This is more akin to eavesdropping than proactive surveys, but it's equally effective.
We also examine the impact of various local herd-level risk factors within each of the trial group (proactive, survey-only and reactive) to identify the most important bTB breakdown risk factors for herds within the RBCT areas.
Overall, when comparing the probabilities of confirmed herd bTB breakdowns during the period under study between proactive and survey-only herds, we found that the best model included effects of triplet (p = 0.04), herd type, herd size, farm area and the historic bTB incidence for that trial area.
Although a log-linear analysis remains the most robust approach to investigate treatment effects inside the RBCT [36], we find non-significant differences in the probability of (p = 0.07) and the time to (p = 0.008) a confirmed bTB herd breakdown between herds in proactive and survey-only areas during the study period, a finding consistent with previous analyses.
Overall, when comparing the time to the first confirmed herd bTB breakdown between proactive and survey-only herds during the period under study, we found that the best model included effects of farm area, herd type, herd size, triplet and the historic bTB incidence within the trial area.
Analysis of these data was undertaken using proportional hazards (PH) models, comparing the time to the first confirmed bTB breakdown for herds recorded inside trial areas subjected to the proactive and survey-only treatments.
Increased numbers of badgers carrying M. bovis and increased numbers of active badger setts significantly increased the probability of a breakdown for herds in proactive and survey-only/reactive areas respectively.
Using the herds that did not experience any bTB breakdown during the period under study as controls, we used logistic regression to compare the probability of one or more confirmed herd bTB breakdowns for each herd recorded inside trial areas subjected to the proactive and survey-only treatments.
In addition, we used logistic regression to model the probability of one or more confirmed bTB herd breakdowns during the period under study for each herd within a particular treatment (proactive, reactive or survey-only).
PH models were also used to predict the time to first confirmed breakdown for herds within a particular treatment group (proactive, reactive or survey-only).
This practice is considered to be a conservative assumption; however, if usual care has significantly lower response rates to the follow-up survey than proactive care then this approach would favor the intervention.
