Seestrasse 55, 16868 Wusterhausen, Germany 5 Companion Animals, 8 Montrose Villas, Cliff Street, Cheddar, Somerset, UK. BS27 3PR 6 National Veterinary Institute, SVA, SE-751 89 Uppsala, Sweden

Executive summary Rabies is a zoonotic viral disease that causes acute encephalitis in humans and other warm-blooded animals; it is the only disease with almost 100% mortality once clinical signs are present. Transmission is usually through saliva via the bite of an infected animal, with dogs being the main transmitter of rabies to humans. The World Health Organization has estimated the annual number of human rabies deaths to be 55,000 with 95% of the deaths occurring in Asia and Africa. Although animal rabies has declined in Europe due to extensive control programs, pockets of infection continue with sporadic increases in some EU Member States (MSs). The United Kingdom (UK) is officially classified (under OIE standards) as being free from terrestrial rabies. This was historically achieved by the investigation of every case of disease and strict controls on dog movements and, later, on cat movements. Since 1897 dogs entering the UK have been subject to 6 months quarantine. Six month quarantine has also been in place for cats since 1928. The last human death from indigenous classical rabies occurred in 1902, and the last case of indigenous terrestrial animal rabies was in 1922. In 2008, a ten-week-old puppy infected with rabies was brought into the UK from Sri Lanka; however, the rabies-free status of the UK was unaffected because the infected animal did not leave the quarantine kennel. In February 2000, amendments were made to the UK quarantine laws for companion animals and the Pet Travel Scheme (PETS) was subsequently introduced for companion animals (dogs and cats) entering the UK from qualifying listed countries; ferrets were added in 2004. The scheme enables companion animals from qualifying listed countries (EU Member States (MSs) and Equivalents 1 and listed third Countries 2 ) to enter and re-enter the UK without 6 months quarantine (movements from unlisted third countries must still enter via 6 month quarantine). In 2004, the EU also introduced a companion animal movement and importation policy to protect against rabies introduction via cats and dogs and ferrets, hereafter referred to in this report as the EU Pet Movement Policy (EUPMP). However, five EU MSs, the UK, Sweden, Finland, Ireland and Malta currently have derogations from the EU policy of companion animal movement controls. This allows the UK companion animal movement policy to differ to the EUPMP. This derogation however, is due to expire in December 2011 and it is therefore prudent to assess the impact of adopting the EUPMP on the risk of rabies entering the UK due to the movement of companion animals. Therefore, in order to assist Defra (the UK Department for Environment Food and Rural Affairs) with their development of Government policy on companion 1

Andorra, Austria, Belgium, Bulgaria, Croatia, Republic of Cyprus, Czech Republic, Denmark (inc Greenland and Faroe Islands), Estonia, Finland, France (inc French Guiana, Guadeloupe, Martinique and Reunion) Germany, Gibraltar, Greece, Hungary, Ireland, Iceland, Italy, Latvia, Liechtenstein, Lithuania, Luxembourg, Malta, Monaco, Netherlands, Norway, Poland, Portugal (inc Azores and Madeira), Romania, San Marino, Slovakia, Slovenia, Spain (inc Balearic Islands, Canary Islands, Ceuta and Melilla), Sweden, Switzerland, Vatican 2 Antigua & Barbuda, Argentina, Aruba, Ascension Island, Australia, Bahrain, Barbados, Belarus, Bermuda, Bosnia-Herzegovina, British Virgin Islands, Canada, Cayman Islands, Chile, Falkland Islands, Fiji, French Polynesia, Guam, Hawaii, Hong Kong, Jamaica, Japan, Malaysia, Mauritius, Mayotte, Mexico, Montserrat, Netherlands Antilles, Singapore, St Lucia, New Caledonia, New Zealand, Russian Federation, St Helena, St Kitts & Nevis, St Pierre & Miquelon, St Vincent & The Grenadines, Taiwan, Trinidad & Tobago, United Arab Emirates, USA (mainland), Vanuatu, Wallis & Futuna

i

animal travel, a quantitative risk assessment (QRA) has been undertaken. The specified risk questions were given as: • •

How would the risk of rabies introduction to the UK via travelling pets from EU Member States and listed third countries change were the UK to apply the current harmonised EU rules for the non-commercial movement of pets? How would risk of rabies introduction from all countries (EU Member States, listed third and unlisted third countries) change: o where rules are followed with 100% compliance? o where rules are followed with varying degrees of less than 100% compliance?

A stochastic QRA was developed using the software package @Risk (© Palisade) Version 4.5, an add-on package within Microsoft Excel (© Microsoft). Only uncertainty is included in the model. The model estimates the probability of one or more infected dogs/cats entering the UK and the number of years between rabies entry for both the current UK movement policy (PETS/Quarantine) and EUPMP. This model only considers companion animals entering the UK from abroad, returning UK animals were not included in the assessment due to data limitations. It is thought that in the UK the level of compliance with PETS is very high, therefore the vast majority of domestic animals travelling abroad will have been successfully vaccinated and serologically tested and therefore would be protected against rabies infection. Furthermore, it is anticipated that the majority of travelling animals will be abroad for only a relatively short period of time, e.g. a two-week holiday, and are therefore unlikely to come into contact with rabies-infected animals in the destination country, particularly as the majority of listed countries have a very low incidence of rabies. Taking these factors into consideration it is anticipated that the risk of a travelling UK animal becoming infected with rabies and subsequently re-entering the UK will be lower than the risks posed by animals entering the UK from overseas. The model was parameterised using data from the published and unpublished literature and databases from Animal Health. Within the time scale of this risk assessment it was not possible to assess the risk of rabies entry for each individual country in the world. Therefore, in order to account for the significant variability in rabies incidence/prevalence between countries and also country categorisation (i.e. EU MS, listed third countries and unlisted third countries) the world was split into 9 country groupings. The results of the QRA, assuming 100% compliance with all regulations, suggest that under EUPMP the annual risk of rabies introduction from non-UK cats/dogs would increase from 7.79 x 10-5 (5.90 x 10-5, 1.06 x 10-4) to 4.79 x 10-3 (4.05 x 10-3, 5.65 x 10-3). This is a 60-fold increase in the mean risk. In terms of the number of years between rabies entry, it was estimated that under the current pet movement policy (with the current number of imports and worldwide rabies situation) there will be one rabies introduction every 13272 (9408, 16940) years but under EUPMP this would increase to one rabies introduction every 211 (177, 247) years. An alternative representation of these results is that under PETS and quarantine there will be one rabies entry every 617,028,552 companion animal entries, while under the EUPMP one rabies entry is expected every 9,809,601 animals. An overall increase in risk has

ii

been predicted, however, within the different country classifications (EU MSs, listed/unlisted third countries) different trends were observed. Under the current scheme (PETS and quarantine) the highest mean risk is presented from animal entries from unlisted third countries, while under the EUPMP the highest mean risk is from listed third countries (see tables below). It is predicted that moving to the EUPMP will result in an approximately two-fold decrease in the mean risk of rabies entry to the UK from unlisted third countries, which is largely due to the use of a serological test with a high specificity. For EU MSs and listed third countries 250fold and 100-fold increases in risk are predicted, respectively. As expected, between the different country groupings (groups 1-3) there is a higher level of risk associated with companion animal entries from countries with higher numbers of reported rabies cases (groups 2 and 3). It should be noted that there is a large amount of uncertainty associated with the EUPMP scheme for unlisted third countries compared to quarantine, consequently the conclusions made about decreased risk from unlisted countries under the EUPMP are less certain than the overall conclusions or those relating to EU MSs and listed third countries. Annual probability of at least one infected cat/dog entering the UK, assuming 100% compliance. Country Classification Entry scheme Listed third Unlisted third EU MSs countries countries PETS/quarantine EUPMP

8.34x10-6 (3.43x10-6, 1.60x10-5) 2.02x10-3 (1.41x10-3, 2.79x10-3)

2.64x10-5 (1.32x10-5, 4.69x10-5) 2.75x10-3 (2.39x10-3, 3.15x10-3)

4.31x10-5 (3.91x10-5, 4.82x10-5) 2.58x10-5 (9.47x10-6, 5.05x10-5)

Predicted number of years between rabies entry to the UK, assuming 100% compliance Country Classification Entry scheme Listed third Unlisted third EU MSs countries countries PETS/quarantine EUPMP

149129 (62683, 291248) 517 (359, 708)

43942 (21299, 75973) 366 (317, 419)

iii

23302 (20738, 25557) 50440 (19792, 105590)

Box plots comparing the risk of rabies entry from each country category under PETS/quarantine and the EUPMP. Note: y-axes are the log10 of the annual probability of rabies entry to the UK

Similar to other risk assessment models for the introduction of rabies, the results of the model are highly sensitive to the compliance of the scheme in question. Under the current scheme at 90% compliance the number of years between rabies entry would decrease to 761 (632, 894), while for the EUPMP one rabies entry would be expected every 170 (146, 195) years. In particular the non-compliance scenarios are sensitive to the fact that if a non-compliant animal is brought into the UK there is virtually no waiting time (1-day is assumed), whereas under PETS a 6 month period of waiting is required for EU and listed third countries and under EUPMP the waiting period is 21 days for EU and listed third countries and 3 (+1) months for unlisted third countries. Importantly, it was concluded that the risk of introduction via EUPMP is less affected by non-compliance compared to the current scheme, as the removal of the serological test (EU MSs and listed third countries) and border checking requirements (EU MSs) means there are less potential areas of non-compliance and therefore less potential for increase in risk under non-compliant scenarios. During the development of the QRA many data gaps and deficiencies were identified. Therefore a full sensitivity and uncertainty analysis was undertaken to assess the influence that these data gaps and deficiencies have on the model. From these analyses it was concluded that many of the data gaps and deficiencies are influential to the QRA and that further research/surveillance activities would be advantageous for an improvement to the estimation of risk; these were: • Data to formulate an improved estimate of prevalence: o Improved worldwide reporting of rabies cases o Estimation of under-reporting in countries worldwide

iv

o More comprehensive estimation of the cat and dog (companion animal) population. • Current level of compliance with PETS • Incubation period for natural cases of rabies Other limitations that have been identified with the model are the high level of uncertainty that is associated with the future level of compliance with the EUPMP in the UK and across Europe, particularly with reduced levels of checking in the EU, and uncertainty over the potential change in the number and frequency of companion animal movements within the EU and into the UK following adoption of the EUPMP. Further analyses identified that the serological test in PETS is an important barrier to the introduction of rabies into the UK. Assuming 100% compliance, if the serological test requirement was removed from PETS, it was predicted that the risk of rabies entry to the UK would increase approximately 10-fold, from 7.79 x 10-5 (5.90 x 10-5, 1.06 x 10-4) to 9.17 x 10-4 (5.94 x 10-4, 1.28 x 10-3), with the number of years between entry decreasing from 13272 (9408, 16940) to 1152 (780, 1682) years. Similarly, the QRA results suggest that a reduction in the waiting period to 90 days does not greatly increase the risk of rabies entry, but if the waiting period is any shorter the risk increases much more substantially. Therefore, in summary, it is concluded that adoption of EUPMP would cause a 60fold increase in the mean risk of rabies entering the UK via the movement of non-UK cats/dogs with, in particular, the risk increasing for movements from EU MSs and listed third Countries. However, in absolute terms the risk of rabies entry under both schemes is low. The mean risk from unlisted third countries is predicted to reduce if the EUPMP were to be introduced however there is significant uncertainty associated with this result. Finally, it was shown that the risk of rabies entering the UK is less affected by non-compliance under the EUPMP scheme, compared to PETS which is highly sensitive to non-compliance.

v

Contents Executive summary ......................................................................................................... i Contents ........................................................................................................................ vi 1. Introduction ............................................................................................................ 1 2. PETS, quarantine and EUPMP .............................................................................. 3 2.1. The current UK system: PETS and quarantine ................................................... 3 2.2. European Union Pet Movement Policy............................................................... 4 2.3. Comparison of the schemes ................................................................................ 5 3. Scope of work ........................................................................................................ 6 3.1. Species included in the QRA .............................................................................. 6 3.2. Country groupings .............................................................................................. 6 3.2.1 Grouping methodology used in previous risk assessments........................... 6 3.2.2. Rabies cases ................................................................................................. 7 3.2.3. Methodology for country groupings ............................................................ 7 3.2.4 Country groups.............................................................................................. 8 4. Methodology ........................................................................................................ 11 4.1. Risk pathway ..................................................................................................... 11 4.2. Model implementation ...................................................................................... 13 4.3. Parameter estimation ......................................................................................... 14 4.3.1. Incubation period (IP) ................................................................................ 14 4.3.2. Probability companion animal from country group j is infected (PI,j) ....... 16 4.3.3. Probability that an unprotected animal passes the serological test (PST+) . 17 4.3.4. Probability that a vaccinated pet is not protected (PNP) ............................. 19 4.3.5. Probability than an unprotected animal becomes infected during waiting period (PI*,j).......................................................................................................... 21 4.3.6. Probability that an infected animal does not display clinical signs during the waiting period (PNCS) ..................................................................................... 21 4.3.7. Probability an animal passes import checks from a country group j (PC+,j) .............................................................................................................................. 22 4.3.8. Compliance parameters .............................................................................. 23 4.4. Number of animals imported per year (NI,j) ..................................................... 24 4.5. Scenario and uncertainty analysis ..................................................................... 25 4.5.1. Scenario analysis ........................................................................................ 25 4.5.2. Uncertainty analysis ................................................................................... 27 5. Results, Uncertainty Analysis & Scenario Analysis ............................................ 29 5.1. Baseline results (100% compliance) ................................................................. 29 5.2. Investigation of compliance .............................................................................. 35 5.3. Other scenario analyses..................................................................................... 39 5.3.1. Checking entries from EU MSs under the EUPMP ................................... 39 5.3.2. Increase in the number of imports ............................................................. 40 5.3.3. Potential poor vaccine efficacy in unlisted third countries ........................ 41 5.3.4. Change in incubation period ...................................................................... 42 5.3.5. Prevalence Scenarios ................................................................................. 43 5.3.6. Protection after vaccination ....................................................................... 46 5.3.7. Serological test specificity ......................................................................... 47 5.3.8. Modifications to PETS ............................................................................... 48 5.4. Uncertainty analysis .......................................................................................... 49 6. Discussion ............................................................................................................ 51 7. Conclusion ........................................................................................................... 59 vi

Acknowledgements ...................................................................................................... 60 Glossary ....................................................................................................................... 61 References .................................................................................................................... 63 APPENDIX 1: Peer review of existing rabies VLA risk assessments and Defra qualitative risk assessment ........................................................................................... 67 APPENDIX 2: Response to “Points to consider during review of previous RA for risk of rabies and for use in the new model.” ..................................................................... 86 APPENDIX 3: Exclusion of ferrets from the QRA ..................................................... 88 APPENDIX 4: Datasets ............................................................................................... 94 4.1. Incubation period datasets................................................................................. 94 4.2. Rabies incidence and cat/dog population datasets ............................................ 96 4.3. Serological test datasets .................................................................................. 144 4.4. Vaccine datasets .............................................................................................. 144 4.5. Cat/dog import datasets................................................................................... 144 APPENDIX 5: Classification of lyssaviruses ............................................................ 151 APPENDIX 6: A review of the risk frameworks used in previous risk assessments on the introduction of rabies into a specific region via the movement of cats and dogs 152

vii

1. Introduction Rabies is a zoonotic viral disease that causes acute encephalitis in humans and other warm-blooded animals. The clinical disease is untreatable and is almost always fatal; it is the only disease with almost 100% mortality once clinical signs are present. Transmission occurs usually through saliva via the bite of an infected animal, with dogs being the main transmitter of rabies to humans. The World Health Organization has estimated the annual number of human rabies deaths to be 55,000 with 95% of the deaths occurring in Asia and Africa (WHO, 2008). Although rabies has declined in Europe due to extensive control programs, pockets of infection continue with sporadic increases in some EU Member States (MSs). The United Kingdom (UK) is officially classified (under OIE standards (OIE, 2008)) as free from terrestrial rabies (Fooks et al., 2004). This was historically achieved by the investigation of every incidence of disease and strict controls on dog movements and, later, on cat movements. Since 1897 dogs entering the UK have been subject to 6 months quarantine. Six month quarantine has also been in place for cats since 1928. The last human death from indigenous classical rabies occurred in 1902, and the last case of indigenous terrestrial animal rabies was in 1922. In 2008, a ten-week-old puppy was imported to the UK as part of a single consignment of thirteen animals rescued by an animal charity dedicated to re-homing stray dogs in Sri Lanka (Fooks et al., 2008). It was housed in a quarantine facility in North London but died after a short illness of diarrhoea and convulsions. Rabies was laboratory-confirmed shortly afterwards. The rabies-free status of the UK was unaffected because the infected animal did not leave the quarantine kennel. In February 2000, amendments were made to the UK quarantine laws for companion animals and the Pet Travel Scheme (PETS) was subsequently introduced for dogs and cats entering the UK from qualifying listed countries; ferrets were added in 2004. The scheme enables companion animals from qualifying listed countries to enter and reenter the UK without 6 months quarantine. In 2004, the EU also introduced a pet movement and importation policy to protect against rabies introduction via cats, dogs, and ferrets, referred to in this report as the EU Pet Movement Policy (EUPMP). However, five MSs, the UK, Sweden, Finland, Ireland and Malta currently have derogations from the EU policy of pet movement controls. This means the UK pet movement policy differs from the EUPMP. However, this derogation is currently due to expire in December 2011 and it is therefore prudent to assess the impact of adopting the EUPMP on the risk of rabies entering the UK due to the international movement of pets. This assessment will form part of the existing evidence-base collected in order to inform Government policy. Therefore, in order to assist Defra (the UK Department for Environment Food and Rural Affairs) with their development of Government policy on pet travel, a quantitative risk assessment (QRA) has been undertaken. The specified risk questions were given as (Defra, 2010c): •

•

How would the risk of rabies introduction to the UK via travelling pets from EU Member States and listed third countries change were the UK to apply the current harmonised EU rules for the non-commercial movement of pets? How would risk of rabies introduction from all countries (EU Member States, listed third and unlisted third countries) change: 1

o where rules are followed with 100% compliance? o where rules are followed with varying degrees of less than 100% compliance? Risk assessments have previously been developed by numerous countries, including the UK, to assess the risk of rabies introduction. The majority of these are quantitative in nature and closely model the different stages of the importation policies being analysed. For the UK, most of the quantitative risk assessments have been carried out by the Veterinary Laboratories Agency (VLA) (Jones et al., 2005a; Jones, 2002a; Jones, 2002b; Jones, 2003; Ramnial, 2009). A review of the VLA rabies risk assessments is included in Appendix 1. Also included in Appendix 1 is a review of a qualitative risk assessment carried out by Defra (Defra, 2009). Reviewer’s comments (supplied by Defra) on the previous VLA risk assessments have been addressed as far as possible in relation to the time and data available; see Appendix 2. This report presents the methodologies and results of the quantitative risk assessment conducted on behalf of Defra between May and August 2010. Section 2 provides a brief summary of the different pet movement schemes considered (i.e. PETS, quarantine and the EUPMP). The scope of the work is detailed in Section 3, which includes a description of the approach to group countries according to the reported number of rabies cases and also the pet species to be included in the risk assessment. The full details of the QRA are provided in Section 4, including the data used and methodologies for parameterising the variables within the QRA. Section 5 presents the results of the QRA, including the further analyses of the model (uncertainty analysis and sensitivity analysis). A discussion of the model and results is given in Section 6 and conclusions in Section 7. The Appendices provide additional information/data, including a description of all the viruses within the lyssavirus genus.

2

2. PETS, quarantine and EUPMP 2.1. The current UK system: PETS and quarantine Since February 2000, dogs and cats from qualifying listed countries have been able to enter the UK via PETS, without the requirement of spending 6 months in quarantine. Table 1 lists the countries that are eligible for PETS. Animals from the unlisted third countries are not part of PETS and must spend 6 months in UK quarantine, where they are vaccinated against rabies. For cats and dogs to qualify for PETS, each animal must meet the following sequential criteria (Defra, 2010b): • • • • •

Fitted with a microchip Vaccinated against rabies with an approved vaccine Blood tested (after a period of time recommended by the vaccine manufacturer) Issued with an official PETS passport (EU MSs) or 3rd country certificate (Listed countries) Treated against a tapeworm and ticks 24-48 hours prior to entry

In 2004, when EU regulation 998/2003 took effect, ferrets were added to PETS. Under PETS ferrets can enter the UK via the process of microchipping and vaccination, with either a 21 day waiting period for entries from EU countries or a 6 month period for listed third countries (Defra, 2010a). Under the current scheme there is no requirement for ferrets to be serologically tested after vaccination to check antibody levels. Figure 1 (amended from Defra, 2010c), diagrammatically describes the processes for EU MSs (& equivalents) and listed third countries. Table 1: Countries classified as EU MSs & equivalents and as listed EU countries, which can travel under PETS (EU regulation 998/2003). Countries not listed are classified as ‘non-listed third countries’ and pets must spent 6 months in isolation in quarantine. Category Country EU MSs & equivalents Andorra, Austria, Belgium, Bulgaria, Croatia, Republic of Cyprus, Czech Republic, Denmark (inc Greenland and Faroe Islands), Estonia, Finland, France (inc French Guiana, Guadeloupe, Martinique and Reunion) Germany, Gibraltar, Greece, Hungary, Ireland, Iceland, Italy, Latvia, Liechtenstein, Lithuania, Luxembourg, Malta, Monaco, Netherlands, Norway, Poland, Portugal (inc Azores and Madeira), Romania, San Marino, Slovakia, Slovenia, Spain (inc Balearic Islands, Canary Islands, Ceuta and Melilla), Sweden, Switzerland, Vatican Antigua & Barbuda, Argentina, Aruba, Ascension Island, Australia, Listed third countries Bahrain, Barbados, Belarus, Bermuda, Bosnia-Herzegovina, British Virgin Islands, Canada, Cayman Islands, Chile, Falkland Islands, Fiji, French Polynesia, Guam, Hawaii, Hong Kong, Jamaica, Japan, Malaysia, Mauritius, Mayotte, Mexico, Montserrat, Netherlands Antilles, Singapore, St Lucia, New Caledonia, New Zealand, Russian Federation, St Helena, St Kitts & Nevis, St Pierre & Miquelon, St Vincent & The Grenadines, Taiwan, Trinidad & Tobago, United Arab Emirates, USA (mainland), Vanuatu, Wallis & Futuna All other countries. Unlisted third countries

3

(a)

(b)

Figure 1: Current UK system for the movement of cats and dogs from (a) EU MSs and 8 ‘equivalents’ and (b) listed third countries. Pets from unlisted third countries must spend 6 months in quarantine, where they are vaccinated against rabies (amended from Defra, 2010c)

2.2. European Union Pet Movement Policy The EUPMP stipulates that companion animals must be identified and vaccinated against rabies at least 21 days prior to entry into another EU member state if the country of origin is a EU MS or equivalent or a listed third country. If the country of origin is unlisted, the animal must be identified, vaccinated, serologically tested at least 30 days following vaccination (in a laboratory approved and listed by the EU (EUROPA, 2010)), and the sample for the test must be taken at least three months prior to entry into the member state. Therefore, there are some similarities between the two (PETS and EUPMP) schemes as both require identification of the animal and vaccination against rabies as a method of control. However, there are important differences. For example, under the EUPMP companion animals from unlisted third countries would no longer have to spend 6 months in quarantine; instead they will be vaccinated, serologically tested and allowed to enter the UK 3 months after a successful blood sample (and 4 months following vaccination). In addition, for companion animals entering from EU MSs and listed third countries, there would be no requirement for serological testing and a reduction in the waiting time (from 6 months to 21 days). The EUPMP system is summarised in Figure 2 (taken from Defra, 2010c). Within this report the notation EUEU&Listed refers to the EUPMP scheme for EU MSs and listed third countries (i.e. Figure 2a) and EUunlisted refers to the EUPMP scheme for unlisted third countries (i.e. Figure 2b).

4

(a)

(b)

Figure 2: EUPMP system for the movement of cats, dogs and ferrets from (a) EU MSs and listed third countries and (b) unlisted third countries. (amended from Defra, 2010c)

2.3. Comparison of the schemes The differences between the policies for cats and dogs are summarised in Table 2. The risk pathways for each policy will be designed to account for these differences. Where a requirement varies between the policies, e.g. waiting time, this will be reflected in the model by different parameter values. Table 2: A comparison of the requirements of quarantine, PETS and the EUPMP for cats and dogs. Requirement Movement policy Quarantine PETS EUPMP EU & listed Unlisted Microchip No Yes Yes Yes Vaccination Yes (once in UK) Yes Yes Yes Blood sample Waiting Period

No 6 months (in UK quarantine)

Yes 6 months*

No 21 days

Yes 3 months*

*A 30-day wait is also required between vaccination and testing

Table 3: A comparison of the requirements of quarantine, PETS and the EUPMP ferrets. Movement policy Requirement PETS EUPMP Quarantine EU Listed EU & listed Unlisted Microchip No Yes Yes Yes Yes Vaccination Yes (once in UK) Yes Yes Yes Yes Blood sample No No No No Yes 6 months (in UK 6 Waiting Period 21 days 21 days 3 months* quarantine) months *A 30-day wait is also required between vaccination and testing

5

3. Scope of work 3.1. Species included in the QRA This risk assessment considers non-UK pet cats and dogs only. Ferrets and domestic UK-resident dogs/cats returning to the UK from abroad are not considered. It was not possible to include UK-resident pets due to a lack of data on where they travel to and also on how long they stay. Cats and dogs are grouped together as many of the parameter estimations in the risk assessment are common to both due to inadequate data for one or both species. This has also enabled the risk assessment to be carried out efficiently and effectively in the time available. Similarly, due to significant data gaps and the very low number of ferrets entering the UK compared to cats/dogs, ferrets are not considered further in this risk assessment. Appendix 3 provides further information. 3.2. Country groupings This risk assessment is on a worldwide scale (i.e. not just restricted to a few countries/regions). Currently, all countries are assigned to one of three categories: EU MSs (and equivalents), listed third countries and unlisted third countries (these countries are defined according to Annex II (as amended) of EC regulation 998/2003). However it is recognised that there is significant variability between countries within these categories in relation to the prevalence/incidence of rabies in their dog, cat and wildlife populations. Within the time scale for this risk assessment (3.5 months) it was not possible to assess the risk to the UK for the introduction of companion animals from each individual country. Therefore countries have been grouped within the three categories according to their yearly number of reported rabies cases, thus overcoming the need to model each country individually. This also allows consideration of those countries for which the data on rabies cases and the animal population are either incomplete or absent; these data are available in Appendix 4.2. 3.2.1 Grouping methodology used in previous risk assessments The concept of grouping countries has been used in previous rabies risk assessments. In the Kennedy report (Advisory Group on Quarantine, 1998), the world was split into regions, predominantly according to WHO classification, with additional division of Europe into the European Union (EU) and Eastern Europe, the Americas into North America and Latin America and the Caribbean, and separating Australia and New Zealand from the rest of the West Pacific region to take account of differing rabies incidence within each region. A 2006 Swedish risk assessment assessed the risk of rabies entering Sweden from the importation of companion animals from other EU countries (Hallgren, 2006), in which the EU was divided into four groups based on the number of reported rabies cases over the two year period from 2003 to 2004 and the risk of rabies introduction from surrounding areas. Finally, the UK Health Protection Agency (HPA) considers the rabies risk to UK travellers, classifying countries as either ‘No risk’, ‘Low risk’ or ‘High risk’ (HPA, 2010). This assessment of risk is based on a document from the Centre for Infections (Rabies Risk Table, February 2010), see Appendix 4.2, which contains qualitative information from the National Travel Health Network and Centre (NaTHNaC), World Organisation for Animal 6

Health’s (OIE) World Animal Health Information Database (WAHID) and the WHOs RABNET database. The NaTHNaC provide health advice for travellers and for each country in the world, give an indication as to whether it is rabies free, has no reported risk, or whether rabies is presumed present or confirmed in domestic and wild animals. Where data gaps exist HPA have used expert opinion to assign a rabies risk classification to each country (Dr Hilary Kirkbride, HPA, pers. comm.). 3.2.2. Rabies cases Quantitative data for rabies cases in cats, dogs and wild animals were required for as many countries as possible. Rabies cases in bats were only considered for countries in the Americas where it is known that bats can transmit classical rabies virus (RABV species 1) to domestic animals (Daoust et al., 1996; Leslie et al., 2006). For countries in the rest of the world, if rabies is reported in bats then it is assumed to be a rabiesrelated lyssavirus (species 2 – 11) strain of the virus. See Appendix 5 for a description of the different lyssavirus species. There is no evidence to support sustained transmission of these viruses within terrestrial mammals, only sporadic spill-over to domestic animals are documented. Therefore, for these countries, rabies cases in bats are not considered in this risk assessment. Data on the annual number of rabies cases were collected from a number of sources: •

•

•

Rabies Bulletin Europe (http://www.who-rabies-bulletin.org/) is sponsored by the World Health Organization (WHO) and describes the reported cases of rabies in individual European countries. Data are documented yearly up to 2009 for a range of species including cats, dogs, wild animals and bats. RABNET (http://apps.who.int/globalatlas/default.asp) is a WHO database and contains data on rabies cases for most countries in the world. Data are presented on a yearly basis up to 2010 for humans and animals (dogs, other domestic animals, wild animals and bats). However, no data specific to cats is documented. The World Animal Health Information Database (WAHID) Interface (http://www.oie.int/wahis/public.php?page=home) provides access to all data held within OIE’s World Animal Health Information System (WAHIS). It contains data on rabies cases for a range of species (including cats, dogs, wild animals) for most countries in the world in the period 2005-2010. Rabies in bats is included in the wild animal dataset.

No published data for rabies prevalence were available, i.e. no data that has been adjusted for the animal population in each country. In addition, no (comparative) quantitative data could be found for rabies surveillance systems in the different countries. It would therefore be challenging to adjust the raw data on clinical cases for under-reporting of rabies in an objective manner. Therefore the number of rabies cases is not adjusted for any potential under-reporting of rabies and consequently the countries were grouped solely based on their annual number of reported rabies cases. 3.2.3. Methodology for country groupings The grouping structure needs to take into account the categories that are present in PETS/Quarantine and the EUPMP import schemes. As a consequence the countries were divided into three main categories: 7

1. EU Member States & equivalents 2. Listed third countries 3. Unlisted third countries Within each of these categories there is freedom to create country groupings based on rabies incidence, prevalence or reported cases. Based on discussions within the project team, it was decided to group countries according to their reported number of rabies cases. In particular, it was felt important that countries with rabies in their wildlife population (not including lyssavirus species 2 – 11 in bats) but with no or very few cat/dog cases (from spill over events) be separated from countries with no reports of rabies. As a consequence of this, the following groups were adopted within each of the 3 main categories listed above: Group 1: No reported cases of rabies in cats, dogs and wildlife between 2007-09. For countries where quantitative data on rabies cases are incomplete or unavailable they must lie in the ‘No risk’ category of the HPA classification and be recorded as either ‘Rabies free’ or ‘No reported risk’ by NaTHNaC. Group 2: Less than or equal to five total rabies cases reported in cats and dogs between 2007-09 and/or rabies present in wildlife This group caters for cases of imported animals with rabies and sporadic spill-over from wild to domestic animals. For countries where quantitative data on rabies cases are incomplete or unavailable they must lie in the ‘Low risk’ category of the HPA classification and be recorded as ‘Rabies presumed present in domestic and wild animals’ by NaTHNaC. Group 3: Greater than five rabies cases reported in cats and dogs between 2007-09. For countries where quantitative data on rabies cases are incomplete or unavailable they must lie in the ‘High risk’ category of the HPA classification, be recorded as ‘Rabies present in domestic and wild animals’ by NaTHNaC or be reported as ‘Disease present but without quantitative data’ by WAHID. The three year period, 2007-09, was selected in order to capture the current situation regarding rabies cases as well as the variation in reporting. Five cases in the three year period 2007-09 was chosen as a threshold between Group 2 and Group 3. This threshold was defined by observing the number of rabies cases that had been reported. The threshold is also in good agreement with the HPA’s ’Low risk’ and ‘High risk’ classification based on those countries for which data were available, therefore validating the use of the HPA classification for the placement of countries without quantitative data into the grouping structure of this risk assessment. Isolated imported cases (detected outside of quarantine) were included in Group 2 even if the country had been previously free from rabies; examples of such countries include France, Germany & Belgium. 3.2.4 Country groups Using data on the number of rabies cases in cats, dogs and wild animals obtained from the databases along with the HPA classification and NaTHNaC advice (when required), each country in the world was considered in turn according to the groupings defined above. It was noted that some unlisted third countries reported zero cases of rabies in most of the years during 2007-09 with no data available for the remaining 8

years. For these countries Group 3 was considered to be inappropriate and so the HPA classification and NaTHNaC advice was used to distinguish these countries between Group 1 and Group 2. The country groupings are detailed below, indexed by j. Groupings for EU Member States and equivalents Group 1 (j = 1) Austria, Canary Islands, Czech Republic, Denmark, Finland, GB, Greece, Guadeloupe, Iceland, Ireland, Liechtenstein, Luxembourg, Malta, Martinique, Monaco, Netherlands, Norway, Portugal, Republic of Cyprus, Reunion, San Marino, Slovakia, Spain (mainland), Sweden, Switzerland, Vatican* * No rabies data are available for the Vatican, however it is placed in Group 1 as it is assumed to be rabies free. Furthermore, no companion animal enter the UK from the Vatican (See Appendix 4), therefore this placing will not affect the overall risk of rabies entry to the UK from EU MSs Group 1 countries.

Group 2 (j = 2) Andorra, Azores, Belgium, Estonia, Faroe Islands, France, French Guiana, Germany, Gibraltar, Greenland, Hungary, Italy, Madeira Islands, Slovenia Group 3 (j = 3) Bulgaria, Ceuta, Croatia, Latvia, Lithuania, Melilla, Poland, Romania Groupings for Listed third countries Group 1 (j = 4) Antigua & Barbuda, Aruba, Australia, Barbados, British Virgin Islands, Cayman Islands, French Polynesia, Guam, Hawaii, Jamaica, Japan, Malaysia, Mauritius, Montserrat, Netherlands Antilles, New Caledonia, New Zealand, Singapore, St Kitts & Nevis, St Lucia, St Vincent & The Grenadines, Taiwan, Wallis & Futuna Islands Group 2 (j = 5) Ascension Island, Bahrain, Bermuda, Chile, Falkland Islands, Fiji, Hong Kong, St Helena, St Pierre & Miquelon, Trinidad & Tobago, United Arab Emirates, Vanuatu Group 3 (j = 6) Argentina, Belarus, Bosnia-Herzegovina, Canada, Mayotte, Mexico, Russian Federation, USA (mainland) Groupings for Unlisted third countries Group 1 (j = 7) Albania, Anguilla, Bahamas, Djibouti, Dominica, Kuwait, Macedonia, Northern Mariana Islands, Palau, Pitcairn Islands, Tokelau, Turks and Caicos Islands, Virgin Islands (USA) Group 2 (j = 8) Belize, Brunei, Cape Verde, Costa Rica, Egypt, Grenada, Honduras, Kiribati, Lebanon, Libya, Maldives, Marshall Islands, Micronesia (Federated States), Montenegro, Nauru, Niue, Panama, Papua New Guinea, Samoa, Sao Tome and 9

Principe, Seychelles, Solomon Islands, Suriname, Svalbard and Jan Mayen Islands, Syria, Tonga, Tuvalu, Uruguay Group 3 (j = 9) Afghanistan, Algeria, Angola, Armenia, Azerbaijan, Bangladesh, Benin, Bhutan, Bolivia, Borneo, Botswana, Brazil, Burkina, Burundi, Cambodia, Cameroon, Central African Republic, Chad, China (People’s Rep. of), Colombia, Comoros, Congo (Dem. Rep. of), Cote d’Ivoire, Cuba, Dominican Republic, Ecuador, East Timor, El Salvador, Equatorial Guinea, Eritrea, Ethiopia, Gabon, Gambia, Georgia, Ghana, Guatemala, Guinea, Guinea-Bissau, Guyana, Haiti, India, Indonesia, Iran, Iraq, Israel, Jordan, Kazakhstan, Kenya, Korea (Dem. People’s Rep.), Korea (Rep. of), Kosovo, Kyrgyzstan, Laos, Lesotho, Liberia, Macau, Madagascar, Malawi, Mali, Mauritania, Moldova, Mongolia, Morocco, Mozambique, Myanmar, Namibia, Nepal, Nicaragua, Niger, Nigeria, Oman, Pakistan, Paraguay, Peru, Philippines, Puerto Rico, Qatar, Rwanda, Saudi Arabia, Senegal, Serbia, Sierra Leone, Somalia, South Africa, Sri Lanka, Sudan, Swaziland, Tajikistan, Tanzania, Thailand, Togo, Tunisia, Turkey, Turkmenistan, Uganda, Ukraine, Uzbekistan, Venezuela, Vietnam, Yemen, Zambia, Zimbabwe

10

4. Methodology 4.1. Risk pathway To develop a quantitative risk assessment model, a risk pathway is required to identify the biological sequence of events that must take place in order for the hazard (i.e. rabies) being modelled to occur; in this case, the risk of at least one cat/dog incubating rabies entering the UK. In order to address the risk question posed, the model must be able to compare three different importation policies; 6 month quarantine, PETS and the EUPMP, a separate risk pathway is therefore required for each of these schemes. In order to aid the design of the risk pathway (and to also inform the team of previous modelling approaches with regards to parameter calculations) a literature review of previous rabies risk assessments was undertaken. This review focused on the risk pathways used, and, where appropriate, the calculation of the parameters within each model; the full report from this review can be found in Appendix 6. Following completion of this review, it became apparent that many of the risk assessments use very similar risk pathways, with a number of studies basing their risk pathways or parameter calculations on previous risk assessment models (see Table 4). The pathways are similar because they all follow the logical steps of the importation policies, which share a number of common requirements. The only significant difference between the risk assessments being the consideration of checking at the importing border in some models. It was decided that the current risk pathway should include all parameters common to previous rabies risk assessments. In addition, the review identified one model that compared the UK PETS and quarantine system with the EUPMP (Ramnial et al., 2009). It was therefore concluded that the risk pathways used in that model should be used as a baseline in the development of the risk pathway for the current risk assessment, and that additional parameters and routes of rabies entry, and modified/updated parameter estimates should be included where appropriate. Table 4: The parameters in the risk pathways involving a serological test in the previous rabies risk assessments considered. Shaded boxes represent parameters that are included in a particular model. Risk Assessment Parameter 1 2 3 4 5 6 P(companion animal is infected) P(companion animal is vaccinated upon entry) P(companion animal is protected after vaccination) P(serological test after vaccination) P(serological test gives a false positive) P(companion animal becomes infected during waiting period) P(clinical signs of rabies not displayed until after waiting period) P(companion animal is checked upon entry) P(companion animal passes entry check) [1] Advisory Group on Quarantine, 1998 [2] EFSA, 2006 [3] Kamakawa et al., 2009 [4] Ramnial et al., 2009 [5] Hallgren, 2006 [6] Napp et al., 2010.

11

The risk pathways for the potential routes of rabies entry into the UK via the movement of a single dog or cat through quarantine, PETS/EUPMPunlisted and EUPMPEU&listed are displayed in Figure 3, Figure 4 and Figure 5 respectively. For each pathway, there are a number of scenarios by which an animal incubating disease can enter the UK. In these figures, the pathways through which non-compliant companion animals could enter the UK are displayed in red italics; animals will only enter through these routes when noncompliance with the regulations is included in the model.

Model formulation R = PI*Pv*PNP*PNCS

Figure 3: Pathway for rabies entry to the UK via a dog/cat movement through the 6 month quarantine system.

Model formulation R1 = PI*PV*PNP*PST*PST+*PNCS*PC*PC+ R2 = PI*PV*PNP*PST*PST+*PNCS*(1-PC) R3 = PI*PV*PNP*(1-PST)*PNCS*PC*PC+ R4 = PI*PV*PNP*(1-PST)*PNCS*(1-PC) R5 = PI*(1-PV)*PNP*(1-PST)*PNCS*PC*PC+ R6 = PI*(1-PV)*PNP*(1-PST)*PNCS*(1-PC)

R7 = (1-PI)*PV*PNP*(1-PST)*PI**PNCS*PC*PC+ R8 = (1-PI)*PV*PNP*(1-PST)*PI**PNCS*(1-PC) R9 = (1-PI)*PV*PNP*PST*PST+*PI**PNCS*PC*PC+ R10 = (1-PI)*PV*PNP*PST*PST+*PI**PNCS*(1-PC) R11 = (1-PI)*(1-PV)*PNP*(1-PST)*PI**PNCS*PC*PC+ R12 = (1-PI)*(1-PV)*PNP*(1-PST)*PI**PNCS *(1-PC)

12

R = ∑ Rn where n = pathway n =1

Figure 4: Pathway for rabies entry to the UK via a dog/cat movement through PETS or EUPMPunlisted. Illegal pathways are denoted by red italics.

12

Model formulation R1 = PI*PV*PNP*PNCS*PC*PC+ R2 = PI*PV*PNP*PNCS*(1-PC) R3 = PI*(1-PV)*PNP*PNCS*PC*PC+ R4 = PI*(1-PV)*PNP*PNCS*(1-PC)

R5 = (1-PI)*PV*PNP*PI**PNCS*PC*PC+ R6 = (1-PI)*PV*PNP*PI**PNCS*(1-PC) R7 = (1-PI)*(1-PV)*PNP*PI**PNCS*PC*PC+ R8 = (1-PI)*(1-PV)*PNP*PI**PNCS*(1-PC)

8

R = ∑ Rn where n = pathway n =1

Figure 5: Pathway for rabies entry to the UK via a dog/cat movement through EUPMPEU&listed. Illegal pathways are denoted by red italics.

4.2. Model implementation A stochastic risk assessment model was developed using the software package @Risk (© Palisade) Version 4.5, an add-on package within Microsoft Excel 2003 (© Microsoft 1985-2003). This allows uncertainty, when known, to be mathematically described by probability distributions from which, on each iteration of the model, a single value is selected. It is good risk assessment practice to separate uncertainty and variability, where uncertainty represents lack of knowledge and variability natural variation. It is important to keep the concepts separated because uncertainty can be reduced, but variability cannot. Advanced simulation techniques, called second order modelling (Vose, 2000) allows the separation of these two characteristics however, within the time scale given for this project this was not possible. Therefore, only uncertainty was considered within the model; variability was omitted. While it is difficult to determine the effect of variability in the overall model, the effect of variability in each parameter was investigated via sensitivity analysis (see Section 4.5), to give an idea of how natural variation may affect the overall results. For each scenario on a particular risk pathway, the risk of an infected pet entering the UK was calculated by multiplying the probabilities associated with the branches of 13

the scenario. The overall risk of a single pet entering through a specific scheme (R) was then calculated by adding the risks associated with all scenarios (See Figure 3 - Figure 5). Consequently the annual probability of importing at least one infected cat/dog (PR,j) via a particular scheme for country group j was estimated by PR , j = 1 − (1 − R)

NI , j

where NI,j is the number of pets imported via the particular scheme for country group j. Therefore, the number of years between rabies entry from a particular group (YR,j) is YR , j =

1 . PR , j

The probabilities of at least one infected animal entering the UK from either an EU MS, listed third country or an unlisted third country are calculated using the equations given below. The number of years between entry from each country classification is calculated by dividing 1 by these probabilities, i.e. 1 PR ,EU , 1 PR ,listed and 1 PR ,unlisted .

PR ,EU = 1 − (1 − PR ,1 )(1 − PR , 2 )(1 − PR ,3 ) PR ,listed = 1 − (1 − PR , 4 )(1 − PR ,5 )(1 − PR ,6 ) PR ,unlisted = 1 − (1 − PR , 7 )(1 − PR ,8 )(1 − PR ,9 ) Therefore, the overall annual probability of at least one infected dog/cat entering the UK is calculated as PR = 1 − (1 − PR ,EU )(1 − PR ,listed )(1 − PR ,unlisted ) . The overall number of years between rabies entry is calculated by dividing 1 by the annual probability of at least one infected animal entering the UK, i.e. 1 PR . 4.3. Parameter estimation 4.3.1. Incubation period (IP)

Data on the incubation period of rabies in both cats and dogs were collected from a number of sources and are presented in Appendix 4.1. These data consist of results from both experimental studies and estimates of the incubation period from naturally acquired cases of rabies. Experimental data on rabies infection and incubation periods in dogs and cats were collected from four sources (Bingham, 1999; Fekadu et al., 1982; Soulebot et al., 1981; Trimarchi et al., 1986). In these cases the animals will have been infected with a viral dose that may be much higher than the dose that would be transmitted in a natural infection. Therefore, it is likely that the disease will manifest more quickly than it would in naturally occurring cases and the true incubation period could be underestimated. Some estimates of the incubation period from naturally acquired rabies cases were available (Advisory Group on Quarantine, 1998; Foggin, 1988). However, it was 14

difficult to estimate the precise incubation period as often the initial date of exposure is unknown. One potential data source that can be of use is data on the manifestation of clinical signs of rabies in quarantine centres (Committee of Enquiry on Rabies, 1971; Fooks et al., 2008). For these data an estimate of the natural incubation period was generated using the date of entry into quarantine and the date of onset of clinical signs; however, this estimate would not take into account the period of time between when the animal was actually exposed and when it entered quarantine. Irrespective of this, although still an underestimate, these estimates are likely to provide an alternative estimate of the true incubation period. Another potential issue with the quarantine data set, however, is that there may have been some risk of transmission of rabies within the quarantine centres. Although this should not happen, there has been a large drop in the number of rabies cases in UK quarantine since the introduction of mandatory vaccination upon arrival and separation of individual dogs in 1971, which supports the conclusion that some cases previously seen in quarantine had been contracted within the quarantine facility. Consequently, some incubation period estimates from quarantine data could be a vast overestimate, especially if the animal was infected near the end of its time in quarantine. This could potentially skew the incubation period data and lead to an overestimate of the true incubation period. As for quarantine, data on imported cases of rabies in rabies-free countries can also be used as an estimate of the incubation period in natural cases (Anonymous, 2008; Le Roux & Van Gucht, 2008). In these cases the date of entry into the country and the date of onset of signs are used as the estimate of the incubation period. Again, the period of time between exposure and movement into the rabies-free country is an unknown value and therefore the incubation period is likely to be underestimated. The experimental and natural infection data sets acquired for estimating the incubation period of rabies in cats and dogs are both highly uncertain. Given the caveats with both sources of data, it is inappropriate to use only one source of data (experimental or natural infection) for the estimation of the incubation period. Therefore all of these data were combined to give an average incubation period of rabies in cats and dogs that encompasses all available information. However, because of the associated uncertainty this parameter is investigated in the sensitivity/scenario analysis (see Section 4.5). The combined data from all sources gives an average incubation period of 35 days with a standard deviation of 36.8 days. These values were used to define a Lognormal distribution. Using this distribution there is a chance (albeit a very small chance) that an animal may incubate rabies for longer than 6 months, thus presenting a risk that an animal incubating rabies may enter the UK through quarantine. The mean and standard deviation of the associated Normal distribution (μ and σ) are given by ⎛

⎞ ⎛ 2 .8 2 ⎞ ⎟ and σ = ln⎜ 35 + 36 ⎟⎟ , ⎜ 2 2 ⎟ 35 2 ⎝ ⎠ ⎝ 35 + 36.8 ⎠

μ = ln⎜⎜

35 2

(Vose, 2000).

15

4.3.2. Probability companion animal from country group j is infected (PI,j)

The probability that a cat/dog from a country group j, (where j = 1,…,9 as defined in Table 5) is infected was estimated by considering those countries (i) within the group for which there are available data on both the number of rabies cases and cat/dog populations. Appendix 4.2 outlines the data on the number of reported rabies cases between 2007 and 2009 and (where available) the cat/dog population data for each country. Table 5: Definition of the group index (j) EU MSs (and equivalents) Listed third countries Unlisted countries Group 1 (j = 1) Group 1 (j = 4) Group 1 (j = 7) Group 2 (j = 2) Group 2 (j = 5) Group 2 (j = 8) Group 3 (j = 3) Group 3 (j = 6) Group 3 (j = 9)

For each country, i, reported rabies cases in cats and dogs are combined to give annual totals for 2007, 2008 and 2009 (denoted Ii(2007), Ii(2008) and Ii(2009) respectively). The largest value, i.e.

I i(max) = max(I i( 2007) , I i( 2008) , I i( 2009) ) is selected in order to (1) represent the current worst case scenario and (2) take into account, to some degree, cases of rabies which are not reported. Summing the maximum number of cases for each country grouping j gives

I j = ∑ I i(max) . i

The probability that a dog/cat is infected is estimated using a Bayesian method (Jones et al., 2005b). It was assumed that new cases of rabies in a country group j follow a Poisson process with rate λj. Using Bayesian inference, the uncertainty associated with λj is derived by assuming an uninformed ‘flat’ prior (Vose, 2000) and a Poisson likelihood. The resulting posterior distribution for λj describes the uncertainty associated with the number of unobserved cases. This posterior distribution is given by

⎛⎛

λ j ~ Gamma⎜⎜ ⎜⎜ I j × ⎝⎝

⎞ IP ⎞ ⎟ + 1,1⎟ , ⎟ ⎟ 365 ⎠ ⎠

where IP is the mean incubation period (35 days). In the instances when there are no reported cases of rabies the distribution reduces to Gamma (1,1). The probability that a dog/cat from a country group is infected (PI,j) is then estimated by dividing λj by the combined cat and dog population of the group. Estimates for cat and dog populations were obtained from the World Society for the Protection of Animals (WSPA), WAHID and various other sources such as pet food manufacturers and veterinary groups. See Appendix 4.2 for the cat/dog population data. The most recent data values were taken for those countries where data were available. For some countries, two values were obtained for the same year. In these cases a Uniform distribution was used to describe the associated uncertainty. 16

Note that the probability, PI,j, was estimated by considering those countries in a group for which there is data available on both number of rabies cases and cat/dog population. Both data sources were available for approximately half of the countries in the world, however it was noted that there were substantially more data gaps for the unlisted countries compared to EU MSs and listed third countries. It was assumed that this probability is also representative of the average situation for the remaining countries in the group for which there were incomplete data. All of the databases used in this QRA, from which data on the number of rabies cases were collected, detail rabies cases and domestic animals separately. Therefore rabies cases in wildlife are not included in the estimates of the probability of a companion animals being infected. RBE is the only database which reports rabies cases in stray dogs separately from the rabies cases in domestic dogs (cases in stray and domestic cats are assumed to be grouped together). All other data sources used (WAHID and RABNET) don’t specifically document rabies cases in stray cats or dogs. Therefore, the estimates that are generated from the data collected from these sources may include rabies cases in stray cats and dogs since no data were available to allow these cases to be removed from the overall total. The WSPA report (unpublished) details pet (dog and cat) populations for a large number of countries worldwide for 2007. It is possible that these estimates may include stray cats and dogs. The WAHID database also contains data on cat and dog populations up to 2009. Again, these figures may include stray animals. It was not possible to separate strays from domestic animals in these data. Where possible, the data used to estimate PI,j is limited to domestic animals, however given the data limitations this has not always been possible and therefore there is uncertainty associated with the final estimate. 4.3.3. Probability that an unprotected animal passes the serological test (PST+)

An animal that is vaccinated against rabies is only considered to be protected from rabies infection if a serological test results in a neutralising antibody titre of greater than 0.5 IU/ml. There are two serological tests approved for pet movement schemes that measure the rabies antibody titre of animals; the Fluorescent Antibody Virus Neutralisation (FAVN) test and the Rapid Fluorescent Focus Inhibition Test (RFFIT). It is unlikely that either of these tests will be 100% specific, therefore some animals, especially those with a titre close to the threshold of 0.5 IU/ml may test positive. Previous rabies risk assessments have used a 1998 study detailing the development of the FAVN test (Cliquet et al., 1998) to estimate the sensitivity and specificity of the serological tests. In this study the FAVN test was compared to the RFFIT and the Mouse Neutralisation Test (MNT). The results from this study are displayed in Table 6 and Table 7 below.

17

Table 6: Comparison of the MNT and FAVN tests for 20 unvaccinated and 30 vaccinated dogs (Cliquet et al., 1998) MNT FAVN Number of sera 20 Unvaccinated + + 16 Vaccinated 10 Vaccinated + 4 Vaccinated + 0 Vaccinated Table 7: Comparison of the RFFIT and FAVN tests for 78 unvaccinated and 77 vaccinated dogs (Cliquet et al., 1998) RFFIT FAVN Number of sera 74 Unvaccinated + 4 Unvaccinated + + 47 Vaccinated 24 Vaccinated + 0 Vaccinated + 6 Vaccinated

The estimates of the diagnostic test performances for each test were derived by using a Bayesian model. The model, based on that, in Branscum et al., 2005 was fitted to the data in Table 6 using WinBUGS 1.4 software and the results were used to inform the prior distributions for input to a similar model fitted to the data in Table 7. Therefore the two stage process was: Stage 1 (Table 6): The prior distributions for sensitivities and specificities of the FAVN test were assumed to be uninformative (Beta (1,1)). However the antibody prevalence in the unvaccinated dogs was assumed to be very low (99% sure that it was below 0.1%) and high in the vaccinated dogs (99% sure that it exceeded 95%). Stage 2 (Table 7): Uninformative priors were specified for the sensitivity and specificity of the RFFIT test but for the sensitivity and specificity of the FAVN test and the antibody prevalences in the unvaccinated and vaccinated dogs the prior Beta distributions were based on the parameter estimates from Stage 1. The final estimates are summarised in Table 8. Table 8: Bayesian model outputs for FAVN and RFFIT test specificities Parameter Mean Median S.D. 2.5% 97.5% 0.991 0.993 0.009 0.967 0.999 FAVN specificity (SpFAVN) 0.948 0.951 0.022 0.896 0.983 RFFIT specificity (SpRFFIT)

The FAVN and RFFIT specificities are described by Beta distributions, which were defined using the values in Table 8 and the BetaBuster software tool (©University of California, Davis).

SpFAVN = Beta(124.8,1.248) and Sp RFFIT = Beta(92.97,5.132) . Limited information was available on the frequency of use of each test. Therefore, assuming that each test is used at the same frequency, both specificities were equally

18

weighted to produce a mean specificity of the serological tests for rabies antibodies (Sp). Using this specificity the overall probability of an unprotected animal passing the serological test (PST+) is estimated as

PST + = 1 − Sp , where Sp =

SpFAVN + Sp RFFIT . 2

No false-negatives were observed in either test. However, a sensitivity of less than 100% does not affect the risk of an infected animal entering the UK, as protected animals that produce a false-negative will either be re-vaccinated and re-tested or removed from the scheme. 4.3.4. Probability that a vaccinated pet is not protected (PNP)

It is unlikely that any vaccine will be 100% protective and therefore a certain proportion of vaccinated animals may fail to mount an immunological response that will confer protection against rabies infection. The current EU and OIE standards for rabies protection after vaccination classify an animal as protected from rabies if it can achieve an antibody titre greater than a threshold value of 0.5 international units per ml (IU/ml) in a serological test. This is a conservative threshold, as animals having antibody titres between 0.1 and 0.5 have a very high survival rate when challenged with rabies (greater than 90%, Aubert, 1992). In this risk assessment, the current EU and OIE threshold of 0.5 IU/ml was used as a worst case assumption, i.e. if an animal has an antibody titre greater than 0.5 IU/ml then it is protected, otherwise it is unprotected. The number/proportion of animals that failed to produce a titre greater than 0.5 IU/ml after vaccination was obtained from a number of published studies on the response of cats and dogs to rabies vaccinations (Bahloul et al., 2006; Kallel et al., 2006; Minke et al., 2009a; Sihvonen et al., 1995b). These data are presented in Table 9.

19

Study number (m) 1 2 3 4

Table 9: Summary of the data from a number of different vaccination studies Serological Number achieving Reference Vaccine Days between Number test used threshold titre used vaccination of animals (>0.5 IU/ml) (k) and testing tested (Sm,k) (nm,k) Sihvonen Rabisin 83 80 et al., 30-40 RFFIT Madivak 47 46 1995a Kallel et Rabisin 30 5 RFFIT 4 al., 2006 Bahloul et Rabisin 35 4 RFFIT 4 al., 2006 Minke et Rabisin 15 14 28 FAVN al., 2009b Nobivak 15 10

For each study, m (m = 1,…, 4), the number of animals with antibody titre >0.5 IU/ml, sm,k was corrected in order to account for a serological test that is not 100% specific; therefore estimating the number of animals that were truly protected after vaccination (NV+,m,k) with vaccine k (k = Rb: Rabisin; k = Nb: Nobivak, k = Md: Madivak). This was achieved by applying a Negative Binomial distribution, which uses the test specificity of the relevant serological test (as estimated in Section 4.3.3) to estimate the number of false-positive animals in each study NV + ,m,k = sm,k − NegBin( sm,k + 1, Spm ) , where Spm is the specificity of the serological test used in study m. Using Bayesian inference with an uninformed prior Beta(1,1), the probability of a particular vaccine inducing protection given the results of the 4 studies, PV+,k, is described by 4

4

4

m=1

m =1

m =1

PV + ,k = Beta(∑ NV + ,m ,k + 1, ∑ nm ,k − ∑ NV + ,m ,k + 1) ,

where nm,k is the number of animals tested in study m with vaccine k. From the vaccines considered in the studies outlined in Table 9 only Rabisin and Nobivak are licensed for use in the UK. Limited information is available on the frequency of use of each vaccine. Therefore, assuming each vaccine is used with the same frequency, the probability of a cat/dog not being protected (PNP) after vaccination in UK quarantine was estimated using the average efficacy of Rabisin and Nobivac PNP = 1 −

PV + , Rb + PV + , Nb . 2

Similarly, for animals vaccinated in the country of origin (PETS and EUPMP), the probability of not being protected after vaccination was estimated using all available vaccine data, giving an average probability that a rabies vaccine will not induce protection

20

PNP = 1 −

PV + , Rb + PV + , Nb + PV + ,Md 3

.

The approach described above gives a mean probability of an uninfected pet not being protected after vaccination of 18.6%, with 5th and 95th percentiles of 11.5% and 26.6% respectively. These values apply only to animals not incubating rabies. For animals that are already infected with the rabies virus it was assumed that vaccination induces no effective immune response and the animal remains unprotected, therefore in these cases PNP = 1. For animals that are not vaccinated it was also assumed that they are not protected (PNP = 1). 4.3.5. Probability than an unprotected animal becomes infected during waiting period (PI*,j)

For the purposes of this risk assessment it is assumed that all companion animals in a particular country/group have the same daily probability of becoming infected, therefore the probability that an unprotected companion animal from country group j becomes infected during the waiting period (PI*,j) is estimated from the annual number of reported rabies cases in that group. Firstly the uncertainty associated with the reported number of rabies cases (θj) is estimated using a Gamma distribution as in section 4.3.2.

θ j ~ Gamma(I j + 1,1) The daily probability of a companion animal in that group becoming infected (PI’,j) is therefore PI ', j =

θj N * 365

where N is the combined cat/dog population of group j. The probability of an unprotected companion animal becoming infected during the waiting period is therefore

PI *, j = 1 − (1 − PI ', j )T where T is the waiting period (212 days for PETS, 121 days for EUPMPunlisted and 21 days for EUPMPEU&listed). Note, the waiting period for PETS and EUPMPunlisted incorporates both the time between vaccination and serological testing (assumed to be 30 days) and the number of days the animal must wait in the country of origin after testing before it can be moved. 4.3.6. Probability that an infected animal does not display clinical signs during the waiting period (PNCS)

Dogs and cats that are infected with rabies are only likely to enter the UK via quarantine, PETS or the EUPMP if they are not yet showing clinical signs of disease. It was assumed that any compliant animals displaying clinical signs of rabies will be

21

detected and removed from the scheme. The probability of an infected animal not displaying clinical signs in the waiting period is dependent on the time between infection and entry (t) and the incubation period of rabies (IP). For animals that are infected before vaccination a worse-case scenario was assumed, that is, the animal was infected immediately before entry into the scheme. Therefore t is dependent on the waiting period T of the entry scheme, 212, 182, 121 and 21 days for PETS, quarantine, EUPMPunlisted and EUPMPEU&isted respectively. The probability of the incubation period being greater than the waiting period in a particular entry scheme (PNCS) is calculated from the cumulative density function of the incubation period, a Lognormal(μ, σ) distribution (see Section 4.3.1)

PNCS = P( IP > T ) . For animals that become infected during the waiting period the probability of not showing clinical signs is calculated by averaging the probabilities of each possible day of infection T

PNCS =

∑ P( IP > t ) t =1

T

.

The resulting probabilities for each entry scheme are shown in Table 10. Table 10: The probability of clinical signs not being displayed during the waiting period for each entry scheme Pet Waiting Probability clinical signs not displayed during waiting period movement period in (PNCS) scheme days (T) Animal infected before Animal infected during waiting vaccination period 182 0.0096 N/A Quarantine 212 0.0059 0.1604 PETS 121 0.0308 0.2703 EUPMPunlisted 21 0.5631 0.8079 EUPMPlisted

4.3.7. Probability an animal passes import checks from a country group j (PC+,j)

Data on the number of cats and dogs presented for entry into the UK via PETS between 2005 and 2009, and the number that passed and failed the entry checks were obtained from Animal Health, Chelmsford. The raw data on the number of dog/cat entries to the UK for the last five years can be found in Appendix 4.5.

22

Table 11: The number of dogs/cats passing the PETS entry checks for each country group 20052009, and the number failing because of problems with vaccination or the serological test

EU MS & equivalents Listed third Countries

Country Group

Group 1 Group 2 Group 3 Group 1 Group 2 Group 3

j

Number Presented (NC,j)

Number failing section IV of the pet passport (vaccination against rabies)

Number failing section V of the pet passport (rabies serological test)

1 2 3 4 5 6

Number passing checks (NC+,j)

63863 91056 7652 7280 2206 19442

59877 85662 6157 7184 2150 17631

126 204 26 4 0 1

128 270 35 0 0 2

For each country group (j = 1,…,6), the probability of an animal from that group passing documentation checks upon entry to the UK is estimated using a Beta distribution PC + , j = Beta( N C + , j + 1, N C , j − N C + , j + 1) . No data were available for the probability of dogs/cats from unlisted third countries passing entry checks as these countries do not currently import animals to the UK via PETS. Therefore, for these countries under the EUPMP, the probability of passing import checks was estimated using a Beta distribution incorporating all the data on animals passing and failing checks upon entry into the UK, i.e. 6

6

6

j =1

j =1

j =1

PC + , j = Beta(∑ N C + , j + 1, ∑ N C , j −∑ N C + , j + 1) , for j = 7, 8, 9.

4.3.8. Compliance parameters 4.3.8.1. Probability companion animal is vaccinated (PV)

The probability that a companion animal is vaccinated, PV, is the only compliance parameter that is present in all of the travel schemes (quarantine, PETS, EUPMPunlisted and EUPMPEU&listed). From the data presented in Table 11 a very small proportion of companion animals fail entry checks due to section IV of the pet passport, which relates to vaccination. Moreover many of these failures may be attributable to clerical errors, e.g. lacking a signature, rather than the animal not being vaccinated. Therefore, full compliance is assumed for the baseline model (PV = 1). 4.3.8.2. Probability companion animal is serologically tested (PST)

It is a requirement in PETS and EUPMPunlisted that a dog/cat is serologically tested in order to determine if it has a level of antibodies considered adequate for protection against rabies. The proportion of companion animals failing entry checks due to section V of the pet passport (serological testing) is very low (Table 11). Hence, the probability that a companion animal is serologically tested, PST, for both PETS and EUPMPunlisted is assumed to be 1. As with vaccination, from all of these failures, many will probably be purely clerical errors.

23

4.3.8.3. Probability companion animal is checked at border (PC)

Border checks are required for all animals that are entering the UK via PETS and for companion animals that are entering the EU from listed third and unlisted countries via EUPMP. By definition, the data on companion animal entries that have passed and failed checks (Table 11) incorporates only those animals that have been checked upon entry. Consequently no reliable estimate can be made on the compliance of checking. It was therefore assumed that the probability that an animal is checked at the border, PC, is 1. There is no requirement for companion animals from EU Member States (and equivalents) entering via EUPMP to be checked. It is therefore assumed that PC is 0 for this scenario. 4.3.8.4. Assessing the impact of compliance

As part of the risk question, Defra require an assessment of the risk of rabies introduction if the rules are followed with 100% compliance, compared to compliance of less than 100%. Previous risk assessments (e.g Jones et al., 2005a; Ramnial, 2009) have used the distributions specified by Kennedy et al. (Advisory Group on Quarantine, 1998) to investigate the impact. However these distributions were deemed to be no longer appropriate for use in this risk assessment, as they were generated from data available in 1998, before PETS was introduced, and are therefore unlikely to be representative of the current level of compliance in the UK. Therefore a scenario analysis was carried out in order to investigate various issues surrounding compliance and to assess the change in risk. The various compliance parameters, PV, PST and PC, were reduced simultaneously in a stepwise manner, first to 90% compliance and then to 80% compliance. When running the compliance analyses the assumption was made that an animal that is knowingly non-compliant (i.e. has not been vaccinated and/or tested) will attempt to enter the UK immediately with false-documentation. Consequently, there is no waiting period in the country of origin under these scenarios. Note, when PV is reduced it is assumed that this does not apply to vaccinations in UK quarantine, and that all animals entering quarantine are still vaccinated. This assumption is made because these vaccinations take place in UK quarantine centres and it is therefore assumed that all regulations will be fully complied with. 4.4. Number of animals imported per year (NI,j)

The number of cat and dog entries into the UK through both quarantine and PETS for each year between 2005 and 2009 were obtained from Animal Health, Chelmsford, see Appendix 4.5. These entries were grouped by classification in the PETS scheme (i.e. listed, unlisted, etc) and then by the three country groupings defined in the prevalence calculations (Table 12). Table 13 represents the data in a slightly different format, showing the annual number of companion animal entries into the UK from (i) each country group (ii) each country classification and (iii) from all countries in the world. The maximum number of annual imports for each group was selected for use in the model as a current worst-case scenario. It should be noted that a substantial number of cats and dogs entering the UK from PETS qualifying countries during this time frame entered via quarantine. These entries are likely to fall into two types. Firstly, cats/dogs whose owners have, for whatever reason, decided to forgo PETS (it is believed that this may be people not wishing to wait for 6 months before entering 24

the UK) and secondly cats/dogs that have been placed in quarantine because they have failed the requirements for PETS. For the purposes of this risk assessment it was assumed that all animals that enter UK quarantine remain there for the full 6 months. This is based on the assumption that the majority of PETS entries that fail and go to quarantine will remain in quarantine for a substantial period of time (nearing the full 6 month requirement), as short-term failures due to incorrect tick and tapeworm treatment/documentation can be corrected for at the border therefore only more serious failures will enter quarantine (Richard Ackroyd, Defra, pers. comm.). Furthermore, 6- month quarantine has previously been shown to be more risky than fully compliant PETS (Jones et al., 2005a) and therefore, if anything, this assumption is likely to provide a slight over-estimate of the risk of rabies entry to the UK. A number of assumptions have to be made in order to effectively model pet entries to the UK via PETS and the potential change in risk for these entries under the EUPMP. Firstly it is assumed that all animals enter the UK immediately after completion of the specified waiting period after the initial vaccination. A second, linked, assumption is that all of the entries considered in the model are first-time entries to the UK, i.e. companion animals that have previously entered the UK do not return at a later date. This allows all animals to be modelled using the first assumption, thereby giving a specified time period after which, if clinical signs have not been displayed, an infected animal could introduce rabies to the UK. Table 12: Maximum annual number of cat and dog entries to the UK through quarantine and PETS between 2005 and 2009 (NI,j) EU MSs

Listed 3rd

Unlisted 3rd

PETS (%)

Quarantine (%)

PETS (%)

Quarantine (%)

Quarantine (%)

Group 1

13434 (97%)

423 (3%)

2657 (92%)

217 (8%)

22 (100%)

Group 2

19830 (98%)

403 (2%)

859 (88%)

121 (12%)

51 (100%)

Group 3

1740 (97%)

52 (3%)

5091 (86%)

819 (14%)

1222 (100%)

Table 13: The maximum number of cat and dog entries into the UK from each country group and PETS region between 2005 and 2009

Group Country group Overall

EU MSs & Equivalents Group Group Group 1 2 3 13857 20233 1792

Listed third Countries Group Group Group 1 2 3 2874 980 5910

Unlisted third Countries Group Group Group 1 2 3 22 51 1222

35882

9764

1295

46941

4.5. Scenario and uncertainty analysis 4.5.1. Scenario analysis

In addition to an uncertainty analysis (see Section 4.5.2), a number of other scenarios were tested. Within a scenario analysis, the impact that a single parameter has on model results is investigated and also the impact of ‘what if’ scenarios (e.g. what would be the change in risk to the UK if the annual number of companion animal entries were to double?). The following section describes in detail the analyses that were tested and the reasons why each scenario was selected. A summary of the

25

scenarios implemented can be found in Table 14. For completeness the compliance scenarios described in Section 4.3.8.4 are also included in Table 14. The legislation for the EUPMP does not stipulate a requirement for the border checking of pet movements between EU member states. Consequently, if the UK were to adopt the EUPMP, companion animals from EU MSs the current checking levels may not be sustained. To provide insight into whether checking of entries from EU MSs under EUPMP would have an impact on the risk of rabies introduction, a scenario was run where these entries were checked, as opposed to the baseline model in which entries from EU MSs were not checked and 100% compliance with all regulations was assumed. This scenario is implemented in the model by assuming Pc = 1 for companion animals entering from EU MSs under EUPMP, as opposed to the baseline model where it was assumed that Pc = 0 for these entries. Since the UK introduced PETS, the annual number of companion animals entering the UK has greatly increased. The Kennedy report stated that in 1996, 7,267 cats and dogs were quarantined in the UK (Advisory Group on Quarantine, 1998). In 2009, 30,268 cats and dogs entered the UK via PETS and a further 1,099 came in through quarantine (not including returning UK pets). This increase in entry numbers could be because it is now easier and cheaper for an owner to bring their animal to the UK, i.e. it can be vaccinated and kept at home rather than having to enter UK quarantine for 6 months on entry. If the UK were to adopt the EUPMP, further simplification of the requirements could lead to entry numbers increasing further. To test the effect of this trend continuing, a scenario analysis was conducted where the entry numbers for each country group (NI,j) were increased in a stepwise manner above the current level (+25%, +50% etc). In addition to this a separate analysis was performed in which only entry numbers from unlisted countries increased, as it was speculated that entries from EU MSs and listed countries may not significantly increase in the future under the EUPMP because for these animals the potential change in policy represents only a further simplification of the requirements, rather than a switch from 6 month quarantine. To qualify to become a listed third country, a country must prove, along with other requirements, that it “complies with certain conditions relating to notification, monitoring, veterinary services, prevention and control of rabies and regulation of vaccines” (EU regulation 998/2003), thus allowing it to adequately meet the requirements of the travel scheme. Under the EUPMP, companion animals could enter the UK from any unlisted country in the world via a process similar to the current PETS. A number of these countries may have inadequate veterinary structures, and may use locally produced vaccines to avoid the high cost of importing vaccines from foreign companies. Experimental studies have questioned the quality of locally produced rabies vaccines, with as few as 10% of vaccinated animals receiving protective immunity after vaccination (Hu et al., 2008). Therefore a scenario analysis was run in which only 10% of vaccinations confer protection against rabies in unlisted third countries under the EUPMP, i.e. PNP = 0.9. Other than reduced vaccine efficacy, 100% compliance with the regulations was assumed. As discussed previously (see Section 4.3.1), there are a number of reasons why neither the rabies incubation periods from experimentally infected cases nor naturally infected cases can be used as accurate estimates for this parameter. In the baseline

26

model all of the data were combined to give an estimate of the average incubation period of rabies. This assumption was tested in a scenario analysis where only one of the data sets (i.e. experimental or natural infection) was used to generate the estimate, thus showing the effect on the overall risk of rabies entry to the UK if the true incubation period was that suggested by the data on natural infection or experimental infection. A number of scenarios were run to analyse the effect of changing the prevalence of rabies in the countries from which dogs/cats enter the UK. Firstly a scenario was undertaken in which the probability of an animal from a Group 1 country being infected was 0. This analysis tested the effect of applying the Gamma(1,1) distribution to Group 1 countries which had the effect of these countries still presenting some, if small, risk of rabies entry to the UK. Under the alternative scenario no risk was associated with companion animal entries from Group 1 countries. Secondly the total number of rabies cases in Group 2 and 3 countries were increased doubled and tripled. This scenario analysis addresses both the effect of potential under-reporting of rabies cases and a change (worsening) in the worldwide situation of rabies. Note, in these scenarios the number of rabies cases in Group 1 countries does not increase. A third scenario analysis with respect to prevalence was run in which only the rabies cases in unlisted third countries were increased (by multiples of five and ten). This analysis was run based on anecdotal evidence suggesting that the under-reporting of rabies cases in many of the unlisted third countries may be widespread and significant. At this stage the only parameters that have not been considered in a scenario analysis are those relating to the probability of not being protected after vaccination (PNP) and the specificity of the serological tests (PST+). Therefore, in order to test the effect of change in these parameters on the model outputs, scenarios analyses were run in which these parameters were set to values above and below the mean, but within a realistic range. Finally, a number of analyses were run in which sections of PETS were removed or adjusted to analyse which elements (if any) of the current regulations have a strong effect on the risk of rabies entry to the UK. Firstly a scenario was run in which the serological test requirement was removed. Subsequent analyses tested the effect of a reduction in the waiting period in the country of origin, from 212 days in the baseline model to 1 day in reductions of 30 days. This was done either individually or in combination with the removal of the serological test. 4.5.2. Uncertainty analysis An uncertainty analysis was conducted using the software @Risk (© Palisade) Version 4.5. This analysis outputs regression coefficients for each distribution in the model, thus allowing identification of the parameters the model results are most sensitive to (or which parameters contribute the most uncertainty to the output uncertainty).

27

Table 14: A summary of the scenarios that were tested in the scenario analysis and the modified parameter values under each scenario. Parameter Compliance parameters (PV, PST, PC)

Scenario 1

Value x0.9

2

x0.8

Checking entries from EU MSs (PC) Number of imports (NI,j)

1

PC set to 1 for EU MSs under EUPMP

1 2 3 4 5 6 7 8 1

x1.25 for all groups (j = 1,…,9) x1.5 for all groups (j = 1,…,9) x1.75 for all groups (j = 1,…,9) x2 for all groups (j = 1,…,9) x1.25 for all unlisted third country groups (j = 7, 8, 9) x1.5 for all unlisted third country groups (j = 7, 8, 9) x1.75 for all unlisted third country groups (j = 7, 8, 9) x2 for all unlisted third country groups (j = 7, 8, 9) Set to 0.9 for all unlisted third country groups (j = 7, 8, 9)

1

Lognormal distribution with mean 39.7 days and standard deviation 41.9 days Lognormal distribution with mean 23.7 days and standard deviation 15 days 0 for country group j (j = 1, 4, 7)

Probability a vaccinated companion animal is not protected (PNP) Incubation period (IP)

2 Probability companion animal from Group j is infected / becomes infected (PI,j, PI*,j)

1

2 3 4

Used to assess the change in risk for quarantine & PETS and EUPMP schemes Used to assess the change in risk from unlisted countries under EUPMP Used to assess the change in risk if poorly effective vaccines are used in unlisted third countries Natural incubation period Experimental incubation period Used to assess the change in risk if it is assumed that there are no infected animals present in Group 1 countries Used to assess the change in risk if there was an increase in the number of rabies cases in (2-3) Group 2 and 3 countries and (4-5) unlisted third countries

1 2 3

Reported cases x2 for country group j (j = 4,…,9) Reported cases x3 for country groups j (j = 4,…,9) Reported cases x5 for all unlisted third country groups (j = 7, 8, 9) Reported cases x10 for all unlisted third country groups (j = 7, 8, 9) Set to 1 for all country groups (j = 1,…,9) Set to 0.9 for all country groups (j = 1,…,9) Set to 0.5 for all country groups (j = 1,…,9)

1 2 3

Set to 0.05 for all country groups (j = 1,…,9) Set to 0.1 for all country groups (j = 1,…,9) Set to 0.2 for all country groups (j = 1,…,9)

Used to assess the effect of the parameter PST+ on the final results. Mean baseline value = 0.032

1

Set to 1 in PETS pathway

Used to assess the change in risk when there is no serological test in the baseline model for PETS

1 2 3 4 5 6

6 months (T = 180 days) for PETS 5 months (T = 150 days) for PETS 4 months (T = 120 days) for PETS 3 months (T = 90 days) for PETS 2 months (T = 60 days) for PETS 1 month (T = 30 days) for PETS

Used to assess the change in risk when the waiting period is reduced for PETS

5 Probability a vaccinated companion animal is not protected (PNP) Probability of companion animal passing the serological test (PST+) Probability of companion animal passing the serological test (PST+) Waiting period (T)

Additional comments Used to test the effect of reduced compliance on the risk of rabies entry to the UK. If not 100% compliant the waiting period is assumed to be 1 day.

28

Used to assess the effect of the parameter PNP on the final results. Mean baseline value = 0.814.

5. Results, Uncertainty Analysis & Scenario Analysis 5.1. Baseline results (100% compliance)

The results of the baseline risk assessment for 6 month quarantine/PETS (current scheme) compared to EUPMP, obtained from 50,000 iterations of the model are presented in this section. The results presented follow the standard form of the arithmetic mean and the 5th and 95th percentile values. Accordingly the latter represent the range of values for which we are 90% certain that the true value lies between, these values represent uncertainty only, variability is not considered in the model. The mean values and 5th and 95th percentiles are given for (i) the annual probability of an infected cat or dog entering the UK, (ii) the number of years between rabies entry and (iii) the annual probability of a single infected cat or dog entering the UK (Table 15, Table 16 and Table 17 respectively). The baseline model assumes 100% compliance with all regulations of the respective pet movement policies and therefore no animals enter the UK via a non-compliant route in this scenario. The results of the baseline model indicate that for the current scheme of 6 month quarantine and PETS, the mean annual probability of an infected cat or dog entering the UK is 7.79 x 10-5 with 5th and 95th percentiles of 5.90 x 10-5 and 1.06 x 10-4 respectively. If the UK were to adopt the EUPMP then it is predicted that this mean probability will increase to 4.79 x 10-3 (4.05 x 10-3, 5.65 x 10-3). For the number of years between rabies entries it is predicted that one infected entry will occur every 13272 (9408, 16940) years through the current system of quarantine and PETS and every 211 (177, 247) years through the EUPMP. This represents a 60-fold increase in the overall mean risk of rabies entry to the UK. Within each country classification (EU MSs & Equivalents, listed third countries, unlisted third countries), Group 1 gives the lowest risk of importing an animal infected with rabies and Group 3 gives the highest risk. This result is to be expected as the group definitions are based on reported rabies cases in each country. The mean risk of importing an infected animal under the current scheme is highest from unlisted third countries, 4.31 x 10-5 (3.91 x 10-5, 4.82 x 10-5), whereas in the EUPMP the greatest risk is associated with companion animal movements from listed third countries, 2.75 x 10-3 (2.39 x 10-3, 3.15 x 10-3). A change in policy to the EUPMP is predicted to reduce the probability of rabies entry from unlisted third countries to 2.58 x 10-5 (9.47 x 10-6, 5.05 x 10-5) and increase the risk of entry from EU Member States and listed third countries from 8.34 x 10-6 (3.43 x 10-6, 1.60 x 10-5) to 2.02 x 10-3 (1.41 x 10-3, 2.79 x 10-3) and 2.64 x 10-5 (1.32 x 10-5, 4.69 x 10-5) to 2.75 x 10-3 (2.39 x 10-3, 3.15 x 10-3) respectively. For unlisted third countries this represents approximately a two-fold decrease in the mean risk, while for EU member states and listed third countries this represents a 240-fold and a 100-fold increase, respectively.

29

Table 15: The annual probability of at least one infected cat or dog entering the UK, assuming 100% compliance Risk group

EU MSs & Equivalents

Listed third Countries

Unlisted third Countries

Scenario Group 1

Group 2

Group 3

Group 1

Group 2

Group 3

Group 1

Group 2

-7

-6

-6

-7

-7

-5

-6

-6

Group 3

Current

2.65x10 (4.02x10-8, 6.82x10-7)

2.66x10 (8.91x10-7, 5.61x10-6)

5.42x10 (2.30x10-6, 1.02x10-5)

1.13x10 (1.71x10-8, 2.88x10-7)

3.28x10 (3.05x10-8, 9.12x10-7)

2.60x10 (1.28x10-5, 4.64x10-5)

1.43x10 (7.36x10-8, 4.30x10-6)

1.82x10 (9.34x10-8, 5.46x10-6)

3.98x10-5 (3.73x10-5, 4.25x10-5)

EUPMP

1.54x10-4 (1.01x10-5, 4.56x10-4)

4.87x10-4 (1.20x10-4, 1.09x10-3)

1.38x10-3 (1.00x10-3, 1.80x10-3)

4.57x10-5 (3.01x10-6, 1.36x10-4)

1.32x10-4 (1.02x10-5, 3.71x10-4)

2.57x10-3 (2.26x10-3, 2.90x10-3)

2.08x10-7 (2.68x10-8, 5.74x10-7)

2.64x10-7 (3.47x10-8, 7.31x10-7)

2.53x10-5 (9.25x10-6, 4.97x10-5)

Group

Current

8.34x10-6 (3.43x10-6, 1.60x10-5)

2.64x10-5 (1.32x10-5, 4.69x10-5)

4.31x10-5 (3.91x10-5, 4.82x10-5)

EUPMP

2.02x10-3 (1.41x10-3, 2.79x10-3)

2.75x10-3 (2.39x10-3, 3.15x10-3)

2.58x10-5 (9.47x10-6, 5.05x10-5)

Country group

Current

7.79x10-5 (5.90x10-5, 1.06x10-4)

EUPMP

4.79x10-3 (4.05x10-3, 5.65x10-3)

Overall

Table 16: The predicted number of years between rabies entry to the UK, assuming 100% compliance Risk group

EU MSs & Equivalents

Listed third Countries

Unlisted third Countries

Scenario Group 1

Group 2

Group 3

Group 1

Group 2

Group 3

Group 1

Group 2

Group 3

Current

8505859 (1464916, 24857680)

514925 (178135, 1121533)

227094 (97539, 435159)

19787170 (3475837, 58571870)

9305773 (1096580, 32770500)

44872 (21535, 78239)

8155119 (232693, 13587890)

6099551 (183264, 10702060)

25143 (23544, 26837)

EUPMP

29001 (2191, 98783)

3207 (917, 8303)

749 (555, 997)

95384 (7374, 332444)

33050 (2693, 98461)

391 (344, 443)

12147290 (1743382, 37355940)

9396825 (1368282, 28845500)

51465 (20126, 108107)

Group

Current

149129 (62683, 291248)

43942 (21299, 75973)

23302 (20738, 25557)

EUPMP

517 (359, 708)

366 (317, 419)

50440 (19792, 105590)

Country group

Current

13272 (9408, 16940)

EUPMP

211 (177, 247)

Overall

30

Table 17: The probability of single infected cat or dog entering the UK, assuming 100% compliance Risk group

EU MSs & Equivalents

Listed third Countries

Unlisted third Countries

Scenario Group 1 -10

Current

2.00x10 (1.76x10-11, 5.80x10-10)

EUPMP

1.11x10-8 (7.30x10-10, 3.29x108 )

Group

Group 2

Group 3

Group 1

-8

-10

5.03x10-10 (1.62x10-10, 1.04x109 )

1.50x10 (1.08x10-8, 1.98x10-8)

2.41x10-8 (5.95x10-9, 5.39x10-8)

7.69x10-7 (5.60x10-7, 1.01x10-6)

Group 2

Group 3

Group 1

Group 2

-8

-8

-8

2.91x10 (2.58x10-11, 8.42x10-10)

1.10x10 (8.24x10-9, 1.51x10-8)

6.52x10 (3.34x10-9, 1.95x10-7)

3.57x10 (1.83x10-9, 1.07x10-7)

3.26x10-8 (3.05x10-8, 3.48x10-8)

1.59x10-8 (1.05x10-9, 4.72x10-8)

1.34x10-7 (1.04x10-8, 3.79x10-7)

4.36x10-7 (3.83x10-7, 4.92x10-7)

9.45x10-9 (1.22x10-9, 2.61x10-8)

5.18x10-9 (6.80x10-10, 1.43x108 )

2.07x10-8 (7.57x10-9, 4.07x10-8)

Current

1.57x10-8 (1.15x10-8, 2.06x10-8)

1.37x10-8 (9.64x10-9, 1.95x10-8)

1.34x10-7 (4.96x10-8, 2.78x10-7)

EUPMP

8.05x10-7 (5.94x10-7, 1.04x10-6)

5.86x10-7 (4.40x10-7, 8.37x10-7)

3.53x10-8 (1.25x10-8, 7.13x10-8)

Country group

Group 3

2.40x10-9 (1.90x10-10, 6.75x109 )

Current

1.63x10-7 (7.87x10-8, 3.08x10-7)

EUPMP

1.43x10-6 (1.15x10-6, 1.77x10-6)

Overall

31

Figure 6: Box plots comparing the annual risk of rabies entry to the UK under the current scheme to the risk of entry through the EUPMP for (top left) EU member states (top right) listed third countries (bottom left) unlisted third countries and (bottom right) all regions combined. Note: all y-axes are the log10 of the annual probability of rabies entry to the UK.

The baseline model results for each country classification and movement policy are also represented in a box plot format (Figure 6). From this it can be seen that a change in policy to the EUPMP greatly increases the risk of rabies entry from EU MSs and listed third countries, whereas the decrease in risk from unlisted third countries is comparatively small. Furthermore, there is a large amount of uncertainty associated with the results for unlisted countries under the EUPMP compared to quarantine. Therefore, while the mean values suggest that the EUPMP is less risky than quarantine for entries from unlisted third countries, when the large amount of uncertainty with the EUPMP is taken into account, this conclusion is less certain. The box plots for the overall results show a similar pattern to the results from EU MSs and listed third countries, implying that companion animal entries from these two sets of countries contribute most to the change in the overall risk of rabies entry under the EUPMP. In the risk pathway for UK PETS (see Figure 4) there are 12 potential scenarios via which an infected animal can enter the UK. Figure 7 shows which scenarios contribute the most risk for a rabies entry through PETS in the baseline model. Under this fully compliant scenario, as expected, only two pathways contribute to the overall risk. Of these, the pathway where an unprotected animal becomes infected during the waiting period contributes approximately 94% of the overall risk. Note that for quarantine there is only one branch in the risk pathway, therefore by definition it contributes 100% of the risk of rabies entry.

32

Figure 7: The percentage contribution of each risk pathway to the overall risk of rabies entry through PETS in the baseline model

Figure 8 and Figure 9 show similar diagrams for the percentage contribution of the routes of rabies entry in the EUPMPunlisted and EUPMPEU&listed pathways for the baseline model, respectively. Interestingly, the percentage contribution of routes for the EUPMPunlisted pathway, despite being the same risk pathway, is different from the results of the PETS model. In the EUPMPunlisted scheme with 100% compliance 35% of the risk comes from the scenario where the animal was infected before vaccination with 65% of the risk comes from the scenario where the pet becomes infected during the waiting period, comparatively less important than the same pathway in the PETS model). This is likely to be due to the shorter waiting period of the EUPMPunlisted scheme, meaning that fewer infected animals will display clinical signs during this period and fewer will become infected during the process. For the EUPMPEU&listed pathway the percentage contribution is different between EU MSs and listed third countries, as entries from EU MSs are not checked whereas entries into the EU from listed third countries are. Under this policy the vast majority of the risk comes through the route where the animal is infected before vaccination, 93% for entries from EU MSs and 89% for listed third countries. Again, this route is contributing a much higher percentage of the risk as the waiting period for listed countries is even shorter than for unlisted countries and therefore there is a smaller probability of a companion animal becoming infected during this period.

33

Figure 8: The percentage contribution of each risk pathway to the overall risk of rabies entry through EUPMPunlisted in the baseline model

Figure 9: The percentage contribution of each risk pathway to the overall risk of rabies entry through EUPMPEU&listed in the baseline model

34

5.2. Investigation of compliance

To investigate the effect of compliance on the risk of rabies entry to the UK a scenario analysis was performed in which the compliance for a number of the requirements (vaccination, serological testing and checking) was reduced simultaneously to various levels below 100% (see Table 14). As shown in Figure 10, the annual risk of rabies entry to the UK under the current systems of PETS and quarantine increases when non-compliance is included in the model. This trend is also seen in the number of years between rabies entry (Figure 11), which decrease when the level of compliance decreases. In these non-compliance scenarios the risk of rabies entry greatly increases as soon as non-compliance is introduced. This is because of the assumption that the non-compliant animals will be moved after 1 day, effectively meaning there is no waiting period and that any animal incubating rabies will not display clinical signs until after they have arrived in the UK. Due to the nature of this assumption it is inappropriate to model the effect of decreasing only a single compliance parameter, as the results are so strongly affected by the movement after one day that a difference cannot be seen between the effects of decreasing either the vaccination or serological test compliance (analyses not shown). It should be noted that when modelling non-compliance in the current scheme this only applies to entries through the PETS scheme. Entries from unlisted third countries are assumed to be unaffected as these animals come in through quarantine and are vaccinated in the UK, in accordance with quarantine regulations. This assumption is also applied to the small number of entries through quarantine from EU MSs and listed third countries. 6.00E-03

Annual risk of rabies entry

5.00E-03

4.00E-03

3.00E-03

2.00E-03

1.00E-03

0.00E+00 100%

90%

80%

Compliance level

Figure 10: The change in the mean probability of rabies entry to the UK under the current system of PETS and quarantine, with varying levels of compliance. Error bars represent the 5th and 95th percentiles.

35

35000

Number of years between rabies entry

30000

25000

20000

15000

10000

5000

0 100%

90%

80%

Compliance level

Figure 11: The mean number of years between rabies entry to the UK under the current system of PETS and quarantine, with varying levels of compliance. Error bars represent the 5th and 95th percentiles.

A similar trend is seen in the EUPMP model (Figure 12 and Figure 13), where an increase in risk is also predicted as compliance is reduced. However, it should be noted that in absolute terms, the increase in risk is much lower for EUPMP than it is for quarantine and PETS. This is because within the EUPMP there are fewer areas where non-compliance can be introduced. For example, in the EUPMP scheme for listed third countries and member states there is no requirement for a companion animal to be serologically tested for rabies antibodies before movement between countries. Therefore, a change in the compliance with the serological test will only affect entries from unlisted third countries, and hence contribute a smaller change in risk than a similar change in compliance under the current scheme (which would affect entries from both EU MSs and listed third countries). Also, companion animal movements between EU MSs are not checked, so a change in this parameter only affects the risk of rabies entry from third countries to the UK. This explains why the overall increase in risk for the non-compliant scenarios under the EUPMP is less than for the current movement policy of PETS and quarantine. 1.6E-02

1.4E-02

Annual risk of rabies entry

1.2E-02

1.0E-02

8.0E-03

6.0E-03

4.0E-03

2.0E-03

0.0E+00 100%

90%

80%

Compliance level

Figure 12: The change in the mean probability of rabies entry to the UK under the EUPMP, with varying levels of compliance. Error bars represent the 5th and 95th percentiles.

36

500

Number of years between rabies entry

450 400 350 300 250 200 150 100 50 0 100%

90%

80%

Compliance level

Figure 13: The mean number of years between rabies entry to the UK under the EUPMP, with varying levels of compliance. Error bars represent the 5th and 95th percentiles.

The percentage contribution of each route in the PETS risk pathway for a noncompliant scenario is shown in Figure 14. The values are taken from the scenario where all three compliance parameters are reduced to 90%. In these scenarios, as would be expected, several potential routes of rabies entry to the UK through PETS contribute some risk. In this case, two of the pathways contribute the majority of the overall risk, the pathway where an infected animal is vaccinated but not tested and the pathway where an infected animal is not vaccinated or tested (41% and 45.5% respectively). Only a relatively minor proportion of the risk is from the non-compliant pathways due to the assumption that any non-compliant animal will enter the UK immediately with false-documentation. Almost identical results are seen in the EUPMPunlisted scheme (Figure 15), suggesting that the effect of changing the waiting period (from 6 to four months) for compliant animals has little effect on the overall risk when a substantial proportion (10%) of animals are not vaccinated or serologically tested. Unlike the PETS and EUPMPunlisted schemes, the percentage contribution for the EUPMPEU&listed scheme under reduced compliance are more similar to the results seen in the 100% scenario (Figure 16). The majority of the risk still comes from the scenario in which the animal is infected before vaccination and is therefore not protected from rabies. However, some risk (13.3%) comes from the non-compliant pathway in which the infected companion animal is not vaccinated but enters with false-documentation. Again, only a tiny percentage of the risk comes through routes in which an uninfected animal becomes infected during the waiting period. This is likely to be because of the short waiting period (21 days) leading to a very small probability of an uninfected animal becoming infected during the waiting period.

37

Figure 14: The percentage contribution of each risk pathway to the overall risk of rabies entry through PETS when 90% compliance is assumed. Note all values have been rounded to 1dp

Figure 15: The percentage contribution of each risk pathway to the overall risk of rabies entry through EUPMPunlisted when 90% compliance is assumed. Note all values have been rounded to 1dp

38

Figure 16: The percentage contribution of each risk pathway to the overall risk of rabies entry through EUPMPEU&listed when 90% compliance is assumed. Note all values have been rounded to 1dp

5.3. Other scenario analyses

Table 14 summarises the scenario analyses considered. For all of the scenarios investigated below 100% compliance was assumed. 5.3.1. Checking entries from EU MSs under the EUPMP

Under the EUPMP there is no requirement for the checking of companion animal movements between EU member states. This was represented in the baseline EUPMP model by giving a probability of 0 to the checking of companion animal entries from EU member states under the EUPMP (PC). To test the effect of potential border checking of entries a scenario was run in which all entries from EU MSs were subject to a documentation check, and the results of this were compared to the baseline model. As indicated by Table 18 below, implementing the checking of companion animal entries from EU MSs under the EUPMP is predicted to have only a slight effect on the risk of rabies entry to the UK, with the number of years between rabies entries reducing from 211 years to 231 years.

39

Table 18: The change in the risk of rabies entry to the UK under the EUPMP when documentation checks are conducted on companion animal entries from EU MSs Annual probability of Number of years rabies entry between rabies entries 211 4.79x10-3 Baseline (177, 247) (4.05x10-3, 5.65x10-3) 225 4.48x10-3 Scenario (189, 263) (3.81x10-3, 5.29x10-3)

5.3.2. Increase in the number of imports

Figure 17, Figure 18 and Table 19 show that, unsurprisingly, increasing the number of imports causes an increase in the risk of rabies entry. The model results also indicate that if the UK were to adopt the EUPMP and entry numbers were to double there could be a rabies entry into the UK within 120 years. It may be more realistic to assume that companion animal entries into the UK under the EUPMP would only significantly increase from unlisted third countries that currently have to go into 6 month quarantine. As shown in Figure 18, an increase in the number of companion animal entries from unlisted third countries does not substantially increase the overall risk of rabies entry to the UK under the EUPMP. This is because the majority of the risk under the EUPMP comes from entries from EU MSs and listed third countries. 4.0E-04

Annual probability of rabies entry

3.5E-04

3.0E-04

2.5E-04

2.0E-04

1.5E-04

1.0E-04

5.0E-05

0.0E+00 Current imports

+25%

+50%

+75%

+100%

Import level

Figure 17: The potential change in the annual risk of rabies entry to the UK if imports numbers from all countries were to increase under the current system of PETS and quarantine. Error bars represent the 5th and 95th percentiles.

40

1.20E-02

Annual probability of rabies entry

1.00E-02

(i) Increase in entries from all countries

8.00E-03

6.00E-03

(ii) Increase in entries from unlisted countries only

4.00E-03

2.00E-03

0.00E+00 Current imports

+25%

+50%

+75%

+100%

Import level

Figure 18: The potential change in the annual risk of rabies entry to the UK under the EUPMP if imports numbers from all countries were to (i) increase for all countries and (ii) for unlisted third countries only. Table 19: The increase in the annual probability of rabies entry to the UK under various scenarios if import numbers were to increase Entry scheme

Current -5

Import level +50%

+25% -5

-4

+75%

+100% -4

PETS/Quarantine

7.79x10 (5.90x10-5, 1.06x10-4)

9.73x10 (7.37x10-5, 1.33x10-4)

1.17x10 (8.85x10-5, 1.60x10-4)

1.36x10 (1.03x10-4, 1.86x10-4)

1.56x10-4 (1.18x10-4, 2.13x10-4)

EUPMP

4.79x10-3 (4.05x10-3, 5.65x10-3)

5.98x10-3 (5.06x10-3, 7.06x10-3)

7.17x10-3 (6.08x10-3, 8.48x10-3)

8.37x10-3 (7.08x10-3, 9.87x10-3)

9.55x10-3 (8.08x10-3, 1.13x10-2)

EUPMP with increase from unlisted thirds only

4.79x10-3 (4.05x10-3, 5.65x10-3)

4.80x10-3 (4.05x10-3, 5.66x10-3)

4.80x10-3 (4.07x10-3, 5.67x10-3)

4.81x10-3 (4.07x10-3, 5.69x10-3)

4.81x10-3 (4.07x10-3, 5.69x10-3)

5.3.3. Potential poor vaccine efficacy in unlisted third countries

Based on the results of experimental studies indicating that some vaccines used in unlisted third countries have reduced efficacy, a scenario was run where only 10% of animals vaccinated in unlisted third countries would be protected from rabies, i.e. the probability an animal is not protected after vaccination (PNP) is set to 90%. These results indicate that there would be a large increase in the risk of rabies entry from unlisted third countries if vaccines with poor efficacy were used in those countries (one entry every 11096 years as opposed to every 50440 years), see Table 20. This is also an increased risk when compared to the current scheme where companion animal entries from unlisted third countries through quarantine (where one entry was expected every 23302 years). Comparing the results for the overall risk of rabies entry (Table 21), it can be concluded that the use of poorly effective vaccines in unlisted third countries has only a minimal impact on the overall risk of rabies entry under the EUPMP. However it is important to note that this is still assuming 100% compliance and that the specificity (Sp) of the test is the same as in EU and listed third countries (mean value of 97%), a reasonable assumption given that the blood tests still have to be performed in labs approved by the EU (EU regulation 998/2003). If this was also reduced then it is likely that the risk of rabies entry would increase.

41

Table 20: Risk of rabies entry from unlisted third countries under the EUPMP with varying vaccine efficacy. The baseline Quarantine results have been included for reference. Entry from unlisted third countries Scenario Probability of Rabies Entry Years between rabies entries 23302 4.31x10-5 Quarantine (20738, 25557) (3.91x10-5, 4.82x10-5) 50440 2.58x10-5 EUPMP (19792, 105590) (9.47x10-6, 5.05x10-5) 11096 1.06x10-4 Reduced vaccine (5517, 20713) (4.83x10-5, 1.81x10-4) efficacy (EUPMP) Table 21: The overall risk of rabies entry to the UK through the current scheme, the EUPMP and the EUPMP when vaccine efficacy in unlisted third countries is assumed to be very low. The baseline PETS/Quarantine results have been included for reference. Probability of Rabies Entry Years between rabies entries 13272 7.79x10-5 PETS/Quarantine (9408, 16940) (5.90x10-5, 1.06x10-4) 211 4.79x10-3 EUPMP (177, 247) (4.05x10-3, 5.65x10-3) 207 4.87x10-3 Reduced efficacy (174, 242) (4.13x10-3, 5.74x10-3)

5.3.4. Change in incubation period

A scenario analysis was performed in which the data on incubation periods are separated into experimentally and naturally infected cases to derive two different estimates of the incubation period (see Table 14). The results from this analysis for PETS/quarantine and the EUPMP are displayed in Table 22 and Table 23 respectively. As expected, when experimental (shorter) incubation periods are used the probability of rabies entry decreases and the number of years between entries increases. This is because more animals will display clinical signs before movement and will therefore be removed from the scheme. Conversely, using an estimate of the natural incubation period increases the risk of rabies entry to one entry every 8874 years under the current scheme, and for the EUPMP the entry rate increases to one entry every 175 years. While these results are somewhat unsurprising, the use of an incubation period estimated only from naturally infected cases, given all the data available, predicts a higher risk. Consequently the baseline model may give a slight under-estimate of the true risk. Table 22: The change in the risk of rabies entry through PETS and quarantine when different incubation periods are assumed Probability of Years between rabies Scenario rabies entry entries 13272 Baseline 7.79x10-5 (9408, 16940) incubation period (5.90x10-5, 1.06x10-4) 8874 Natural incubation 1.15x10-4 (6749, 10781) period (9.28x10-5, 1.48x10-4) 72237 Experimental 1.84x10-5 (26930, 154675) incubation period (6.46x10-6, 3.71x10-5)

42

Table 23: The change in the risk of rabies entry through the EUPMP when different incubation periods are assumed Probability of Years between rabies Scenario rabies entry entries 211 Baseline 4.79x10-3 (177, 247) incubation period (4.05x10-3, 5.65x10-3) 175 Natural incubation 5.78x10-3 (148, 203) period (4.93x10-3, 6.76x10-3) 332 Experimental 3.05x10-3 (269, 400) incubation period (2.50x10-3, 3.72x10-3)

5.3.5. Prevalence Scenarios

The probability that a dog or cat from Group j is infected, PI,j (j = 1,…,9), is a highly uncertain parameter, because of uncertainty associated with the true number of rabies cases in each country (due to under-reporting) and relatively poor information on the companion animal population in each country. Therefore, a number of scenarios were considered to assess the effect this parameter has on the overall risk of rabies entering the UK from imported pets. 5.3.5.1. Zero cases of rabies for Group 1 countries

To investigate the effect that the Gamma distribution has for this particular grouping on the overall risk of rabies introduction to the UK, the probability of a companion animal being infected for Group 1 countries was set to zero, with no associated uncertainty. Consequently, the annual probability of an infected pet entering the UK decreases from a mean value of 7.79 x 10-5 (5.90 x 10-5, 1.06 x 10-4) to 7.61 x 10-5 (5.7 5 x 10-5, 1.04 x 10-4) for the current scheme (quarantine and PETS) and from 4.79 x 10-3 (4.05 x 10-3, 5.65 x 10-3) to 4.59 x 10-3 (3.90 x 10-3, 5.41 x 10-3) for EUPMP. The number of years between an infected pet entering the UK increases from 13272 (9408, 16940) to 13600 (9580, 17384) for the current scheme and from 211 (177, 247) to 220 (185, 257) for EUPMP. These results indicate that the uncertainty associated with the probability of an imported companion animal being infected from Group 1 has very little impact on the overall risk of rabies entry for both the current scheme (quarantine & PETS) and EUPMP. It can also be concluded that companion animal movements from Group 1 countries contribute very little to the overall risk of rabies entry to the UK, which is to be expected. 5.3.5.2. Rabies cases are doubled and trebled for Group 2 and Group 3 countries

To investigate the potential impact of under-reporting the number of reported rabies cases for Group 2 and 3 countries was doubled and trebled to assess the increase in risk of importing a pet infected with rabies. This scenario also tests the effect of a change in the worldwide situation of rabies. The results are presented below (Table 24 and Table 25). There is a linear increase in the annual probability of rabies entry when the number of rabies cases is increased. This highlights that the results of the QRA are sensitive to the reporting of rabies in different countries and that a world-wide increase in the number of rabies cases would increase the risk of rabies entry into the

43

UK. Conversely, a reduction in world-wide rabies cases in pet cats/dogs would reduce the risk to the UK. Table 24: Annual probability of rabies entry for increasing number of rabies cases Annual probability of entry Number of cases Quarantine & PETS EUPMP 4.79x10-3 7.79x10-5 Baseline (5.90x10-5, 1.06x10-4) (4.05x10-3, 5.65x10-3) -4 9.36x10-3 1.54x10 Doubled -4 -4 (1.17x10 , 2.11x10 ) (7.94x10-3, 1.10x10-2) -4 1.39x10-2 2.30x10 Trebled -4 -4 (1.74x10 , 3.15x10 ) (1.18x10-2, 1.63x10-2) Table 25: Number of years between rabies entry for increasing number of rabies cases Number of years between entry Number of cases Quarantine & PETS EUPMP 13272 211 Baseline (9408, 16940) (177, 247) 6716 108 Doubled (4748, 8567) (91, 126) 4497 73 Trebled (3174, 5740) (61, 85)

Figure 19: Box plots representing the change in the risk of rabies entry to the UK under PETS/Quarantine and the EUPMP when the number of rabies cases is increased. Note: all y-axes are the log10 of the annual probability of rabies entry to the UK.

As seen in the box plots representing the change in risk if the number of worldwide rabies cases were to increase (Figure 19), the effect on the overall risk of rabies entry to the UK is similar under both the current scheme and the EUPMP, i.e. there is an increase, with very similar patterns seen on both graphs. However it should be noted that even with three times the current number of reported cases, the annual risk of 44

entry under the current scheme is still much lower than it is under the EUPMP with the baseline number of rabies cases. 5.3.5.3. Number of rabies cases increased for only unlisted third countries

For the majority of unlisted third countries the data on rabies cases is comparatively poor when considered alongside the same data from EU MSs and listed third countries. Therefore to test the possibility that the vast majority of rabies cases in unlisted third countries go unreported scenarios were run where the number of rabies cases in these countries was increased by multiples of 5 and 10 respectively. As expected, the results shown in Table 26 and Table 27 indicate that if the number of rabies cases in unlisted third countries is actually five or ten times the number of cases currently reported then the risk of rabies entry to the UK from these countries would increase under both policies. However, as shown in previous scenarios in which modifications are made to the risk from unlisted third countries, the overall risk of entry through the EUPMP is not greatly affected by the increase in rabies cases in unlisted countries, a decrease from 211 to 201 years between entries when the number of cases is multiplied by 10. Comparatively the overall risk through the current scheme greatly increases with this scenario change, with the number of years between rabies entries decreasing from 13272 years to 2158 years. This is because in the current scheme the majority of the risk comes from rabies entries from unlisted countries when animals incubate rabies for longer than the 6 month quarantine, whereas in the EUPMP much more risk is associated with entries from EU member states and listed third countries. Table 26: The change in the annual probability of rabies entry to the UK from unlisted third countries when the number of rabies cases in unlisted countries is increased Annual probability of entry Number of cases Quarantine & PETS EUPMP 2.58x10-5 4.31x10-5 Baseline (3.91x10-5, 4.82x10-5) (9.41x10-6, 5.08x10-5) -4 1.29x10-4 2.15x10 x5 -4 -4 (1.96x10 , 2.41x10 ) (4.69x10-5, 2.55x10-4) -4 2.57x10-4 4.31x10 x10 -4 -4 (3.91x10 , 4.82x10 ) (9.51x10-5, 5.06x10-4) Table 27: The change in the number of years between rabies entry to the UK from unlisted third countries when the number of rabies cases in unlisted countries is increased Annual probability of entry Number of cases Quarantine & PETS EUPMP 23302 50440 Baseline (20738, 25557) (19792, 105590) 4661 10108 x5 (4149, 5112) (3925, 21335) 2331 5050 x10 (2076, 2556) (1977, 10513)

When the results are displayed in box plot format (Figure 20) it can be seen that while the risk of rabies entry to the UK is lower under the EUPMP, for this scenario the uncertainty associated with this result is considerably higher (the results are spread over a much wider range). This high level of uncertainty is likely to be due to the large amount of uncertainty associated with the input parameters for this scheme (particularly the rabies cases in unlisted countries). 45

Figure 20: Box plots representing the change in the risk of rabies entry to the UK from unlisted third countries under the current 6 month quarantine and the EUPMPunlisted scheme when the number of rabies cases is increased. Note: all y-axes are the log10 of the annual probability of rabies entry to the UK from unlisted third countries.

5.3.6. Protection after vaccination

In the baseline model, there is unquantified uncertainty relating to the probability that a vaccinated pet is not protected as this was estimated from a number of studies on the seroconversion of vaccinated animals above a threshold titre of 0.5 IU/ml. This parameter was therefore included in the sensitivity analysis (see Table 14). As expected, when the vaccine efficacy is increased from the value used in the baseline model the risk of rabies entry to the UK decreases. However, within the range of values in which we believe the true efficacy lies (81-100%, given that we believe the baseline value to most likely be an underestimate), the change in risk does not have a large effect on the model results, and all conclusions would remain unchanged, therefore suggesting the overall model results and conclusions made are not affected by the unquantified uncertainty associated with this parameter. Table 28: The change in the annual probability of rabies entry to the UK when the vaccine efficacy is altered Probability of rabies entry Percentage of uninfected pets protected after vaccination PETS/Quarantine EUPMP 4.28x10-3 5.13x10-5 100% (4.71x10-5, 5.66x10-5) (3.58x10-3, 5.12x10-3) -5 4.56x10-3 6.52x10 90% -5 -5 (5.57x10 , 7.69x10 ) (3.85x10-3, 5.40x10-3) -5 4.79x10-3 7.79x10 81.4% (baseline) -5 -4 (5.90x10 , 1.06x10 ) (4.05x10-3, 5.65x10-3) -4 5.64x10-3 1.21x10 50% -5 -4 (8.20x10 , 1.71x10 ) (4.93x10-3, 6.49x10-3)

46

Table 29: The change in the number of years between rabies entry to the UK when vaccine efficacy is altered Years between rabies entry Percentage of uninfected pets protected after vaccination PETS/Quarantine EUPMP 19568 236 100% (17683, 21240) (195, 279) 15496 222 90% (13002, 17953) (185, 260) 13272 211 81.4% (baseline) (9408, 16940) (177, 247) 8707 178 50% (5850, 12197) (154, 203)

5.3.7. Serological test specificity

The results from this scenario analysis suggest that under the current scheme of PETS and quarantine the annual risk of rabies entry to the UK is quite sensitive to the test specificity, with a reduction in the number of years between rabies entry to the UK from 13272 years in the baseline scenario for PETS/quarantine to 4640 years with 80% specificity. Comparatively, a reduction in the vaccine efficacy to 50% only reduced the number of years between rabies entry to 8707. The relatively small decrease in risk in the vaccination scenarios is likely to be due to the fact that a highly effective serological test was still assumed to be in place, meaning that the majority of animals with failed vaccinations would not be allowed to proceed in the scheme. Conversely, for the EUPMP the reduced test specificity has relatively little effect on the overall risk. This result is to be expected as only companion animals from unlisted third countries are subject to a serological test, and therefore entries from EU MSs and listed third countries are unaffected in this scenario. Table 30: The change in the annual probability of rabies entry to the UK when the serological test specificity is altered Probability of rabies entry Test Specificity PETS/Quarantine EUPMP 4.79x10-3 7.79x10-5 96.8 % (5.90x10-5, 1.06x10-4) (4.05x10-3, 5.65x10-3) (baseline) -5 4.80x10-3 9.34x10 95% -5 -4 (7.66x10 , 1.12x10 ) (4.06x10-3, 5.67x10-3) -4 4.84x10-3 1.37x10 90% -4 -4 (1.04x10 , 1.73x10 ) (4.10x10-3, 5.70x10-3) -4 4.92x10-3 2.24x10 80% -4 -4 (1.58x10 , 2.97x10 ) (4.17x10-3, 5.81x10-3) Table 31: The change in the number of years between rabies entry to the UK when the serological test specificity is altered Years between rabies entry Test Specificity PETS/Quarantine EUPMP 13272 211 96.8 % (9408, 16940) (177, 247) (baseline) 10848 210 95% (8914, 13060) (176, 246) 7489 209 90% (5770, 9615) (175, 244) 4640 205 80% (3366, 6312) (172, 240)

47

5.3.8. Modifications to PETS

Due to the potentially large increase in the risk of rabies entry that has been predicted as a consequence of adopting the EUPMP it was deemed appropriate to run some scenarios to test which elements of the current PETS scheme are contributing most to the lower levels of risk associated with this policy. If the serological test requirement was removed from PETS it is predicted that the overall annual risk of rabies entry to the UK would increase approximately 10-fold from 7.79 x 10-5 (5.90 x 10-5, 1.06 x 10-4) to 9.17 x 10-4 (5.94 x 10-4, 1.28 x 10-3), with the number of years between entry decreasing from 13272 (9408, 16940) to 1152 (780, 1682) years. These results indicate that the serological test is having a substantial reductive effect on the overall risk of rabies entry to the UK under PETS. The increase in overall risk of rabies entry to the UK through the current scheme when the waiting period for companion animals is reduced is shown in Figure 21. These probabilities indicate that a reduction in the waiting period to 90 days does not greatly increase the risk of rabies entry, but if the waiting period is any shorter the risk increases much more rapidly, likely to be because these shorter waiting periods are around for which it is estimated that the majority of animals will incubate rabies. When the waiting period is reduced to 30 days and the serological test is removed the annual risk of rabies entry increases to 3.14 x 10-3 (2.66 x 10-3, 3.70 x 10-3), with the number of years between entries decreasing to 322 (270, 376) years. This risk is similar to the predicted risk if the UK were to switch to the EUPMP, which is to be expected as the EUPMP scheme for EU MSs and listed countries has a short waiting period (21 days) and no requirement for a serological test. Note that the risk is not exactly the same as the EUPMP results as a slightly different waiting period is used (30 days as opposed to 21) and companion animal entries through quarantine (from unlisted third countries and the small number from EU MSs and listed third countries) are unaffected by the modifications made to the PETS scheme in these scenarios. In addition, it is important to note that these results are dependent on the duration of the incubation period for rabies which is uncertain (see Section 4.3.1). 7.0E-04

Probability of rabies entry

6.0E-04

5.0E-04

4.0E-04

3.0E-04

2.0E-04

1.0E-04

0.0E+00 212

180

150

120

90

60

30

1

Waiting Period (days)

Figure 21: The change in the probability of rabies entry to the UK through the current system as the waiting period for PETS is decreased

48

5.4. Uncertainty analysis

A regression analysis was undertaken using the software in @Risk. This analysis identifies (form those represented by a distribution) the parameters in the model that a particular output variable (in this case the annual probability of rabies entry) is most sensitive to. The model only incorporates uncertainty, therefore the analysis shows which inputs contribute most to the output uncertainty. For each of the inputs that are being considered, regression coefficients were calculated. An input parameter with a relatively high absolute value of the regression coefficient contributes more to the uncertainty relating to the output variable. The results of these analyses are displayed as tornado charts (Figure 22 and Figure 23). These plots show the parameters that are the largest contributors to the output uncertainty as identified by the uncertainty analysis along with their respective regression coefficients. On each chart the parameter most strongly affecting the final output (i.e. the parameter with the regression coefficient of greatest magnitude) is shown at the top of the chart; the other parameters are then displayed in order of decreasing importance. This ordering gives the chart its characteristic ‘tornado’ shape. Figure 22 shows a tornado chart displaying the regression coefficients for the parameters most strongly affecting the overall uncertainty of the results for the current system of PETS and quarantine. These coefficients indicate that the uncertainty associated with the model results are most strongly affected by the specificity of the serological test for rabies antibodies and the average efficacy of the rabies vaccines. This supports the results of the scenario analyses in sections 5.3.6 and 5.3.7 which showed the model results to be sensitive to these parameters, particularly the serological test specificity. Other parameters that strongly affect the uncertainty associated with the overall results are the number of unobserved infected animals (and consequently prevalence of rabies) in unlisted third countries. This may be because there is high uncertainty with these parameters due to large data gaps, or because the largest risk of rabies entry under the current scheme is associated with entries through quarantine (from unlisted third countries). Interestingly, the regression coefficients for the EUPMP scheme (Figure 23) indicate the uncertainty in this model is most strongly affected by the number of unobserved infected animals from EU member states and listed third countries, while the number of unobserved infections in unlisted third countries does not appear to have a large effect on the output uncertainty. This is likely to be because the majority of the risk of rabies entry via the EUPMP is attributable to companion animal entries through the listed scheme with a 21 day waiting period (with annual probabilities of rabies entry 2.02 x 10-3 and 2.75 x 10-3 for EU MSs and listed third countries respectively) and so the uncertainty surrounding the number of unobserved infections in unlisted third countries becomes less important.

49

Sp (RFFIT)

-0.505

P (V+, Nobivak)

-0.353

Sp (FAVN)

-0.254

P (V+, Madivak)

-0.193

P (V+, Rabisin)

-0.139

λ (unlisted group 2)

0.12

λ (unlisted group 3)

0.103

λ (unlisted group 1)

0.095 -1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

Std b Coefficients

Figure 22: Tornado chart displaying the regression coefficients for the parameters the model outputs are most sensitive to for PETS and quarantine λ (EU MSs group 2)

0.632

λ (EU MSs group 3)

0.493

λ (listed group 3)

0.353

λ (EU MSs group 1)

0.307

λ (listed group 2)

0.244

P (V+, Nobivak)

-0.211

P (V+, Madivak)

-0.108

λ (listed group 1)

0.091 -1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

Std b Coefficients

Figure 23: Tornado chart displaying the regression coefficients for the parameters the model outputs are most sensitive to for the EUPMP

50

6. Discussion In order to help inform future policy decisions regarding the entry of companion animals into the UK a QRA has been developed to estimate the potential change in risk of rabies entry to the UK should the UK adopt the EUPMP. Based on the results of this risk assessment, the risk of rabies entry to the UK under the current scheme of PETS and quarantine is very low, with an annual probability of entry of 7.79 x 10-5 (5.90 x 10-5, 1.06 x 10-4), or one entry every 13272 (9408, 16940) years. The model predicts that if the UK were to harmonise with the EU and adopt the EUPMP the annual probability of rabies entry would increase to 4.79 x 10-3 (4.05 x 10-3, 5.65 x 103 ), or one entry every 211 (177, 247) years. This is approximately a 60-fold increase in the mean risk of rabies entry to the UK. This result is consistent with a previous rabies risk assessment for cat/dog movements from Turkey, that also predicted that EUPMP would pose a higher risk of rabies entry compared to the current policy of PETS and quarantine (Ramnial, 2009). An alternative method of presenting the results is to say there will be one rabies entry in a given number of animal entries. In the model there are 46491 companion animal entries to the UK each year, not including returning UK animals (see Table 13). Therefore, for the current scheme of PETS and quarantine, one rabies entry every 13272 years can also be represented as one animal incubating rabies every 617,028,552 companion animal entries to the UK. Similarly, for the EUPMP, one entry every 211 years is one entry every 9,809,601 animals. Using this interpretation of the results it can be concluded that while the EUPMP represents substantially higher risk than PETS and quarantine, the absolute risk under both schemes is low. A similar risk assessment undertaken for Sweden also predicted that a change to an import policy with no titre checks would increase the likelihood of rabies entry over a 10 year period from 0.05 (0.02 -0.1) to 0.13 (0.05 – 0.34) (Hallgren, 2006). While it is problematic to directly compare the two assessments (no change in waiting period was considered in the Swedish risk assessment as Sweden have already harmonised with this section of the legislation and only entries from the EU were considered) the general conclusions concur, both suggesting the EUPMP policy for EU countries leads to a higher risk because of the removal of the serological test requirement. This conclusion was also reported by the analyses carried out here (see Section 5.3.8) and Jones, 2002b, in which it was stated that “the addition of serology within PETS reduces the risk of importing rabies”. With respect to the country classifications, a policy change from PETS to the EUPMP is predicted to increase the mean risk of rabies entry to the UK for companion animal imports from EU MSs and listed third countries, with the highest risk associated with entries from listed third countries. The reasons for this larger increase for listed third countries compared to EU MSs is most likely due to the substantially larger number of animals that enter the UK from listed Group 3 countries (5910 per year) than from Group 3 countries in the EU (1792 per year). The increased risk associated with entries from these regions will be due to the specific requirements of each policy. The EUPMPEU&listed policy has fewer requirements than PETS (no serological test) and also a reduced waiting period, therefore it is logical to conclude that there would be an increase in risk due to the change in policy from the current PETS/quarantine scheme. 51

One of the most interesting results of this risk assessment is that the mean risk of entry from unlisted third countries is predicted to decrease under the EUPMP. This result can be verified as previous risk assessments have shown the EUPMPunlisted policy to represent a lower risk than quarantine (Ramnial, 2009). PETS, which bears many resemblances to EUPMPunlisted, has also been shown to be less risky than quarantine (Jones et al., 2005a). The decrease in risk from unlisted third countries is most likely due to the combination of the various preventative measures that occur in the country of origin (vaccination, serological test, waiting period) under EUPMP. The scenario analysis in Section 5.3.8 showed that reducing the waiting period for PETS to 3 months did not greatly reduce the risk of rabies entry to the UK. Therefore, based on waiting time alone, it could be hypothesised that moving to the EUPMP would not greatly increase the risk of rabies entry from unlisted third countries (as this represents a reduction in waiting time from 6 to 3 months). However, it is the use of the serological test with high specificity that contributes most to the risk of rabies entry for EUPMPunlisted being lower than quarantine. The scenario analyses carried out in Sections 5.3.7 and 5.3.8 confirm that the model is sensitive to this parameter. Modifying EUPMPunlisted further by removing the serological test results in an increase in the annual probability of rabies entry to 8.06 x 10-4 (5.40 x 10-4, 1.10 x 10-3), with the number of years between rabies entry decreasing to 1300 (908, 1850). This is a considerably higher risk than quarantine, where one infected entry was expected every 23302 (20738, 25557) years, thus confirming that the serological test in EUPMPunlisted is the driving factor behind the decrease in risk associated with this scheme. However, it should be remembered that there is a much greater amount of uncertainty associated with the EUPMPunlisted scheme than quarantine (see Figure 6) and when this uncertainty is taken into account the difference in the risk for each movement policy is less clear. The QRA presented here was commissioned to quantify any potential change in the risk of rabies entry to the UK following harmonisation with the EU legislation for companion animal movement (Defra, 2009) and to help inform any future policy decisions. Defra has previously conducted a qualitative study to assess the change in risk; the three main conclusions from the Defra risk assessment are compared to the results of the QRA described here. It was firstly concluded by Defra that the risk of rabies entering the UK due to companion animal entries from EU MSs will remain negligible. This QRA predicts a 250-fold increase in the mean risk under the EUPMP from 8.34 x 10-6 (3.43 x 10-6, 1.60 x 10-5) to 2.05 x 10-3 (1.41 x 10-3, 2.79 x 10-3) or one entry every 149129 (62683, 291248) years as opposed to every 517 (359, 708) years. This is one rabies entry every 24,035,847 animals and therefore the results of this QRA still suggest a very low or negligible risk of rabies entry to the UK from EU MSs under the EUPMP. Secondly, it was concluded that there would be no change to the current negligible risk of rabies entry to the UK from unlisted third countries whereas we conclude, from the mean results, that this risk will actually decrease as a result of the policy change. The third conclusion of the qualitative risk assessment was that if the UK were to adopt the EUPMP there would be an increase to a very low risk of rabies entry from listed third countries. This conclusion is strongly supported by this QRA, which predicts that under the EUPMP the mean risk of rabies entry from listed third countries will increase to 2.75 x 10-3 (2.39 x 10-3, 3.15 x 10-3) or one entry every 366 (317, 419) years. This is the highest mean risk predicted for any of the country classifications (EU MSs, listed/unlisted third countries) under the EUPMP in

52

this QRA. The higher risk associated with entries from unlisted third countries is likely to be because of the large number of annual companion animal entries into the UK from listed Group 3 countries. Therefore, in summary, the overall results of the two risk assessments concur, in that moving to the EUPMP will lead to an increase in the annual risk of rabies entry. In quantitative terms we predict that under the EUPMP there will be a rabies introduction to the UK within 211 (177, 247) years. The baseline model for this QRA assumes 100% compliance, i.e. all animals have the correct vaccination and serological test, this approach was taken as the best available data (see section 4.3.7) for compliance indicated that the current level of compliance with the vaccination and serological test requirements is nearing 100%. However, it is possible that some cat/dog owners would abuse these systems and attempt to enter the UK using falsified documentation. Defra requested an investigation of the impact of compliance on the overall risk of rabies introduction, under both the current pet movement policy and EUPMP. This was done as a scenario analysis, where the level of compliance was reduced in a stepwise manner, i.e. 90% compliance, 80% compliance etc. This approach differs from previous risk assessments (Jones et al., 2005a; Ramnial, 2009) which model compliance based on estimates from the Kennedy report (Advisory Group on Quarantine, 1998). However, the Kennedy estimates were published in 1998, before the PETS scheme was introduced, and are therefore unlikely to accurately represent the current level of compliance within PETS. Despite this change in methodology, the results are consistent with previous assessments, demonstrating that if the schemes were followed with less than 100% compliance the risk of rabies entry to the UK would substantially increase. As expected the QRA predicts that as compliance decreases the risk of rabies entry through both PETS/quarantine and the EUPMP increases, however, decreased compliance affects the current scheme more than the EUPMP. Under scenarios with 100% compliance, a very high percentage of the risk associated with PETS entries comes from entry routes where an unprotected animal (due to vaccine failure) becomes infected during the waiting period. It is anticipated that it would be difficult to effectively mitigate this risk, as the majority of procedures take place in the country of origin. However, providing the exporting country has an effective, regulated veterinary infrastructure, which uses highly effective vaccinations (and where required, serological tests), the risk of rabies introduction to the UK through PETS should be minimal. For the EUPMPunlisted and especially the EUPMPEU&listed policies, the risk is greater through the pathway where the animal was infected before vaccination, likely because the waiting periods in these schemes are shorter and therefore there is less opportunity for an unprotected pet to become infected between the failed vaccination and movement to the UK. As expected, as the level of compliance is reduced, a higher proportion of the risk of rabies entry to the UK comes through non-compliant pathways. However, in these scenarios very little risk is associated with animals that become infected during the waiting period, due to the assumption that owners/animals that are knowingly noncompliant with the regulations will acquire false (or inaccurate) documentation and move to the UK immediately. Again, non-compliance within the country of origin is hard to control, with the only potential control point being a documentation check at the border upon entry, although these checks have been shown to have minimal effect on reducing the risk of rabies entry to the UK from EU member states under the

53

EUPMP. It is possible that there may be less compliance with the vaccination and testing requirement in unlisted third countries (in the EUPMP), as many of these countries may not have an effective veterinary infrastructure. Therefore, the checking of entries from unlisted countries might have more effect on reducing the risk of rabies entry associated with these movements, although this has not been explicitly tested in the model. During the development of this model a number of assumptions had to be made; either due to the lack of available data, or the uncertainty associated with the data that is available. Many of the assumptions, data gaps and uncertainties are now discussed below. The risk assessment is on a worldwide scale, i.e. it considers companion animal movements into the UK from all countries in the world. This is in contrast to previous rabies risk assessments which have considered companion animal movements from either a single country or a particular region (Hallgren, 2006; Jones et al., 2005a; Ramnial, 2009). In order to effectively model the risk of rabies entry to the UK and to reduce the computational requirements in the time available to undertake the QRA, the countries of the world were separated into different groups. These groups were defined based on the number of reported rabies cases in each country over the last three years, and where quantitative data were not available, expert opinion from the HPA risk assessment was incorporated. Once the groups were finalised the probability that a dog/cat is infected from each group was estimated using the maximum number of combined cat and dog rabies cases reported in the period 2007-09 and the combined cat and dog population for each country where quantitative data were available. In effect, each group was treated as one very large country, with an average rabies prevalence that is representative of the overall rabies situation in all the countries in the group. Therefore in this QRA the groupings of the countries are according to the reported or assumed rabies cases. The quality of the veterinary system may also reduce the potential risk of animals transmitting the disease, which may be particularly relevant for some listed third countries such as the USA, could not be considered. It is acknowledged that there may a substantial number of unreported cases of rabies in some countries (Knobel et al., 2005). However with no comprehensive, comparative, data it was not possible to adjust the number of cases for under-reporting or reporting delay, although using 2009 as the latest year should reduce reporting delay (assuming that all cases that were detected in 2009 had been reported by the time that data were collected for this risk assessment in May and June 2010). It is likely that under-reporting is more prominent in the unlisted third countries that have a poor veterinary infrastructure. Under-reporting has, to some degree, been incorporated in the model by using the maximum number of rabies cases over the last 3 years. The effect of the increased number of cases, as a result of under-reporting, was therefore considered in the scenario analysis. This analysis concluded that the risk of rabies entry to the UK is highly dependent on the rabies prevalence in the rest of the world. Therefore, the under-reporting of rabies cases in unlisted third countries is an important factor in the estimates of the risk of rabies entry to the UK. This is a large data gap that has been identified by the project.

54

All of the policies for companion animal movements considered in this risk assessment include rabies vaccination of the animal. Although there is some evidence that vaccination may have some protective effect on infected animals (there has only been one rabies case in UK quarantine since the introduction of vaccination, and post exposure prophylaxis has been demonstrated by prompt active vaccination in experimental animals), it has been assumed in the development of the model that animals already incubating rabies are unaffected by vaccination and that they will continue to incubate and then display clinical signs of rabies in a similar manner to unvaccinated, infected animals. If vaccination does induce any sort of protective response the risk of rabies introduction to the UK would be lower than the levels suggested here. Regardless, for a certain proportion of uninfected animals, it is likely that vaccination will not be protective. This value has been estimated in the model using a number of studies on the seroconversion of animals following vaccination (Bahloul et al., 2006; Kallel et al., 2006; Minke et al., 2009b; Sihvonen et al., 1995a) with the assumptions that all vaccines were equally efficacious and that animals with an antibody titre >0.5 IU/ml will be protected from rabies (informed by current EU, WHO and OIE guidelines (OIE, 2009)). Consequently it was assumed that animals with a titre less than the 0.5 IU/ml threshold are not protected. However, it has been shown in some studies that the lack of neutralising antibodies in vaccinated animals before challenge with rabies virus does not indicate the animal is unprotected from rabies challenge (Aubert, 1992; Moore & Hanlon, 2010). Therefore the risk assessment may be overestimating the probability of an animal not being protected following vaccination. The effect of the vaccine efficacy on the overall results was tested in a scenario analysis, which showed that, although the efficacy does affect the final results, a change in efficacy around the region in which the true value is expected to lie, has a minimal impact on the overall annual probability of rabies entering the UK. To provide an estimate of the incubation period, data were collected from a number of sources, including data on both experimentally and naturally infected cases. A number of data limitations made it inappropriate to estimate the incubation period using only one of the data sets (either experimental or natural cases). In particular, companion animals infected with rabies virus in experimental studies are likely to be infected with a much higher dose of virus than would be transmitted in natural cases. Additionally, the site of infection (a bite in natural cases) is also likely to affect the incubation period, as clinical signs do not occur until the virus has reached the CNS/brain of the animal. Conversely, reliable data on incubation periods naturally infected cases of rabies are rare, and for the majority of available data on natural cases of rabies the incubation period had to be estimated as the time period between entry (into quarantine or a country) and the date of clinical signs. In these cases the time between the initial infection and the data of entry into quarantine or the country could not be included in the overall estimate. In light of the reliability issues with the majority of the data on the incubation period of rabies, the data were combined to provide an average estimate of the incubation period. This gave an average incubation period of 35 days, which is consistent with estimates from other published sources even though new data has been collected for this QRA (Jones et al., 2005a; Weng H.-Y., 2004; Wilsmore et al., 2006). The sensitivity of the QRA results to the incubation period was investigated in the scenario analysis, where simulations using data from either experimental or naturally infected

55

cases were run. As would be expected this demonstrated that with a longer incubation period (from natural infection) the risk was increased through all schemes, as it increased the likelihood of clinical signs of rabies not being displayed until after the duration of the waiting period. This confirms that the incubation period of rabies is an important parameter within the QRA and acquiring further information on the incubation period will increase the accuracy of the final risk estimate. The data collected on the number of companion animal entries to the UK through PETS between 2005 and 2009 contained both entries from EU countries, third countries and returning UK animals. Returning UK dogs (and cats) have most commonly accompanied their owners for short term vacations, shows, breeding exchange or occasionally for work e.g. rescue dogs. However, a large number of the animals that return accompany owners on work placements or those with two homes; one in the UK and the other elsewhere. These animals spend considerable periods of time out of the UK. Returning UK animals make up a substantial number of PETS entries into the UK, for example over half of the entries in the UK in 2009 were returning UK dogs/cats. However, there is no central recording of the countries in which these animals have been domiciled or the duration of their stay in those countries. Therefore, returning UK animals were removed from the assessment and the risk assessment model was built to analyse only the change is risk associated with companion animal entries from non-UK countries (i.e. those entries listed in the Animal Health data as being entries from non-UK countries). The Defra qualitative risk assessment also did not consider returning UK pets for this reason (Defra, 2009). To accurately model the risk of returning UK pets bringing rabies into the UK detailed data would be needed on the destination, duration and frequency of the visits. Under PETS ferrets can enter the UK via the process of microchipping, vaccination and checking, with either a 21 day waiting period for entries from EU countries or a 6 month period for listed third countries (Defra, 2010a). Note that under the current scheme there is no requirement for ferrets to be serologically tested after vaccination to check antibody levels. In general, there is a lack of data specific to ferrets for the majority of the parameters in the risk assessment model (see Appendix 3). Therefore the decision was made to remove ferrets from the model and only focus on cats and dogs. Furthermore the data collected on companion animal entries to the UK indicated that ferrets accounted for only a small percentage (0.5 IU/ml) will be protected from rabies infection • Animals with a serological titre 0.5 IU/ml) Sihvonen Rabisin 83 80 30-40 RFFIT et al., 1995 Madivak 47 46 Kallel et Rabisin 30 5 RFFIT 4 al., 2006 Bahloul et Rabisin 35 4 RFFIT 4 al., 2006 Minke et Rabisin 15 14 28 FAVN al., 2009 Nobivak 15 10

4.5. Cat/dog import datasets

Country

Table 44: Imports from EU MSs & equivalents (Animal Health, 2010) PETS Quarantine Animal 2009 2008 2007 2006 2005 2009 2008 2007 2006

2005

Group 1 Austria Cyprus Czech Republic Denmark Finland

CAT DOG CAT DOG CAT DOG CAT DOG CAT

12 173 118 304 16 358 23 329 12

37 226 364 841 31 296 51 303 55

39 190 263 570 42 299 67 349 35

33 156 228 421 21 215 40 232 37

144

39 115 221 401 13 95 50 276 33

2

2

5

4

9

65

40

39

45

43

7

8

10

8

14

3

4

5

5

5 3

Greece Guadaloupe Iceland Ireland (Republic Of) La Reunion Liechtenstein Luxembourg Malta Martinique Monaco Netherlands North Cyprus Norway Portugal San Marino Slovakia Spain Sweden Switzerland

DOG CAT DOG DOG CAT CAT DOG CAT DOG DOG CAT CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG

Total Group 2 Andorra Belgium Estonia France French Guiana Germany Gibraltar Greenland Hungary

CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT

140 79 247

217 110 317 1

182 60 262 1

1 4 22 518

49 442

1 3 24 394

13 115 29 37

2 14 101 49 59

1 2 16 108 22 58

164 93 230

157 59 264

4 14 289

2 12 260 1

14 112 31 39

8 111 27 38

35

40

43

38

42

2

3

1

7

3

2

4

11

7

32

13

19

24

21

24

30

13

30

4

7

4

5

6

10

22

26

10

20

27

1

2

7

1 6 5

223 2,832

5 307 2,969 28 164 87 346

1 261 2,858 32 151 69 341

360 2,847

12 133 66 298 2 2 3 109 737 3,974 29 359 23 456 11807

37 145 87 330

3 2 93 742 4,032 65 433 83 509 13434

8 75 805 3,976 64 425 81 423 12558

4 40 739 3,666 58 442 55 392 11575

1 6 116 1,591 1 74 1,059 7,643

1 6 148 1,505 6 30 1,146 10,218

7 127 1,535 7 24 1,158 8,981

318 3,814 2 47

2 518 4,523 31 17

1 470 4,070 20 37

17

1 45

34

3 157 1,551 5 8 1,117 8,503 1 2 566 4,020 41 52 3 4 19

2 353 2,714 14 100 85 244

2

145

27 728 3,184 52 372 68 378 10503 9 202 1,459 0 3 931 6,660 3 548 3,823 22 57 1 8 3

5 151

132

98

159

204

9

6

4

10

7

18

7

14

16

17

393

321

274

386

423

15

11

15

10

1

2

213

182

217

265

195

48

39

37

53

90

1

4

2

3

3

4

2

3

8

19

Italy Slovenia

DOG CAT DOG CAT DOG

Total Group 3 Bulgaria Croatia Latvia Lithuania Poland Romania

Group 1 Antigua and Barbuda Aruba Australia Barbados British Virgin Islands Cayman Islands French Polynesia Grenadines Hawaii Jamaica Japan Malaysia Mauritius Montserrat Netherlands Antilles New Caledonia

238 194 1,137 2 62 19830

191 183 1,000 2 47 17894

17 45 1 20 8 38 20 105 123 1,213 11 62 1663

10 34 0 11 19 33 8 62 208 1,282 27 46 1740

1 12 1 10 8 9 12 56 202 1,154 11 25 1501

104 145 855 6 29 17191

61 157 643 1 24 14615

26

20

32

60

4

46

2

307

262

307

403

5

11

15

10

3

1

2

2

2

1

5

4

1

1

3

9

23

6

12

10

16

5

2

5

5

5

41

28

23

39

52

Table 45: Imports from listed Third countries (Animal Health, 2010) PETS Quarantine Animal 2009 2008 2007 2006 2005 2009 2008 2007 2006

2005

CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG

Total

Country

250 127 1,004 7 55 16132

CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG

4

8

2 8 4 6

640 1,125 3 20

2

2 17 11

1 2

2 9 12 15 6 37 140 737 1 1 960

2 5

320 652 4 11 3 3 6 9

6 2 1 3 8 61 212

293

1 2

301 556 5 17 2

326 566 3 12

3 10

3 3

1

3

2

4

99

86

81

59

2

9

3

3

7

5

7

3

12

21

18

1 91 1

5

2

1 3

3

1

1

1

1 72 61

7 2 0

8 1 1 2 2

2 40 65

38 71

2 1 9

0 2 1 2 1

2 9 17

0 38 46

1 1 3 1

146

353

15

3

7

4

19

14

8

19

7

11

7

10

19

14

2

2

3

3

2

1

New Zealand Singapore St Kitts and Nevis St Lucia St Vincent Taiwan Total Group 2 Ascension Island Bahrain Bermuda Chile Falkland islands Fiji Hong Kong St Helena Trinidad and Tobago United Arab Emirates Vanuatu

CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG

Belarus Bosnia Canada Mexico Russia USA Total

137 156 70 72

99 147 65 52

36

41

27

30

20

21

16

19

19

12

3 1

5 12

15 11 2657

64 55 22 19

2 4 1647

37 23 22 25

0 2 1 1 1460

23 17 23 28

2 2 2 1379

18 41 22 26

3

2

1

1

1

1

3

10

206

191

188

217

163

2

2

2

13

1

36 48

80 99

57 79

0

2

43 76 1 1

5 246 257 6 5 859

1 566

424

3 22

12

1 3

153 167

40 46 4

87 123 1

342 433 9 48 53 1,607 2,574 5091

7

2

3

6

2

2

1

1

2

15

14

30

17

28

4

1

1

1

1

1

2

11

6

12

79

54

45

56

57 2

198

103

297 436 0 2 67 83 1,304 2,170 4371

147

248 391 1 2 45 35 1,295 2,049 4070

87

6 1

13 41 1 6 9 44 102 306 522

7 1

1 1

2

1 5

1

103 CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG CAT DOG

138 203 65 77

2

41

Total Group 3 Argentina

6

158 268 86 145

222 331 4 7 1,172 1,841 3577

2

100

86

121

1

1

8

1

1

1

3

1

3

54

64

61

97

68

7

6

6

9

12

8

9

22

24

28

345

414

376

593

699

421

495

470

726

819

Table 46: Imports from unlisted Third countries (Animal Health, 2010) Country 2009 2008 2007 2006 2005 Group 1 Bahamas 2 5 10 6 3 Kuwait 11 3 11 15 Macedonia 1 1 Turks and Caicos 6 4 Islands Total 13 5 20 22 18 Group 2 Belize 1 1 Brunei 1 3 1 6 4 Costa Rica 1 5 2 2 3 Egypt 32 19 25 19 23 Grenada 1 2 1 8 Honduras 1 Lebanon 2 2 1 Libya 4 1 1 3 Maldives 1 Panama 9 4 3 Seychelles 1 4 3 2 2 Syria 1 1 2 1 Total 51 41 32 38 47 Group 3 Afghanistan 6 1 4 Angola 2 5 1 1 Azerbaijan 11 11 9 5 2 Bangladesh 2 1 2 Bolivia 1 2 Borneo 1 Botswana 5 5 11 7 Brazil 12 22 14 33 23 Burundi 1 Cambodia 2 1 2 Cameroon 1 China 31 36 21 14 23 Colombia 5 3 4 1 1 Congo 1 1 2 Cuba 2 4 Dominican Republic 1 3 3 5 2 Ecuador 1 2 6 El Salvador 1 Eritrea 1 Ethiopia 1 4 1 1 Gabon 1 Gambia 3 5 4 3 Georgia Republic 3 2 1 Ghana 10 2 1 5 7 Guatemala 1 Guyana 1 India 40 38 24 43 27 Indonesia 12 6 4 13 1 Iran 3 2 8 5 Iraq 4 6 5 2 8 Israel 7 9 6 13 19 Ivory Coast 1 Jordan 9 8 5 7 3

148

Kazakhstan Kenya Kosovo Liberia Macau Madagascar Malawi Mongolia Morocco Mozambique Namibia Nepal Nigeria Oman Pakistan Paraguay Peru Philippines Puerto Rico Qatar Saudi Arabia Serbia Sierra Leone South Africa South Korea Sri Lanka Sudan Tanzania Thailand Tunisia Turkey Turkmenistan Uganda Ukraine Uzbekistan Venezuela Vietnam Yemen Zambia Zimbabwe Total

1 16

2 20

2 13 2

2 19

2 12 3

1 1 3 0 6 3 1 5 8 18 21 0 4 6 36 22 1 463 15 15 4 7 57 2 105 2 5 8 3 4 3 1 3 20 1035

1 4 13 5 3 5 12 6 4 15 26 1 5 703 16 25 4 8 39 2 85 4 10 1 3 5 11 16 1222

149

7 1 2 3 1 8 10 8 2 3 16 30 1 542 5 11 3 40 4 71 3 1 3 1 2 6 2 3 24 945

1 2

2 4

5

7 3 3 2 15 13 7

2 6 6 15 12 4 9 4 29 36 2 591 12 25 4 3 43 58 2 4 3 8 4 1 5 34 1132

22 33 1 1 584 10 19 3 30 3 61 2 5 5 2 3 5 1 4 49 1061

Bahloul, C., Taieb, D., Diouani, M.F., Ahmed, S.B., Chtourou, Y., B'Chir B, I., Kharmachi, H., Dellagi, K., 2006. Field trials of a very potent rabies DNA vaccine which induced long lasting virus neutralizing antibodies and protection in dogs in experimental conditions. Vaccine 24, 1063-1072. Cliquet, F., Aubert, M., Sagne, L., 1998. Development of a fluorescent antibody virus neutralisation test (FAVN test) for the quantitation of rabies-neutralising antibody. J Immunol Methods 212, 79-87. Fekadu, M., Shaddock, J.H., Baer, G.M., 1982. Excretion of rabies virus in the saliva of dogs. J Infect Dis 145, 715-719. Kallel, H., Diouani, M.F., Loukil, H., Trabelsi, K., Snoussi, M.A., Majoul, S., Rourou, S., Dellagi, K., 2006. Immunogenicity and efficacy of an in-house developed cell-culture derived veterinarian rabies vaccine. Vaccine 24, 4856-4862. Minke, J.M., Bouvet, J., Cliquet, F., Wasniewski, M., Guiot, A.L., Lemaitre, L., Cariou, C., Cozette, V., Vergne, L., Guigal, P.M., 2009. Comparison of antibody responses after vaccination with two inactivated rabies vaccines. Vet Microbiol 133, 283-286. Sihvonen, L., Kulonen, K., Neuvonen, E., Pekkanen, K., 1995. Rabies antibodies in vaccinated dogs. Acta Vet Scand 36, 87-91. Soulebot, J.P., Brun, A., Chappuis, G., Guillemin, F., Petermann, H.G., Precausta, P., Terre, J., 1981. Experimental rabies in cats: immune response and persistence of immunity. Cornell Vet 71, 311-325. Trimarchi, C.V., Rudd, R.J., Abelseth, M.K., 1986. Experimentally induced rabies in four cats inoculated with a rabies virus isolated from a bat. Am J Vet Res 47, 777-780.

150

APPENDIX 5: Classification of lyssaviruses Species

Abbreviation Serotype Genotype (ICTV)a

Potential vector(s)/reservoirs

Distribution

Lyssavirus (Rabiesvirus)

RABV

I

I

Lagos-BatVirus

LBV

II

II

MokolaVirus

MOKV

III

III

?

Duvenhage Virus

DUVV

IV

IV

Insectivorous bats

European Bat Lyssavirus 1 European Bat Lyssavirus 2 Australian Bat Lyssavirus Aravan virus

EBLV 1

V

Insectivorous bats (Eptesicus serotinus)

Europe

EBLV 2

VI

Insectivorous bats (Myotis sp)

Europe

ABLV

VII

Frugivorous/insectivorous bats (Megachiroptera/Microchiroptera)

Australia

ARAV

?

?

Insectivorous bats

Central Asia

Khujand virus

KHUV

?

?

Insectivorous bats

Central Asia

Irkut virus

IRKV

?

?

Insectivorous bats

East Siberia

West Causcasian bat virus

WCBV

?

?

Insectivorous bats

Caucasian region

a

Carnivores (worldwide); bats (Americas)

Worldwide (except several islands) Frugivorous bats (Megachiroptera) Afrika

ICTV = International Committee on Taxonomy of Viruses.

151

SubSaharan Africa Southern Africa

APPENDIX 6: A review of the risk frameworks used in previous risk assessments on the introduction of rabies into a specific region via the movement of cats and dogs 1. Introduction

Great Britain (GB) has been rabies-free for almost 100 years, the last indigenous case of terrestrial animal rabies occurred in 1922. This was achieved primarily through the introduction of strict veterinary control measures and the introduction of 6-month quarantine for cats and dogs entering the United Kingdom (UK) from overseas. Recent years have seen numerous scientific advances that have cast doubt on the necessity and viability of the 6-month quarantine scheme in the UK. In 2000, the UK introduced the Pet Travel Scheme (PETS), which replaced the quarantine system for some pet movements with a process involving microchipping, vaccination and serological testing. The decision to move to the PETS system was based partly on the findings of a quantitative import risk assessment (Advisory Group on Quarantine, 1998). In 2004, the EU adopted a harmonised Pet Movement Policy (EUPMP) which consists of a similar, but arguably less stringent process, of microchipping and vaccination. Here we present a summary of the risk assessments that have analysed the risk of rabies entry through various pet movement protocols. Using the review as a basis, a pathway will be developed for the 2010 VLA risk assessment for rabies entry into the UK from all countries. A summary of each model will detail the pathways identified through which rabies may enter the importing country, and also the data used to populate these pathways. This will be followed by a discussion highlighting the similarities and differences between the different models, thus providing a starting point for the development of the risk pathway for use in the current VLA risk assessment. 2. Review of previous risk assessments 2.1 The Advisory Group on Quarantine (Kennedy Report)

In 1998 the Advisory Group on Quarantine published a review of the UK quarantine laws, known as the “Kennedy Report”, that included a risk assessment (Advisory Group on Quarantine, 1998). The PETS scheme was implemented based on the contents of this review. Within the risk assessment risk was expressed in two ways; absolute risk, where the likelihood of the hazard event occurring was detailed as one event every so many months or years, and relative risk, which allows a comparison of the risk of the hazard occurring under alternative policies. The principal outcome (or hazard) of this risk assessment was the risk of an imported cat or dog developing rabies. The risk assessment considered scenarios including the (then) present quarantine system and a number of alternative options. The fraction of imported animals infected with rabies was determined to be a function of the incidence of rabies at source and the incubation period of the disease, estimated to be 42 days (20 days, 60 days). To determine the rabies incidence at source the risk assessment divided the world into regions predominantly according to WHO

152

classification, with additional division of Europe into the European Union (EU) and Eastern Europe, the Americas into North America and Latin America and the Caribbean, and separating Australia and New Zealand from the rest of the West Pacific region. The most likely incidence rates of rabies infection in different parts of the world were defined by a triangular distribution around low, best and high values (Table 1). It was noted that the incidence data for the EU and North America was based on relatively reliable surveillance data while incidences in other countries were based on special studies. Table 47: The regions of the world detailed in the Kennedy report and the estimated incidence rates of rabies Region Incidence, million animals per year. Best (Low, High) European Union (EU) 1 (0, 10) Eastern Europe (EEU) 10 (1, 100) North America (NAM) 1 (0, 10) Latin America and the Caribbean (LAC) 100 (10, 1000) Sub-Saharan Africa (SSA) 100 (10, 1000) South-East Asia (ASI) 100 (10, 1000) Eastern Mediterranean (EMR) 100 (10, 1000) Australia and New Zealand (AZL) 0 (0, 0) Western Pacific Region (WPR) 100 (10, 1000)

The risk assessment represents the possible routes by which imported cats and dogs can enter GB using a table view, where each column of a table represents one route. Figure 1 shows all the possible routes of entry in relation to the 6-month quarantine regime.

Figure 1: Routes by which animals can enter Great Britain under the quarantine system. Successful checking indicated an attempted smuggled animal being checked and subsequently placed under quarantine. Black squares indicate the pathways via which the majority of rabid animals enter Great Britain. (Advisory Group on Quarantine, 1998)

Each pathway can be quantified using a formula representing the number of animals entering via that route. Equations 1 and 2 represent the number of infected animals entering via routes 1 and 5 in Figure 1, where I represents the number of infected imported animals, q the fraction of animals entering via quarantine, rq the fraction of animals developing rabies whilst in quarantine and c the fraction of animals checked at port of entry. N1 = Iq(1 − rq )

(1)

153

N 5 = I (1 − c)(1 − q)

(2)

The majority of the alternative policies tested in the risk assessment are more complex than the quarantine system. The most important of the alternative policies combined vaccination and serological testing before entry, supported by quarantine for animals which could not comply with the requirements. This policy formed the basis of the PETS system. The alternative policy consisted of 47 possible routes of entry compared to the 8 seen in the quarantine system (Figure 2). The equations describing each entry route are not given in the report, however, following inspection of the @Risk spreadsheet model the equation quantifying route 3 in Figure 2 can be deduced (equation 3), where v is the fraction of animals vaccinated before entry, s is the fraction of animals blood tested, t+ is the fraction of infected animals testing (false) positive and rp is the fraction of infected animals which develop rabies during the preentry period. All other parameters are the same as in equations 1 and 2. N 3 = Ivst + (1 − rp )(1 − c)

(3)

154

Figure 2: Routes by which animals can enter Great Britain under an alternative policy based on vaccination and serological testing. Black squares indicate the pathways via which the majority of rabid animals enter Great Britain. (Advisory Group on Quarantine, 1998)

155

In 1996, 7267 cats and dogs were quarantined in GB, the majority of which came from the EU and North America (NAM). However, dominant sources of animal infection were Sub-Saharan Africa (SSA), the West Pacific Region (WPR) and the Eastern Mediterranean Region (EMR). Therefore, the chance of an infected European animal developing rabies on holiday in GB was considered to be negligible. The model results after 1,000 iterations indicated that the most likely interval between rabies entry through quarantine was 36 (21, 87) years, representing an annual risk of 0.027/year or one case in 324,000 animals. Under the alternative policy of vaccination and serological testing six months prior to entry the relative risk of rabies entry was determined to be 2% (1%, 7%) higher than under the present policy. This was seen as a marginal increase on the initial very low risk. The difference was deemed to be attributable to the fact that vaccination does not protect all animals against infection. In absolute terms the best estimate of risk was one case every 28-34 years. 2.2 Update of the Kennedy model for North America, Laurenson, Woolhouse and others (2002).

In 2002, the model developed by the Advisory Group on Quarantine was modified in order to assess if there would be an increased risk to the UK if pets from North America were allowed to enter through the PETS scheme. The model used was essentially the same as the one described in the Kennedy report, and as such will not be discussed in detail here. However it should be noted that peer review of this model by risk assessors at the VLA highlighted a number of issues with the model, including lack of transparency and an overly complex model structure with respect to the new risk question. However, that is not to say that the risk pathway itself was inappropriate. In conclusion, the model framework from the Laurenson (and Kennedy) model should be considered when designing the risk pathway for use in the current VLA risk assessment, but care should be taken to ensure the risk pathways are clearly defined, the model is transparent and parameter estimations use current data, where possible, and are biologically plausible. 2.3 Weng (2004) The risk of rabies introduction into Taiwan

A risk assessment developed by the Graduate Group in Epidemiology at the University of California assessed the risk of introducing rabies to Taiwan though the importation of dogs (Weng H.-Y., 2004). This risk assessment only considered dogs, as it assumed the only risk of introducing rabies to Taiwan is through the importation of dogs. A stochastic model was developed to assess the risk of rabid dogs entering Taiwan via different combinations of safeguards and smuggling (Figure 3). The parameter values are detailed in Table 2. Where necessary a Pert distribution (using most likely, maximum and minimum values) was used to allow for uncertainty in model parameters. The incidence of rabies in the country of origin is modelled by a Pert distribution with 0, 1.5x10-4, 10-3, i.e. country specific data are not used in this risk assessment and all results are presented as the risk related to imports from a fictitious country. The results suggest rabies infection will occur every 17.2 years for no quarantine to every 138.6 years for 180-day quarantine with requirement of a verified vaccination

156

certificate and control of smuggling. The results indicate that control of smuggling and requirement of a verified vaccination certificate were economically favourable and significant measures that could be used to prevent rabies introduction into Taiwan. Countries of origin

Import routes

Quarantine

Travel