The proportion infected (NSP positive with Asia-1 SP titre ≥32) was 86% in the uninhibitors vaccinated (222/257), 65% in the TUR 11 vaccinated cattle (211/327) and 89% in the Shamir vaccinated cattle

(129/145). Vaccine coverage of animals over four months was 84% (Ardahan investigation), 42% (Afyon-1 investigation), 83% (Denizli investigation) and 60% (Afyon-2 investigation). The Shamir vaccine was only used in the Ardahan investigation except for eleven cattle in the Afyon-1 investigation. Table 2 shows both descriptive statistics and univariable associations with clinical FMD with more details in table S2 (a) and (b). All factors except trimester of pregnancy (p = 0.3) showed some degree of association with clinical FMD (p < 0.1) (i.e. vaccination

status, age, use of common check details grazing, breed, sex, herd size, time since vaccination, herd vaccine coverage see more and the investigation). Of the 394 animals with clinical FMD on examination, farmers reported disease in 283 (detection sensitivity of 72%). This showed little variation with herd size (p = 0.1). Failure to detect FMD will result from mild disease or limited farmer observation and recall. Cases where the farmer reported disease but clinical signs were not apparent on examination (47/371 [13%]) will result from recovery or recall error. The remaining 87% where both the farmer and the examination did not detect disease gives a pessimistic estimate of farmer specificity. Detection rates were similar for vaccinated and unvaccinated cattle (p = 0.6), so misdiagnosis should not bias vaccine effectiveness estimates. Accurate government vaccine records were available for 372 animals. From these, 280/287 were correctly reported as vaccinated by the farmer (98% accuracy [95% CI = 95%–99%]). This error rate was unaffected by FMD status (p = 0.25). Farmer reporting was correct for 83/85 unvaccinated cattle (98% [95% CI = 92%–100%]). Again, FMD status

did not affect this misclassification (p = 0.14). After exclusion of MycoClean Mycoplasma Removal Kit young calves, only one vaccinated and one unvaccinated animal were misclassified from 263 vaccinated and 57 unvaccinated cattle. After multiple doses of the Shamir vaccine, risk of FMD fell from 89% in single vaccinated cattle to 40% in those with more than five doses over their lifetime (see Table 3). Crude estimates for VE are presented stratified by different variables (Table 2), according to different clinical outcomes (table S3) and for infection assessed by different serological criteria (table S4). However, due to confounding limited conclusions can be drawn from crude VE estimates (see regression model below). VE varied with time between vaccination and the outbreak. For the TUR 11 vaccine VE appeared to decline markedly after 100 days (Table 2).