Submission to the Temporary Committee on Foot and Mouth Disease

R P Kitching       National Centre for Foreign Animal Disease


Relevant Dates:    Early February: Probable introduction of FMD virus to  Index Case
   20th Feb:  First case identified
   23rd Feb:   Animal movement ban

     21st March:  Prof. Roy Anderson in interview with the BBC News night
claims that the outbreak is "not under control" and that it would not peak
until early May, the date set for the General Election.
      27th March: Peak of new reported cases
      22nd March: Latest peak for new infections (27th March minus 5 day
incubation period)

     26th March: First meeting of the Science Committee
     27th March: Introduction of Ring Cull

The predictions presented by Prof Roy Anderson, in my opinion, heavily
influenced the decision to move policy decisions for the management of the
outbreak from MAFF to the Science Committee which directly advised COBR.
However the calendar of events listed above would indicate that the outbreak
was being brought under control before the introduction of the policy
recommended by the Science Committee, in particular the 48hr deadline for
the contiguous cull.  This policy initially required a cull of premises
within 3 km of infected premises and was never fully implemented - later it
was acknowledged that "the epidemic appeared to be decreasing more quickly
in Devon than in areas where the implementation of the contiguous cull had
been more effective." (Minutes of the 16th meeting of the FMD official
Science Group: 2nd May)

The early predictive models produced were deficient in a number of input
parameters, in particular:

1. The PanAsia strain causing the outbreak is not spread significantly as
an aerosol. Visits by myself to Sth Korea and Japan where the same strain of
virus has caused virgin outbreaks has not been characterised by aerosol
spread; and the index case in the UK which had involved 500 adult pigs,
infected for at least 3 weeks,  had not spread extensively to neighbouring
farms, even though initial predictions using aerosol production data derived
from pigs infected with another FMD virus strain had indicated virus
production sufficient to infect animals on the coast of Denmark.  The low
aerosol production from animals infected with this strain was later
confirmed by experimental data from IAH Pirbright.

2. The  models relied on identification of the first clinical case on a farm
in order to calculate when the virus first entered the premise.  Many of the
infections were in sheep in which  clinical disease was difficult to
identify, and in many cases it was clear the virus had been present a
considerable time before being recognized eg the first case in Shropshire
was identified on 16th March, but had probably been introduced on 19th or
20th Feb (FMD 2001 outbreak - descriptive epidemiology, HQ Epidemiology
Team, 6/4/01).

3. Although the disease was predominantly in the sheep, the models relied on
data generated from  outbreaks in cattle and pigs, both species in which the
virus spread rapidly within the herd and in which  clinical disease was easy
to recognize.  In sheep the disease spread only slowly, affecting only a
small percentage of the flock at any one time, this further reduced the
potential for aerosol spread, to a level that made it very unlikely that
aerosol transmission could occur over more than a few metres.   The presence
of disease in sheep was many times probably not seen until it spread to
cattle, or had been present some time in the sheep flock.  However, because
so little epidemiological investigation was taking place, this cannot be
proven; on the farm that sent sheep to Newcastle through Hexham and Longtown
markets, which I did visit (FMD/06), the disease had been recognized in the
cattle, but I was able to bleed the sheep on the farm, and we identified a
group of sero-positive sheep outside the cattle yard.  It was this group of
sheep that had been sent to market.  No such  epidemiological investigations
were subsequently carried out on other infected premises, other than the one
I visited and sampled in Devon the following week (FMD/07)

4. Because the models did not accommodate the delayed diagnosis of FMD,
their predictions of the rate of spread to new premises was inaccurate.

5. "The modellers also agreed that the epidemic was coming under control
faster than predicted by the models presented in previous weeks" (Minutes of
the 13th Meeting of FMD official Science Group: 19th April).

6. When questioned how the virus was spreading in spite of the movement ban,
Prof Anderson explained that this was not a function of the models, and no
explanation was required.


Following implementation of the policy recommended by the models, all
infected premises were required to be slaughtered within 24 hours, and
contiguous premises, initially up to a radius of 3 Km from the infected
premise, within 48 hours.  There was no opportunity for those responsible in
the affected areas to use discretion, based on local knowledge or previous
experience.  Diagnosis was on clinical evidence without laboratory support,
and whereas this was acceptable  for cattle and pigs, this clinical evidence
without laboratory support, and whereas this was acceptable for cattle and
pigs, this was not possible for sheep.

On 1st May, I asked for a summary of results generated at Pirbright; of 1876
premises that had been slaughtered, classified as VDR, DCF and SOS, samples
from 52.76% were negative on laboratory tests.  This was reported to the
Science Committee on 2nd May.

On numerous occasions during meetings of the Science Committee, both myself
and Dr Alex Donaldson expressed concern about the validity of the policy
derived from the models.  This was also transmitted in a minute to Minister
Spellar by Mr Richard Kornicki on 16th April, for submission to COBR.

The implementation of the rapid cull also prevented any detailed
epidemiological investigations, and sometimes even the collection of any
samples form and "infected premises" - a concern expresses on the 6th April
by the HQ Epidemiology Team.

At no time was the diagnostic capability of IAH overloaded, although we did
state that it was not necessary to submit large numbers of samples from each
of the suspect farms.  Because of the difficulty in making a clinical
diagnosis in sheep, and because of the probability than the FMD virus had
been present in a suspect flock for sometime, I queried the necessity of
using the 24 hour cull policy, as there would be no disadvantage to waiting
for laboratory confirmation.  A blood sample from a suspect infected sheep
would be either virus positive or antibody positive, sometimes both  - this
advice was ignored.  Later the question was raised that the sensitivity of
the tests being used at Pirbright was not sufficient to identify al infected
animals.  In my opinion, samples collected, following recommended
procedures, from infected animals in an unvaccinated population , would be
close  to 100% sensitive.

The consequence of adopting the policy recommended by the models was:

1 Excessive slaughter of healthy animals(during the 1967/8 outbreak in the
UK, there were approximately 2500 affected farms, and, 500.000 animals were
slaughtered, in the 2001 outbreak, there were 2026 declared infected farms
in the UK, and over 4,000,000 animals slaughtered, plus 2,500,000
slaughtered for welfare).

2  Inability to remove carcasses.

3 Necessity to transport carcases through uninfected areas

4 Loss of confidence of local Veterinarians and farmers

Public perception of control programme both in the UK and abroad was
severely damaged

Conclusion.

In my opinion, MAFF were bringing the 2001 FMD outbreak in the UK under
control before any of the policy changes recommended by the Science
Committee were implemented and that while predictive models can be a useful
tool in helping to formulate an epidemic disease control policy the takeover
of the programme from MAFF by the Science committee, which was heavily
influenced by the modellers with very limited practical experience of  FMD,
resulted in the unnecessary slaughter of possibly as many as 2 million
animals.

In my opinion, most of the spread of the FMD virus occurred before the
imposition of the movement ban on 23rd Feb, but because this was in sheep
and clinically not obvious,  it was not seen.  Therefore,  new cases were not added to the daily total,
until the virus spread into cattle, often on the same farm, or had cycled a
number of times in the sheep flock.  This concept was presented to the
Science Committee in April as an alternative interpretation to the models
being used, one that did account for the observation  that apparent new
introductions were occurring after the movement ban (Fig 1)

My proposal was considered unlikely by the modellers.