Saturday, April 08, 2006

The role of cattle movements in the bovine TB epidemic

Part IV: Mycobacterium bovis, the English bovine tuberculosis outbreak, badgers and the culling debate.

Although bovine tuberculosis (BTB) is known to spread through wild-animal populations and is readily transmissible between cattle, the role of large scale cattle movements could be an important indicator for the spread of BTB. The current epidemic is concentrated in a large central area in the south west of England and Wales, with a large number of outlying foci of infection outside of this core area. Essentially, within the core region it’s predicted that internal factors such as animal reservoirs (badgers), environmental conditions and relaxation of adequate control measures maintain BTB rates. The question is if the movements of infected cattle or material can explain the distribution of BTB. Conveniently, a study published in Nature sought to answer precisely this question and will be the subject of this post.

Gilbert et al., 2005. Cattle movements and bovine tuberculosis in Great Britain.

In their study Gilbert et al., (2005) made a computer model designed to predict the spread of BTB to new areas from the ‘core’ region by using movement data obtained from the British Cattle Movement Service (BCMS). The BCMS was set up in 1996 after the BSE outbreak in Britain and is essentially a large scale monitoring system to trace the births, deaths and movements of cattle across the UK. Part of this system is the Cattle Transfer System (CTS), which keeps the record of cattle movements and was what the authors based their study on.

Importantly they avoid a problem with the way that the CTS records the cattle movement data, which may have compromised the papers results considerably (See supplemental material). You see, every cattle ‘movement’ recorded by the CTS is in reality recorded twice. Once for the “on” movement and another for the “off” movement meaning that all cattle movements have to be paired to make sense of it. When this procedure is actually done, a large amount of the data from before 2000 is shown to be rather inaccurate (there aren’t two movements recorded correctly for example), but is very accurate data for the period from 2000-2003.

Another limitation of the study that the authors put forward (again, supplemental material) is the reliability and time disparity between tuberculin testing. Depending on when an animal is tuberculin tested, it may show to be positive for BTB before or after it has moved, giving an inaccurate impression of where the animal actually was infected and if it was pre or post movement. This would particularly affect the results from regions with less BTB than other areas of the UK, as these areas tend to perform tuberculin tests less frequently. The authors of the paper discount this as a problem reasonably, as the disparity in time between testing is not going to be a uniform problem affecting all of the cattle herds uniformly. Basically, some herds will be tested very regularly and a few will not, which while this adds some further random variation doesn’t affect the overall strength of the association.

Bearing these two factors in mind the question really is: Does their predictive model correlate with the observed BTB spread? The answer in short terms is that it does and it does so with quite a high degree of accuracy. For example, take one of the figures showing the models BTB prediction against the actual observed rates of BTB:

Image hosting by Photobucket

Figure adapted from Gilbert et al., 2005. Cattle movements and bovine tuberculosis in Great Britain. This is one of four comparative maps generated from their model in the paper, in this case showing the result obtained from "2002" against the observed BTB incidence from the same year. This is also performed again for 2003, 2004 and for 2005. 2004 and 2005 do not show the smaller map because BTB incidence data wasn't currently available when the authors did the study.

Here you can see that their modeled BTB distributions and the projections from their model on the larger map of the UK for 2002. The darker the colour on the map, the higher the predicted probability that there will be a presence of BTB in that area. The smaller inset map is the observed frequency of BTB. The figure above and others from the paper demonstrate there is a high correlation between cattle movements and the presence of BTB in a region. Generally speaking, where cattle are being moved around from ‘core’ areas particularly it is associated with spreading BTB to ‘remote’ areas.

What the model doesn’t show and how to experimentally confirm the study

One place where the model shows abnormalities is that there are regions that have numerous cattle imports, yet the disease fails to persist for any meaningful period of time. It’s important to bear in mind that this could have several explanations, the first of which being that the cattle are only temporarily held in the region. For example, they could arrive and be immediately slaughtered on arrival, such as to an abattoir, which wouldn’t give any time to spread the disease. Another possibility is the lack of wildlife reservoirs, namely badgers in any meaningful capacity in the region, which is also the explanation I would favor, in particular given the association between badgers and BTB (see part III). A final explanation may be that these regions are not prone to many movements from ‘core’ areas, which reduces their risk considerably.

This study could also be interesting to contrast with the known spread of BTB in general regions. As it predicts BTB spread based on cattle movements, it does immediately present a means to experimentally verify it on the ground (so to speak). Given their model, it should be expected that movements from areas with BTB should be associated with the detection of certain Mycobacterium bovis spooligotypes (basically strains). This is because the imported infected cattle should bring with them their M. bovis types and therefore spread that to the uninfected herd. Over time, based on what cattle movements went into a region and what spooligotypes infected cattle bought with them, you would expect to see those spooligotypes in subsequent herd breakdowns.

Here is where I become somewhat skeptical. It’s known from previous experiments which have analysed the spooligotypes of M. bovis from badgers and cattle that these tend to be shared between the two species (see part III). This also tends to be isolated by geographical regions, with a mixture of spooligotypes but only a few ‘oddball’ ones that aren’t shared between badgers and cattle. In the model predicted by Gilbert et al., 2004, where cattle movement is the primary motivator for BTB spread it also implies cattle to cattle spread. The lack of spooligotype mixing between regions from studies conducted in Ireland (for example) have shown spooligotypes of M. bovis tend to be similar in a region but not between them, raises concerns over cattle movement as a predictor of M. bovis. It may be likely that cattle movement helps to spread the infection to a new region, but is not sufficient to determine if the disease will be able to establish in the new region.


It’s clear from the Gilbert paper that cattle movements are playing some role in the spread and establishment of BTB in the UK. The question is still open as to what that exact role is and how the movement of cattle would spread BTB to other regions. For example, how well does such a model deal with the rates of BTB spread among ‘closed’ herds in the UK that do not actively import cattle from elsewhere? It would be interesting to compare the incidence in BTB among “closed herds” with the regions predicted by this study as being at risk for BTB from “on” cattle movements. In any event, if cattle movement is an important indicator, it immediately opens up simple ways of reducing the spread of BTB such as by implementing pre-herd tuberculin testing before movements (which I think is being implemented now).

In any event, this does not resolve the problem with badgers as a large reservoir for M. bovis. Ultimately, this debate has become rather invective and polarized over one key issue: If the control of M. bovis requires the culling of badgers in order to be accomplished. In the next part of this series, this immensely important aspect of the control program will be discussed in Part V: Will culling badgers have any effect on the control of bovine tuberculosis?


Gilbert M., A. Mitchell, D. Bourn, J. Mawdsley, R. Clifton-Hadley and W. Wint (2005). Cattle movements and bovine tuberculosis in Great Britain. Nature, 435:491-496. Also see the supplementary materials published with the papers as well.