Flawed Analysis May Hide Benefits of Badger Culling

Bovine tuberculosis (TB) in cattle is linked to a wildlife host, the European Badger (Meles meles), which can carry the bacterium Mycobacterium bovis. In England, large-scale badger culling has been deployed under a licensed Badger Control Policy (BCP) since 2013 to try to reduce transmission to cattle. While several studies have reported an association between the BCP and reductions in cattle TB, a recent paper by Langton et al. (2022) did not find evidence that culling reduced TB incidence. That contradiction prompted a fresh look. Andrew Robertson, the corresponding author of the critique, summarised the main methods used in Langton et al. and identified two major problems with their approach. Langton et al. compared cattle TB incidence in culled areas of SW England with land where culling did not occur, but they pooled data from different cull areas into a single culled-versus-unculled treatment without accounting for when culling actually started. Robertson also found evidence that the choice of land enrolled for culling was biased: areas that entered culling during the study period had higher baseline TB rates than those that did not cull. These two issues are central to understanding why different studies reach different conclusions.

To test how those methodological choices might affect results, Robertson ran a Monte Carlo analysis using simulated data. He created scenarios with differing cull effects, including cases where culling produced a relatively rapid and large reduction in TB, and then applied the same degree of data grouping, the same selection bias, and the same statistical methods that Langton et al. used. The simulations reproduced the core features of the original analysis: culled areas in SW England were compared to unculled land in a combined treatment, and areas that started culling had higher baseline incidence. The results showed that even when a true culling benefit was present in the simulated data, there was a substantial chance that Langton et al.’s analytical approach would fail to detect it. In some simulated runs the analysis even produced a misleading result suggesting culling was associated with an increase in TB. Robertson’s simulation work therefore demonstrates how grouping data across different cull timings and enrolling higher‑incidence areas into culling can obscure real effects.

The critique has direct implications for how researchers and policy makers interpret studies of wildlife control and disease. If the analytical method does not account for when culling began in each area, and if areas chosen for culling already had higher TB levels, comparisons between culled and unculled land can be biased. Robertson’s findings help explain why Langton et al. (2022) reached a different conclusion from other studies that reported disease benefits from the Badger Control Policy (BCP). The work highlights the importance of rigorous statistical analysis that explicitly adjusts for temporal and spatial patterns in disease control data so that true impacts are not missed or reversed by artifact. For decision makers considering badger culling as part of bovine TB control, this critique suggests that careful study design and analysis are crucial to avoid drawing incorrect conclusions from complex surveillance data.