Examination of witnesses (Questions 400-419)|
WEDNESDAY 25 APRIL 2001
400. Is it still the view of your advisers that
the original source of the outbreak was the case at Heddon-on-the-Wall,
at the pig farm?
(Professor King) Why do I not pass it to one of the
(Dr Donaldson) I have no information to the contrary.
That premise had animals with lesions which were 12 days old,
or thereabouts, and the virus isolated from there was the same
as that isolated from the abattoir at Essex; we have characterised
the particular viruses involved.
401. You are aware of statements made by others
that there appear to have been cases in sheep, including one case
which was sheep sent for export, I think, which indicated an earlier
provenance of the virus within the sheep flock?
(Dr Donaldson) I am aware of information, through
our World Reference Laboratory, that French scientists claim that
they tested sheep originating in the United Kingdom and found
that there were antibodies in them; antibodies would be formed
as a consequence of infection. We have asked for those antisera
to verify the test results; that is more than a month ago, and
we have not received the antisera.
402. So, when you said you had no evidence,
you meant that perhaps you might be able to obtain evidence through
that request but it has not yet arrived?
(Dr Donaldson) That is correct.
403. Assuming that the origin of the outbreak
was at Heddon-on-the-Wall, can all of the outbreaks that have
happened since be traced back, through some contact or other,
to that source, or are there any mysteries that we have not been
able to trace back?
(Dr Ferguson) Certainly, for the early stage of the
epidemic, MAFF did a very good job of doing exactly that sort
of tracing, particularly through the livestock markets, Longtown,
Welshpool and the other ones, which were so influential in causing
the epidemic to spread. And just returning to the question of
infected areas, I think one of the reasons the infected areas
were made so wide early on was because it was known that the markets
were playing such an important role and the infection was so disseminated,
even if we had not seen it yet in areas. I think, more recently,
there have been the occasional outbreaks geographically distant
from existing ones, particularly around the Cumbria region but
also elsewhere, going into the Forest of Dean and Gloucestershire,
which have not had such a clearly identifiable source. But given
that we know the virus can spread a considerable distance, whether
wind-borne or whether on vehicles, or being transported by man,
then that is not all that surprising.
404. Have you been able to identify the most
common sources of transit, transmission, from this; you have mentioned
that there are a variety, and at various times statements have
been made that wind-borne transmission has actually been surprisingly
low, in incidence, in this particular outbreak, and that it appears
likely that relatively close contact is the most common transmission
(Dr Donaldson) I think this particular epidemic is
no different from any other involving foot and mouth disease,
in that the most common mechanism of spread is contact between
infected and susceptible animals. In this particular epidemic,
sheep, in particular, have played a major role in that, and, I
think, if one is looking for a major factor in spread, it has
been the involvement of sheep, in which the disease is often clinically
very mild, or even inapparent. Enormous numbers of sheep moved
in the early stages and, I think, contributed both to the scale
of the epidemic and the speed of its spread.
405. One might say that that might indicate
the importance of tracking down this French data with some degree
of urgency, because, as you have said, and certainly from my own
experience, this disease is extremely difficult to spot in sheep,
and I have had many cases drawn to my attention where the disease
has been and gone without anyone noticing it, because it is so
similar to other infections that a sheep can have. If we turn
to the modelling exercise that was done on 23 March, I assume
that the reason why it was done only some month and a bit after
the first outbreak was the poverty of data that would have been
available to produce any adequate models; is that right?
(Dr Ferguson) In part. Personally, I had started developing,
and I am sure the other groups were looking at the sorts of models
which could be done, but in the initial meeting at the Food Standards
Agency, which I think was 8 March, we did discuss the fact that
more data would be required. And, in fact, by the time we actually
started the analysis properly, about 14 March, or so, when we
received the data, we did actually receive quite good quality
data despite the fact case numbers were small, particularly the
contact tracing data; that data took time to collect, and it was
vital to the analyses, and I am sure Dr Woolhouse would agree
with that. And so, yes, there is a limit to what could have been
done earlier. Possibly, had we been involved very slightly earlier,
we could have done something a week or so before, but certainly
no more than that.
406. What other factors are put into the model?
We have mentioned that there are a large number of variables affecting
the way in which this disease is spread: weather conditions, the
geography of the area, the precise breed of animal that is infective,
and so on. What assumptions are placed in the various models?
I recognise also there are at least three different models that
are being used?
(Professor King) I wonder if I could just stress that
point, first of all. There are four teams of modellers involved
and each of them is working quite independently, although there
is a common source of information. John Wilesmith, at MAFF, has
been the common source of information, and John himself is a modeller
and so he has been doing modelling as well. But, in terms of the
input, I am going to leave you to these two here.
(Dr Ferguson) I would say I agree. It depends on exactly
which model you are talking about, and you have trade-offs, basically,
with modelling. You can build a very complex model which describes
every little bit of detail; the problem is, there, you need very
big computers to run it on, and quite often, with the level of
detail you have to put into it, you do not have actually the precise
quantitative parameter values to make that model worthwhile. So
there is a trade-off in how much complexity you put in. In terms
of the model I have been involved in developing, it is one of
the simpler ones, it allows partially for the spatial spread,
so it allows you to look at ring-culling, comparing policies;
it does not go down to the details of looking at all differences
between individual animal breeds or species, necessarily, but
does account for some. You can then apply it though to look at,
say, one of the key differences, which is regional heterogeneity,
by applying it to the different regions. Other modelsand
Bryan Grenfell, from Cambridge, is not here today, so I will try
to speak for himhave put more detail in, in terms of describing
transmission between individual farms, as a kind of random process,
allowing for more heterogeneity: differences between farms, in
terms of numbers of animals, different species. That is a much
more complex model; potentially it has more power, but it takes
a lot more computer power to run it, and you can do many fewer
things with this in a fixed amount of time. So there is a trade-off.
One of the encouraging things though is, in terms of the broad
conclusions of all the modelling efforts, and I will hand over
to Mark, for the last two, to describe what he has been doing,
they have been so broadly similar, in terms of their . . .
407. I was going to say; from what you have
said, in spite of the fact they are using different techniques
(Dr Ferguson) Yes; they have been broadly similar.
408. From the graphs you produced, they produced
roughly the same outcomes?
(Dr Ferguson) Roughly the same results. It is, of
course, easy, to some extent, and I understand this both from
the lay person's point of view and from the experimentalist's
point of view, to say: "Well, you haven't taken account of
this key factor". And I think one of the long experiences
of doing this sort of mathematical modelling for infectious diseases
and other systems is you can get a surprisingly good insight into
the behaviour of these diseases, actually leaving out a lot of
the detail, and sometimes it is best to do that if you want a
rapid answer. But I will let Mark Woolhouse also speak to the
(Professor King) What I wanted to say is the following,
that the group represented by the gentlemen on my left are the
most fleet-footed of the modellers, and the reason is not only
that they work right through the night but that their model is
a relatively simple one, it is what we call a `mean-field' model,
and the field there is not spaces on a farmland. On the other
hand, what gave me confidence was the fact that the more detailed
models, such as the Cambridge model and the Wilesmith model, were
producing results which were broadly very compatible with the
Imperial College model. This allowed me then to talk in great
detail to the Imperial College modellers and say, "Would
you please try a variation, for example, the 24-hour cull, what
if we let it go up, what if we bring it down; equally, the 48-hour
cull, what if we let that relax down to seven days, what effect
does it have; what if we don't have a ring-cull and just a 24-hour
cull?" So all of these results were available because of
the fast capability of the Imperial College model. But the essential
thing to me was that I was getting the same answers from the others,
however slow. I do not know if Mark wants to add something.
(Professor Woolhouse) I hope we are not one of the
slower ones. I think the crucial thing about using mathematical
models to advise policy-makers is that different models have different
strengths in answering different questions. The question we were
all asked, at the meeting on March 23, was a very simple question
for an epidemiologist, "Is this epidemic out of control?"
The answer was a unanimous "Yes," by our definition,
which Professor King gave you earlier. The next question we were
asked was, "Would a culling policy be appropriate for controlling
this epidemic?", and the unanimous answer there was that
a 24-hour culling policy was essential. And, in fact, going back
to an earlier question of yours, that was actually known before
any of the modellers got in, there was published literature to
that effect. But "Was a 24-hour culling policy essential?",
answer "Yes," unanimously; and "Would it be enough?",
answer "No, not on current estimates." And that, of
course, was the basis for introducing the contiguous cull.
409. Without taking the slightly political tone
of Owen, opposite, one of the factors in this, well, the key factor,
is the data that you put into your model, and, clearly, there
must be some influence, if you are tracking the number of outbreaks,
which is what you are doing, if you classify outbreaks in slightly
different ways. So, through the success of the policy that has
been engaged in, which is to cull out neighbouring, contiguous
farms, where the outbreak has not happened but would otherwise
have done if those measures had not been taken, to some extent,
you produce a self-fulfilling process, in which the number of
outbreaks appears to go down because you are culling out the ones
that would otherwise have been counted in that data. It might
be useful for you to explain?
(Professor Woolhouse) I can answer that specifically.
What we have been able to do is to explore different interpretations
of the data, to test the data for robustness regarding the models;
we have done that extensively. Epidemiologists are very cynical,
we do not always believe field data collected in a rush, we always
test the data, we check for quality, we say, "Well, what
if this wasn't true, would the scenario still hold?". And,
again, after extensive simulations, we would argue, yes. One important
thing about our results is that they are conservative, in the
sense that we would not claim the epidemic was under control,
for example, when it was not. We would not claim that a control
policy would be more effective than it would be; we think this
is very important for helping policy-makers. So all our estimates
and projections are conservative, in those senses.
410. I want to talk about control policies.
One of your colleagues, Dr Paul Kitching, said, on Saturday evening
on the Channel Four News, that there were big differences between
this and the '67 outbreak. He said production of aerosol viruses
from this particular virus is considerably reduced, work carried
out at your institution had already shown that aerosol production
of the strain was much less from pigs than from previous strains.
And yet it does seem that the '67 outbreak has been the main driver
of policy on this occasion, even though it does seem that there
could be a case to argue that the virus had been on farms for
a long time, in sheep. So the two questions I have got really
are, to what extent is it appropriate for us to base our treatment
of this outbreak on the '67 experience, and, secondly, if sheep
may have had it for much longer than just a few days, why the
urgency on the 24-hour cull?
(Professor King) Can I stress, first of all, that
it would be incorrect to say that our policy was based on the
'67/'68 outbreak. We came inand, I have to say, Chairman,
I had the benefit of the most remarkable scientific help in this
process; some of those people are represented here. I do not know
that there was anything like that understanding of the outbreak
in '67/'68. And we have been able to advise the policy-makers
on how to proceed. So I do not think it would be right to say
that. It has been based on the modelling. Now, before I pass you
over to the others, what I have also reproduced for you are some
of these graphs that I presented to the media during the outbreak,
and the first, which is labelled `Two' in the top right-hand corner,
each of these are the output of the nimble-footed group, the Imperial
College group, and you will see that this was the stage of the
outbreak when we were getting involved and saying it was out of
control. Curve A represents the `out of control' epidemic, in
which the turnaround is related to the fact that you are losing
all the animals, you are not getting it under control. Curve B
corresponds to the 24-hour cull being introduced; and Curve C
is the output of the model with this 24- plus 48-hour. Now at
the point when we first showed this data we had no clear indication
as to which way this epidemic would proceed; but if you turn the
page you will see where we are at the present day, so that the
black points represent how the epidemic has proceeded. And I think
you would have to agree that the modellers have done remarkably
well, that since the policy was introduced we have followed Curve
C rather more closely than we have followed Curve B, or A. The
criticism, Chairman, of the scientists would be that the epidemic
is going below Curve C, they did not get it quite right, the epidemic
is decaying faster than they predicted.
411. The Bank of England is in difficulty because
inflation is lower than predicted, but I do not think you will
be in too much difficulty if the epidemic is lower than predicted,
(Professor King) I am not avoiding your question but
I am trying to answer it in a slightly different way. What we
have is theory, predicting an output, and then the measurements
coming in; and, as an experimental and theoretical scientist,
that is often how I proceed. What I would say here is, "Well,
that model wasn't bad; now what was the input to the model?"
and the input to the model was the best input that was available
at the time, as has already been described to you, based on the
tracing, the detailed tracing, that MAFF had been doing in the
early stages. Now Dr Donaldson's team, at Pirbright, together
with Paul Kitching, who is a member of my committee, have discovered
more about the disease; and the net result is that, surprise,
surprise, the epidemiologists, particularly the group on my left,
are getting a better description of the way the disease is spreading.
So if you turn to not Curve 4 but Curve 5 you will see the red
curve is their latest prediction about how the epidemic proceeds,
based on all of the most recent inputs, so they have now tweaked
the model to include the most recent input from the animal biologists.
Now, at this point,
412. It seems to be a very fortuitous arrival
at June 7, if I may say so.
(Dr Ferguson) This was not enforced on the model!
413. Karl Popper would be proud of this, by
(Professor King) I think it would be fair to ask Dr
Donaldson to respond to your questions.
(Dr Donaldson) Going back to the comparison between
the '67/'68 epidemic and the current one, a feature of what happened
in '67/'68 was that there was an explosive start to the epidemic,
I am sure you are aware of that; that was attributed subsequently
to the fact that pigs were infected initially, and probably released
very potent, or highly concentrated, plumes of airborne virus,
which spread across the Cheshire Plain, from near Oswestry, which
was the source, over an area of high cattle density. We know now
that, when we are looking at airborne spread of foot and mouth
disease, which is a feature of the disease, and a very important
feature of the disease, the pattern of airborne spread is most
commonly from pigs, at source, they are the powerful emitters,
to cattle downwind, who are the most sensitive species to infection.
Those features were also part of the current epidemic at the early
stages, from the abattoir at Essex and also from the farm at Heddon-on-the-Wall,
leading to the infection of sheep, from Heddon-on-the-Wall, and
then the distribution through the rest of the country, particularly
to Cumbria and then to Devon, and other parts. You referred to
Paul Kitching's statement, that this current strain is excreted
by pigs to a lesser amount than we have found with previous strains,
that is correct; however, although it is excreted to a lesser
extent, it is still excreted at a very significant level. My view
about the modelling, and I have said this within the group, and
my colleague Paul Kitching has also said it, is that we feel that
the modelling does not take sufficient account of the very great
differences there are between different species in the doses required
to infect them and the amounts they subsequently excrete. It is
not a surprise that the modelling has close agreement for different
groups because they are modelling what I would call an hypothetical
species, and I think that needs to be borne in mind. Since they
are not taking great account of species, I would suggest that
there is an inherent inaccuracy in the forecasts. Going back to
the question, earlier, from Mr Mark Todd, about the input data,
my colleague Paul Kitching and I have reservations about the quality
of the input data, and we have expressed that within the science
group also, and it relates primarily to the role of sheep; we
have spoken about sheep earlier and the fact that identification
of infected sheep is difficult, therefore, it is very hard, we
would say, or almost impossible in certain instances, to know
what the lead-in time has been on certain infected premises. There
has been a trend of sheep to be infected, to spread the infection
silently and then to infect cattle, and cattle are the indicator
species. But putting a time-frame to the period from the recognition
of disease in cattle back to the introduction of infection of
the sheep is, I would suggest, very difficult and could have a
wide amount of variation; and I think those factors need to be
considered, in looking at the modelling.
Mr Öpik: That is a very useful clarification,
and if I may follow up with a few short questions relating to
some of the tactics around the issue. First of all, really for
the record, what would happen if we took no action at all, in
the United Kingdom, would the disease eventually become endemic,
die out, would there be an immunity in the stock; leaving aside
any questions about export, I am simply asking from a technical
414. Just a dose of flu; that syndrome?
(Professor King) If we took no action at all, in particular,
not even vaccinating cattle, then I believe the answer is that
we would move into an endemic situation.
415. What would be the long-termonce
again, leaving aside any export question?
(Professor King) Long term, you would have the disease
popping up continuously; that is what we would mean by endemic.
416. And how bad would that be for the animal
population, in terms of health?
(Dr Donaldson) I think someone has used the term `common
cold'. The only common feature between common cold in man and
foot and mouth disease in animals is that the mechanism of spread
from infected to susceptible in contact animals is that the spread
is very like common cold, it is spread in droplets and droplet
nuclei, and, in susceptible animals, in most cases but not all,
most commonly they become infected by inhalation and infection
starts in the respiratory tract and then goes elsewhere, eventually
producing vesicles. If we were to relax the control measures,
I think, almost inevitably, as Professor King has indicated, the
virus would become endemic in the United Kingdom; then you would
get into a situation of a pattern of spread of the virus through
the population, the establishment of a natural immunity, the waning
of that immunity over time, as animals are replaced, and then
another wave of infection. In countries, in Africa and South America,
where they have not had the resource to control the disease, it
tends to cycle at periods of every three to five years, and that
reflects a susceptible population coming through. Might I add
one comment, going right back to the very first questions? "Why
have the slaughter out policy?" The justification for the
slaughter out policy, horrible as it may be, is that it is a well-proven
method of eliminating the possibility of convalescent carrier
animals, and I mean by that animals that could be persistently
infected. Total stamping out of infected herds is the only way
of guaranteeing that that will not happen. If it is not done,
if there is partial slaughter, and by that I mean if only the
diseased animals are killed, then there is a high probability
that the incubating animals, and perhaps some recovered animals,
will continue to harbour the virus, and so you are tipping the
balance very much towards an endemic situation.
417. I have got two other short ones, but just
to conclude on this particular heretical thought; should we be
considering, as one possible option, the acceptance of this disease
as an endemic disease in the United Kingdom, and perhaps across
Europe, in your judgement?
(Dr Donaldson) Absolutely not. The traditional methods
of controlling foot and mouth disease are well proven, and if
properly applied should eliminate the virus.
418. Thank you for that. Just one, very short,
other question. We were informed, on Monday, in our Select Committee,
that the laboratory tests were not necessarily very reliable;
how reliable are the laboratory tests that you are using currently
to determine whether the infection, in fact, was present in a
(Professor King) These are serology tests that you
are now asking about, as distinct from the PCR tests?
419. Correct; yes.
(Professor King) PCR tests for the virus, the serology
tests for the antibody, and the serology tests are performed at
the Pirbright Laboratory, so it is back to Alex.
(Dr Donaldson) There is no doubt in my mind that we
are using the most appropriate, sophisticated, sensitive and specific
tests which are available. We have something like 70 years' experience
of developing and improving on tests for foot and mouth disease.
I am not sure what the basis of your question is.