What was the RBCT?
The Randomised Badger Culling Trial (RBCT) was a field experiment conducted in England between 1998 and 2005, to assess whether culling badgers affected levels of bovine tuberculosis (bTB) in cattle herds.
After a badger with bTB lesions was found in 1971, badgers were blamed for slowing the eradication of bovine TB in some areas of England and Wales. Following on from small-scale culling trials that couldn’t properly assess the effects of badger culling, a larger experiment was proposed in 1997. This trial involved thirty large areas divided into three groups: 10 areas where badgers were culled proactively to quickly lower their numbers in at-risk areas; 10 where they were culled reactively near confirmed bTB outbreaks; and 10 control areas where badgers were not culled at all.
To be comparable, all 30 areas, or triplets as they became known, would be around the same size of 100 sq.km. Unfortunately, it was not possible to standarise the number of herds in each area, but it was expected that this could be adjusted for in the final analysis. Due in part to disruption to the trial as a result of the Foot and Mouth disease outbreak in 2001, the culled triplets were culled for very different lengths of time and some with a gap of one year. But again, it was hoped that this could be adjusted for in the final analysis.
The triplet series where culling was implemented reactively was concluded early. This was because it was claimed that there was a detrimental herd breakdown trend emerging. Badgers disturbed by culling activity were thought to be dispersing outside their usual home ranges & spreading the disease to neighbouring cattle and badgers; the so-called perturbation effect became a hypothesis.
And the main result was…..
The results of proactive culling in the RBCT were first published in a multi-authored paper in Nature Journal in 2006 (Donnelly et al. 2006). The two statisticians on the paper were David Cox (giant of twentieth century statistics) and Christal Donnelly at Imperial College. Using a Poisson model, they reported an approximate 20% net disease benefit in proactively culled areas, and notably at a high level of significance (P<0.005).
Due to the perception of ‘negative effects’, overall, the Independent Scientific Group (ISG) that ran the RBCT experiment concluded that badger culling was not worthwhile. In their final 2007 report, they stated that badger culling could make “no meaningful contribution to cattle TB control in Britain.” Despite this, the Chief Scientific Advisor at the time Sir David King recommended badger culling to larger areas with ‘hard boundaries’, so that negative effects were minimized. In 2011 academic experts for the government concluded that such a strategy could produce a 16% disease benefit.
What happened next?
Following years of wrangling, legal challenge, political interference, even private encouragement from royalty, and no doubt much else, a policy of intensive four-year badger culling began in 2013. This continued, together with follow-up ‘supplementary’ badger culling for 5 years. Some targeted culling of ‘hotspots’ was initiated in the Low Risk Area, which ran until the end of January of 2026. Around one quarter of a million badgers were culled. The majority of them were healthy, and a high proportion were killed inhumanely.
What’s changed?
In 2022, a paper was published in Veterinary Record journal, that analysed the results of up to three years of the industry-led badger cull to 2020 (here). It could not detect any difference in herd breakdown between culled and unculled areas. In other words, there was no clear evidence of a benefit from badger culling from detailed examination of government data, despite its predictions.
So why was industry-led badger culling not showing the clear disease benefit predicted by academics around 2011-2013. Professor Paul Torgerson, who had been one of four reviewers on the Vet Record paper, decided to re-examine the Donnelly et al.2006 paper, to see why the results were so different.
He found that the RBCT analysis had not taken the differences in number of herds between the triplets (sample areas) into account in the conventional way. This was despite the text implying that it had. The differences in the number of herds per unit area, and the differences in the ‘time at risk’ for each area, (the length of time badgers were culled in that sample area), were not adjusted for using the methods normally employed in epidemiology. Instead, each triplet was treated as a variable, leading to an unorthodox model with almost as many variables as samples. The questionable conclusion that badger culling had a significant effect (both positive and negative) on bTB in cattle was simply a function of what experts have subsequently commented was ‘naïve at best’ or ‘unnatural’ approaches (see here).
New publications
In 2022, Professor Torgerson and a group of scientists formed around these problems, and wrote up their findings. They produced more conventional and credible analyses of the data, preprinted them for all to see that year and published them in a paper in Nature Scientific Reports (Torgerson et al 2024). The new analyses, using conventional epidemiological approaches, showed no effect (positive or negative) of badger culling on cattle bTB incidence from the RBCT data. Reversing the original finding.
Badger Crowd has presented a number of blogs on the subsequent publications (see here), but in brief, there were responses to Torgerson et al 2024 with two papers (Mills et al 2024 I & II), which were in turn was responded to in a Comment paper in 2025 (Torgerson 2025), see review here.
The 2025 review of science publications since the (2018) ‘Godfray review’.
The 2025 Godfray review panel (see here), included a new statistician to replace the recused Professor Christl Donnelly; Professor Sir Bernard Silverman. Silverman had accepted that the Donnelly et al (2006) paper had not been the ‘natural’ way to deal with the data. But instead of accepting Torgerson’s statistics, he set about doing his own analysis using a new approach; the binomial model. He found a low-significance effect of badger culling (<0.05) with his preferred model, apparently throwing the RBCT a thin lifeline.
However, as its name suggests, a binomial model models events with exactly two possible outcomes. The problem is that bTB breakdowns in cattle herds, both during the RBCT and now, are not binary; a herd may break down multiple times even in a single year, as they did during the RBCT. In fact around one third had repeat (two or three) breakdowns over the period of study. This means that the binomial approach is simply not appropriate, which is why it was ruled out when the RBCT was designed.
There are other problems with Silverman’s efforts. These include reporting incorrect Information Criteria on models, not using ‘time at risk’ as a variable, and not taking note of problem residuals on his preferred model. Professor Torgerson has posted a preprint outlining just some of these major problems (here).
But wasn’t Silverman’s model peer-reviewed?
As a government report, the Godfray review peer-review process does not appear to have been as rigorous as it would have been for a standard journal. Content approval seem to have been handled by the panel themselves in an informal way.
Enquiries suggest that the ‘peer reviewers’ don’t seem to have been told about the all-important ‘time at risk’ data, and had not received the correct information criteria output for the Silverman model. They do not seem to have known about the repeat breakdowns which make the binomial inappropriate. One reviewer has since expressed a view that the results of the RBCT experiment are inconclusive due to its design limitations.
A different peer reviewer suggested that a herd-based analysis (rather than the triplet based analysis that has been used in all analyses) could be used on the data (see reviews here). They appeared not to know, that this has already been done (see Vial et al 2011), and that no effect of badger culling was detected using this approach. This result was dismissed by the RBCT authors in favour of the log-linear analysis model in the original Donnelly 2006 paper, which has now been shown to be inferior.
Where are we now?
Both Silverman and Godfray have responded in writing to say that ‘reasonable people can agree to disagree’, and declining to engage further, even when Defra have suggested that they should. The tactic now is to say that the science is ‘old’ and of little relevance. But they are still unable to provide any other evidence that badger culling can result in any positive and/or negative effects on bTB in cattle. They point to more recent whole genome sequencing (WGS) papers which in fact just report a wide variety of qualified estimates. Ironically many of these are paramatised by and repeat flawed RBCT analytics, are often contradictory and are not definitive in estimating quantitative transmission rates, usually at their own admission.
Is Defra listening to this debate?
Over the last seven years, Defra have refused to engage with any authors of publications which suggest the RBCT had a flawed analytical process or that badger culling might not have any measureable effect on bTB in cattle. Defra must be aware of the debate in the scientific literature and are said now to lean on the WGS evidence, despite its lack of clarity. It appears that they have commissioned their favoured contractors to continue to look for ways to blame badgers.
Why are Defra so reluctant to accept that badgers are not a significant part of the bTB epidemic in cattle? Are they worried about the consequence of admitting that over a quarter of a million badgers were killed on the back of bad science while £Billions of public and industry funds have been lost due to the distraction? Are they afraid of the reaction of farmers who have been living with the consequences of undetected bTB in their herds? These farmers were told, in error, that badgers were to blame for their diseased cattle, so they have gone out and paid for them to be mass-killed. It is now hard-wired into their understanding of bTB – how might you change that? Is it even possible after all these years of misinformation?
Will Defra be prepared to perpetuate the mistakes of the past by recommending in a government response, a ‘delivery plan’ plan that includes a form of targeted badger culling? And by implementing relatively small scale badger vaccination as some kind of bizarre short term placebo? Will they stand by & watch culling policy being implemented in Northern Ireland and Wales based on their continued flawed scientific advice? The new Minister Stephen Morgan, or perhaps his replacement under a new Prime Minister, has a major challenge on his hands that the last two Ministers have failed to resolve.
