Download The UK foot-and-mouth disease outbreak — the

PERSPECTIVES
33. Costerton, J. W., Lewandowski, Z., Caldwell, D. E.,
Korber, D. R. & Lappin-Scott, H. M. Microbial biofilms.
Annu. Rev. Microbiol. 49, 711–745 (1995).
34. Hall-Stoodley, L., Stoodley, P. & Costerton, J. W.
Bacterial biofilms: from the natural environment to
infectious diseases. Nature Rev. Microbiol. 2, 95–108
(2004).
35. Williams, H. N., Kelley, J. I., Baer, M. L. & Turng, B.-F. The
association of bdellovibrios with surfaces in the aquatic
environment. Can. J. Microbiol. 41, 1142–1147 (1995).
36. Koval, S. F. & Bayer, M. E. Bacterial capsules: no barrier
against Bdellovibrio. Microbiology 143, 749–753 (1997).
37. Koval, S. F. & Hynes, S. H. Effect of paracrystalline
protein surface layers on predation by Bdellovibrio
bacteriovorus. J. Bacteriol. 173, 2244–2249 (1991).
38. Westergaard, J. M. & Kramer, T. T. Bdellovibrio and the
intestinal flora of vertebrates. Appl. Environ. Microbiol.
34, 506–511 (1977).
39. Lenz, R. & Hespell, R. B. Attempts to grow bdellovibrios
surgically injected into animal cells. Arch. Microbiol. 119,
245–248 (1978).
40. Scherff, R. H. Control of bacterial blight of soybean by
Bdellovibrio bacteriovorus. Phytopathology 63, 400–402
(1973).
41. Varon, M. & Shilo, M. Attachment of Bdellovibrio
bacteriovorus to cell-wall mutants of Salmonella spp. and
Escherichia coli. J. Bacteriol. 97, 977–979 (1969).
42. Varon, M. Selection of predation-resistant bacteria in
continuous culture. Nature 277, 386–388 (1979).
43. Alexander, M. Why microbial predators and parasites do
not eliminate their prey and hosts. Annu. Rev. Microbiol.
35, 113–133 (1981).
44. Shemesh, Y. & Jurkevitch, E. Plastic phenotypic
resistance to predation by Bdellovibrio and like
organisms in bacterial prey. 6, 12–18 (2004).
45. Huang, S. S., Labus, B. J., Samuel, M. C., Wan, D. T. &
Reingold, A. L. Antibiotic resistance patterns of
bacterial isolates from blood in San Francisco County,
California, 1996–1999. Emerg. Infect. Dis. 8, 195–201
(2002).
46. Schwudke, D. et al. The obligate predatory Bdellovibrio
bacteriovorus possesses a neutral lipid A containing
α-D-mannoses that replace phosphate residues:
similarities and differences between the lipid As and the
lipopolysaccharides of the wild-type strain B.
bacteriovorus HD100 and its host-independent derivative
HI100. J. Biol. Chem. 278, 27502–27512 (2003).
47. Wilkinson, M. H. F. Predation in the presence of decoys:
an inhibitory factor on pathogen control of
bacteriophages or bdellovibrios in dense and diverse
ecosystems. J. Theor. Biol. 208, 27–36 (2001).
48. Stolp, H. & Petzold, H. Untersuchungen uber einen
obligat parasitischen Mikroorganismus mit lytischer
aktivitat fur pseudomonas bakterien. Phytopathogishe
Zeithschrift 45, 364–390 (1962).
49. Stolp, H. & Starr, M. P. Bdellovibrio bacteriovorus gen. Et
sp. N., a predatory, ectoparasitic, and bacteriolytic
microorganism. Antonie Van Leeuwenhoek 29, 217–248
(1963).
Acknowledgements
The authors would like to thank R. Chaudhuri of Colibase,
University of Birmingham, UK, for assistance with genome analysis. Their work is funded by the Wellcome Trust.
Competing interests statement
The authors declare that they have no competing financial interests
Online links
DATABASES
The following terms in this article are linked online to:
Entrez: http://www.ncbi.nlm.nih.gov/Entrez/
Bdellovibrio bacteriovorus HD100 | Buchnera aphidicola |
CAE77837 | CAE78299 | CAE78505 | CAE78865 | CAE78875 |
CAE79180 | CAE79394 | CAE79452 | CAE79454 | CAE80233 |
CAE80242 | CAE80483 | CAE80640 | CAE81224 | RadA |
RecA | RecG
FURTHER INFORMATION
B. bacteriovorus strain W:
http://www.micro-gen.ouhsc.edu/b_bacter/b_bacter_home.htm
Bacteriovorax marinus strain J:
http://www.sanger.ac.uk/Projects/B_marinus/
Bdellovibrio genome project:
http://www.eb.tuebingen.mpg.de/schuster/research_bd.htm/
R. Elizabeth Sockett’s laboratory:
http://www.nottingham.ac.uk/biology/contact/academics/
sockett/research.phtml
Access to this links box is available online.
NATURE REVIEWS | MICROBIOLOGY
OPINION
The UK foot-and-mouth disease
outbreak — the aftermath
Daniel T. Haydon, Rowland R. Kao and R. Paul Kitching
The 2001 epidemic of foot-and-mouth
disease in the United Kingdom triggered a
livestock culling campaign that involved
the slaughter of more than 6.5 million
animals. Three years later, management of
the epidemic remains controversial. Some
believe that untried control methods
based on unvalidated models replaced
well-established policy, motivating an
unnecessary slaughter. Others hold that
rigorous quantitative approaches provided
the basis for new incisive policies that
significantly curtailed the epidemic. Now,
new and more flexible control policies
have been adopted throughout Europe.
For these policies to receive the full
confidence of scientists, veterinarians and
the general public, it is necessary that we
improve both our understanding of where,
how and why control measures initially
failed in 2001 and how new policies should
be implemented.
Foot-and-mouth disease (FMD) is a highly
infectious viral disease of cloven-hoofed
animals. Its main epidemiological features are
described in BOX 1. The Office Internationale
des Epizooties (OIE) recognizes countries (or
regulated regions within countries) to be in
one of three disease states: FMD free without
vaccination; FMD free with vaccination; and
FMD present with or without vaccination.
Countries move through these stages, typically accessing wider and more profitable
export markets as the level of disease control
is progressively improved. Countries with
endemic FMD generally start with extensive
vaccination programmes, which, in combination with movement regulations and other
sanitary measures, reduce the incidence of
new infections, ultimately to zero, after which
time the use of vaccination can be stopped
altogether. Such programmes are costly, can
take many years to succeed and often experience substantial setbacks1. Consequently,
the development of alternative regulatory
practices that recognize regional disease-free
status as not necessary for safe trade of some
livestock commodities should be strongly
encouraged2. Over the past five decades most
European countries have used vaccination programmes, but in 1992 the widespread use of
vaccination in the European Union (EU) was
banned for reasons we describe below. Since
then, with the exception of occasional outbreaks, the EU has moved to a disease-free
unvaccinated state. In the United Kingdom
vaccination has never been used, and in a fully
susceptible host population control has always
been undertaken using ‘traditional methods’3.
Historical perspectives on control
Traditional control of FMD involves the
imposition of regional movement bans, the
disinfection of infected properties and rapid
slaughter of all animals on premises that are
identified as being infected (IPs), as having
had ‘dangerous contact’ with an IP (DCs)
or as being particularly infectious if, and at
risk of being, infected. For decades most
FMD outbreaks have been well controlled
by traditional methods. For example, the
Northumberland Report4 refers to 180 primary outbreaks of FMD in the United
Kingdom between 1954 and 1967, all of
which were contained using traditional methods: 169 resulted in no more than 20 further
cases (FIG. 1a), and only four epidemics generated more than 50 cases — one of these is the
1967 outbreak that resulted in more than
2,000 cases.
The 2001 UK outbreak was different to
previous outbreaks in many respects, foremost of which was that in the early stages of
the epidemic the authorities were unable to
apply traditional control methods to a high
standard in all areas5,6. For the first time
mathematically based computer models
were used in the management of an outbreak. In contrast to empirically derived
‘tried and tested’ FMD control policies that
have been in use over the past several
decades, the results of this modelling indicated that the epidemic would be controlled
more effectively if widespread culling of
apparently uninfected herds, and, in particular, culling based purely on geographical
proximity was adopted. These observations
beg two questions. First, why did the models
produce such a contrasting FMD control
VOLUME 2 | AUGUST 2004 | 6 7 5
PERSPECTIVES
policy to that apparently endorsed by the
historical record? Second, what would have
happened in 2001 if traditional methods had
been efficiently applied from the outset?
Surprisingly, these are questions that have
received little attention in the literature. A
good understanding of where, how and why
traditional methods failed is vital if the correct lessons are to be learned from this epidemic. It is particularly important that these
issues are subject to scrutiny because the new
UK Department for Environment, Food and
Rural Affairs (DEFRA) contingency plan for
FMD outbreaks is a particularly flexible one
that will require the authorities to choose
from a range of control measures that
include culling of both IP and DC animals,
pre-emptive FIREBREAK CULLING, and emergency
7
VACCINATION TO KILL and VACCINATION TO LIVE .
A summary comparison of the outbreaks
of 1967 and 2001 is thought-provoking. In
1967, estimates of the number of IPs that
were infected directly from primary cases
vary from 38 to 59 (REFS 8,9), and all were
contained in a relatively small area. In 2001,
disease entered the United Kingdom in early
February and by the time disease was confirmed on 20 February, at least 30 premises10,11, and possibly as many as 79 (REF. 12),
were infected. Unlike in 1967, the result was
widespread outbreaks from Dumfries and
Galloway in the north, to Devon in the south,
as well as overseas to Northern Ireland, France
and the Netherlands. Late on the evening
of 23 February a national movement ban
was imposed. Following this, most spread was
‘local’, that is, limited to within 3 km of an
IP10,13. Both the 1967 and 2001 epidemics
were characterized by lengthy final phases. In
1967 the epidemic lasted 212 days and caused
outbreaks on 2,364 different premises4. In
2001 the epidemic lasted 214 days and resulted
in the identification of infection and culling
of herds on 2,026 premises6.
Glossary
FIREBREAK CULLING
The culling of animals for the purpose of preventing
spread of infection beyond an area, even though the
animals are not believed to have been exposed to
infection.
VACCINATION TO KILL
Or suppressive vaccination. A vaccination policy
adopted within the protection zone that anticipates
that vaccinated individuals will be destroyed as soon
as circumstances allow.
VACCINATION TO LIVE
Or protective vaccination. A vaccination policy that
anticipates that vaccinated individuals will not be prematurely slaughtered, and will enter the food chain as
normal.
676
| AUGUST 2004 | VOLUME 2
Box 1 | Foot-and-mouth disease — the basics
Foot-and-mouth disease (FMD) is a disease of cloven-hoofed animals such as cattle, pigs, sheep
and goats. The pathology of FMD includes fever, vesicles in the mouth, feet and udders, loss of
milk production in adult animals and death in young animals. Infected cattle, sheep and goats
can become carriers of FMD that are persistently infected (and occasionally infectious) for up to
3.5 years depending on the host species37.
FMD virus is an RNA virus of the family Picornaviridae (genus Aphthoviridae). It has seven
distinct serotypes, between which there is no immunological cross-reactivity. The FMD virus
genome contains 8,400 nucleotides and 12 genes, four of which encode capsid proteins. FMD
virus genomes are diverse: capsid genes of the same serotype can differ by more than 30% of
nucleotides and distant subtypes within the same serotype may only elicit weak cross-reactivity,
so the choice of (inactivated) vaccine strains must be carefully matched to outbreak strains
against which protection is sought. The virus can evolve rapidly (for example, 1.5% of capsid
gene nucleotides can change per year, or at an estimated fixation rate of one nucleotide in these
genes over the course of an individual infection42), but the Pan-Asia strain has shown
remarkable genetic stability over several years.
The virus is released in all secretions and excretions of an infected animal, especially in their
breath and secretions that are associated with ruptured vesicles. Transmission of FMD virus can
take place mechanically by people who have handled infected animals, on straw or hay that is
contaminated by infected animals, on farm vehicles or milk tankers carrying infected milk, or even
on the surgical equipment of veterinary surgeons. FMD virus can also spread as an aerosol — the
1981 outbreak on the Isle of Wight, UK, was caused by virus from an outbreak in pigs in Brittany, a
distance of more than 250 km43; however, over land, spread of the virus rarely exceeds 10 km44.
One of the most intriguing aspects of the epidemiology of FMD is the seemingly high
variability in the transmissibility of the virus. It is thought to have one of the lowest infectious
doses of any virus45 has enormous potential for infectiousness within herds46 and, under the
right conditions, a remarkable capacity to spread by aerosol over considerable distances8,43.
Oddly however, under other conditions, FMD virus seems to be less infectious — for example,
there is some evidence that FMD virus cannot be maintained over long time periods in sheep
populations47 and when the Pan-Asia virus spread to the Netherlands, studies showed that it
failed to spread between calves that were in direct contact with each other48. More than half (98)
of the 180 primary cases reported in the United Kingdom between 1954 and 1967 failed to infect
any other premises and in the 2001 epidemic the first case — a large pig farm left infectious for at
least 2 weeks during meteorological conditions favourable to transmission28 — is thought only
to have infected 1–10 neighbouring farms before it was finally culled27.
Despite these similarities, some important
differences must be recognized when comparing the two epidemics. In 1967 slaughter was
mostly confined to animals from IPs, whereas
in 2001 animals were culled on a further 8,131
premises that were close to, or in some other
way associated with, IPs 6. In 1967, approximately 442,000 animals were slaughtered to
control the epidemic, whereas in 2001 at least
4 million animals were slaughtered for the
purposes of disease control, with at least a further 2.5 million animals destroyed in ‘welfare
culls’6. Nationally, prior to the 2001 epidemic,
total numbers of the two most important
host species for the FMD virus (cattle and
pigs) were 8% less than 1967 figures. The average dairy herd size had doubled since 1967
(REF. 5), but the effects of fewer larger herds on
disease control are unclear (although obviously the number of individuals culled per IP
must increase). In addition, sheep numbers
had increased by 46% and the physical movement of infected sheep was responsible for
much of the early dissemination of infection
in 2001 (REF. 14) — sheep were present on 80%
of all IPs, including 15% that were solely sheep
farms. By contrast, the 1967 epidemic was
mostly restricted to cattle. Diagnosis in sheep
is much more difficult, which probably led to
delays in the identification of IPs in 2001.
However, sheep are less infectious than both
pigs and cattle15. The modern livestock
industry involves the movement of many
more animals than in 1967, but the epidemiological significance of this diminishes after
the imposition of movement bans. The effects
of these changes on the efficacy of traditional
measures in controlling FMD outbreaks
remain largely unexplored.
The extent to which pathogens are infectious prior to the onset of discernable
pathology is obviously an important determinant of the extent to which epidemics can
be efficiently controlled16. The FMD virus is
genetically diverse and it is plausible that
differences in both the transmissibility and
the route of transmission might exist between
strains, which could alter the effectiveness of
www.nature.com/reviews/micro
PERSPECTIVES
a Sizes of FMDV outbreak
b Time to slaughter for 1967 and 2001 epidemics
70
50
Number of
secondary outbreaks
0
1–5
6–10
11–20
21–50
>50
50
(67)
Percentage of IPs
Number of primary outbreaks
40
2001 Reporting to slaughter
2001 Confirmation to slaughter
1967 Confirmation to slaughter
60
30
(243)
20
(61)
40
30
20
10
10
(2,364)
0
0
54 55 56 57 58 59 60 61 62 63 64 65 66 67
19 19 19 19 19 19 19 19 19 19 19 19 19 19
Year
0
1
2
3
4
5
6
7
8
9
10
11
Interval (Days)
Figure 1 | Past performance and implementation of traditional methods. a | Primary outbreaks of
FMD from 1954 to 1967, together with the number of associated secondary outbreaks4. Primary
outbreaks are those that cannot be linked with any known source in Great Britain and are therefore
attributed to FMD introduction from abroad. Secondary outbreaks are all those that arose by the spread of
infection from primary outbreaks. Numbers in brackets refer to the actual number of outbreaks for those
epidemics with more than 50 secondary cases. b | Comparison of the time between reporting or
confirmation of FMD and slaughter of animals, for the epidemics in 1967 and 2001. Time intervals refer to
the difference between the dates of reporting/confirmation and slaughter from 25 October onward for
1967 (REF. 4) and IPs with confirmation dates on or after 24 February for 2001. This figure was
constructed using the most recent DEFRA data. IP, infected premises.
Although successful control can be considered to be the prevention of endemic disease,
the aims of a successful control strategy may
have changed since traditional methods were
last tested. Control strategies might seek to
minimize various quantities, such as total animal loss, duration of the epidemic (which is
currently the main objective in England and
Wales)7, regional spread, financial loss (to
several economic sectors) or animal suffering.
Moreover, a small outbreak that is under control, and therefore in decline, is likely to be
viewed as acceptable, but when the number
of extant IPs is high, a substantial and potentially unacceptable number of new cases can
still arise from an epidemic that is — technically — considered to be under control and
might therefore necessitate further control
measures18. Historically, traditional methods
that have been viewed as successful because
FMD has not become endemic in the UK
might now be rejected on the grounds that
they might be unable to contain FMD outbreaks under climatic conditions that are
favourable to viral transmission and in areas
of high livestock density. However, controversy persists because assembling evidence
that alternative methods of control might
work any better is not straightforward.
The models — strengths, weaknesses
a traditional control policy. However, experimental evidence for increased transmissibility of the Pan-Asia FMD virus strain that
was responsible for the 2001 outbreak is statistically weak17. What we do know is that, for
some reason, there was little aerosol spread
during the 2001 outbreak and this should
have made the outbreak easier, and not more
difficult, to contain.
Between 1967 and 2001, the capacity of the
State Veterinary Services (SVS) in the United
Kingdom had been greatly reduced. The
number of SVS field staff was reduced from
about 600 in 1967, to 220 in 2001 (REF. 6), and
by 2001 few investigators had any practical
experience of dealing with FMD. The scale
of the outbreak prior to the imposition of
movement controls stretched the SVS beyond
their capacity to implement a functional traditional response. Between implementation
of a national movement ban and adoption of
a contiguous premises (CP) cull (in which
animals on neighbouring premises to an IP
were culled), traditional control methods were
implemented to a standard far below that of
the 1967 outbreak. Fewer than 10% of IPs
were subject to culling on the day disease was
reported, and only 35% were subject to
culling by the following day (FIG. 1b). Although
the number of reported IPs was increasing
NATURE REVIEWS | MICROBIOLOGY
exponentially, only 0.8 DCs were identified
per IP5. This response was inadequate — the
success of traditional methods depends on
having the necessary resources to identify
and cull IPs and DCs as early as possible.
Controlling transmission
Traditional control measures, directed at
culling animals from IPs and DCs within 24
hours of reporting disease, were attempted
up to 23 March, after which additional
measures were introduced that included:
slaughter on suspicion of infection; culling
of pigs, sheep and goats on premises within
3 km of an IP in both Dumfries and Galloway
and in Cumbria; and culling all premises
contiguous to an IP within 48 hours. This
last measure became known as the ‘24/48’
IP/CP cull policy and would prove highly
controversial. There are instructive regional
variations in the extent to which this policy
was implemented; it seemed to be somewhat discretionary in Scotland, and may
not have been “more than 50% implemented” elsewhere6. Indeed, although the
CP cull is in principle easy to define, what is
truly ‘contiguous’ is subject to interpretation, and the most appropriate definition is
contingent on the assumed mode of disease
transmission.
There is one other difference between the
management of the control strategy in 2001
to that of previous outbreaks. Soon after the
discovery of the first case of FMD in 2001,
members of the ‘FMD Science Group’ oversaw the construction and analysis of three
independently developed epidemiological
models of FMD spread. These models, which
were based on computer simulation and
mathematical techniques, were fitted to data
as they were collated over the course of the
epidemic and used to predict the future
course of the epidemic under several control
scenarios13,19–21. Although some very good
quantitative epidemiologists were present in
the United Kingdom in 1967, there is no record of any of them advising the FMD control
policy at that time. The new and important
role for quantitative modelling in real-time
disease-control management reflects technological developments (such as powerful
computers and spatial data), the maturing of
quantitative epidemiology as an academic
discipline and unusually direct communication between leading epidemiologists and
senior government scientific advisors6. More
importantly, it reflects a growing awareness of
the need for rigorous data analysis, which was
highlighted by the experience with BSE in the
United Kingdom22.
VOLUME 2 | AUGUST 2004 | 6 7 7
PERSPECTIVES
The models are compared in detail elsewhere23. As they adopted different approaches
yet still obtained similar important results24, it
has been argued that the conclusions derived
from them are likely to be robust. However,
this confidence should not be exaggerated
because all the models share certain fundamental similarities. One is that they assumed
that the location of the farming premises as
recorded in the agricultural census of 2000
was an appropriate surrogate for the location
of livestock. Discrepancies between the location of livestock and the location according to
the census were well known, and would limit
the spatial resolution at which these models
could determine local culling strategies. For
example, the value of quantitative comparisons amongst a 3 km, 2 km, 1.5 km or CP
cull would have been unclear21. Another similarity is that they were all parameterized using
the same epidemiological data (provided by
DEFRA), and it was assumed that these data
were sufficiently accurate to allow detailed,
predictive mathematical modelling at a level
adequate to advise policy. Although models
are valuable because they allow the objective exploration of data, all models must
inevitably make assumptions and it is part
of the modelling process to explore, question and review the consequences of these
assumptions in an attempt to improve understanding of model behaviour and acquire
greater and more widespread confidence in
their predictions. We believe there are a number of points arising from the use of the outbreak data that require further examination
in the published literature.
First, contact-tracing data indicated that
premises were infectious as early as 3.5 days
after the estimated date of infection, with a
constant probability of transmission until all
the animals were slaughtered13,21. Unless transmission to an IP has occurred by animal
movement, IPs would not always be expected
to be infectious so quickly, and infectiousness
might be expected to increase over time as the
number of infected animals in the herd
increases. The high estimated levels of infectiousness soon after the date of infection could
be due to consistent error in identification
of the source IP or in estimation of the date of
contact with the infection. The date of contact
is estimated by the date of the earliest known
contact with an IP or by the age of lesions;
however, in pigs and cattle older lesions have
an error estimate of ± 3 days25. Although other
factors, such as increased biosecurity on uninfected farms close to IPs might reduce the
effects of rising infectivity over time, the models could be sensitive to changes in the infectiousness profile, and overestimation of the
678
| AUGUST 2004 | VOLUME 2
early infectiousness of IPs would lead to
models exaggerating the importance of CP
culling. The sensitivity of the models to this
assumption remains unclear.
Second, the relationship between the
probability of transmission and the distance
to an IP (known as the ‘transmission kernel’)
is based directly on the contact-tracing data13.
Although the data represents the best estimate
of ‘who infected whom’, underlying biases in
these estimates are mostly unexplored in the
literature. Simulations of the epidemic in
Cumbria overestimate the number of IPs
close to locations at which the epidemic originated26. This might have been the result of
excessive wave-like progression of the simulated epidemics, caused by an overestimation
of the importance of local spread, and could
lead to an exaggerated estimate of the value of
CP culling.
Third, it is assumed that identification of
the disease on farms was accurate. Detailed
analysis of the proportion of IPs that were
confirmed as infected by laboratory tests is
yet to be published, but given the difficulty
of diagnosing disease in sheep, it is anticipated to be substantially less than 100% —
for example, in 1967, fewer than 20% of
doubtful cases were subsequently found to
be infected4. Subsequent analysis12 has
revealed that the role of animal movement
in the early spread of infection might have
been underestimated, leading to a potential
overestimation of the number of premises
that were infected after the movement ban
was imposed. Both sources of error could
cause the models to overestimate the control
effort that was required13.
Fourth, although it is well known that after
movement restrictions were imposed in late
February the transmission kernel changed to
reflect much higher levels of local transmission21, a constant transmission kernel was
assumed in all models thereafter. The transmission kernel only describes the effect of
distance on infectious contact without distinguishing amongst airborne, animal contact
or mechanical transmission. Early on in the
course of the epidemic, aerosol transmission
may have been important27, but environmental conditions28 and the lack of transmission to
pigs reduced its later impact10. Anecdotally, the
proportion of infectious contacts that were
assigned to animal movements or human
activity increased over the time course of the
epidemic29. Transmission is assumed to be
independent of the control policy but changing the control policy might affect logistics,
farmer compliance or the implementation of
biosecurity. Ferguson et al.20 have suggested
that significant increases in the transmission
rate might have occurred towards the end of
the epidemic. Although difficult to prove, the
extensive movements of people, animals and
vehicles owing to the intensive slaughter policy
could have exacerbated transmission.
Finally, the models assume that the linear
distance between farms was the main determinant of transmission risk. If two farms
have the same composition and are equidistant from an IP, then the models assume that
they are equally likely to become infected, and
so a purely spatially motivated policy like the
CP cull is favoured. However, if one is more
likely to become infected than the other (for
example, if two farms are connected by a wellused road), then identifying the high-risk
property becomes more worthwhile. This is
the question that lies at the core of the controversy over the control policy — what is the
effectiveness of a DC cull compared with a CP
cull? Although the models predict that, as
implemented, the cull was superior to the
other options considered, a comparison with
a well-managed traditional policy is yet to be
published13,20. More recent analyses indicate
that even very precise DC culling would only
have been valuable if the time between disease
diagnosis and slaughter could have been
reduced26, and that the overall number of
premises subject to culling may be fairly
insensitive to overculling at the local level18.
This indicates that the superior performance
of the CP cull in reducing the epidemic duration might be more important than any
apparent over-culling. However, a better
understanding of the risk factors that are
associated with transmission of infection is
critical for a comprehensive evaluation of the
benefits of DC culling30.
The main factor responsible for the end of
the epidemic probably varied regionally, but
there is evidence that factors other than the
change in control policy could have been the
most important (BOX 2). Whatever future
analyses may tell us, unprecedented numbers
of animals were slaughtered in a new and
untested control procedure, largely formulated
and justified with the use of necessarily hastily
developed computer models. Given uncertainties in the data and the reliance of these models
on assumptions that are necessarily crude and
also difficult to verify, it is difficult to make the
argument that mathematical models showed
that implementation of widespread and intensive culling was the only tenable option.
Models did show clearly, and at a relatively
early stage, that a traditional policy, as previously implemented, was not sufficient to prevent the development of a very large epidemic.
However, the main arguments in favour of a
CP cull are simpler decision-making and ease
www.nature.com/reviews/micro
PERSPECTIVES
Box 2 | Reduction of disease transmission
IP Incidence
04
/0
5
27
/0
4
20
/0
4
13
/0
4
06
/0
4
30
/0
3
23
/0
3
16
/0
3
09
/0
3
02
/0
3
Date
b
10
9
8
7
6
5
4
3
2
1
27
/0
4
20
/0
4
13
/0
4
06
/0
4
0
30
/0
3
Reduction in the infectious period
0
23
/0
3
Movement restrictions and biosecurity both
contribute to a reduction in the rate at which
disease is transmitted.
5
16
/0
3
Reduction in the transmission rate
10
09
/0
3
In the case of the 2001 epidemic, fortuitous
weather conditions might have limited aerosol
transmission28.
15
02
/0
3
Environmental factors
20
23
/0
2
If the pool of susceptible premises becomes
sufficiently small the disease will die out. As the
range of FMD transmission is limited, the
reduction in the number of susceptible premises
in the vicinity of IPs may also cause the disease
to die out locally. Any culling of premises that
are not infected (whether as part of a dangerous
contact (DC), contiguous premises (CP) or
other cull) will contribute to the depletion of the
susceptible population with CP culling likely to
contribute most to local depletion. Movement
restrictions decrease the effective range of FMD
transmission21, making local depletion effects
more important.
30
25
IP Incidence
Depletion of the susceptible population
a
23
/0
2
In the 2001 epidemic several factors might have
contributed to reducing transmission. These can
be categorized by the nature of their effects, with
some factors having multiple effects.
Date
All culling that removes infected premises
(including IP, DC and CP culling strategies)
contributes to a reduction in the infectious period. The rationale behind the CP cull is that the increased probability of culling uninfected premises is
compensated for by a reduction in the infectious period of those CPs that are infected but have not yet been identified.
When a combination of these factors results in a reduction of the average number of new infections caused by an IP to below one, the epidemic is
classed as being under control49, although there might be many more infection events before the epidemic is over. What caused the 2001 epidemic
to end? This is likely to have varied between regions. Reducing the period of time before an animal is slaughtered and increasing detection rates no
doubt contributed to the decline of the epidemic, and the revised policy measures were designed to facilitate this. However, reconstructions of the
epidemic indicate that the rate at which new infections were arising peaked between 19 March and 21 March11, and the number of reported cases
peaked on 26 March — before these new policy measures were implemented13,23. Therefore, the switch to more stringent control procedures could
not have been responsible for this initial reduction. Population depletion may have had a greater effect. In Cumbria (figure part a) the epidemic
(shown in red) is compared to an average of 100 simulated epidemics using the algorithm of Kao26 (shown in green) and with modified simulations
with the same parameters but double the number of premises and double the area (shown in blue). This analysis addresses what could have
happened if Cumbria covered an area twice as large. The divergence of the two curves in late March suggests that the main cause of the downturn
was geographical isolation — the epidemic simply ran out of room. Thus, in Cumbria, although the revised control policies are likely to have
contributed to the decline of the epidemic18,26, the epidemic was arguably never under control, except in the sense that movement restrictions and
biosecurity prevented the transmission of disease to new areas while the epidemic burned itself out. In Devon (figure part b), the epidemic (shown
in red) is compared to the average of 100 simulations recreating the actual culling policy in Devon (shown in green) and a policy in which culling is
restricted to animals on IPs only (shown in blue). The incidence in the two simulations declines at similar points. This corroborates a previous
result50, which indicated that IP culling alone might have been sufficient to bring the disease under control. Although requiring more rigorous
investigation, this was likely to have been true in other affected regions in the UK where infection was less widespread.
of management, together with the benefit that,
in a time of great chaos and uncertainty, a
clearly defined policy with simple goals can be
of both logistical and political value.
Looking to the future
The direct and indirect economic tolls of the
2001 epidemic are estimated to have been at
least UK £3 billion and UK £5 billion,
NATURE REVIEWS | MICROBIOLOGY
respectively31,32. This, together with widespread public disquiet at the visible slaughter
of at least 6.5 million animals, has prompted a
major revision of outbreak contingency planning. Many excellent recommendations have
been made by commissioned reports5,6,33 and
have been incorporated into future contingency plans7. An obvious recommendation is
the imposition of an automatic nationwide
movement ban on all livestock immediately
after confirmation of the first case. In 2001,
such action could have halved the size of the
epidemic11. More radical is the recommendation that emergency vaccination “should
now be considered as part of the control
strategy from the start of any outbreak”5. This
recommendation arises partly from the positive outcomes of vaccination campaigns
VOLUME 2 | AUGUST 2004 | 6 7 9
PERSPECTIVES
carried out in Uruguay34 and experience in
the Netherlands35 in 2001.
Culling of animals on IPs will continue to
remain part of any control programme, and
therefore any outbreak will always involve
slaughter regardless of whether emergency
vaccination is implemented. Nevertheless, the
recommendation to vaccinate is welcome
provided several important difficulties are
overcome. Vaccinated animals will always
require time to acquire protective immunity,
although this time can be substantially
reduced by the use of high-potency vaccines36.
Furthermore, vaccination might not always
prevent infection or the establishment of a
long-term asymptomatic but potentially
infectious carrier state in cattle and sheep37
(although recent results seem to also confirm
the effectiveness of high-potency vaccines in
inhibiting the carrier state38). For this reason,
OIE regulations had required a delay of
1–2 years before countries that had an FMD
outbreak, and which had used emergency
vaccination, could reapply for full disease-free
status (in contrast to just 3 months if culling
alone was used). This delay has recently been
reduced to 6 months provided adequate
numbers of vaccinated animals are subject to
a test for antibodies to viral non-structural
proteins (NSPs) that can, in principle, distinguish between animals that have been vaccinated and those that are, or have been,
infected. However, there are some problems
with this approach— for example, it is doubtful that putative carrier animals always produce detectable quantities of antibody to
NSPs39 and the development of more sensitive
tests will require the use of more highly purified and expensive vaccines. Policy on emergency ‘vaccination-to-live’ is now included in
the EU directive 2003/85/EC (Article 61)40,
which also requires member states to “make
all arrangements necessary for emergency
vaccination” on confirmation of the first
identified case of disease (Article 14.3). Article
64.1 of this directive prohibits the movement
of vaccinated animals between member states
and, in all likelihood, the movement of such
animals would also be restricted nationally,
anticipating the difficulty that authorities
would have in identifying virus carriers using
existing NSP tests. In the UK, a further consequence has been the extension of the legal
authority to cull, which now includes “any animal the Secretary of State thinks should be
slaughtered with a view to preventing the
spread of foot-and-mouth disease”, as outlined
in the Animal Health Act of 2002.
The UK authorities were poorly prepared
for the 2001 outbreak, and the consequent
renewal of investment in attempts to develop
680
| AUGUST 2004 | VOLUME 2
better vaccines and diagnostic methods, and
to improve contingency planning is both
welcome and long overdue. The addition of
epidemiological models to the range of techniques used in the formulation of FMD control policy is a significant advance. However,
we must learn how information from quantitative models should be incorporated into
policy formulation in a balanced way, mindful of its persuasive but often illusory level of
numerical precision. It is essential that models are developed and used in a manner that
allows both their strengths and inevitable
short-comings to be recognized and widely
understood33,41. The implications of limited
logistical and human resources need to be
integrated into modelling of control scenarios and we need to understand how such
limitations influence the choice of policy
options. The aims of a successful control
policy need to be defined more precisely and
the ability to coordinate policy centrally,
without losing capability to tailor control
tactics locally needs to be developed. We
need to understand more about the precise
mechanisms that allow the local spread of
disease. We need to develop rigorous protocols for exploring phenotypic variability
which might characterize different viral
strains, and not simply track, but react to,
the locations of strains worldwide. In addition, although traditional methods have
worked for most occasions when infection
has been introduced into the United
Kingdom, we need to learn why and when
these measures can fail.
The current DEFRA FMD contingency
plan recognizes the need for a flexible set of
control procedures, but the timing of the
decisions regarding control options is crucial, and criteria are required with which the
seriousness of outbreaks can be evaluated
early so that an appropriately measured
response is selected. On the basis of analyses
of published data from the 2001 epidemic,
we cannot conclude that traditional methods of control no longer work, only that, as
implemented in 2001, they did not work
to an acceptable standard. What is now
required is a marriage of the value of the
expert advice so staunchly defended by the
veterinary practice, with the benefits of
modern surveillance, diagnostic and data
management technologies and the analytical
capabilities of theoretical modelling at the
strategic level. This will require drive, focus
and coordinated cross-disciplinary communication, and patience, good listeners, and
open minds. Properly resourced, FMD contingency planning should provide a model
for twenty-first century disease control.
Daniel T. Haydon is at the Division of
Environmental and Evolutionary Biology,
University of Glasgow, Glasgow G12 8QQ, UK.
Rowland R. Kao is at the Department of Zoology,
University of Oxford, South Parks Rd,
Oxford OX1 3PS, UK.
R. Paul Kitching is at the National Centre for
Foreign Animal Diseases. 1015 Arlington Street,
Winnipeg, Manitoba R3E 3M4, Canada.
Correspondence to D.T.H.
e-mail: [email protected]
All authors contributed equally to this work.
doi:10.1038/nrmicro960
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
Correa Melo, E., Saraiva, V. & Astudillo, V. Review of the
status of foot and mouth disease in countries of South
America and approaches to control and eradication.
Rev. Sci. Tech. 21, 429–436 (2002).
Thomson, G. R. et al. Shifting paradigms in international
animal health standards: the need for comprehensive
standards to enable commodity based trade. [online]
<http://www.eldis.org/fulltext/shiftingparadigms.pdf>
(2004).
Woods, A. To vaccinate or vacillate? The British response
to foot and mouth disease, 1920–2001. Hist. Res. (in the
press).
Northumberland. Report of the Committee of Inquiry on
Foot-and-Mouth Disease. (Her Majesty’s Stationery
Office, London, 1969).
The Royal Society. Inquiry into Infectious Diseases in
Livestock. (The Stationery Office, London, 2002) [online],
<http://www.royalsoc.ac.uk/inquiry/index.html> (2002).
Anderson, I. Foot and Mouth Disease 2001: Lessons to
be Learned Inquiry [online], <http://www.defra.gov.uk/
corporate/inquiries/lessons/index.htm> (2001).
DEFRA. Foot and Mouth Disease — contingency plan.
[online] <http://www.defra.gov.uk/footandmouth/
contingency/index.htm> (2004).
Tinline, R. Lee wave hypothesis for the initial pattern of
spread during the 1967–68 foot and mouth epizootic.
Nature 227, 860–862 (1970).
Haydon, D. T., Woolhouse, M. E. J. & Kitching, R. P.
An analysis of foot-and-mouth-disease epidemics in the
UK. IMA J. Math. Appl. Med. Biol. 14, 1–9 (1997).
Gibbens, J. C. et al. Descriptive epidemiology of the
2001 foot-and-mouth disease epidemic in Great Britain:
the first five months. Vet. Rec. 149, 729–743 (2001).
Haydon, D. T. et al. The construction and analysis of
epidemic trees with reference to the 2001 UK foot-andmouth outbreak. Proc. R Soc. Lond. B Biol. Sci. 270,
121–127 (2003).
Mansley, L. M., Dunlop, P. J., Whiteside, S. M. & Smith,
R. G. Early dissemination of foot-and-mouth disease
virus through sheep marketing in February 2001.
Vet. Rec. 153, 43–50 (2003).
Keeling, M. J. et al. Dynamics of the 2001 UK foot and
mouth epidemic: stochastic dispersal in a heterogeneous
landscape. Science 294, 813–817 (2001).
Gibbens, J. C. & Wilesmith, J. W. Temporal and
geographical distribution of cases of foot-and-mouth
disease during the early weeks of the 2001 epidemic in
Great Britain. Vet. Rec. 151, 407–412 (2002).
Donaldson, A. I., Alexandersen, S., Sorensen, J. H. &
Mikkelsen, T. Relative risks of the uncontrollable (airborne)
spread of FMD by different species. Vet. Rec. 148,
602–604 (2001).
Fraser, C., Riley, S., Anderson, R. M. & Ferguson, N. M.
Factors that make an infectious disease outbreak
controllable. Proc. Natl Acad. Sci. USA 101, 6146–6151
(2004).
Alexandersen, S. & Donaldson, A. I. Further studies to
quantify the dose of natural aerosols of foot-and-mouth
disease virus for pigs. Epidemiol. Infect. 128, 313–323
(2002).
Matthews, L. et al. Neighbourhood control policies and
the spread of infectious diseases. Proc. R. Soc. Lond. B
Biol. Sci. 270, 1659–1666 (2003).
Morris, R. S., Wilesmith, J. W., Stern, M. W., Sanson, R. L. &
Stevenson, M. A. Predictive spatial modelling of alternative
control strategies for the foot-and-mouth disease epidemic
in Great Britain, 2001. Vet. Rec. 149, 137–144 (2001).
Ferguson, N. M., Donnelly, C. A. & Anderson, R. M.
Transmission intensity and impact of control policies on
the foot and mouth epidemic in Great Britain. Nature 413,
542–548 (2001).
www.nature.com/reviews/micro
PERSPECTIVES
21. Ferguson, N. M., Donnelly, C. A. & Anderson, R. M.
The foot-and-mouth epidemic in Great Britain: pattern of
spread and impact of interventions. Science 292,
1155–1160 (2001).
22. Phillips, N., Bridgeman, J. & Ferguson–Smith, M. Report
of the Inquiry into BSE and Variant CJD in the United
Kingdom. (Her Majesty’s Stationery Office, London, 1999).
23. Kao, R. R. The role of mathematical modelling in the
control of the 2001 FMD epidemic in the UK. Trends
Microbiol. 10, 279–286 (2002).
24. Woolhouse, M. E. Foot-and-mouth disease in the UK:
what should we do next time? J. Appl. Microbiol. 94,
S126–S130 (2003).
25. Alexandersen, S., Kitching, R. P., Mansley, L. M. &
Donaldson, A. I. Clinical and laboratory investigations of
five outbreaks of foot-and-mouth disease during the
2001 epidemic in the United Kingdom. Vet. Rec. 152,
489–496 (2003).
26. Kao, R. R. The impact of local heterogeneity on
alternative control strategies for foot-and-mouth disease.
Proc. R. Soc. Lond. B Biol. Sci. 270, 2557–2564 (2003).
27. Gloster, J. et al. Airborne transmission of foot-and-mouth
disease virus from Burnside Farm, Heddon-on-the-Wall,
Northumberland, during the 2001 epidemic in the United
Kingdom. Vet. Rec. 152, 525–533 (2003).
28. Mikkelsen, T. et al. Investigation of airborne foot-andmouth disease virus transmission during low-wind
conditions in the early phase of the UK 2001 epidemic.
Atmos. Chem. Phys. Disc. 3, 677–703 (2003).
29. Wingfield, A. S., Middlemiss, C. H. & Eppink, L. FMD
control strategies. Vet. Rec. 152, 479–480 (2003).
30. Taylor, N. M., Honhold, N., Paterson, A. D. &
Mansley, L. M. Risk of foot-and-mouth disease associated
with proximity in space and time to infected premises and
the implications for control policy during the 2001
epidemic in Cumbria. Vet. Rec. 154, 617–626 (2004).
31. National Audit Office. Report by the Comptroller and
Auditor General: The 2001 Outbreak of Foot and Mouth
Disease. (The Stationery Office, London, 2002) [online],
<http://www.nao.org.uk/publications/nao_reports/01-02/
0102939part1.pdf>, <http://www.nao.org.uk/publications/
nao_reports/01-02/0102939part2.pdf> (2002).
NATURE REVIEWS | MICROBIOLOGY
32. Thompson, D. et al. Economic costs of the foot and
mouth disease outbreak in the United Kingdom in 2001.
Rev. Sci. Tech. 21, 675–687 (2002).
33. Royal Society (Edinburgh). Inquiry Into Foot and Mouth
Disease in Scotland. (Royal Society of Edinburgh,
Edinburgh, 2002).
34. Sutmoller, P. & Olascoaga, R. C. in Evidence for the
Temporary Committee on Foot-and–Mouth Disease of
European Parliament. Meeting 2nd Sept. (European
Parliament 2002).
35. Bouma, A. et al. The foot-and-mouth disease epidemic in
The Netherlands in 2001. Prev. Vet. Med. 57, 155–166
(2003).
36. Barnett, P. V. & Carabin, H. A review of emergency footand-mouth disease (FMD) vaccines. Vaccine 20,
1505–1514 (2002).
37. Donaldson, A. I. & Kitching, R. P. Transmission of footand-mouth disease by vaccinated cattle following natural
challenge. Res. Vet. Sci. 46, 9–14 (1989).
38. Barnett, P. V. et al. Evidence that high potency foot-andmouth disease vaccine inhibits local virus replication and
prevents the ‘carrier’ state in sheep. Vaccine 22,
1221–1232 (2004).
39. Mackay, D. K., Forsyth, M. A., Davies, P. R. & Salt, J. S.
Antibody to the nonstructural proteins of foot-and-mouth
disease virus in vaccinated animals exposed to infection.
Vet. Q. 20, S9–11 (1998).
40. The Council of the European Union. Council Directive
2003/85/EC. Official J. European Union L 306, 1–41
(2003).
41. Eddy, R. Vets asked valuable questions about foot-andmouth measures. Nature 412, 477 (2001).
42. Haydon, D. T., Samuel, A. R. & Knowles, N. J. The
generation and persistence of genetic variation in foot-andmouth disease virus. Prev. Vet. Med. 51, 111–124 (2001).
43. Donaldson, A. I., Gloster, J., Harvey, L. D. & Deans, D. H.
Use of prediction models to forecast and analyse
airborne spread during the foot-and-mouth disease
outbreaks in Brittany, Jersey and the Isle of Wight in
1981. Vet. Rec. 110, 53–57 (1982).
44. Donaldson, A. Foot-and-mouth disease: the principal
features. Ir. Vet. J. 41, 325–327 (1987).
45. Alexandersen, S., Zhang, Z., Donaldson, A. I. &
Garland, A. J. The pathogenesis and diagnosis of
foot-and-mouth disease. J. Comp. Pathol. 129, 1–36
(2003).
46. Woolhouse, M. E. J., Haydon, D. T., Pearson, A. &
Kitching, R. P. Failure of vaccination to prevent outbreaks
of foot-and-mouth disease. Epidemiol. Infect. 116,
363–371 (1996).
47. Hughes, G. J. et al. Serial passage of foot-and-mouth
disease virus in sheep reveals declining levels of
viraemia over time. J. Gen. Virol. 83, 1907–1914
(2002).
48. Bouma, A., Dekker, A. & de Jong, M. C. No foot-andmouth disease virus transmission between individually
housed calves. Vet. Microbiol. 98, 29–36 (2004).
49. Woolhouse, M. E. J. et al. Foot-and-mouth
disease under control in the UK. Nature 411, 258–259
(2001).
50. Kao, R. R. Landscape fragmentation and foot-andmouth disease transmission. Vet. Rec. 148, 746–747
(2001).
Acknowledgments
We thank the anonymous referees for their helpful comments
and many colleagues for advice that has greatly improved the
various drafts of this manuscript. R.R.K. is funded by the
Wellcome Trust.
Competing interests statement
The authors declare that they have no competing financial
interests.
Online links
FURTHER INFORMATION
DEFRA Summary report of the foot-and-mouth disease
modelling workshop: http://www.defra.gov.uk/science/
Publications/FMD_modelling_Summary_Report.pdf
Daniel T. Haydon’s laboratory:
http://www.gla.ac.uk:443/ibls/staff/staff.php?who=PQdGSP
Access to this links box is available online.
VOLUME 2 | AUGUST 2004 | 6 8 1