Wednesday, January 21, 2015


We gave a fair hammering to the paper released by Moustakas & Evans (M&E) in this posting - [link] and the Guardian - [link] was amongst many media outlets to run with the abstract.

Mathematical modeling is a useful tool, but only if the input data is solid. And even though M&E have listed a hundred references, as we pointed out, many are previously modeled assumptions based on or around papers by the ISG or their followers, involved in the RBCT Badger Dispersal Trial or, in the case of other stuff, so historic as to be meaningless after almost 20 years of inertia on badger control.

So let's look at some of the input data, entered by M&E:
Their model area is approx 67 miles square (or 11,500 km2) and contains 1.4 million (dairy) cattle. That's an area slightly more than the size of Devon plus Somerset. The total cattle population of GB at the end of 2013 was 9.7m, of which the dairy herd was said to be 1.8m and the beef herd 1.6m; the rest are calves, young stock and fattening cattle. So they've got almost the entire national dairy herd squashed into two counties, and they're not allowing for sheep, arable, woodland etc, never mind the occasional town or village?

They've also assumed 5.7 adult badgers per km2, which seems on the low side for what appears to be a lovely area of wall-to-wall dairy farms and much lower than FERA have published for Gloucestershire where they trapped 844 in 55 sq km during their vaccination 'Elf n' Safety' trial in 2007/8. That is over 15 per sq km. (or 15.5 if one counts half a badger). And in parts of Oxfordshire, population densities of 38 per sq km have been recorded.
Following those badger numbers, the M&E model starts out with badgers in only 3113 out of 16384 of their "cells" - which seems unlikely.

Cattle-to-cattle infection: M&E say that cattle can transmit TB at even early stages of the disease and so their model reflects this belief. But actual hands on research on reactor cattle [SE3013] -[link] referred to by the ISG - but not in detail for obvious reasons, found that of 1543 nasal samples taken, and a further 1000 in a parallel study [SE3033] not a single one was capable of onward transmission.

 To avoid confusion, we'll quote again the conclusion of this £2.8m study:
"M.bovis was not detected by bacterial culture in any of the nasal mucus samples." and "The results suggest that large concentrations of M.bovis are not present in the nasal passages, and the shedding of M.bovis, if it occurs, is rare in naturally infected GB cattle."
This is also born out by pairings of reactor cattle with in contacts, done in Ireland - [link]Ireland by Eamon Costello and Dr. O'Reilly.

Thus M&E's 'Winter housing' assumption is just that, and not the experience of studies which housed reactors together for months, slaughtered, salami sliced and collected samples to prove or disprove their work.

Their model also seems to assume that all cattle on a farm are housed together in one place. This is not what happens in practice, as is their assumption of cattle moving round their entire farm 'cells' randomly all of the time. Again, not what happens in practice, particularly on a dairy farm.

But the 'assumption' we appreciate most, is that of badger-to-cattle transmission: M&E do not seem to have a clear idea what the rate of badger-to-cattle transmission is, even claiming that it is a "little studied variable". Whaaaaat???

 This has been 'studied' to death - in fact the death of up to 40,000 cattle a year, even if it did not need to be, and as we pointed out in this posting, where we listed the 'postulates' of disease transmission - [link] , a table used by epidemiologists and veterinary scientists - if not modelers.

All papers written over the past four decades - Zuckerman,  Dunnett, Krebbs etc - have concluded that infected badgers pose a significant risk to cattle. And even dear old Defra, in a booklet issued in 2001, instructing readers about reducing the TB risk [to cattle] state:
"The consensus of scientific opinion is that badgers are a significant source of TB infection in cattle...."
So as the conclusions of the model M&E have used, seems to turn all this on its head, one has to question the input data they have used. On the other hand, in a few short months, an election is looming...

But we digress: on badger culls, their model seems to assume that if just one badger is shot in a cull, then all the other badgers in its group will move out? That is taking the perturbation myth a tad too far we think. Particularly as at least twice annually, the badger groups have a shake out of the old, the sick and the young males. And they seem to 'leave' without causing a problem or even a space?

A paper by 'Gilbert' is quoted many times as a source of the M&E assumptions on cattle movements as 'a significant source of TB spread'. This paper is modeled on post FMD restocks, when a handful of cattle were found to have moved, carrying TB. The point here is that their post move skin test identified them, they were slaughtered and that was that. They did not pass their TB on to other cattle on new farm. A point lost on M&E, as is the compulsory pre movement testing of all cattle moving around or from annual testing areas. A policy in place since 2006.

Thus based not on any of the epidemiological work which Defra has produced, taxpayers have funded we have quoted, but assumptions modeled on previous assumptions, M&E published that the "three main factors influencing TB were the percentage of cattle movements, the frequency of testing, and the badger-to-cattle transmission rate."

But hidden within the paper they also say that:
"Culling of badgers does seem to be a strategy that will eventually lead to a lower incidence of TB in cattle."

MAFF / DEFRA knew this thirty years ago, when the Thornbury area of Gloucestershire was cleared so successfully, leading to over a decade of TB free cattle. We asked why:
" The fundamental difference between the Thornbury area and other areas [] where bovine tuberculosis was a problem, was the systematic removal of badgers from the Thornbury area. No other species was similarly removed. No other contemporaneous change was identified that could have accounted for the reduction in TB incidence within the area" [157949]
So from a professional modeler,  a comment on computer modeling generally:
" There are usually a few key variables and you can adjust them to give you almost any answer you want."
 And finally a reminder that less than a year ago, the Guardian - [link] ran a story on computer generated guff, which was published and regurgitated, because no-one was prepared to challenge it or admit they couldn't understand a bloody word of it.

If the input data is 'unsafe' then its modeled conclusions, however seductive to the politics of the day, are even more so.

No comments: