Statistical modelling of the FMD outbreak
4th August 2001
R.S. MORRIS, J. W. WILESMITH, M. W. STERN, R. L. SANSON, M. A. STEVENSON
Summary:
A spatial simulation model of foot-and-mouth disease was used in March and
early April 2001 to evaluate alternative control policies for the 2001 epidemic
in Great Britain. Control policies were those in operation from March 20, 2001,
and comprised a ban on all animal movements from February 23, 2001, and a
stamping-out policy. Each simulation commenced with the known population of
infected farms on April 10, 2001, and ran for 200 days. For the control policy
which best approximated that actually implemented from late March, the model
predicted an epidemic of approximately 1800 to 1900 affected farms, and
estimated that the epidemic would be eradicated between July and October 2001,
with a low probability of continuing beyond
October 2001. This policy included the slaughter-out of infected farms within 24
hours, slaughter of about 1-3 of the surrounding farms per infected farm within
a further 48 hours, and minimal interfarm movements of susceptible animals.
Delays in the slaughter of animals on infected farms beyond 24 hours after
diagnosis slightly increased the epidemic size, and failure to achieve
pre-emptive slaughter on an adequate number of at-risk farms substantially
increased the expected size of the epidemic. Vaccination of up to three of the
most outbreak-dense areas carried out in conjunction with the adopted control
policy reduced the predicted size of the epidemic by less than 100 farms.
Vaccination of buffer zones (designed to apply available vaccine and manpower as
effectively as possible) carried out in place of the adopted control policy
allowed the disease to spread out of control, producing an epidemic involving
over 6000 farms by October 2001, with no prospect of immediate eradication.
Vet. Record 149, 137-144.
This study was funded by DEFRA and involved many well known and respected authors.
The authors used advanced and well respected techniques to model the FMD outbreak. They predicted the length and size of the outbreak given the application of different control strategies, including vaccination.
In the first series of strategies the effect of varying the speed of stamping-out and varying the degree of contiguous culling was examined. They assumed a 24 hour killing time for infected premises and 72 hours for contiguous premises.
In the next series of strategies the time to the slaughter of both infected and contiguous premises was progressively increased.
In the third series of strategies the effect of a band vaccination strategy was examined. This strategy was as follows:
The final series of strategies examined involved a combination of slaughter and vaccination. The effects of ring vaccination of the highest density outbreaks was assessed. Again only cattle were vaccinated and vaccination tool place at the rate pf 8 000 cattle / day.
Their findings are summarised below:
The 'buffer zone' vaccination strategy proposed by the authors was ineffective at controlling the outbreak and indeed led to rapid spread of the disease throughtout the UK
The combined slaughter/vaccination strategy proposed by the authors was little more effective than slaughter alone.
They predict that the model closest to the actual strategy adopted in the UK will lead to eradication of FMD by October 31st 2001 with a total number of 1800-1900 infected farms.
Some Comments on the above paper
We are polling 'expert' opinions as to the usefulness of the above paper. However the comments below provided by Andrew Stephens, a UK Veterinary surgeon who has very actively been following the FMD crisis since the very beginning, give a useful overview of some of the problems we see with this work.
"It's a very extensive and complex article - but the only ( I think)
vaccination strategy modelled was one of buffer zones/bands that subdivided GB
into 6 zones. ie a sort of multiple firewall type strategy. The authors claim
this represented the only vaccination strategy that could have been mounted at
the time (end March) given available resources.
I'm not surprised that the InterSpread model predicted a very large epidemic if
this - buffer zone/band vaccination - was the *sole* control measure undertaken
- after all the epidemic(s) could run wild within each of the zones where
infection had already got a grip until it reached the vaccinated firewall
buffer. For example, the whole of Northern England was one zone.
This modelling study does NOT say that
vaccination per se would not and could not have been effective - instead it says
that this particular buffer zone/band style of vaccination - which it is claimed
is all that could have been done at the time for resource reasons would not have
been effective.
They did not for example model blanket vaccination using high load vaccine
within and around emerging hotspots combined with slaughter of definitive IPs,
or any of many other possible vaccination strategies that could have been
considered had resources not been the constraint they claim they were. Were
these assumed constraints real?
The InterSpread model - although FAR superior to the Ferguson, Anderson,
Donnelly model - also used early field data - up to April 10 in this study -
which is known to have been 'dodgy'.
It's important to consider exactly what this paper does and does not examine in
its correct, now historical, context.
No doubt -sadly - there will be many who cite this paper to claim that it says
that vaccination doesn't work in controlling FMD.
----------------------------------------------------------------------------
Footnote:
The Morris/EpiCentre team at Massey Uni, NZ has an excellent and deserved
reputation for VETERINARY epidemiological modelling and use of computers in
ANIMAL disease control. We should not throw out the baby with the bathwater.
The InterSpread model - part of the EpiMan information system for emergency
disease control - has a long history and is has a good reputation.
InterSpread models species separately, model actual locations of farms, and
accommodates the effects of chance when forecasting patterns of disease. It is a
FAR more sophisticated and well validated system than that hacked
together by Anderson et al.
Models of this sort are actually very helpful in exploring likely patterns of
disease and disease spread in POPULATIONS of animals - as opposed to lab
experiments that typically only explore development of disease in individual
or very small numbers of animals in a lab setting. They often reveal insights
that cannot be gained any other way - the human brain just doesn't handle
'calculating/forecasting' how a disease behaves at the population level very
well - and often what appears to be 'common sense' turns out not to be when
explored with computer simulation models. Epidemiology is a branch of medicine
that 'looks at' the 'natural history' of disease at the population level -
rather than at the individual level.
The study only explores one (in my opinion daft) vaccination strategy. The
results risk being interpreted as saying 'vaccination doesn't work' - but that
is NOT what the paper says - it says that one (absurd) strategy for vaccination
probably would not have worked. In fact the vaccination strategy modelled is so
absurd that even the 'man in the street' could have figured out that it would be
ineffective.
It's also true to say that the field data - from the early weeks of the epidemic
- used to parameterise this model - and others - is known to be of somewhat
dubious quality, but it's not clear (to me anyway) how much impact
this would have on the results of the modelling studies.
-------------------------
(Updated comments since Andrew's first quick look at the article)
This study (funded by DEFRA) only modelled one strategy for vaccination - a
strategy that I've never ever seen mentioned elsewhere as a possibility - no
doubt because it's patently daft. It divided UK into 6 large zones by bands
of vaccination ('firewalls') and the zones are often HUGE.
For example, the whole of Northern England was one zone. Only cattle - not sheep
or pigs - would be vaccinated in the modelled strategy (so the 'firewall' would
of course be permeable!) and no slaughter - even of infected animals - would
take place anywhere - just this permeable band vaccination of cattle alone.
One doesn't need to be a rocket scientist to realise that such a vaccination
strategy would be almost entirely useless in preventing the spread of the virus.
The 'justification' for only modelling this absurd use of vaccination is the
claim that this is/was the only form of vaccination campaign that could have
been mounted at the time under consideration - around the end of March - due
to resource (vaccine and manpower) constraints. In my opinion this is complete
rubbish.
I hate to say it but it does appear that this paper (The Veterinary Record) has
two goals -
a) to retrospectively justify the (unjustifiable) contiguous cull policy and
b) to " trash" vaccination.
I had thought that the major authors had more integrity than that - but perhaps
not. Of course there's no way DEFRA - who funded the study - would have allowed
publication of any results that did not support their policies.
I cannot imagine that the absurd vaccination strategy for which results are
presented was the only vaccination strategy modelled - but perhaps DEFRA blocked
publication of any other results?
........
One whole 'line' of criticism could be I think that their results do not appear
to consider the total numbers of farms culled out nor to talk number of animals
culled out - merely measure 'efficiency' by estimating the number of 'infected
farms' which I take to mean IPs. This is clearly a very crude measure of
efficiency vis a vis various strategies. One of the worst
aspects of current policy is the 'collateral damage' - to use a military
euphemism from the Vitenam war - is. the number of farms culled out as CPs and
DCs without proper risk assessment, and the massive overkill and subsequent
disposal problems not to mention the disproportionate effects all this had on
rural communities and other businesses and industries (of which tourism is but
one).
One cannot make rational choices between possible strategies if one doesn't
factor in such things - and there appears to have been no effort made to do so -
nor does their appear to have been any social cost benefit analysis of the
possible strategies although they may say there was not time to do this I'd
disagree because I think a rough and ready estimate could have easily have been
drive from their models and would have shown huge differences between possible
strategies that would have influenced decision making.
The InterSpread model is plainly infinitely superior to the model used by
Anderson et al. but I do wonder whether key parameters were credibly estimated -
eg. the likelihood of farm to farm spread - which is vitally important.
I was wrong in claiming they only modelled one vaccination strategy - they also
modelled the superimposition of vaccination in hotspots on top of the 24/48 cull
policy - needless to say it didn't reduce no. of infected farms much - of course
it didn't because most would have been culled out with this policy anyway. So
that was an equally silly policy to model IMO. The zones/bands vacc. strategy
was just plain daft.
One of their results showed that delaying culling of index cases beyond 24 hours
didn't have much effect - Hmmmmm.......that would suggest that there would be
time to do diagnostic tests on suspected cases - which much
greater spread of virus - which in turn would have avoided misdiagnoses and the
sequelae of those.
As I say one could pick away at their paper for years.
Andrew Stephens BVetMed MRCVS
36 Falstaff Avenue, Earley, Reading, RG6 5TQ, UK
Tel: +44-(0)118-9756574 Fax: +44-(0)870-1337217
Email: andrews@mediavets.co.uk
Web: http://www.mediavets.co.uk
top of page