General enquiries on this form should be made to:Defra, Science Directorate, Management Support and Finance Team,Telephone No. 020 7238 1612E-mail: research.competitions@defra.gsi.gov.uk

SID 5 Research Project Final Report

SID 5 (Rev. 3/06) Page 1 of 24

NoteIn line with the Freedom of Information Act 2000, Defra aims to place the results of its completed research projects in the public domain wherever possible. The SID 5 (Research Project Final Report) is designed to capture the information on the results and outputs of Defra-funded research in a format that is easily publishable through the Defra website. A SID 5 must be completed for all projects.

This form is in Word format and the boxes may be expanded or reduced, as appropriate.

ACCESS TO INFORMATIONThe information collected on this form will be stored electronically and may be sent to any part of Defra, or to individual researchers or organisations outside Defra for the purposes of reviewing the project. Defra may also disclose the information to any outside organisation acting as an agent authorised by Defra to process final research reports on its behalf. Defra intends to publish this form on its website, unless there are strong reasons not to, which fully comply with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.Defra may be required to release information, including personal data and commercial information, on request under the Environmental Information Regulations or the Freedom of Information Act 2000. However, Defra will not permit any unwarranted breach of confidentiality or act in contravention of its obligations under the Data Protection Act 1998. Defra or its appointed agents may use the name, address or other details on your form to contact you in connection with occasional customer research aimed at improving the processes through which Defra works with its contractors.

Project identification

1. Defra Project code SE3026

2. Project title

Bovine TB in restocked herds: Risk factors and dynamics

3. Contractororganisation(s)

University of WarwickCoventry CV4 7AL               

54. Total Defra project costs £ 1.15 million(agreed fixed price)

5. Project: start date................ 01 June 2003

end date................. 31 March 2006

SID 5 (Rev. 3/06) Page 2 of 24

6. It is Defra’s intention to publish this form. Please confirm your agreement to do so...................................................................................YES NO (a) When preparing SID 5s contractors should bear in mind that Defra intends that they be made public. They

should be written in a clear and concise manner and represent a full account of the research project which someone not closely associated with the project can follow.Defra recognises that in a small minority of cases there may be information, such as intellectual property or commercially confidential data, used in or generated by the research project, which should not be disclosed. In these cases, such information should be detailed in a separate annex (not to be published) so that the SID 5 can be placed in the public domain. Where it is impossible to complete the Final Report without including references to any sensitive or confidential data, the information should be included and section (b) completed. NB: only in exceptional circumstances will Defra expect contractors to give a "No" answer.In all cases, reasons for withholding information must be fully in line with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.

(b) If you have answered NO, please explain why the Final report should not be released into public domain

Executive Summary7. The executive summary must not exceed 2 sides in total of A4 and should be understandable to the

intelligent non-scientist. It should cover the main objectives, methods and findings of the research, together with any other significant events and options for new work.

The foot and mouth disease (FMD) epidemic in 2001 was used to examine bovine tuberculosis (bTB) transmission dynamics in cattle. Bovine tuberculosis (bTB) testing was reduced and cattle movements were altered during the FMD epidemic. On some farms herds were restricted whilst the FMD epidemic continued whilst on others herds were depopulated and subsequently restocked, forming new cattle herds. We have used this perturbation in movement (both less than usual and more than usual) to test the hypothesis that cattle to cattle transmission of bTB occurs. We have collected and assimilated data through farmer interview of 148 farmers and through the use of previously collected data on bovine TB (stored in VetNet) and data on cattle movements (stored in the British Cattle Movement Scheme (BCMS) database) to quantify this effect. We have used statistical and mathematical approaches and undertaken a study of bTB in GB cattle using three approaches:

1. A cohort study of 148 herds in the south west of England. We collected prospective and retrospective data and incorporated the Randomised Badger Culling Trial groupings (RBCT) into the results. These data have been used to assess herd and animal level risks of herd breakdown / becoming a bTB reactor respectively.

2. A retrospective cohort study of restocked herds across Great Britain.We gathered data from the FMD, VetNet and BCMS databases relating to cattle movement and bTB incidence. These data have been used to assess the farm level risks of bTB breakdown with respect to bTB history of farms and animals bought during restocking.

3. Mathematical models of bTB transmission dynamics.We have developed two approaches based on the interaction between surveillance and control, and the numbers of reactors detected on breakdown.

The key results are as follows:

1. Approximately 6% of newly formed cattle herds (in Study 2) broke down at their first bTB test after restocking. The associated risks for this were a farm history of bTB before 2001,

SID 5 (Rev. 3/06) Page 3 of 24

purchasing cattle from high frequency testing herds (> 6 tests in 12 years) and the herd size at testing. The risk associated with history of HBD on the farm reduced the greater the time since the last HBD. If the last HBD was five years previously the risk was not increased.

2. Approximately 50% of the 148 cohort farms (Study 1) broke down during the study. Associated risks with bTB exposure were farm history of bTB, storing manure or slurry in a closed container compared with storing in the open or spreading immediately, purchasing cattle from market and purchasing steers.

3. Approximately 1000 / 55,000 cattle that were tested on the farms between 2001 and 2004 were reactors. The strongest associated risk for being a reactor was the number of reactors cattle were exposed to before the test. bTB history of the farm and herd size were also associated with reactors.

4. Considering the GB cattle population as a whole, and combining bTB testing data (VetNet) and cattle movement data (BCMS) showed that in 3 randomly selected cohorts 72-84% of animals died prior to being tested for bTB.

5. Mathematical models indicate that the qualitative patterns of changes in bTB incidence can be explained by the dynamic consequences of changing testing interval.

6. Mathematical models indicate that the distribution of reactors within herds may be explained by a delay in cattle becoming infectious so that some reactors are infectious and others are not.

The results of this study to date indicate that:

1. There is an associated risk of HBD from the purchase of cattle from herds with a history of bTB. Infection can be imported on to farms through movement of infected cattle.

2. There is an associated (stationary) risk of HBD from persistence of bTB in the farm environment, outside of cattle on the farm. This risk decays with time since HBD.

3. The distribution of numbers of reactors after a period without testing is highly suggestive of on-farm transmission between cattle.

4. Farm management practices associated with the risk of HBD included purchasing practices, management of manure and diet.

5. The frequency of testing for bTB determines the success of both control and surveillance. We hypothesise that increasing testing coverage / effort will result in disclosure of more infection and greater control.

Project Report to Defra8. As a guide this report should be no longer than 20 sides of A4. This report is to provide Defra with

details of the outputs of the research project for internal purposes; to meet the terms of the contract; and to allow Defra to publish details of the outputs to meet Environmental Information Regulation or Freedom of Information obligations. This short report to Defra does not preclude contractors from also seeking to publish a full, formal scientific report/paper in an appropriate scientific or other journal/publication. Indeed, Defra actively encourages such publications as part of the contract terms. The report to Defra should include: the scientific objectives as set out in the contract; the extent to which the objectives set out in the contract have been met; details of methods used and the results obtained, including statistical analysis (if appropriate); a discussion of the results and their reliability; the main implications of the findings; possible future work; and any action resulting from the research (e.g. IP, Knowledge Transfer).

SID 5 (Rev. 3/06) Page 4 of 24

Scientific Objectives

1. The enrolment of a cohort of 270 cattle herds.This objective includes the identification of appropriate farms and the collection of base-line (enrolment) data. A total of 270 farms were to be recruited into a cohort study with exposure being destocking during the FMD epidemic of 2001 and restocking (practising a wide range of restocking practices) and non exposed (non destocked farms, practising a wide range of purchasing replacement activities). The hypothesis was that variable sourcing of cattle would be able to be used to investigate risk of HBD from purchasing cattle. These farms were to be matched within a triplet of the RBCT but not with adjoining properties.2. Routine longitudinal follow-up of cohort.The cohort studies will be followed for 3 years. Further data will be collected annually.3. Event sampling of cohort.Data will also be collected either when the farm experiences an outbreak / epidemic event (e.g. abortion storm) or when tuberculosis breakdown is reported (through usual surveillance).4. Setting up of system to identify farms for the case control studyThis objective includes liaison with VLA to set up an information line to inform Warwick of the location and contact details of one case and two matched controls within a feasible time period of HBR.5. Case and control farm visitsA case and control study of restocked farms only with a case defined as a herd where a HBD occurred at the first visit after restocking and a control no HBD at the first bTB test after restocking was proposed. Farms will be visited within 4 weeks of each other. Details of restocking practices will be recorded on standard forms and will be used to gather information about these farms and cows on the farms. Where possible the affected cow will be examined and control cows on the case farm and control farms selected to identify cow risks for breakdown. 6. Data analysis. From the data collected in objectives 1 - 3 we will develop statistical models using survival analysis and frailty models (should sufficient herds have more than one breakdown) to test the hypothesis that repopulation affects the risk of HBR. 7. Development of mathematical models of bovine TB transmission dynamicsUsing the available data, deterministic and stochastic models will be developed and validated that include the following processes:a. within and between herd transmission; b. dose-dependent infection outcomec. age-dependent infection outcome. These will provide insight and enhanced understanding of quantitative aspects of TB epidemiology and potential control options. We do not intend to develop spatially explicit models.

Meeting the objectives

A total of 148 farms were enrolled under objective 1. This was lower than anticipated because the total number of restocked (exposed farms) provided by VLA was 80 and 40 of these had restocked and agreed to participate in the study. There were no further known restocked (exposed) farms to enrol and using epidemiological principles it was not logical to recruit more unexposed farms since the ratio of exposed to unexposed was approximately 1:4. It was also apparent that the incidence of bTB was far higher than anticipated from discussion with VLA at 50% (rather than the estimated 15%) so there was no loss of statistical power from the original estimate that factors with a relative risk of 3 were likely to be significant with 95% confidence and 80% power. The lower number of farms was offset by the higher incidence of HBD (herd breakdown) with bTB.The objectives above were achieved with the exception of objective 5, where the objective was altered (with approval from DEFRA) from a case control study with farm visits to a retrospective cohort study using VetNet and BCMS data (see below). This was done because the whole population of restocked farms could be studied rather than a sub sample and because at the time of the start of the project many restocked herds had been tested up to a year previously, so recall bias would have been a concern.

Finally, we could use BCMS data to investigate the impact of purchasing behaviour on the risk of HBD with bTB. This raised the statistical power of the analysis and a retrospective cohort study was more powerful a study than a case control study. This redesigned study provided extremely useful results.

Introduction

Bovine tuberculosis (bTB) is caused by Mycobacterium bovis, a zoonotic pathogen that can cause infection and disease in many host species including humans (Ashford, et al. 2001).

In GB there has been an increase in the incidence of herd breakdown (HBD) in the south west of England since the late 1970s. Since the late 1980s, HBD have occurred outside the south west of England, moving northwards from the south west to areas where bTB had not occurred for many years (Bourne, et al. 2000; Pollock, et al. 2001; DEFRA 2005; Green and Cornell 2005). Currently in GB and Ireland (where the incidence of HBD has remained static for decades (Griffin et al., 2005)) the control of bTB in cattle is based on testing of herds using the comparative (comparing with M. avium) intradermal skin test to identify positive (reactor) cattle which are then removed from the herd. Herds with reactor cattle are placed under movement restriction: cattle may be sold direct for slaughter but cannot otherwise leave the farm until the herd is deemed free from infection, when trading may continue as normal. From the 1970s to 2005 there has been no restriction of cattle sales between herds based on history of HBD with bTB in GB (i.e. the accredited herd system was disbanded). The herd inter test interval ranges from one to four years depending upon the perceived risk for the herd; there is more frequent testing where more bTB is detected in neighbouring herds usually within the same parish.

The comparative skin test has an estimated sensitivity of 77 to 95% (Monaghan, et al. 1994; Costello, et al. 1997) and a specificity in excess of 99% (Leslie, et al. 1975). It is therefore likely that infected cattle are undetected when a herd is tested, either resulting in a negative herd test, or in infected animals remaining in the herd after a positive test. The more animals that are infected in a herd, the less likely that some infection will not be detected when these cattle are tested for bTB, but also the more likely that some infection will remain after testing for bTB.

One possible hypothesis for the persistence and spread of bTB in GB includes cattle-to-cattle transmission of M. bovis within and between herds because of the less than perfect sensitivity of the test and because of the interval between herd tests (Medley, 2003) and because the majority of cattle are never tested for bTB in their life time (Mitchell et al., 2006) due targeted testing (see Study 3, page 17, below). There is published evidence of an association between movement of cattle and HBD with bTB in Ireland, Italy, Canada and Great Britain, (Griffin 1993; Marangon, et al. 1998; Munroe, et al. 1999; Johnston 2005). Long distance spatial clustering of bTB between herds has also been reported (Green and Cornell, 2005) which can be explained plausibly only by cattle movements. Gilbert et al. (2005) reported that movement of cattle from areas where bTB was common to areas where bTB is considered rare was the best predictor for the introduction of bTB into a naïve geographical area.

Another hypothesis is that there is transmission between cattle and wildlife (particularly badgers) within farms. Much of the remaining risk for herd breakdown, and the lack of progress in GB towards the elimination of bTB, has been attributed to a reservoir of bTB infection in badgers (Phillips, et al. 2003; Griffin, et al. 2005). There is extensive literature reporting the occurrence of bTB in badgers and possible routes of spread of bTB from badgers to cattle (Cheeseman, et al. 1989; Gallagher and Clifton-Hadley 2000; Scantlebury, et al. 2004). The most recent and strongest evidence for the role of badgers in persistence of bTB in cattle in England comes from the results of the Randomised Badger Culling Trial (RBCT) where proactive culling of badgers within a geographically defined area was associated with a reduction in HBD in herds in the centre of the spatial area compared with farms located in a control area (Donnelly et al., 2005). Similar results, although with a different definition for controls, were reported from Ireland (More et al., 2005). Nearly all other vertebrates can be infected with M. bovis. However, the role of non-badger wildlife as potential reservoirs for bTB is not well researched and documented in

GB. In mainland Europe and America many wildlife species including deer and wild boar (Hermoso de Memosa et al., 2006) have been reported as infected with bTB.

The mode of transmission of M. bovis between bovines and a non-bovine host is unknown, but it probably be indirect, or at least partially indirect, since viable M. bovis organisms have been detected on farms around badger setts and latrines and have led to the hypothesis that M. bovis may persist on farms outside any host for some period of time (Courtenay et al., 2006). This persistence may also occur from M. bovis shed by cattle; possible sources of M. bovis might include slurry (Johnson et al., 2005), soil or dust in farm buildings. The existence of a persistent stationary reservoir of M. bovis would also explain the apparent geographical stability of bTB risk.

Purpose of this Project

The purpose of this research was to identify risk factors associated with the introduction of and transmission of bTB following the recent foot and mouth epidemic and associated culling of whole herds of cattle. The 2001 outbreak of foot and mouth disease (FMD) resulted in the culling of cattle on approximately 8000 premises. This tragedy provided a natural experiment that has benefited our understanding of the introduction and transmission of bTB among cattle. Approximately 6000 farms that were cleared of cattle (because of FMD) were restocked by the end of 2002. Farmers purchased cattle at varying intervals from July 2001 and from a variety of sources, both affected and unaffected by bTB.

Methods

Study design

There were three approaches to this study. We used two cohort studies to investigate the bovine and environmental risks for bTB breakdown. First, we recruited 148 cattle farms situated within the Randomised Badger Culling Trial (RBCT), principally in Devon and Gloucestershire, and collected data from these farms for three years. Second, we collected data relating to 6000 restocked herds throughout Britain using previously collected data from national databases. The third approach was to use these data and other published results to investigate the transmission dynamics of bTB by construction of mathematical models.

The data

We used three national databases for this research in all three approaches.

1. The FMD database (Keeling et al. (2003) to identify cattle farms that were depopulated in 2001. We acknowledge the contribution of Dr Keeling in making these data available to us.

2. The British Cattle Movement Scheme (BCMS) database of cattle movements. In 2000 it became compulsory to report all cattle movements and all cattle births. Data from before 2000 are less reliable. We acknowledge the help of BCMS staff in giving us access to these data.

3. The VetNet database that stores all bTB testing and test results in Britain. Data are complete from approximately 1995. We acknowledge the help of the Veterinary Laboratory Agency (VLA) staff in making these data available.

We accessed these data with appropriate permissions from Defra.

Study 1. A prospective study of 148 cattle farms from within the Randomised Badger Culling Trial.

Approximately 450 farmers located within the RBCT were contacted via addresses acquired from VetNet. Note that this included all the farms which we believed to have been restocked following FMD. From these, 148 farmers were recruited. Approximately 25% of the 148 had restocked after FMD

(exposed) and 75% had stock restricted during FMD (unexposed). The exposed and unexposed farms were non-contiguous but were matched within treatment regimes in the triplets of the RBCT. Over 90% of farmers remained in the study for the full 3 years.

Each farm was visited on at least three occasions. A sample of blood was collected from all adult cattle at each visit. The farmer was interviewed once using a detailed questionnaire that requested data from 2000–2003. The interview took approximately three hours to complete, and included detailed data on spatio-temporal grouping of all cattle on the farm as well as farm management, disease and disease control. All buildings were surveyed at a separate visit. Cattle movements were obtained from cattle passports on the farm and from the BCMS electronic records of cattle movements. All data were entered on a purpose-built database. Farmers were contacted where discrepancies or inconsistencies were noted on questionnaires. These data are secure and archived.

The data link into a related BBSRC funded project on endemic diseases of cattle. The 30,000 sera collected from the cohort study have been bar coded and stored in a walk-in freezer at the University of Warwick. These are currently being analysed for antibodies to BVDV, IBR, Johnes’ disease, leptospirosis and Neospora, with funding from BBSRC (BBS/B/04854). This project is due for completion in 2008, and will not be presented in this report.

Results

A total of 75 (50.6%) of the 148 herds had at least one HBD by November 2004. There were 59/112= 52.6% of the continuously stocked herds and 16/36= 44.4% of the restocked herds with at least one HBD. This difference was not statistically significant. There were 56 confirmed HBD; 52 were culture positive. A total of 26 herds had at least two HBD in this time (Table 1). Whilst restocked herds appeared to be at reduced risk of HBD immediately after FMD, the two groups had similar bTB incidence during the whole study (Figure 1).

The significant associations (p<0.05) between herd breakdown and farm management practices were: purchasing steers, purchasing cattle from markets (versus private sales), increasing herd size, storage of manure in enclosed spaces, not feeding supplementary minerals and vitamins (Table 2). There was no statistical association between farmer reporting of presence of, and direct contact with, badgers, deer or cats.

Animal risks for becoming a reactor

The previous analysis addresses the risk of a farm breakdown, i.e. there being reactor animals on the farm at the time of test. However, since it is the cattle which become infected, and cattle move between farms, it is more appropriate to address the risks of cattle being reactors at each test performed. To investigate the risks of cattle reacting to the bTB skin test a three level hierarchical model was developed (tests within animals within herds). The outcome variable was that a bovine became a reactor at a particular test (referred to as an animal-test). Hence the data were clustered by repeated tests within bovine within herd (the three random variables in the hierarchical model). There were approximately 120,000 animal-tests on 55,500 cattle that disclosed 1000 reactors in the cohort farms during the study. Full movements were traced for 838 reactors. The results are presented in Table 3.

The factors associated with increased risk of an animal becoming a reactor were: age when tested, number of animals tested on the same day, previous exposure to reactors and farm history of TB. Factors that were protective were animal-test on a restocked farm compared with a continuously stocked farm and a short interval test (i.e. following a HBD). In all the models fitted, the RBCT triplet type of the farm did not reach a statistical significance of 0.05, with confidence intervals including 1, but being in a reactive triplet increased and a proactive triplet decreased the risk of becoming a reactor.

Since age itself is not explanatory, we presume that increasing age is a surrogate variable for exposure. We counted the number of reactors that each bovine had been exposed to on the study farm,

and, for purchased animals, on the immediate source farm before each animal-test. For many purchased cattle there was no record of a test being conducted whilst the bovine was on their previous farm. When the number of reactors that a bovine was exposed to in its life was cumulated and tested in the model in Table 3, the majority of other risks for becoming a reactor became non-significant (Table 4), indicating that previous exposure to reactor cattle was a major risk for a bovine to react to the skin test.

Figure 1. Kaplan-Meier survival curve for 148 cohort farms from start of unrestricted status to first positive test (HBD).

Table 1. The number and percent of HBD and number of reactors among all 148 farms and among the 115 with adult cattle.

# HBD (Oct 01-Nov 04)

All 148 farms in study 1

 

Farms with adult breeding cattle

n = 115 

Farms with no adult breeding cattle

n = 33Number Percent Number percent Number Percent

none 73 49.32 56 48.70 12 36.36at least 1 75 50.68 59 51.30 21 63.64

only 1 49 65.33 38 64.412 22 29.33 17 28.813 4 5.33 4 6.78

number of reactors per HBD

0 6 5.71 1 1.19 5 23.811 28 26.67 21 25.00 7 33.332 16 15.24 15 17.86 1 4.763 14 13.33 12 14.29 2 9.524 9 8.57 6 7.14 3 14.295 3 2.86 2 2.38 1 4.76

6-10 11 10.48 9 10.71 2 9.5211 to 20 9 8.57 9 10.71 0 0.00 21- 40 7 6.67 7 8.33 0 0.00

> 40 2 1.90 2 2.38 0 0.00Total HBD 105 100.00 84 100.00 21 100.00

Table 2. Cox proportional regression multivariable model of management factors associated with HBD in 148 cattle farms

ExposureNumber of. herds with exposure

Hazard ratio Coefficient. Standard

errorSignificance

Value (P)

95% CI(significant when

they do not include 1)

Herd had TB before 2001 (yes / no)

75 0.47 0.14 0.29 0.64 0.65-2.03

Farm restocked after FMD 36 0.92 -0.09 0.34 0.8 0.47- 1.79

log number herd tests prior to first HBD 148 2.33 0.85 0.3 0.005* 1.30- 4.20

Herd had dairy cattle 57 1.86 0.62 0.31 0.042 1.02- 3.39log herd size 148 1.38 0.32 0.17 0.062 0.98- 1.93Manure stored in enclosed space 83 1.98 0.69 0.26 0.009 1.19- 3.31

minerals lick and vitamins provided 72 0.42 -0.87 0.26 0.001 0.25- 0.69

Big bale wheat straw used on farm 22 1.56 0.45 0.33 0.17 0.82- 2.98

Bulls hired in 40 0.48 -0.73 0.29 0.014 0.27- 0.86Farmer stated abortions occurred in cattle 16 1.97 0.68 0.37 0.064 0.96- 4.02

Purchase cattle from market 65 1.77 0.57 0.25 0.022 1.08- 2.88

Purchase steers 40 1.96 0.68 0.32 0.033 1.05- 3.65

P < 0.05 indicates that this result is expected to occur with a probability less than 5% if the variable is unrelated to HBD. This is interpreted as statistically significant within this report.

Table 3. Multivariable model of animal level risks of becoming a reactor

Exposure Category of exposureNumber of reactor cattle

Percent of reactor cattle

Odds ratio95% confidence intervals (not significant when include 1)

Restocking status Continuously stocked 747 0.69Restocked 91 0.21 0.19 0.13 – 0.29

RBCT triplet Survey 219 0.48Reactive 328 0.57 1.11 0.88 – 1.39Proactive 219 0.46 0.83 0.66 – 1.05

Age tested Up to 3 years 261 0.223 – 6 years 344 1.86 6.39 5.27 – 7.74Over 6 years 233 1.64 8.73 6.32 – 12.07

Type of test Whole herd test 213 0.49Short interval test 423 0.64 0.44 0.35 – 0.56Other tests 202 0.50 0.70 0.55 – 0.89

Number of cattle tested Up to 200 323 0.42

More than 200 515 0.69 1.71 1.41 – 2.08Previous exposure to reactor cattle No exposure 160 0.43

Exposed to reactor cattle and born on farm 336 0.68 1.75 1.38 – 2.24

Exposed to reactor cattle and purchased 249 1.11 1.65 1.25 – 2.17

TB on study farm no 166 0.30yes 672 0.70 1.44 1.16 – 1.80

The 95% confidence interval will not contain the true odds ratio on 5% of occasions. An odds ratio of unity indicates no effect. If the confidence interval encapsulates unity, then the effect of the variable on animal-test is not statistically significant.

Table 4. Model 3 amended to include cumulative number of reactors exposed to prior to current test

Exposure Category Number reactors

Percentreactors

Odds Ratio 95% confidence intervals (not

significant when include 1)

       restocked status          continuously stocked 762 0.42  restocked 95 0.18 0.79 0.46 – 1.36RCBT survey 222 0.31  reactive 342 0.40 1.48 0.88 – 2.48

proactive 293 0.39 1.24 0.70 – 2.18type test test type

WHT 216 0.34SI 434 0.39 0.28 0.23 – 0.34other tests 207 0.35 0.58 0.47 – 0.71

age up to 1year old 33 0.04>1-2 95 0.15 3.20 2.15 – 4.75>2-3 137 0.47 9.20 6.23- 13.59>3-4 131 0.72 12.29 8.23 – 18.37more than 4 years old 434 0.92 16.05 10.89 – 23.67

log number tested 857 0.37 1.02 0.94 – 1.11

Born on study farm with no previous reactors to bTB 8 0.02

purchased with no tests for bTB done on source farm and not reactors on study farm 10 0.04 1.07 0.49 – 2.34

purchased with no tests for bTB on source farm and reactors on study farm 272 0.86 17.32 9.45- 31.74

Purchased with no reactors on source or study farm 10 0.07 1.81 0.80 – 4.12

Purchased with reactors on source farm and no reactors on study farm 2 0.04 1.05 0.29-3.75

Born on to study farm with reactors 402 0.42 16.84 9.37-30.27Purchased with reactors on source farm and study farm 60 1.14 20.88 10.79-40.43

The 95% confidence interval will not contain the true odds ratio on 5% of occasions. An odds ratio of unity indicates no effect. If the confidence interval encapsulates unity, then the effect of the variable on animal-test is not statistically significant.

Study 2. A retrospective cohort study of restocked farms. During the FMD epidemic approximately 6,000 cattle farms were depopulated (Davies 2002). After the epidemic, many depopulated cattle farms were repopulated. Since 2000 it has been a legal requirement to report all cattle movements off and onto farms to the British Cattle Movement Scheme (BCMS) where they are electronically stored. It is also a legal requirement that all newly formed cattle herds have a bTB test, the herd check test. In this research we used BCMS and VetNet to investigate the associations between herd breakdown with bTB in restocked farms and the source of purchased cattle.

Study population and study period

A total of 6,213 cattle farms located in 58 counties of GB that were depopulated during the FMD epidemic of 2001 were identified from a database held by Keeling et al. (2003) that has been compiled from data relating to the FMD epidemic. The county, parish, holding (CPH) number was used to identify restocked herds and compared with bTB test results stored in VetNet and to identify data on the source of restocked cattle from the British Cattle Tracing System (BCTS) (Mitchell et al., 2005).

There were cattle movement records from both before and after the epidemic for approximately 4200 restocked farms. A total of 2,491 of these 4200 herds were tested at least once by August 2004 (after the FMD epidemic) with a herd test (indicating that the herd was not restricted with bTB at the start of FMD). The herd types included check tests (VE-CT), routine herd tests (VE-RHT) and whole herd tests (VE-WHT). These herds were used for further analysis.

Statistical analyses

The binary HBD was the outcome variable. All continuous explanatory variables were transformed using loge(number +1) and categorised into deciles to test for linearity. Colinearity with other continuous variables was also investigated. Source farms where cattle were tested at an average interval of less than once per two years in the six years prior to FMD were defined as from “high frequency tested herds”. The categorical explanatory variables tested were calendar period from the start of restocking, a history of bTB in the restocked farm before depopulation from 1996 – 2000 and whether dairy, beef or both dairy and beef breeds of cattle were purchased.

All statistical analyses were carried out using MLwiN 1.0006 (Rasbash et al., 1999). A one level binomial logistic regression model with herd as the unit of comparison with second order MQL estimation was developed. Time to first test from restocking was forced into the model and then each

variable was tested in turn and the one with the greatest improvement in the likelihood was left in the model. This process was repeated until all variables significant at p<0.05 were included. Goodness of fit was assessed using the Hosmer Lemeshow statistic (Dohoo et al., 2003).

Results

Restocking, testing and breakdownsThe first study farms restocked in April 2001. The majority of farms were restocked from

November 2001 (Figure 2a). Testing of restocked herds was concentrated around spring 2002 and November 2002-February 2003 (Figure 2b). The maximum number of herds restocked but untested occurred from January to March 2002 (Figure 2c). The first test on each farm occurred between 30 th

September 2001 and 17th August 2004.

History of bTB on the restocked farmThere were 177/2941 (6.0%) farm HBD at the first test after restocking. There were 246 farms where bTB had been detected before the FMD outbreak and 52 (21.1%) of these herds broke down at the first test compared with 4.6% of HBD in herds on 2695 farms with no history of bTB (Table 5)

Table 5 Number of farms with herd breakdown (HBD) with bovine tuberculosis by year and the proportion that broke down at the first test after restocking.

Year of last farm HBD

Number with HBD

Number that broke down in first test

Proportion that broke down in first test

Crude odds ratio

Lower 95% CI

Upper 95% CI

2000 53 17 0.32 9.708 5.244 17.9751999 31 8 0.26 7.149 3.080 16.5931998 34 6 0.18 4.406 1.759 11.0341997 42 6 0.14 3.425 1.392 8.4231996 70 8 0.11 2.654 1.224 5.755

Figure 2. a) Distribution of herds by date of start of restocking. b) Distribution of herds by the month of testing. c) Number of herds in the process of restocking and untested by month.

Date of start of restocking

No.

her

ds

010

030

0

All herdsHerds north of 54º 30' 42''Herds south of 54º 30' 42''

June 2001 Dec 2001 June 2002 Dec 2002 June 2003 Dec 2003 June 2004

Date of testing

No.

her

ds

010

030

0

June 2001 Dec 2001 June 2002 Dec 2002 June 2003 Dec 2003 June 2004

No herds restocked untested

No.

her

ds

050

015

00

June 2001 Dec 2001 June 2002 Dec 2002 June 2003 Dec 2003 June 2004

Purchasing practicesThere were 343,141 animals purchased; 21% of these had left the restocked herd (sold or died) before the herd was tested for bTB. The restocked cattle were sourced from 18,670 herds. Each source herd (the previous herd that the bovine was in before joining the restocked herd) provided a median of 6 animals to each restocked herd (inter-quartile range 2-19). Approximately 70% of purchased animals were born in the source herd. The source herd could not be determined for 7,980 (2.3%) cattle, and a further 2.1% of cattle spent less than 30 days in the source herd and their previous location was unknown. Both these groups were excluded from the analysis because their bTB testing history was unknown. Approximately 43% of cattle were purchased from within the same county as the restocked herd.

Factors associated with time to testingVariables positively associated with early time to test were number of cattle purchased and tested, number of cattle purchased from high frequency testing herds, and number of cattle potentially tested (on the farm and within the correct age range) in the source herd. Variables negatively associated with early time to test were number of animals purchased and then sold prior to test, per cent of dairy breed animals in herd, number of animals from source herd tested less than one year previously and start of restocking later than December 2001. There was no significant difference in time to test between herds with and without a history of bTB.

Binomial logistic regression model (Table 6)

There were three principal factors associated with an increased risk of HBD. First, the number of cattle sourced from high frequency testing farms with an OR = 1.35 for each log increase in cattle number. That is, when 10 cattle were purchased from a high frequency farm the odds ratio = log e (10+1)* 1.35 = 3.24 and similarly for 100 such cattle purchased the OR for HBD = 6.22. This purchasing behaviour was highly skewed, with few farmers purchasing many such cattle. However, a total of 48% farms purchased at least one bovine from a high frequency testing herd, and the risk of HBD among those purchasing at least one such bovine was OR = 10.8.

Table 6. Multiple binomial logistic regression analyses of associated risks with herd breakdown with bovine tuberculosis in GB from 2002 – 2004 in 2941 restocked herds

Explanatory variable Number of herds coefficient s.e. OR lower 95% CI upper 95% CI

Restocked before June 2001 179 -4.652 0.484

July - Dec 2001 1559 -0.106 0.303 0.899 0.491 1.649Jan - June 2002 970 -0.57 0.34 0.566 0.287 1.116July - Dec 2002 170 -0.264 0.527 0.768 0.268 2.203After Dec 2002 63 -0.907 1.066 0.404 0.048 3.404Log number of cattle sourced from high frequency testing area 0.299 0.049 1.349 1.223 1.487

never had HBD with bTBLast HBD in 2000 1.779 0.565 5.924 1.914 18.338Last HBD in 1999 1.532 0.336 4.627 2.363 9.061Last HBD in 1998 1.327 0.449 3.770 1.536 9.253Last HBD in 1997 1.079 0.486 2.942 1.113 7.776Last HBD in 1996 0.681 0.487 1.976 0.746 5.233Last HBD in 1995 0.617 0.402 1.853 0.829 4.141Log number of cattle tested 0.323 0.086 1.381 1.163 1.640OR = odds ratio, CI = confidence intervals, s.e. = standard error

A second associated risk was that the restocking farm had a history of bTB. Although a history of bTB was a rare event (crude estimates in Table 5) the associated risk was large. The adjusted risk ranged from an OR = 5.9 when the last herd breakdown was in 2000 to OR = 2.9 when the last HBD was in 1997; farms where the previous HBD was before 1997 were not at significantly greater risk than herds with no history of HBD although the trend was still a positive risk.

The third key risk was that the greater the number of cattle tested, the greater the risk of HBD with an OR = 1.38 for each increase in log (cattle numbers +1).

The number of cattle tested was positively correlated with the number of cattle sold before the bTB test, the number of source herds from which cattle were purchased, the proportion of source herd purchased and the proportion of cattle purchased from high frequency testing areas. These variables were not significant in the multivariable model.

Model fitThe Hosmer Lemeshow statistic was not significant indicating that the model was not a bad fit to the data.

DiscussionThe FMD epidemic provided a natural experiment to study the impact of purchasing cattle on the occurrence of herd breakdown with bovine tuberculosis in cattle herds on farms that restocked after the epidemic. All farms in Study 2 were depopulated, remaining empty for a variable period of time during and after the FMD epidemic, consequently, there was no possibility of bTB positive cattle that had escaped detection at previous bTB tests on the restocked farm. Any infection in the herds after restocking must necessarily have come from either purchased cattle or another source of M. bovis which may include cattle on neighbouring farms (Green and Cornell, 2005), badgers (Donnelly et al., 2005) or

persistence in pasture, manure, feed, (Ramirez et al., in preparation (see Study 1 above); Johnson et al., 2004). The results indicate that infection in restocked herds came from both cattle and “environmental” sources.

The number of cattle purchased from high frequency testing herds was a significant associated risk in the final model, indicating that sourcing cattle from herds of high testing (and therefore where bTB has been detected but not cleared) was a risk for HBD. This highlights the associated risk of purchasing cattle from potentially infected farms. It also demonstrates that testing did not reduce the prevalence of infection in the source herds to that of less frequently tested herds. This suggests that testing frequency is targeted at farms more likely to be infected, but fails to clear the infection.

Restocked farms with a history of HBD before 2001 were a risk to the herds that were formed on them. This was a stationary risk associated with the farm as described above. This risk waned exponentially as the time since last HBD increased to a non-significant risk when HBD had not occurred for five years (Tables 5 & 6, Figure 3). This risk period far exceeded the time for which farms were empty of cattle as a result of FMD. There are two interpretations of this waning. First is that the “environment” becomes contaminated with M. bovis, but is unable to support a sustained risk to cattle. Thus, after a period of 5 years, the environmental reservoir decays to zero, and the horizontal axis in Figure 3 should be interpreted as time. Second is that farm level intrinsic risk, perhaps due to management or local geographical factors, is heterogeneous, i.e. some farms are more likely to breakdown than others and this difference is constant. Farms that are more likely to breakdown will have a shorter time between breakdowns, and the horizontal axis in Figure 3 should be interpreted as time between breakdowns, representing the risk of breaking down.

Figure 3. The odds ratios of herd breakdown estimated from the final multiple binomial regression model (Table 3). The decline with time since last breakdown is indistinguishable from an exponential decay of risk with rate 0.25 yr-1.

0

5

10

15

20

1 2 3 4 5 6

Time since previous HBD (t) (years)

OR

of H

BD

There was an increased risk of HBD associated with the number of cattle tested, in line with previous reports (Griffin 1996; Marangon, Martini et al. 1998; Green and Cornell, 2005). This may be explained by an increased probability, of detecting at least one reactor after purchasing at least two exposed bovines with increasing number purchased. That is, with a test sensitivity of 70% there is, by definition, a 70% probability of detecting one truly positive bovine but a 91% probability of detecting at least one of two positive cattle at a test, thus increasing the probability that at least one infected bovine is detected at the bTB test.

Bearing in mind that this sample is a relatively large proportion of restocked herds nationally, these three variables explain much of the pattern of HBD in the first test after FMD Table 7. There is an overlap in risk exposure between purchase and environmental risks: there are few farms that purchased animals from high testing frequency herds and had had a HBD since 1995 (105 vs 246). This is presumably because most cattle are bought locally and bTB is geographically clustered. The effect of the purchase risk on the adjusted estimates of environmental risks (Table 3 vs. Table 4: the adjusted OR is 3 times smaller than the crude OR) is suggestive of the “environmental” risk being at least related to purchasing behaviour and/or the amount of bTB in cattle locally. If 3% is the background HBD rate for unattributable risk (60/2128 herds without bTB history and lower purchasing risk), this can be compared to 11% (65/567) with higher risk purchasing and 15% (105/706) with higher risk purchasing and environment.

Table 7 Descriptive statistics of the number of herds by the three main associated risks identified in model in Table 6

  <= 9 Animals from HFT 10+ Animals from HFTHerd size Herd size

LastHBD <21 <61 <131 130+ All <21 <61 <131 130+ All Totala) number of herdsnone 742 521 503 362 2128 32 107 178 250 567 26952000 2 2 0 2 6 0 0 3 7 10 161999 2 6 3 2 13 2 10 12 16 40 531998 2 7 0 4 13 1 3 6 8 18 311997 4 1 5 3 13 3 5 5 8 21 341996 6 7 1 6 20 1 7 6 8 22 421995 11 9 12 10 42 2 5 9 12 28 70Total 769 553 524 389 2235 41 137 219 309 706 2941

b) Number of herd breakdownsnone 11 8 16 25 60 2 6 18 39 65 1252000 0 0 0 0 0 0 0 2 5 7 71999 0 0 0 1 1 0 2 6 8 16 171998 2 0 0 1 3 0 1 2 2 5 81997 0 0 0 1 1 0 1 1 3 5 61996 1 1 0 0 2 0 1 0 3 4 61995 0 1 2 2 5 0 2 0 1 3 8Total 14 10 18 30 72 2 13 29 61 105 177

c) Proportion of herd breakdownsnone 0.01 0.02 0.03 0.07 0.03 0.06 0.06 0.10 0.16 0.11 0.052000 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.67 0.71 0.70 0.441999 0.00 0.00 0.00 0.50 0.08 0.00 0.20 0.50 0.50 0.40 0.321998 1.00 0.00 0.00 0.25 0.23 0.00 0.33 0.33 0.25 0.28 0.261997 0.00 0.00 0.00 0.33 0.08 0.00 0.20 0.20 0.38 0.24 0.181996 0.17 0.14 0.00 0.00 0.10 0.00 0.14 0.00 0.38 0.18 0.141995 0.00 0.11 0.17 0.20 0.12 0.00 0.40 0.00 0.08 0.11 0.11Total 0.02 0.02 0.03 0.08 0.03 0.05 0.09 0.13 0.20 0.15 0.06

Further work

By the end of 2005 not all restocked herds had been tested for a second time. Of those that had, herds were more likely to have had a second test if they had broken down at the first bTB test after restocking and these herds were more likely to breakdown than those tested that had not broken down at the first test. When a full set of tests is available there are several further hypotheses to test.

1. There are two explanations for the pattern of reduced risk with time since last HBD on the farm. The first is that for each year that a farm does not fail the risk of HBD decreases, the second is that there is between farm heterogeneity in risk and the exponential pattern describes that heterogeneity. Simulations show that both hypotheses can produce the pattern above, although the interpretation is complicated by variation in time between tests. We expect that when all herds

have had at least one more test there will be more information to enable us to clarify these hypotheses, and possibly test them formally.

2. By definition of the study, all cattle that were reactors at the first test after restocking were imported, at second and subsequent tests each farm will have homebred and / or purchased cattle. We can compare four groups of cattle, those purchased at restocking, those born before the first test, those purchased since the first test and those born on the farm since the first test. We can then test the hypothesis that cattle are at greater risk from introduction into infected herds from naïve herds and vice versa.

Study 3. Mathematical Modelling

Many approaches have been considered throughout the project, and we present the two most informative here. They take very different perspectives: a national level (i.e. a group of interconnected herds in which herds may be positive or negative), and a herd level (i.e. a group of animals in which animals may be infected or not).

Theoretical Framework of Surveillance / Control Interaction

A simple farm-level mathematical framework was developed which included the process of diagnosis, clearance and restriction of infected farms. The model was used to investigate the interactions between testing (surveillance) effort and control. The results show that success of the bTB control programme is dependent on the frequency of testing in relation to the duration of infectiousness, and not on the prevalence of infection: reducing the frequency of testing results in increased prevalence of infection but reduced ability to detect it. Increasing testing decreases prevalence and increases detection. Ability to control infection is constrained by the sensitivity of the test, the effectiveness of clearance of infected herds and the effectiveness of movement restrictions (Medley, 2003). See Figure 4 for example of the expected trends over time in numbers of infections and disclosed infections (restricted herds) as a consequence of changing testing interval.

Empirical analysis of testingThe results of the previous model indicated that testing frequency is critical to both surveillance and control of bTB, and consequently a study was undertaken in collaboration with A. Mitchell (VLA) to investigate exposure of cattle to bTB testing by age, sex and location. The VetNet and BCMS databases were interrogated to find cohorts of cattle, their movements and the testing of the herds on which they resided. The data indicate that the majority (>70%) of cattle are never in a herd when it is tested, and are therefore untested during their lifetime. The majority of the testing effort is concentrated in a small minority of herds and animals (Mitchell et al., 2006) (Table 8). Testing of the cattle population is based on history of HBD in the locale rather than homogeneous and is targeted to take into account the geographical gradient of TB incidence in the population. The large number of untested cattle is partly due to the testing programme being targeted at farms believed to be at higher risk, so that the majority of farms are tested on a 4 year cycle, meaning that many animals can be born and slaughtered in between scheduled tests. There did not appear to be any clear relationship between testing and animal movement.

Figure 4. The expected (model) outcome in numbers of herds infected and restricted as a result of changing testing frequency (from Medley, SVEPM 2003).

Explanation: The model is started at equilibrium with testing interval every 3 years in a population of 1000 herds: approximately 1/3 are infected (heavy dotted line), 50 are restricted (heavy solid line) of which 45 are true positive restrictions (dashed line) and 5 are false positives (solid line). At 1 year, the testing interval is reduced to annual, with a consequent increase in the number of diagnoses and greatly improved infection control. At 11 years, the testing frequency is reduced to every 4 years, with a consequent reduction in control effectiveness and detection (both indicated by reduced number of restrictions).

Table 8. Frequency distributions of total number of tests per animal per premise for a cohort of 96,862 animals born in 2000 (from Mitchell et al. SVEPM, 2006)

No. Tests

Number of locations on which cattle were tested1

Number of Animals

% of Animals

0 1 2 3 4 Total0 82682 82682 85.41 7424 0 0 0 7424 7.72 2477 529 0 0 3006 3.13 1239 263 19 0 1521 1.64 733 168 9 0 910 0.9

5+ 1048 227 33 11 1319 1.4Total 95603 1187 61 11 96862

1 Cattle locations are defined as County-Parish-Holding-Herd (CPHH) identifier

Effect of Test SensitivityA critical issue with regard to understanding the epidemiology of bTB is the non-disclosure of infection in infected cattle due to the relatively low sensitivity of the comparative skin test. Tables of probabilities of outcomes with different numbers of reactors, sensitivies and specificities were created and compared to distributions of numbers of reactors found at HBD from VetNet. The results suggest that the sensitivity is in the range reported (70%-90%), and that specificity is close to 100%, indicating that false positives are a rare event.

Years

0 5 10 15 20 250

50

100

150

200

250

300

350

Number of Herds

Total, true infections

Restricted herds

Farm-level Transition ModelFrom the combined VetNet / BCMS data, 7014 herds were identified that had been tested within 3 months prior to FMD, and were unrestricted at the start of FMD (Table 9). A herd-level infection prevalence in the first test post-FMD for all herds was 15% (1233 / 7014), whereas the animal-level prevalence was 1.1% (9255 / 821216), and in four herds, all animals tested were positive.

The comparison between pre-FMD and post-FMD status was not random: in particular there were too many herds with reactors at both times (observed 157, expected 84) and too few with reactors before and no reactors after FMD (observed 325, expected 397). Herds made all possible transitions, indicating that infection must have been gained from external sources and spontaneously lost (e.g. movement of animals can achieve both).

Interpretation of these data depends on the sensitivity of the test, i.e. to what extent the negatives are false negatives. A simple transition model was used to explore the effect of a range of sensitivities. The prevalence of infection pre-FMD ranged between 6.9% (for 100% sensitivity) and 21.2% (for ~35% sensitivity). The probability of spontaneous loss of infection varied between 70% (for 100% sensitivity) and 0% (for ~35% sensitivity). The probability of introduction of infection varies between 17% (for 100% sensitivity) and 40% (for ~35% sensitivity).

Table 9. Contingency table of herd status in last test before FMD (prior to declaration of clearance) and first post-FMD test

No reactors after Reactors after

No reactors before 5456 1076 6532

Reactors before 325 157 482

5781 1233 7014

Within-Farm Transmission Model

These data can be used to parameterise a model for within farm transmission, by comparing to the observed distribution of number of reactors at the first test after FMD on bTB clear herds that were not culled during FMD. The distribution of numbers of reactors at the first test post-FMD is highly skewed with a large number of zeros, but some herds with many (>10) reactors. In particular, there are a relatively large number of herds with singleton reactors.

A number of transmission dynamic models have been fitted to these data. Briefly, Monte Carlo simulations are performed to give an expected distribution of numbers of infected animals at the time of testing. Particular issues include the herd size (and its variation) and the initial condition (i.e.. zero reactors can result from non-zero infection). The distribution of true reactors is generated from the infection data based on individual animal probabilities of reacting. The log-likelihood is calculated directly from the probability of observing the number of reactors, but since the expected distribution of reactors can only be simulated, the likelihood can only be calculated with some (simulation) error, so direct, formal fitting procedures are not possible. However, several general conclusions can be drawn from those models that fit the data (an example fit is shown in Figure 5).

Figure 5 Comparison of expected & observed distributions of numbers of infections (top row) and reactors for herd sizes up to 50 (bottom row). Note that the likelihood is calculated from individual observations not these summary distributions. Right graphs omit the zero reactors and are on the linear scale

0 5 10 15 20 2510

-2

100

102

104

Infections

0 5 10 15 20 250

20

40

60

80

100

120

0 5 10 15 20 2510

-2

100

102

104

Reactors

0 5 10 15 20 250

20

40

60

80

100

In order to generate the observed distribution of reactors several factors need to be included. First, in several herds the prevalence of reactors (and therefore infection) reaches approximately 100% of animals, so that on these farms transmission (from whatever source) must be relatively homogeneous and high. Including cattle-cattle transmission shows that R0 (the number of secondary infections from one infection in a susceptible herd) increases with herd size and is well above unity. The increase is approximately linearly related to log herd size. Recent estimates of R0 (e.g. Cox et al., 2005) suggest that the overall (within and between herd) value is just above unity, which is consistent with within herd transmission being higher. Whilst it is theoretically possible to create the same distribution of reactors without cattle-cattle transmission, the required spatio-temporal variation in infection rate is large. Second, since the prevalence of infection is relatively low, introduction of infection is infrequent, and the rate of introduction increases with herd size. Third, the large number of singleton reactor herds suggests that there must be some delay between infection and infectiousness, i.e. individual animals can be infected and remain infected without causing an epidemic.

Modelling Conclusions and Further Work

The pattern of distribution of reactors disclosed at HBD is consistent with a transmissible infection within cattle herds incorporating a delay of infectiousness. The pattern at a single test might also be explained by herd/farm heterogeneity with variation in the probability of introduction of infection (i.e. at least one reactor) and transmission potential within each farm (R0) creating probabilities of zero vs. single vs. many reactors. In order to distinguish these two hypotheses, further modelling of repeated breakdowns is required.

Overall Conclusions

The results of this study to date indicate that:

1. There is an associated risk of HBD from the purchase of cattle from herds with a history of bTB.

2. There is an associated (geographically stationary) risk of HBD from persistence of bTB on farm, outside of cattle, and this risk decays with time since HBD. The possible sources of this risk include wildlife, neighbouring cattle and inanimate reservoirs.

3. Modelling suggests that the distribution of numbers of reactors after a period without testing is highly suggestive of on-farm transmission between cattle.

4. Farm management practices associated with the risk of HBD included purchasing practices, management of manure and diet.

5. The frequency of testing for bTB determines the success of both control and surveillance. We hypothesise that increasing testing coverage / effort will result in disclosure of more infection and greater control.

References cited or referred to during the project

Bourne, J., Donnelly, C.A., Cox, D.R., Gettinby, G., McInerney, J.P., Morrison, I., Woodroffe, R., 2000, Bovine tuberculosis: towards a future control strategy. Vet Rec 146, 207-210.

Caffrey, J.P., 1994, Status of bovine tuberculosis eradication programmes in Europe. Vet Microbiol 40, 1-4.

Cheeseman, C.L., Wilesmith, J.W., Stuart, F.A., 1989, Tuberculosis: the disease and its epidemiology in the badger, a review. Epidemiol Infect 103, 113-125.

Costello, E., Egan, J.W., Quigley, F.C., O'Reilly, P.F., 1997, Performance of the single intradermal comparative tuberculin test in identifying cattle with tuberculous lesions in Irish herds. Vet Rec 141, 222-224.

Courtenay O., Reilly L.A., Sweeney F.P., Hibberd V., Bryan S., Ul-Hassan A., Newman C., Macdonald D.W., Delahay R.J., Wilson G.J. and Wellington E.M.H. 2006. Is Mycobacterium bovis in the environment important for the persistence of bovine tuberculosis? Biological Letters pp1 -3.

Davies, G., 2002, The foot and mouth disease (FMD) epidemic in the United Kingdom 2001. Comp Immunol Microbiol Infect Dis 25, 331-343.

DEFRA 2005. Tuberculosis statistics.Dohoo I.R., Martin, S.W., and Strynn H. 2003. Veterinary Epidemiologic Research. pp 360. AVC Inc,

Charlottetown PEI, CanadaDonnelly, C.A., Woodroffe, R., Cox, D.R., Bourne, F.J., Cheeseman, C.L., Clifton-Hadley, R.S., Wei,

G., Gettinby, G., Gilks, P., Jenkins, H., Johnston, W.T., Le Fevre, A.M., McInerney, J.P., Morrison, W.I., 2006. Positive and negative effects of widespread badger culling on tuberculosis in cattle. Nature. 439:843-6

Evans, H., and Thompson, H., 1981, Bovine tuberculosis in cattle in Great Britain. I: Eradication of the disease from cattle and the role of the badger (Meles Meles) as a source of Mycobacterium bovis for cattle. Animal Regulation Studies 3, 191-216.

Gallagher, J., and Clifton-Hadley, R.S., 2000, Tuberculosis in badgers; a review of the disease and its significance for other animals. Res Vet Sci 69, 203-217.

Green, L.E. and S.J. Cornell. 2005. Investigations of cattle herd breakdowns with bovine tuberculosis in four counties of England and Wales using VETNET data. Prev. Vet. Med. 70:293-311.

Gilbert, M. Mitchell, A, Bourn D. Mawdsley J., Clifton-Hadley R. and Wint. W. 2005. Cattle movements and bovine tuberculosis in Great Britain. Nature 435:491-496.

Griffin, J. 1993. The role of bought-in cattle in herd breakdowns due to tuberculosis in part of County Cavan during 1989. Irish Veterinary Journal 46, 143-148.

Griffin, J., Martin, S., Thorburn, M., Eves, J., Hammond, R., 1996, A case-control study on the association of selected risk factors with the occurrence of bovine tuberculosis in the Republic of Ireland. Prev Vet Med 27, 217-229.

Griffin, J.M., Williams, D.H., Kelly, G.E., Clegg, T.A., O'Boyle, I., Collins, J.D., More, S.J., 2005. The impact of badger removal on the control of tuberculosis in cattle herds in Ireland. Prev Vet Med 67, 237-266.

Hermoso de Mendoza J, Parra A., Tato A., Alonso, J.M. Rey J.M., Pen˜a, J. Garcı´a-Sa´nchez A., Larrasa, J. Teixido, J., Manzano, G. Cerrato, R. Pereira, G. Ferna´ndez-Llario, P. Hermoso de Mendoza M. 2006. Bovine tuberculosis in wild boar (Sus scrofa), red deer (Cervus elaphus) and cattle (Bos taurus) in a Mediterranean ecosystem (1992–2004) Prev Vet Med 74 239–247

Johnston, W., Gettingby, G., Cox, D., Donnelly, C., Bourne, J., Clifton-Hadley, R., Le Fevre, A., McInerney, J., Mitchell, A., Morrison, W., Woodroffe,R., 2005. Herd-level risk factors associated with tuberculosis breakdowns among cattle herds in England before the 2001 foot-and-mouth disease epidemic. Biol. Lett. 1, 53-56.

Keeling, M.J., Woolhouse, M.E. May, R.M. Davies G. and Grenfell, B.T. 2003. Modelling vaccination strategies against foot-and-mouth disease. Nature 421:136-142.

Lesslie, I.W., Herbert, C.N., Barnett, D.N. 1975. Comparison of the specificity of human and bovine tuberculin PPD for testing cattle. 2. South-eastern England. Vet Rec 96, 335-338.

Marangon, S., Martini, M., Dalla Pozza, M., Neto, F., 1998. A case-control study on bovine tuberculosis in the Veneto Region (Italy). Prev Vet Med 34, 87-95.

Medley, G. 2003. The design of test and clearance programmes. In: Meeting of the Society for Veterinary Epidemiology and Preventive Medicine, University of Warwick, pp. 60-74.

Mitchell, A., Bourn, D., Wint, W., Clifton-Hadley, R., Gilbert, M. 2005, Characteristics of cattle movements in Britain - an analysis of records from the Cattle Tracing System. Animal Science 80, 1-9.

Mitchell, A.P., Green, L.E., Clifton-Hadley, R. , Mawdsley, J. , Sayers, R., & Medley, G.F.2006 An analysis of intradermal comparative cervical test (SICCT) coverage in the GB cattle population. SVEPM Exeter.

Monaghan, M.L., Doherty, M.L., Collins, J.D., Kazda, J.F., Quinn, P.J., 1994, The tuberculin test. Vet Microbiol 40, 111-124.

Munroe, F.A., Dohoo, I.R., McNab, W.B., Spangler, L., 1999. Risk factors for the between-herd spread of Mycobacterium bovis in Canadian cattle and cervids between 1985 and 1994. Prev Vet Med 41, 119-133.

O’Brien, D.J., Schmitt, S.M., Fierke, J.S., Hogle, S.A., Winterstein, S.R., Cooley, T.M., Moritz, W.E., Diegel, K.L., Fitzgerald, S.D., Berry, D.E., and Kaneene J.B. 2002. Epidemiology of Mycobacterium bovis in free-ranging white-tailed deer, Michigan, USA, 1995–2000 Prev. Vet. Med. 54, 47-63

Phillips, C.J., Foster, C.R., Morris, P.A., Teverson, R. 2003. The transmission of Mycobacterium bovis infection to cattle. Res. Vet. Sci. 74, 1-15.

Pollock, J.M., Buddle, B.M., Andersen, P., 2001. Towards more accurate diagnosis of bovine tuberculosis using defined antigens. Tuberculosis (Edinb) 81, 65-69.

Robinson, C.J., and Whitehead, P., 2003, Cross-cultural management of pest animal damage: a case study of feral buffalo control in Australia's Kakadu National Park. Environ Manage 32, 445-458.

Scantlebury, M., Hutchings, M.R., Allcroft, D.J., Harris, S. 2004. Risk of disease from wildlife reservoirs: badgers, cattle, and bovine tuberculosis. J Dairy Sci 87, 330-339.

Wedlock, D.N., Skinner, M.A., de Lisle, G.W., Buddle, B.M. 2002. Control of Mycobacterium bovis infections and the risk to human populations. Microbes Infect 4, 471-480.

References to published material9. This section should be used to record links (hypertext links where possible) or references to other

published material generated by, or relating to this project.

Publications generated from the project to date

Medley, G.F. (2003) The design of test and clearance programmes. SVEPM, April 2003.Mitchell, A. Green, L.E., Clifton-Hadley, R., Mawdsley, J., Sayers, R.& Medley, G.F. An analysis of intradermal comparative cervical test (SICCT) coverage in the GB cattle population. SVEPM, April 2006.Carrique-Mas, J., Medley, G.F. & Green, L.E. Risk of bovine tuberculosis breakdowns in post-foot-and-mouth disease restocked cattle herds in Great Britain. SVEPM, April 2005.Green, L. (2005) Modelling disease – a guide for the practitioner. Cattle Practice 13, 243-248.J. Carrique-Mas and A. Ramirez have presented their work orally and on posters at a number of different conferencesPapers are being prepared for publication and will be submitted to Defra for approval in the coming months.The results of this research will be publicised on our website: www.warwick.ac.uk/go/e2