Defra

4.0.1 The field trial is the most contentious component of our work. It is, however, essential to future disease control policies that we obtain accurate data on the prevalence of TB in badgers and are able to relate the underlying pattern of TB infection within badger populations to the incidence of TB in cattle. It is also necessary to study the distribution of TB in badger populations, and how this is affected by variables such as population density, social group size and structure, disposition of badger territories, and past badger removal operations. Unfortunately there is at present no reliable diagnostic test for TB infection in badgers that can be used in the live animal, and accurate diagnosis can therefore only be made at post mortem. Only by carrying out the field trial can essential epidemiological data be obtained, and, critically, the field trial is the only means by which we can answer two key questions: (a) what is the quantitative contribution of badgers to TB in cattle; and (b) is culling effective in controlling the disease in cattle, and if so, in what circumstances?

4.0.2 The trial will evaluate the effects of badger culling on the incidence of TB in cattle by comparing three treatments applied in trial areas enrolled as 'triplets':- proactive culling (initial removal of as many badgers as feasible consistent with welfare constraints and thereafter maintaining numbers at a low level); reactive culling (removal of badger social groups with access to a farm in response to an outbreak of TB on that farm); and survey only (no culling).

4.0.3 If badgers cause a substantial proportion of herd breakdowns in the trial areas, then the areas subjected to the proactive strategy should be the first to show reduced TB incidence in cattle. The reactive strategy is designed to determine the effect on TB incidence in cattle of removing badgers located around TB breakdown farms, which are more likely to be infected with M. bovis, while leaving other social groups relatively undisturbed. The survey-only areas are also vital to the trial as they establish the effects of no culling and allow comparison with the two culling strategies.

4.0.4 The trial is one of the largest scientific field exercises ever undertaken and hence has needed meticulous planning and an ordered approach to maximise its effectiveness. In addition to designing the trial to generate the rigorous scientific data needed, and constructing the intellectual framework within which those data will be analysed, our approach has involved careful consideration and review of operational factors, and of the welfare, ethical and environmental implications of the work.

4.1 Trial design

4.1.1 The design of the trial required the specification of the method of treatment allocation, the number of farms to be enrolled and the timescale of the trial.

Treatment allocation

4.1.2 In theory, the three strategies could be allocated individually to farms enrolled in the trial, but many badger social groups may have territories overlapping more than one farm and, thus, potentially be allocated to more than one treatment. To avoid this and, furthermore, to reduce the interference between different strategies, all of the farms in relatively large trial areas (roughly 100 km²) are assigned the same treatment.

4.1.3 Trial areas are grouped into triplets (with each of the three treatments being allocated to one area within each triplet). As far as possible the similarity of trial areas within a triplet is maximised, but, as explained in our first report, precise matching is not always feasible in practice; nor do we consider it essential.

4.1.4 Treatments are assigned randomly to each triplet to avoid any biases that might lead to inherent differences between the three areas. The randomisation procedure is conducted at the latest possible stage so that neither the level of consent given by occupiers, nor the determination of area boundaries, nor the intensity of work across the areas concerned, are influenced in any way.

Statistical power

4.1.5 An important feature of the planning of the trial was to ensure that the size of the trial was appropriate for its intended purpose. The statistical power calculations for the trial, originally presented in the Krebs report and adopted by the ISG, were based on the simple but reasonable assumption that the variability of numbers of observed cattle TB breakdowns is essentially that found in the Poisson distribution, the statistical distribution governing the count of events occurring totally at random.

4.1.6 Based on the historical incidence of TB in cattle across Great Britain between 1992 and 1996 inclusive, the Krebs report recommended that a minimum of 30 100km² areas should be included in the trial. It also illustrated that the expansion of the total trial area beyond 3000 km² would yield diminishing returns in terms of the number of repeat and contiguous breakdowns that would be covered.

4.1.7 The statistical power is the probability of detecting a reduction in incidence of TB in cattle, if it exists. It depends primarily on the total number of TB breakdowns (i.e. the cumulative incidence) in the survey-only (control) areas and the percentage reduction in the breakdown rate in the treatment (proactive and reactive) areas. It was suggested in the Krebs report that if all 10 triplets were implemented immediately, and incidence of TB in cattle remained at the level observed over the previous five years, that a reduction as low as 20% in the trial areas subject to reactive culling (deemed likely to be the less effective of the two badger removal treatments) should be detectable within five years. For logistical reasons, it was not possible to implement all the triplets simultaneously. Based on the original calculations of the Krebs team this would have the effect of extending the timeframe necessary for the trial to deliver the "target" referred to above. However, the disease situation since Professor Krebs reported has not been static.

4.1.8 Figure 4a below illustrates the total number of breakdowns required for the trial to have a 90% probability of detecting a given percentage reduction. In trial design terms it is accepted that 100% probability is extremely difficult to achieve, if not impossible. The minimum probability for scientific trials is generally accepted to be 80%. With 90% probability, the statistical power for the trial exceeds this. Working from the Krebs reference to an average 20% reduction in TB incidence, 385 breakdowns in the survey-only areas would be required to yield a 90% chance of detecting such an effect on the incidence rate of TB in cattle.

Figure 4a

4.1.9 To translate the total number of breakdowns into a timetable for the trial requires the consideration of 'triplet years'. Each triplet year represents one year's observation in a triplet once initial proactive culling is complete. (On the Krebs team's assumptions, 50 triplet years would have accumulated by the end of five years - 10 triplets implemented simultaneously, multiplied by five years' observation). Table 4b maps out the implementation timetable for the trial and projects the accumulation of triplet years over the calendar years ahead.

4.1.10 To calculate when a given reduction in the incidence rate of TB in cattle might be detectable, the total number of TB breakdowns required is divided by the average breakdown rates per survey-only area per year. This results in the number of triplet years necessary. Thus, if the average breakdown rate per survey-only area were 8 breakdowns per area per year, then 48 triplet years (= 385/8) would be required to yield a 90% chance of detecting a 20% reduction in the incidence rate of TB in cattle. From table 4b it can be seen that 48 triplet years should have accrued by the end of 2005.

4.1.11 Higher TB incidence in the trial areas (but the same ratio of incidence rates between treatments) would reduce the number of triplet years required to detect a difference. For comparison, if the average breakdown rate per survey-only area were 12 breakdowns per area per year, then 32 triplet years (= 385/12) would be required to yield a 90% chance of detecting a 20% reduction in the incidence rate of TB in cattle. 32 triplet years should have accumulated by early 2004. The tables at Appendix B show the three year and one year TB incidence figures on the basis of which Triplets B (Devon/Cornwall) and C (East Cornwall) were selected.

4.1.12 Obviously, if a consistent statistically significant effect of culling were to be observed earlier than either of these points, valid results could be obtained sooner than originally suggested. However, in order to evaluate the consistency of the ratios of breakdown rates between triplets, the minimum coverage necessary would be for all 10 triplets to have been observed for at least a year, and preferably two (to allow for cattle testing timetables, and any lag in the impact of culling). From Table 4b, it can be seen that on the projected timeline for the trial, the earliest this could occur would be the end of 2002.

Assumptions

4.1.13 The assumption that TB breakdowns occur totally at random will tend to underestimate the variability to be encountered in practice. For example, there is some evidence of clustering of TB breakdowns in space and time, though this is likely in part to be an artefact of the testing regimes used. The final analysis of variation will be based on the observed consistency of the ratios of breakdown rates (for example, the ratio within a triplet of the TB incidence in the proactive area divided by the TB incidence in the survey-only area) between triplets, adjusted for herd and cattle numbers and possibly other features. The amount of variability encountered will only become apparent as the trial progresses.

4.1.14 While 10 is the minimum number of triplets advisable for effective error control, the viability of the trial does not, as such, depend on the validity of the original power calculations. These were formulated in terms of detecting the presence of a 20% or greater reduction in breakdown rates and in terms of specific assumptions about the variability to be analysed. An equivalent and equally satisfactory formulation, would be the estimation of the ratio of breakdown rates with appropriate confidence limits.

4.1.15 Non-compliance with the trial through interference with culling operations, denial of access for survey or culling teams (particularly in the proactive and reactive areas) and illegal killing of badgers (especially in the survey-only area) would all reduce the differences between treatment areas. Depending on the circumstances, such factors could serve to mask the true effect of culling treatments. The statistical power of the trial is sufficient to deal with non-compliance of all these types, but if significant problems were experienced the duration of the trial might have to be extended. This would be in no-one's interests: it would delay the accumulation of the scientific data required to inform policy, and could result in more badgers having to be culled than would otherwise have needed to be the case.

4.2 Welfare and ethical issues

4.2.1 In designing and implementing the trial we have given as much weight as possible to animal welfare considerations.

Capture methods

4.2.2 We remain committed to the use of humane methods to capture badgers. Cage traps, designed specifically for the capture of badgers, are a widely accepted form of capture and allow easy identification and release of non-target animals. On our recommendation, preliminary trials have been carried out to assess alternative capture methods, such as padded leg-cuffs, but these investigations are not at a stage which would lead us to contemplate their use in the trial.

4.2.3 Standard operating procedures stipulate that traps should be set as late in the day, and then checked as early in the day, as possible, to minimise the period in which trapped badgers are held in captivity, and to limit any non-target capture. Trap visit times are kept under continual review.

4.2.4 Procedures have been formulated for humane despatch of trapped badgers by shooting, and checks have been established to ensure that instant death has occurred. A record is kept of any injuries observed during post mortem examinations. Non-target species are released except on the rare occasions when they are severely injured, in which case they are despatched humanely.

4.2.5 Humane capture and despatch operations have been subject to internal checks and external audit by an independent expert on animal welfare. The auditor's final report will be published.

The closed season

4.2.6 As described in our first report, in designing the trial we moved away from Professor Krebs's original recommendation and adopted a closed season to minimise the risk of capturing females with dependent cubs below ground.

4.2.7 The closed season (1 February - 30 April) was selected on the basis of the experience that, once cubs are old enough to appear above ground regularly (usually by late April), they can be easily captured and despatched humanely. Changes to working practices for reactive culling operations have led to the suspension of trapping over weekends (whereas proactive operations require continuous capture over a set period). For the months of May and June, when cubs might still be expected to be dependent, this raised the possibility that cubs could be left unattended if their mothers were captured late in the first week of trapping. We therefore recommended that for those particular months reactive operations should include continuous trapping over weekends.

Ethical issues

4.2.8 The ISG has considered the ethical questions posed by the field trial, and principally whether the culling of a protected wildlife species can be justified either for scientific purposes or for the benefit of another (non-protected) animal. We explored these issues at length in drawing up our first report, and have kept them under review since, discussing them with key individuals, including the independent welfare auditor.

4.2.9 The broad-based Government decision to proceed with the trial was based on a lengthy public consultation and, crucially, on the ISG's design for the trial which incorporated the extensive welfare provisions referred to above. The trial is aimed at producing conclusive results as quickly and efficiently as possible and we are satisfied that it withstands an ethical examination.

4.2.10 Until the trial has been completed it is impossible to say how many badgers will be killed. Using the assumption that average badger density is around 5/km² the Krebs report suggested that some 12,500 might be taken during a five-year trial. However, experience from culling operations to date suggests the Krebs figure might prove an over-estimate. Trial culling operations will be conducted over less than 1% of Great Britain's surface area and no more than 4% of the South West land mass (specifically, the area covered by the State Veterinary Service's West Region). On this basis we consider that it will therefore not have a significant effect on the national badger population, estimated during the last national badger survey, co-ordinated by Bristol University, to be in the order of 310,000.

On to Challenges faced and lessons learned

Site navigation

• © Crown copyright
• Terms & conditions
• Privacy
• Access keys
• Help
• Feedback

Department for Environment, Food and Rural Affairs