Bovine tuberculosis (bTB) circulates between cattle and a wildlife reservoir in parts of the United Kingdom, with badgers acting as a sylvatic host and forming a bi-directional system with cattle. Reducing bTB in badgers is a component of the national strategy to achieve Official TB-Free status. Historically, control approaches have included proactive and reactive culling and vaccination. Culling reduces the number of infectious individuals quickly but can cause a perturbation effect that may increase movement and transmission. Vaccination reduces susceptibility and disease prevalence over time without disrupting social structure and is often regarded as more publicly acceptable and useful for maintaining low disease levels after culling.
This study uses a simulation model to quantify how different vaccination schedules and field conditions affect population-level outcomes. The goal is to support operational managers in estimating likely success of vaccination and to identify when reduced-frequency vaccination (every second or third year) might provide similar epidemiological benefits while improving operational efficiency.
The TBi model is an individual-based, stochastic simulation implemented in Python 3. It operates with a two-month timestep and simulates individual badger behaviour, demography and disease processes to produce a stable population and disease epidemiology. The model has undergone sensitivity analysis and was validated against a Test, Vaccinate or Remove (TVR) pilot trial in Northern Ireland, supporting its use for predicting vaccination effects at the population level. Full parameter lists and an ODD protocol description are provided in the supplementary material.
Simulations run on a 100 x 100 grid wrapped as a torus to eliminate edge effects. Three hundred badger social groups are randomly and homogeneously distributed at a mean density of 0.75 km-2. A central circular core of 100 km2 containing 85 social groups is defined as the managed area where vaccination is targeted. Around the core is a larger non-controlled area; both regions start with the same density and background disease prevalence.
Farms (approximately 215) are randomly placed across the arena to determine participation in management; each farm has a 70% probability of participating. Whether a social group receives management depends on the participation status of overlapping farms. Within participating land in the central core, vaccination attempts occur in June of each year for badgers that are trapped.
Each simulated badger carries state variables for social group, sex, age class (cub, yearling, adult), health status and immunity status. Health states include healthy, infected, single-site excretor and multi-site excretor, with progression probabilities informed by Woodchester Park field data. Reproduction is density dependent; litter size is drawn from a distribution with a mean of 2.94 cubs and an assumed 1:1 sex ratio. Mortality is age- and sex-specific with elevated early-life mortality for cubs and higher mortality among bTB excretor classes. Dispersal can be permanent to any social group, with movement probabilities influenced by group size and sex.
Vaccination in the model represents the intramuscular BadgerBCG protocol used in the UK, which requires trapping and restraint. The simulation compares three operational schedules: annual vaccination, vaccination every second year, and vaccination every third year. Trapping (capture) probability was assumed to be 50% per badger per year; this aligns with field estimates of 50–70% capture efficacy.
Vaccine-induced immunity follows three discrete outcomes based on an expert review of BCG experiments: 70% of vaccinated badgers gain full protection (zero infection probability), 10% gain partial protection (probability of becoming infectious is halved), and 20% gain no protection (same infection probability as unvaccinated animals). These immunity states are assigned independently per individual. Further vaccination while an individual is still immune has no additional effect; if immunity is lost, re-vaccination can restore protection with the same probabilities as initial vaccination.
Population processes (births, deaths, dispersal) and disease progression are simulated using parameters from the literature and Woodchester Park data. Mortality increases for animals in excretor disease classes. The duration of vaccine-induced immunity (DoI) was a key parameter evaluated in scenarios because badger lifespans average three to four years, and few reach 10 years of age. Short DoI (one to two years) would reduce the durability of population-level protection.
A range of scenarios were run to explore the impact of vaccination frequency, population density, background disease prevalence and vaccine characteristics, including duration of immunity and capture efficacy. The model structure and parameters allow managers to simulate local field conditions and operational constraints to estimate probable outcomes for different deployment strategies.
Model results indicate that population density, initial disease prevalence, and the duration of vaccine-induced immunity substantially influence vaccination outcomes. In many scenarios, reduced-frequency vaccination (every second or third year) remained effective at reducing disease prevalence at the population level, particularly when DoI and capture rates were favourable. Lower-frequency schedules can create operational efficiencies by enabling a single vaccination team to treat multiple areas on an alternating-year timetable, shortening the delay before additional areas are treated compared with sequential annual coverage.
Because trapping campaigns are resource-intensive and constrained by access, habitat and regulation, reducing treatment frequency can be an effective way to stretch limited operational resources. By treating multiple areas in rotation, the cost per unit area can be reduced and more acreage may be covered sooner than treating areas sequentially. Managers can use the model outputs to select strategies that balance epidemiological benefit and operational practicality for local conditions.
Key uncertainties remain, notably the true duration of BCG-induced immunity in badgers, which has not been robustly quantified in controlled longitudinal studies. Capture efficacy is another operational variable that can vary with terrain, land access and team skill. While the model has been validated against field trial data and all parameters and model descriptions are supplied in supplementary files, results should be interpreted in the context of these uncertainties and local conditions.
All relevant data and supporting information are provided with the paper to permit further exploration and replication of scenarios.
An individual-based simulation model demonstrates that vaccination schedules less frequent than annual can still be effective in reducing bTB among badger populations under many plausible field conditions and vaccine assumptions. Reduced-frequency vaccination offers potential operational efficiencies that may allow more extensive area coverage with the same resources. Model outputs can help operational managers and decision makers choose strategies appropriate to local population density, disease prevalence and likely vaccine duration of immunity, while acknowledging remaining parameter uncertainties.