Examination of Witnesses (Questions 440
WEDNESDAY 24 OCTOBER 2007
BOURNE CBE, PROFESSOR
Q440 Chairman: The message I am getting
from that is that you adjudge, as somebody who is a real expert
on the statistical modelling in the context of this work, that
you had a very sound methodology and that you firmly rebut the
line of criticism which Sir David King has identified.
Professor Donnelly: In terms of
the unsoundness or weakness in the external effect, yes.
Q441 Chairman: Would you agree, just
looking at it as a layman, that it is possible for somebody looking
at your data to draw the kind of alternative conclusions which
Sir David King has done, because the thing that I struggle with
in this is that we are not dealing with absolutes. There is not,
for example, an absolute measure of temperature that there is
in certain scientific experiments. What you are dealing with is
your interpretation of some field data, so is it unfair for anybody
to come to a different conclusion than you did?
Professor Donnelly: I think it
is unfair for them to come to a different conclusion on the basis
of counting how many of these confidence intervals include zero.
There are always going to be uncertainties in terms of epidemiology.
When you find an increased heart attack risk associated with some
risk you cannot identify usually which individuals had the extra
heart attacks and it is similar here in that you cannot identify
which of the herd breakdowns are the extra ones that would have
become breakdowns had we not done culling or got the extra breakdowns,
so there is that difficulty. However, the consistency between
the strength of the statistical results is then strengthened considerably
by the analysis of ecological data, which is what I think Rosie
can speak to more clearly.
Chairman: I just want to stop you there
because there are going to be limits to what I and my colleagues
can absorb for this and I want to give them a chance.
Q442 Mr Williams: The article in
one of the veterinary academic journals by some Irish authors,
I think, was critical of your report on the basis that the perturbation
effects could not possibly have taken place in such a short time
as you have indicated on the basis that the time the badgers take
to move, the time that they need to infect other badgers and for
those badgers to infect other cattle and for that infection to
be evidenced in tests would have been a very long time and much
longer than you claimed in your work. That was not something that
was actually raised by the Government Chief Scientific Adviser.
Professor Woodroffe: We are in
the process of responding to that paper and I am very happy that
you have raised it. There are two separate issues here. One is
saying, was there a detrimental effect, and I think that the statistical
evidence that there was a detrimental effect is very strong. What
they were criticising in that case was our interpretation of the
mechanism leading to that detrimental effect. We had proposed
in brief that when you cull the badgers that disrupts their territorial
organisation, they mix around with one another, range more widely,
meet more cattle herds, and if they are infected they have got
the opportunity to infect more cattle. The contention was that
we could not possibly see such a rapid effect because that chain
of events could not take place fast enough.
Q443 Mr Williams: Was that in The
Professor Woodroffe: That was
in The Veterinary Record, precisely. This is an issue that
has been raised in the past, and in fact when we published these
detrimental effects around proactive culling areas in Nature
in 2005 we addressed exactly this issue in the supplementary information.
I sympathise with Sir David King and his committee at having to
read our fat report, and also not just all of our scientific papers
but also many pages of supplementary information that were provided,
and perhaps the need to read such a large volume of information
might explain some of the deficiencies that we have seen in their
report. The series of events is, you cull the badgers. How quickly
do they start ranging more widely? Essentially, in ecological
terms, immediately; it happens within a few days. We know that.
In fact, Professor Tim Roper was one of the advisers to Sir David
King, he was the person that did that work, so we know very confidently
that badgers ecologically respond rapidly. Then they are presumably
contacting cattle and that is the sort of area where you are not
quite sure how often they are contacting cattle but, after being
exposed to M. bovis, evidence from experimental infections carried
out, I believe, at the Institute for Animal Health suggests that
responsiveness to the tuberculin test takes about three weeks
to develop. Therefore, in theory, within a couple of months you
should be able to detect an elevated number of infections, so
we have contested very strongly, and are continuing to do that,
the idea that these results are implausible. They are perfectly
plausible based upon the evidence that is available to us.
Professor Bourne: We deal with
this in the report. The suggestion is that taking experimental
data from cattle with respect to the timing of development of
a tuberculin reaction is not the thing to do; we should use field
material. Well, of course, you cannot use field material because
you do not know when these animals have been infected. I suppose
the nearest we can come to this is an experiment that was done
at CVL over the last 12 months when they took a group of tuberculin-positive
cattle and housed these with sentinel cattle, and within seven
months 50% of the sentinel cattle were infected.
Q444 Chairman: But I can equally
quote you an example that we heard when we went to Devon last
week of 29 infected calves within a closed herd being put in with
other cattle and there was no cattle-to-cattle transmission.
Professor Bourne: I can also quote
you a situation in Australia where infection in cattle took seven
years to be detected by the tuberculin test. That probably tells
me more about the tuberculin test than anything else. We can use
only experimental data, and that is experimental data we have,
which are consistent with this finding.
Q445 David Taylor: Professor Donnelly,
I saw on page 99 of the report and the bottom half of the portrayed
data that it says the distance from the trial area boundary is
four to five kilometres in a 100 square kilometre area. In a 300
square kilometre area you cannot get much more than eight to nine
kilometres, I would calculate, from that boundary. Is it an incorrect
inference to think that that graph might extrapolate down slightly
as you get further and further away from a boundary and that the
beneficial effect would get significantly closer to 100%? It stops
at 50% in that bottom graph, does it not, at the five-kilometre
Professor Donnelly: Yes, about
that. If you are doing it linearly you have got a pretty good
idea that you get ten in and then get 100%, which would not be
plausible if we think that badgers are responsible for
Q446 David Taylor: No, exactly.
Professor Donnelly: less
than 100%. While you could do that, as a statistician I would
be extremely cautious about doing that. For one thing, the linear
trend, even among the data that we have observed, is just on the
borderline of significance, so it is not overwhelming even for
the data that we have and then extrapolating. The difficulty is,
of course, that that would be extrapolating a linear trend on
that basis but what you really want to extrapolate is the ecology.
Q447 David Taylor: So how has David
King come up with 300 square kilometres as the magic area?
Professor Donnelly: I think he
is looking at the graphs that we did. We did two different scenarios.
Professor Bourne: This is the
graph on page 105.
Professor Donnelly: You have the
full report there. It shows two scenarios, the bottom of which
assumes a constant benefit across the whole culled area, no matter
how big it is, and then counters that with a detrimental effect
outside. The one above, A, assumes that you have this trend with
distance from the boundary, so the lower one is more conservative
but the reality is that it is difficult to say beyond four to
five kilometres, which is what we are able to observe.
Q448 Sir Peter Soulsby: I want to
take you back to your very helpful explanation of the data and
what it meant and take you particularly to what has already been
mentioned as to what Sir David King said about perturbation. He
saidlet me see if I can get this rightthat there
is a reasonable possibility that the disruption is transient,
and then he takes things on from there and develops that quite
significantly. Is he right? Is there a reasonable possibility
from what you said to us and what your studies show that as a
disruption the perturbation is merely transient?
Professor Donnelly: It is quite
possible. I will read you the sentence of the results. Having
looked at this detrimental effect outside, it looks, like a linear
would be, quite reasonable; it wobbles around a bit but it has
a general linear trend down. The linear trend suggested a 7.3%
decrease in the detrimental proactive effect with each cull, but
the P value for that is 0.17. Statisticians for significance produce
0.05, so it is a suggestion of the transients, but it is far from
convincing in and of itself. That is when you would also look
toward the other ecological data.
Professor Woodroffe: Yes. If I
can just come in on that, it was quite frustrating, reading this
report, where they say that the possibility that this effect was
transient was something which the ISG ignored, when we had made
repeated mention of it, and that paper that Professor Donnelly
has just circulated actually had quite a detailed ecological explanation
for why you might expect to see these effects diminish over time.
I can go into the ecological explanation if you like but it is
a little intricate. Suffice it to say that there are suggestions
of reasons for thinking that you might see such a reduction over
time, but, as Professor Donnelly says, it is not a statistically
significant effect. One issue that I did want to raise is that
later on in their report, buried somewhere deep inside, there
is discussion of reopening the door on localised reactive culling.
We have already discussed this issue of whether the first year
is irrelevant or not. Sir David King's group admitted claiming
that we could not really interpret anything from the first year
of data, and that happens to be the year when you see these particularly
severe detrimental effects, and ecologically that is exactly what
you would expect. That is the period when you have caused the
massive disruption in badger society. As Christl has argued, if
anything, that might even be an underestimate of how severe the
detrimental effects are because it has to some extent been diluted
by what happened the year before, before you started culling.
If you are going to do culling over many years, for example, over
the first few years of the studyand Christl will correct
me on the numbersthat initial detrimental effect is so
severe that you get a statistically significant effect even over
two, three, four years. It is only when you start going for much
longer periods that it begins to make the overall effect non-significant.
If you were, say, going to do 15 years of culling, you could begin
to say, okay, this is a cost you have to pay early on, but when
it is diluted over many years perhaps you can discount it or think
it is going to get diluted. If you go back and think about something
more like reactive culling, something more like licensed culling
to individual farmers, which is likely to be small-scale and,
critically, was always conducted as one-off culls, all you are
ever having is the first year when you have these very severe
effects, so it is very important to bear in mind that, even if
you think there is an argument for discounting the first year
when you consider proactive culling, which I would contest, at
least on the sorts of timescales we are talking about, it is absolutely
critical for understanding what the effects of localised culling
Q449 Chairman: If you put it the
other way round that reinforces one of your principal findings,
that if it is going to work at all it has got to be big.
Professor Woodroffe: Yes.
Q450 Chairman: Big and continuous.
Professor Woodroffe: Big and continuous.
These detrimental effects in the first year are so severe that
they overwhelm any of the benefits and the net impact is negative.
That graph that you have seen is based upon assuming that you
cull over five years. If you were to do the same thing for one
or two years you would see an overwhelming detrimental effect
more or less, almost however big an area you cull.
Professor Bourne: But there are
other criteria for culling and one is that it should be done simultaneously.
Professor Woodroffe: Yes.
Q451 Mr Cox: It should be co-ordinated.
Professor Bourne: Yes, absolutely.
Also, Sir David says here that the removal process must be effectively
done by competent operators.
Chairman: I am going to draw matters
to a conclusion. Two more colleagues have caught my eye.
Q452 Lynne Jones: My question was
really to follow up on the same thing. Sir David King only says
that there is a reasonable possibility that the disruption is
transient and that the detrimental effect appeared to reduce.
In a way, that is true, is it not?
Professor Woodroffe: Yes.
Q453 Lynne Jones: And he acknowledges
that the reduction in the detrimental effect is not statistically
significant, so I cannot see anything wrong with what he is saying
there. What I think is wrong is then to take that particular statement
and say that as a result of that culling is a reasonable strategy.
Professor Woodroffe: Absolutely.
If I could just raise one issue that I think is very important
and that I would like you all to think about, it is comparing
what we have been saying with what Sir David King has been saying.
In the report he says, and I quote early on in the report, I think
at point four, "TB control will require interventions that
reduce the prevalence of disease in both cattle and wildlife".
That is what they are hoping to achieve, and certainly when Sir
David King was sitting in this seat here and talking about this,
he repeatedly talked about the need to eliminate TB and that unless
you deal with the issue in badgers you are never going to eliminate
TB. Unless you eradicate badgers you cannot eliminate TB in badgers
by culling them. There is no current method. This paper, which
was published in Proceedings of the National Academy of Sciences
of the United States of America and which I will provide to
you, is called Culling and cattle controls influence tuberculosis
risk for badgers, and shows very clearly that culling increases
the prevalence of disease in badgers, and it is very consistent
and there is really very little debate about it. I think it is
quite a fundamental misunderstanding that when you read this report
you could come away with the idea that you can eliminate TB in
the cattle/badger system by culling badgers. You cannot.
Professor Bourne: "Elimination"
is not a word we have ever used.
Professor Woodroffe: Unless you
eliminate badgers completely.
Professor Bourne: No. Defra eliminated
the word from their vocabulary two years ago.
Q454 Chairman: But I do not think
in either of the reports, whether you cull and on whatever basis,
anybody has actually ever said you eliminate TB. In fairness to
both sets of reports, that is not a conclusion you drew.
Professor Woodroffe: Sir David
King used it just this afternoon, and if you check the transcript
you will see.
Q455 Chairman: One has got to be
fair on what he said because I read his report and that is not
what he said.
Professor Woodroffe: He said it
Chairman: He may have said those words
but that is not what is written in his report. The argument is
about effectiveness within the scenarios of culling. I am pleased
to see that the handout in this day and age of electronic education
has not gone away. Your lectures must be full of people with bits
Q456 Mr Drew: The ISG work was heavily
criticised throughout for not taking out sufficient numbers of
badgers to really be able to feel that there was a robustness
in the way in which you were tryingand I apologise if I
used the word "experiment"to carry out the particular
study that you were carrying out, and we got that last week in
our trip to Devon where there were lots of accusations that you
did not take out enough badgers regularly enough to get a real
purchase on what you were trying to achieve. Is trapping an effective
enough way to get the sort of removal that the King report is
alluding to, or have you got to consider other measures?
Professor Bourne: We were committed
on welfare grounds to using cage traps, as you know, with the
expectancy that we could achieve an 80% trapping efficiency, and
that is consistent with the data we have, that we did trap around
that percentage and we achieved something like 75% or a little
less removal of badgers at the end of the day, taking into account
badgers coming back into the piece. Much of the confusion with
respect to culling activity in the trial resulted from figures
that were presented in the consultation document which were erroneous.
I think there has been an attempt by authors such as Chris Cheeseman
and Graham Smith to put the record straight on that, using a model
based upon first cull only, and we have hard data to support our
evidence. It is described in this document as a guess. I think
that is a silly statement because the data we have is the data
that you would collect to give you an assessment of numbers of
badgers on the ground and we have done it by using three or four
criteria to give us that figure. With respect to other methods
of trapping, in Ireland, as you know, they use snares. We were
told we could not use snares and we did not use snares. With respect
to gassing, just consider the Thornbury area where we do have
experience. This was an area of 110 square kilometres with a much
lower badger population than we are experiencing now, maybe a
quarter of that we are experiencing now but certainly very much
lower. It took them over five years to clear that area of badgers
using cyanide gas, involving over 500 gassings, so I think to
achieve elimination or near elimination is extremely difficult.
To achieve the level of removal that we did in the trial is extremely
demanding and requires skill from the operators doing it, given
the amount of land that one does not have access to, and I have
no reason to believe that that will change. Thirteen per cent
of the inaccessible land was because one could not identify the
owners, so it was illegal for us to get onto that land. To answer
your question, there may well be better trapping methods but we
do not have any data on that. We can only recall what has happened
Professor Woodroffe: If I could
just embellish what Professor Bourne has said, in chapter 10 of
the final report, which I am sure you have got copies of, we thought
in great detail about precisely this issue and an array of other
issues. We thought of every way that we could to improve the effectiveness
of badger culling in the broadest sense, or other badger management
methods, and everything that has ever been suggested to us we
went through and thought very carefully, based upon our own and
others' evidence, would this work, would this make it better.
Clearly, improving trapping efficiency was one of the first things
to think about, and, as Professor Bourne says, we achieved about
a 70% reduction in badger activity, and multiple different measures
of badger activity triangulate and show the same sort of effectiveness.
The other thing that that evidence showed was evidence of substantial
badger immigration into the trial areas, and in fact there is
a paper which has just come out today in Molecular Ecology
showing genetic evidence of very widespread movement of badgers
into these areas over substantial distances, and I am sure we
will be very keen to share that recently published paper with
you. As we culled these areas successively, so an increasing proportion
of badgers would be getting caught on the boundaries, showing
that there was immigration into these areas. What that means is
that if, say, rather than removing 80% of the badgers by cage
trapping, you used snaring and removed 90%I am making these
numbers upthat does not mean that you are going to have
an equivalent reduction in the standing crop of badgers because
every time you remove badgers from an area others are coming in
from the outside, so if you remove a higher proportion of what
was there it may be that you would just have even more coming
Q457 Lynne Jones: What about boundaries?
Professor Woodroffe: Precisely,
which is why we then looked at boundaries. The sorts of boundaries
which Sir David King and his team discussed in their report, motorways,
rivers, coastline, came from us and we showed indirect evidence
suggesting that what we called hard boundaries, which are pretty
substantial boundaries to badger movement, do seem to have an
effect because these increases in the prevalence of infection
in badgers, which are described in that paper I just circulated,
are most substantial where those boundaries are missing, so you
do not see such a big increase where there are boundaries, suggesting
that badgers immigrating in are contributing to this perturbation
and therefore to the increased prevalence of infection. At the
same time we did another analysis in which we looked at soft boundaries,
other smaller water courses and other little boundaries, and found
no effect from those. That was why we said ourselves in the final
report that you have in front of you that perhaps culling badgers
would be more effective if you could do it within these geographically
circumscribed areas, which is what was done in Ireland. The four
Irish areas were chosen to be these areas which were comparatively
isolated by geographic boundaries to badger movement. They are
not representative of the broader areas where TB is an issue,
either in Ireland or in Britain, but if you look at a map of the
south west area and you look at the distribution of motorways
and large rivers, there are not that many.
Q458 Mr Cox: I can think of one straightaway.
There is a very large area between the north coast, the Tamar
and the A30.
Professor Woodroffe: The northern
reaches of the Tamaractually, that is not a very big river,
and at the southern end of the Tamar is a big estuary. I am Cornish.
Q459 Mr Cox: You know it as I do.
I represent that area and I can tell you that I would not like
to swim it. Have you done it?
Professor Woodroffe: I have not
swum the Tamar, no. There are very few areas which are geographically
defined in that way, and if you are looking for a policy which
is going to affect a large proportion of Britain's cattle farmers
that is not a solution.