The Epidemiology of Mycobacterium bovis Infection in Cattle 47
progression (i.e. from infected compartments O,
R, I, A).
Empirical estimates of the sensitivity and
specificity of tuberculin tests rely on post-
mortem comparison to culture of M. bovis from
test-positive and -negative animals, respectively.
The discrepancy between these relative mea-
sures of diagnostic test performance and the
true sensitivity and specificity (with respect to
disease status) will depend on the extent of
misclassification of test-negative animals. In
turn, this hidden burden of infection due to
test-insensitive animals will vary with the
prevalence of infection and past history of
transmission. Herds with a recent high rate of
transmission will have a disproportionately high
proportion of occult animals, whereas a herd
that has been heavily infected for a long period
of time may have a greater proportion of anergic
animals.
Quantifying the burden of infection missed
by testing therefore requires dynamic models of
transmission that track the history of infection
and removal of animals (Conlan et al., 2012).
From the modeller’s perspective, this relation-
ship means that the parameters governing test
characteristics in mathematical models of
transmission, where we know the true infec-
tion status of animals, cannot be simply
equated to relative measures from visible
lesions. More fundamentally, alternative
diagnostic tests for bovine tuberculosis have
different windows of sensitivities with gamma-
interferon tests more likely to pick up earlier
infection (de la Rua-Domenech et al., 2006)
and antibody tests more likely to pick up later
infection (Whelan et al., 2010). Simple com-
parison of the performance of bovine TB tests
with respect to lesions could therefore be highly
misleading, a concern reinforced by recent
latent class analysis (Nuñez-Garcia et al., 2017)
that suggests that the true sensitivity of the
single intradermal comparative cervical tuber-
culin skin (SICCT) test may be as low as 50%
(26–78%, 95% credible intervals) compared to
the commonly accepted median estimate of
relative sensitivity of 80% (range of estimates
from 50–100%; de la Rua-Domenech et al.,
2006). However, as we discuss in Section 4.3,
the way we use testing is just as important – if
not more so – than the performance of the test
in individual animals.
4.1.3 Routes and mechanisms of
transmission between cattleBovine tuberculosis in cattle is primarily a pul-
monary infection (Francis, 1947; Liebana et al.,
2008), which suggests that respiratory trans-
mission is the primary route of cattle-to-
cattle transmission (Menzies and Neill, 2000;
Goodchild et al., 2015). However, pseudo-
vertical transmission through milk – the main
historic zoonotic risk for humans – is also a likely
alternative route of infection for calves.
Although only a relatively small proportion of
infected animals in an uncontrolled situation
(1–2%) develop infection within the udders,
bulk milk feeding practices may pose a high
risk. Alimentary infection is rarely detected,
particularly in managed populations, although
bacteria have also been recovered from faeces
from experimentally challenged animals (Neill
et al., 1988).
Pathology can suggest the likely routes of
infection, and is often appealed to justify the
importance of a particular route of transmis-
sion. However compelling these narratives may
be, the reality is that no study to date has suc-
cessfully quantified the relative importance of
direct contact, aerosol spread and indirect envi-
ronmental spread through fomites to the risk of
acquiring infection. Historical studies in the
south of England found that M. bovis could
remain alive and virulent on pasture for at least
49 days during the summer (Maddock, 1933).
Environmental factors are likely to affect this
viability with more recent studies in New Zea-
land (Jackson et al., 1995) and Michigan, USA
(Fine et al., 2011) finding considerable variabil-
ity in the viability of bacteria between seasons.
For chronic, relatively poorly transmissible
pathogens such as M. bovis, the duration of
exposure is as important a component as the
dose. While the exposure from direct contact
might be orders of magnitude greater than con-
tamination on pasture, the frequency of expo-
sure through grazing may amplify the overall
risk of acquiring infection.
Environmental transmission was largely
ruled out in the 1930s based on experimental
infection studies, which demonstrated that
transmission to calves was possible from pas-
tures sprayed with artificially high concentra-
tions of M. bovis (Maddock, 1933), but