60 R.A. Skuce et al.
et al., 2015) of M. tuberculosis to support out-
break analysis and contact tracing is now valued
and well established in several countries, notably
the USA (Centers for Disease Control and Pre-
vention, 2012) and the Netherlands (Borgdorff
and Soolingen, 2013), although the cost versus
benefit of some systems has been questioned
recently (Mears et al., 2015).
5.4.2 Data sources and surveillanceIn countries with advanced test and control pro-
grammes, animal-level testing, at prescribed
intervals, is well established and confirmation by
specialist bacterial culture provides an opportu-
nity for structured M. bovis molecular type sur-
veillance. In order to understand the diversity in
the local M. bovis population it is important that
representative sampling, whether retrospective
or prospective, be achieved (Skuce et al., 2010).
Comprehensive herd-level, and in some cases
animal-level, surveillance of M. bovis types in
confirmed outbreaks has become established,
most notably in the UK, where very large datas-
ets have been assembled and are being analysed
(Skuce et al., 2010; Broughan et al., 2015), and
also in the USA, France, Spain, New Zealand
etc., as discussed later. In order to effectively
index the genetic diversity in a particular region
it remains important to configure molecular
epidemiology tests at a resolution that matches
the epidemiological questions being addressed.
Until relatively recently, the bovine TB
molecular epidemiology literature was domi-
nated by method development and validation
studies, rather than applications of mature tech-
nologies. While the bovine TB molecular type
data are apparently ‘simple’ on a superficial
level, it is important that such data are inter-
preted with caution and only in the context of
other relevant epidemiological and genetic data,
particularly retrospective data gathered over
several years. Outbreak investigators find the
data useful in the outbreak setting where it can
support, or challenge, a hypothesis. Stakehold-
ers expect competent authorities to deploy all
available data and tools in a cost-effective man-
ner to investigate local bovine TB breakdowns.
Relations between all parties should improve
where this is seen to be the case.
The epidemiology of bovine TB is highly
complex and many of the processes driving the
current epidemic are not fully recorded or
observed (O’Hare et al., 2014; Trewby et al.,
2016). An important example of such a process
is animal movements between herds and via
markets, which can act as a means of dissemi-
nating infection (Gilbert et al., 2005). In addition,
herds and animals also tend to be geo-referenced
to the centre of the main farm holding. In reality
the location attributed to cattle may be a system-
atic error, although the trend will be evident
(Durr and Froggatt, 2002; Enright and Kao,
2016). Farm fragmentation and unrecorded
cattle movements have the potential to signifi-
cantly increase the size of epidemics driven by
livestock movements and undermine the value
of cattle tracing systems (Enright and Kao,
2016). Even in countries with exceptionally
well-recorded animal test and movement data
there may still be significant deficiencies in terms
of data resolution in both space and time. An
important challenge in understanding how,
when and where bovine TB transmits to, from
and between cattle, wildlife and the environment
is that infections are not immediately apparent.
The ante and post mortem cattle tests used in
official schemes have high specificity but only
moderate sensitivity (Nuñez-Garcia et al., 2017;
Lahuerta-Marin et al., 2015), and countries dif-
fer significantly in the extent of culture confir-
mation undertaken (de la Rua-Domenech et al.,
2006; Clegg et al., 2011; Abernethy et al., 2013;
Bermingham et al., 2015). This impacts the
representativeness of subsequent molecular sur-
veillance of confirmed cases where epidemiolo-
gists will only ever see a reduced part of the
transmission tree and have to make at least some
assumptions (Biek and Real, 2010; Biek et al.,
2015).
Due to the slow growth of the organism,
molecular typing data are often only available
well after the disease control decisions need to be
made. However, while real-time data are most
useful, inference of trends in the data can inform
on ways to modify and enhance approaches to
disease control. As increasingly sophisticated
mapping and database solutions are imple-
mented to help exploit these data (Rodriguez-
Campos et al., 2012a), novel insights are being
gained. Advances in bacterial physiology
(O’Connor et al., 2015), direct detection and