A restatement of the natural science evidence base relevant to the control of bovine tuberculosis in Great Britain†

Bovine tuberculosis (bTB) is a very important disease of cattle in Great Britain, where it has been increasing in incidence and geographical distribution. In addition to cattle, it infects other species of domestic and wild animals, in particular the European badger (Meles meles). Policy to control bTB is vigorously debated and contentious because of its implications for the livestock industry and because some policy options involve culling badgers, the most important wildlife reservoir. This paper describes a project to provide a succinct summary of the natural science evidence base relevant to the control of bTB, couched in terms that are as policy-neutral as possible. Each evidence statement is placed into one of four categories describing the nature of the underlying information. The evidence summary forms the appendix to this paper and an annotated bibliography is provided in the electronic supplementary material.


Introduction
Bovine tuberculosis (bTB) is a major disease of cattle that can also affect humans, and many other livestock and wild animal species [1,2]. Human infection has not been a major public health problem in developed countries since the introduction of milk pasteurization [3]. Advanced cases in cattle experience loss of condition, and this directly affects the economic value of the animal, but in most developed countries detection of infection leads to movement restrictions being placed on the herd, mandatory slaughter and considerable indirect losses for the farmer [4].
The incidence and geographical distribution of bTB in Great Britain has been increasing for the last two decades [5] (see also appendix; box 1), and the English and Welsh governments estimate that they have spent £0.5 billion in the last decade on testing, compensation and research with further costs being borne by the agricultural industry. All cattle herds are tested regularly for bTB, more frequently in areas of high incidence. Confirmation of infection triggers restrictions on cattle sale and movement, and the withdrawal of 'Official Tuberculosis Free Status' [4]. To reduce the risks of infection, farmers are encouraged to adopt preventive biosecurity measures. Much attention has also been paid to reducing the risk of transmission from wildlife reservoirs, of which the most important in the British Isles is the European badger, Meles meles [1,2]. There are vaccines available for bTB that provide some protection to badgers and cattle, variants on those used to protect against human tuberculosis [6]. EU law currently prohibits the vaccination of cattle as it can mask the detection of infection. The vaccination of badgers is the subject of intense current research [6,7], and vaccination has been under way in Wales since 2012 [8].
One strategy intended to reduce infection in wildlife reservoirs is culling. Badger culling was used routinely in the past [2], and its effectiveness was the subject of a major experiment, the Randomised Badger Culling Trial (RBCT), which ran from 1998 to 2006 [1,9]. Since then there has been no official badger culling, though the UK government has indicated its intention to allow culling in England, and badger culling at two pilot sites has been authorized for the summer of 2013 [10].
The prospect of badger culling has resulted in bTB policy becoming one of the most contentious areas of policy-making that involves science in the UK. The natural science evidence base is used by different sides to support different arguments, and exactly what constitutes natural science evidence has been called into question. The aim of the project described here is to provide a restatement of the relevant natural science evidence base written in a succinct manner comprehensible to non-expert readers and providing an entry into the technical literature. We have tried as far as possible to be policy-neutral, though realizing that this can never be absolute (the mere discussion of a strategy implicitly assumes it is a possible intervention). We hope that restating the scientific evidence will reveal the clear distinction between the science base, which is largely agreed, and the policy implications of that science base, which are hotly debated. This baseline summary also provides a natural starting point for a future review of evidence gaps.

Material and methods
The relevant literature on bTB in Great Britain was reviewed and a first draft evidence summary produced by a subset of the authors. At a workshop, most authors met to discuss the different evidence components and to assign to each a description of the nature of the evidence. Using existing systems such as GRADE [11], a tool for grading the quality of evidence used to support decisions in healthcare, we explored the restricted vocabulary used by the International Panel on Climate Change [12] to describe uncertainty associated with global environmental change, and ranking of evidence used by a study on bTB commissioned by the Welsh Government [13]. However, none precisely matched what we needed and instead we defined the following categories: These are explicitly not a ranking as, for example, some projections are firmly rooted in rich datasets, while some expert opinions are very much less so.
A revised evidence summary was produced and further debated electronically to produce a consensus draft. This was sent out to 25 scientists involved in bTB research, as well as to representatives from the farming industry, non-governmental organizations concerned with culling and Defra, the UK government department responsible for bTB policy. The document was revised in the light of much helpful feedback.
The project was funded by the Oxford Martin School ( part of the University of Oxford), and though many groups were consulted, the project was conducted completely independently of any stakeholder.

Results
The summary of the natural science evidence base relevant to bTB policy-making in Great Britain is given in the appendix, with an annotated bibliography provided as the electronic supplementary material.

Discussion
We note several limitations of our project and how it might be extended.
First, the project considered only the natural science evidence base. There are very important social science issues involved with bTB policy-making that would also benefit from a formal evidence summary. For example, there are complex behavioural and behavioural economic aspects to the implementation of bTB control measures by the farming industry. Furthermore, the spectrum of possible interventions available to government is moulded by debate in civil society. The European Union's Common Agricultural Policy, and how each member state interprets it, shapes the economics of the livestock industry in Europe. The way agriculture is supported in England and Wales affects the structure of the countryside, including the wild animals that can harbour bTB. An entry into the social science literature on bTB is provided in the electronic supplementary material.
Second, the review concentrates on the evidence base from Great Britain. bTB is also a major problem in the Republic of Ireland, where badgers are a major reservoir. In Australia and New Zealand, successful efforts to control bTB have included targeting, respectively, introduced water buffalo (Bubalus bubalis) and brush-tailed possum (Trichosurus vulpecula), which act as reservoirs of infection. Differences in the regulatory and social structure of farming, the countryside, and the ecology of the different reservoirs all mean that lessons from other countries have to be taken with great caution, but the approach taken in this project might be usefully extended to consider more evidence from other countries. An entry into the literature on bTB control outside Great Britain is provided in the electronic supplementary material.
Finally, the review has largely concentrated on bTB epidemiology. We have not tried to summarize the evidence base relevant to the technical or operational logistics of culling or vaccination campaigns, nor the animal welfare consequences of different interventions.
We finish by stressing this is a consensus document written by the authors, and that we accept that a different group might have included or omitted different statements and might have categorized them in different ways. Policy-makers have to integrate evidence from the natural and social sciences, as well as to make political judgements about weighing the interests of different stakeholders. We hope the current summary will make it easier for evidence from the natural sciences to rspb.royalsocietypublishing.org Proc R Soc B 280: 20131634 contribute to policy-making, and clarify where there is agreement and where dissent. We also hope that this restatement of the current evidence base will stimulate discussion about how to prioritize investment to address remaining uncertainties.    (4) This document concentrates on the natural science evidence base; evidence from social sciences and economic analysis is also of great importance for policy-makers but is not included here. The document also largely concentrates on the evidence base from Great Britain. There is a need for a careful review of how lessons from bTB control in other countries with different farming systems or wildlife reservoirs can inform policy in Great Britain. (5) Despite the substantial progress that has been made in understanding bTB the natural science evidence base cannot alone determine policy to control or eradicate the disease. All policy options have costs, benefits and risks that affect the stakeholders involved in different ways. Policy-makers inevitably have to consider and weight the interests of these stakeholders, as well as balancing uncertain benefits and potential risks in deciding what actions to take. Different weightings and balances can lead to different decisions. Nonetheless, it is critically important that all policy be informed by the evidence base and that policy-makers clearly distinguish the scientific and other (economic, social, ethical and political) inputs into the decisions that have to be made.
(b) Epidemiology (6) The risk of bTB varies geographically within Great Britain; some areas have a consistently high incidence in cattle while infection has remained low or practically absent elsewhere. Annual herd testing for bTB is currently (2013) carried out over a large area of England (60 000 km 2 ) and the entirety of Wales, though disease incidence varies within this region (box 1) [D ata ]. (7) Since the mid-1980s, the incidence and geographical distribution of bTB in cattle has increased markedly in England and 96-99%) in the meta-analysis cited above). As implemented, IFNg identifies some exposed cattle not identified by the skin test and has a median estimated animal-  Although the early suspension of reactive culling prompted debate over the causal interpretation of these primary results, subsequent analysis of data from within the reactive culling areas found that the presence and extent of badger-culling activity were associated with significantly increased risk of a confirmed herd breakdown on nearby farms, and that when compared with no-cull areas the breakdowns were more prolonged [D ata ]. (d) Culling in the RBCT had no effect (positive or negative) on the incidence of unconfirmed breakdowns [D ata ].
(31) Culling badgers is known to disrupt badger social structure, and this has been shown to cause badgers to move more frequently and over longer distances [D ata ]. This effect is known as perturbation. The idea that perturbation may result in increased disease transmission (to other badgers and to cattle) has been termed the 'perturbation hypothesis' or a 'perturbation effect' [E xp_op ]. proactive culling similar to that carried out in the RBCT predicted that over a 9.5 year period with proactive culling in the first 5 years there would be a relative reduction in confirmed herd breakdowns of 20-34% (central figure 27%) within the culled area. When the additional herd breakdowns in a peripheral 2 km-wide area are included, the overall impact falls to 3-22% (central figure 12%) or 8-24% (central figure 16%), depending on assumptions. 6  The averages involving the post-trial period include 5 years of data; choosing a different time span would affect their values. (iii) The figures above are a comparison of cull and non-cull sites, and hence represent relative differences. As background incidence was rising throughout the monitoring period, absolute reductions in rates of new confirmed cattle herd breakdowns (compared with historical rates) would be smaller than the relative reductions shown here, and absolute increases would be larger than the relative increases shown here. (iv) RBCT culling had no impact on approximately 30% of cattle herd breakdowns, which are unconfirmed.

Endnotes
1 The term 'eradication' is often used in the context of bTB to refer to local reductions in incidence to a defined level. However, the correct term for this level of ambition for an intervention is 'control': the reduction of disease incidence to a locally acceptable level as a result of deliberate efforts. 2 These classifications are explicitly not a ranking. 3 Cattle herds free of bTB are described as 'Officially Tuberculosis Free'; the presence of an animal testing positive for bTB results in an unconfirmed herd breakdown and the herd is described as 'Officially Tuberculosis Free Status Suspended', leading to movement restrictions. Post-mortem or laboratory demonstration of the presence of M. bovis is called a confirmed herd breakdown with 'Officially Tuberculosis Free Status Withdrawn'. 4 In proactive areas, badger culling was carried out annually on all accessible land; in reactive areas, culling was carried out once on and near farmland where bTB had been confirmed in cattle. 5 22% higher with 95% confidence intervals 2.5-45% ( p ¼ 0.025). 6 The first figure including the peripheral area assumes the baseline risk of herd breakdowns is the same in culling and surrounding areas, and the second that the baseline risk is 50% lower in the surrounding areas on the assumption that culling would take place in areas of particularly high incidence. These are relative rates and changes in background incidence rates will affect absolute differences in herd breakdowns (see box 2, paragraph (iii)). 7 The four main techniques were assessed using three criteria (accuracy, ease of conduct, affordability), in all cases 'high' being desirable: (i) counting setts extrapolated to badger numbers (low, high, high); (ii) latrine surveys extrapolated to badger numbers (medium, high, medium); (iii) mark-release-recapture (high, low, low); and (iv) DNA profiling of hair samples (high, low, low).