Description of recent foot and mouth disease outbreaks in nonendemic areas: Exploring the relationship between early detection and epidemic size

Abstract

The objective of this investigation was to describe the detection of foot and mouth disease (FMD) outbreaks in nonendemic areas, and to consider how events early in an epidemic influence the epidemic’s course. We identified 24 epidemics that occurred between 1992 and 2003 in areas officially considered free of FMD. We obtained information about these epidemics from many sources, including the scientific literature, the grey (non peer-reviewed) literature, and individuals involved with the outbreaks. While most of the epidemics consisted of fewer than 150 infected premises, there were 4 extremely large epidemics, each consisting of more than 2000 infected premises. There was no direct relationship between the time to detection and either the total number of infected premises or the number of animals killed for disease control purposes. We believe that the movement of infected animals through markets was the most critical factor that contributed to the unusual magnitude of the very large epidemics.

Introduction

Since 1992, there have been very large epidemics of foot and mouth disease (FMD) in Taiwan, Uruguay, Argentina, and the United Kingdom (UK), all countries previously free of the disease. Substantial economic losses were associated with these epidemics due to the cost of the control measures, the loss of revenue in the trade of animals and animal products, and in the UK, the considerable negative impact on tourism. Additionally, particularly in the UK, there was a public outcry in response to the destruction and disposal of millions of animals (1). These events, combined with concerns about bioterrorism, have renewed interest in finding ways to minimize the extent and impact of future epidemics (2).

The virus that causes FMD is highly contagious and spreads through direct and indirect contact. The magnitude of FMD epidemics is believed to be associated with many factors, including time to detection of the incursion (3–5), the scale of the original infection, livestock and herd density, possibility of airborne spread, the effectiveness of control measures (5,6), normal patterns of animal movement in the affected area (7–9), the species infected, and the strain of virus involved (10).

The identification of an incursion of FMD virus into a non-endemic area typically relies on a producer, a meat inspector, or a veterinarian reporting suspect clinical cases. This method of detection is commonly referred to as passive surveillance. One of the benefits of passive surveillance is that it covers, at low cost, the entire susceptible animal population under owner or veterinary observation (11). However, passive surveillance may not be effective at detecting pathogens that have been absent from a country or region for several years, as is the case with FMD in many countries. This is because those people working with animals on a regular basis (producers, meat inspectors, and veterinarians) may fail to recognize the clinical signs of FMD because of lack of awareness and experience with the disease (2,12).

Despite the perceived importance of early detection of FMD epidemics, little has been published in the scientific literature about how or when outbreaks are normally detected, or about the relationship between time-to-detection and the size of the epidemic. The objective of this paper was to describe the events leading to the detection of outbreaks of FMD in nonendemic areas in the past 10 y. We considered how these events may have influenced the course of the epidemics, and we looked for patterns to guide the development of recommendations that might reduce the impact of future epidemics.

Materials and methods

We identified all epidemics of foot and mouth disease that occurred between 1992 and 2003 in countries or zones considered free of foot and mouth disease. A review article (13) and the World Organisation for Animal Health’s (OIE’s) annual World Animal Health publication were used to identify epidemics subject for inclusion in the study. In a given year, if more than 1 epidemic occurred in a particular area or country and the epidemics were clearly unrelated (separated in time and due to different incursions of virus), then only the 1st epidemic was included in the study.

Since 1996, the OIE has published an official list of countries and areas considered FMD-free, and only epidemics that occurred in a country or zone on this list were eligible for inclusion in our study. In order to determine whether to include particular epidemics that occurred prior to 1996 in our study, we considered the epidemiology of FMD in the region, the year of the most recent previous outbreak in that country or region, and whether the OIE included the country on its list of free countries after 1996.

Information about the selected epidemics was obtained primarily from OIE publications, including Disease Information, the Bulletin, World Animal Health, Scientific and Technical Review, and the Handistatus II prototype. Disease Information, released weekly, consists of notifications sent to the OIE by Member Countries about the principal epidemiological events that have occurred. The Bulletin is published quarterly, and contains news and articles about various items of interest. World Animal Health is a compilation of reports submitted annually by countries on their animal health status with a description of significant epidemiological events. The Scientific and Technical Review is a peer-reviewed journal devoted to current scientific and technical developments in animal health and veterinary public health, worldwide. Finally, the Handistatus II prototype is a Web application (14) containing information about epidemics of animal diseases from 1996 to the present. Data available through the Handistatus II prototype include the number of outbreaks per month, and the numbers of animals susceptible to infection, infected, vaccinated, and destroyed in association with these outbreaks, by species.

We also consulted the scientific literature, ProMed-mail, government publications and Web sites, and on-line newspapers. For the 1993 epidemic in Italy, we accessed some original files during a visit to the Istituto Zooprofilattico Sperimentale in Brescia, Italy. To address specific issues that arose, such as conflicting information, individuals involved in government or research in the respective countries were contacted, initially by e-mail. If they responded, communication proceeded through e-mail or by telephone. If they did not respond, a follow-up e-mail was sent. If this also failed to yield a response, similar attempts were made to contact alternate individuals.

For each epidemic, we recorded the dates of the following: 1) viral incursion, 2) 1st report of suspicion of disease to the veterinary authorities, 3) official confirmation of disease presence, 4) 1st implementation of control measures (usually slaughter of infected animals and animal movement restrictions), 5) official confirmation of last infected premises, and 6) freedom of disease recognized by the OIE. If more than 1 estimated date of incursion of the virus had been published, we used the estimation that appeared to have been made in light of epidemiological understanding of the epidemic. This was usually the most recent estimation.

The “time to detection” was calculated as the difference between the estimated date of incursion of the virus and the date of initial report to the authorities. The “time to implementation of restrictions” was calculated as the difference between the estimated date of incursion of the virus and the date that any movement restrictions were first implemented.

We also recorded the total number of infected premises and the number of animals destroyed in the disease control process, including vaccinated animals that were subsequently destroyed. Details surrounding the detection of FMD were described, including who suspected FMD, the species in which FMD was first detected, the type of premises on which disease was initially detected, and if there was evidence that FMD-infected animals had passed through a market prior to detection or the implementation of movement restrictions. Measures taken to control the epidemic were also noted.

Results

Twenty-four epidemics were identified that fit the criteria described previously. These were located in Europe, S. America, Asia, and Africa. All but 4 of these epidemics consisted of fewer than 150 infected premises, and fewer than 5 infected premises were identified in 11 of the 24 epidemics (Figure 1). Four very large epidemics, each of which involved more than 2000 infected premises, occurred in Taiwan (1997), Argentina (2000–2002), Uruguay (2001), and the UK (2001). General information about each epidemic is presented in Table 1, and details about the initial detection of each epidemic are presented in Table 2.

Figure 1.

Number of infected premises (outbreaks) recorded in epidemics of foot and mouth disease (FMD) between 1992–2003 in countries and zones considered FMD-free by the World Organisation for Animal Health (OIE).

Table 1.

Details of foot and mouth disease (FMD) epidemics that occurred in countries and zones considered FMD-free by the World Organisation for Animal Health (OIE) between 1992–2003.

	Time intervals
Epidemic	Incursion of virus to initial report to authorities (d)	Incursion of virus to first restrictions (d)	Report of first infected premises to last (d)	Initial report to authorities until re-establishment of freedom of disease by OIE (mo)	Since most recent outbreak (y)	Viral strain	Number of infected premises	Number of animals destroyedfor disease control purposes	Vaccination
Italy 1993	2543,44	3143,45	10543	1043, A	443	O44	5744	11 81944	N44
Bulgaria 1993	?	?	N/A	?	245	O44	144	204344	Y44
Greece 1994	~10045,46	~10145,46	5946,47	1446,48,B	1049	O49	9549	18 57345	N49
Greece 1996	1050	?	8947,50	1445,47	2	O51	3914	20 18951	N51
FRY 1996C	?	?	2647,51	N/A	2251	A51	10151	349651	N51
Bulgaria 1996	?	?	N/A	1047,52	3	O51	151, D	149651	N51
Taiwan 1997	2153,54	2753,54	12354	7555, E	6854	O54	614754	4 030 00056	Y54
Japan 2000	2057, F	2457	5157	657,58	9258	O58	458	74057	N58
S. Korea 2000	2159,60	2259,60	2159,60	1860,61	6658	O58	1558	862 91662	Y58
Greece 2000	863,64	863,64	6664,65	764,66	4	Asia 158	1458	800065	N58
Argentina 2000	1167	2067,68	53117	35G	669	A, O70	251917	18115	Y70
Brazil 2000H	?	?	?	2871, I	772	O58	2258	11 06773	N73
S. Africa 2000	574	974,75	5470,74	2174,76, J	4358	O58	475	677375	Y58
Uruguay 2000	877	977	N/A	370,77	1014	O58	158	20 00069	N58
Swaziland 2000	178	178	N/A	N/A	3158	SAT158	158	11078	Y58
UK 2001	2479,80	2579,81	22370,80	1180,82	2083	O70	203070	4 000 00083	N70
France 2001	2484	1084, K	1184,85	661,84	2086	O70	284, L	63 00084	N70
Netherlands 2001	1987	−387,88, K	3887,70	661,87	1789	O70	2688	260 00088	Y70
Ireland 2001	2390–92	390,91,92, M	N/A	661,92	6070	O70	170	60 00047	N70
Uruguay 2001	793,94	−693,94, N	12070,94	2555,94, E	1	A2470	205770	710895	Y95
Botswana 2002	776,96	896,97	1676	1055,76, J	2176	SAT276	276	12 19776	Y96
S. Korea 2002	1098,99	1098,99	5299,100	776,99	2	O76	1676	160 08014	N76
Paraguay 2002	1676,101, O	?	N/A	28102	814	O76	176	719103	NP
Botswana 2003	14104	15104	18104,105	1155,104, Q	1	SAT155	455	3922106	Y107

^ADeclared free by European Community44

^BGreece declared its own freedom

^CFederal Republic of Yugoslavia (FRY). Disease confirmed at National Laboratory of Veterinary Faculty in Belgrade. Virus never isolated by World Reference Laboratory, and there was no evidence of seroconversion51

^DInfected village

^EDeclared free country where vaccination is practiced

^FFirst case detected believed to be true index farm34. Date of incursion of virus was estimated based on an incubation period of seven days108

^GFMD-free zone where vaccination is not practised established 28 May 200271. FMD-free zone with vaccination established 10 July 2003 (lost again in September 2003)109

^HOutbreak occurred in state of Rio Grande do Sol (OIE designated FMD-free region where vaccination is practised)

^IEstablished free zone with vaccination

^JEstablished FMD-free zone where vaccination is not practiced

^KMovement restrictions imposed in response to UK outbreak

^LSeropositive animals on 6 other premises110

^MDate of implementation of movement restrictions. There was a ban of markets and other gatherings of susceptible species from 23 February 200163

^NBan on livestock movement since 10 April 2001 because of outbreak in Argentina94

^OHowever, it took 59 d until presence of FMD was confirmed111

^PDid not vaccinate to control outbreak, already vaccinating

^QTwo of the infected premises were crush pens

Table 2.

Details surrounding the detection of epidemics of foot and mouth disease (FMD) that occurred between 1992–2003 in countries and zones considered FMD-free by the World Organisation for Animal Health (OIE).

Epidemic	Premises where FMD initially detected	Species first suspected of FMD	Suspected method of viral incursion	Details surrounding initial detection
Italy 1993	Farm with 21 cattle and 11 swine44	Beef cattle43	Cattle entered Italy with forged import certificates45	Private veterinarian called by owner to examine sick animals, diagnosed pneumonia. He returned 5 d later as animals had not responded to treatment, and found the cattle were lame. He suspected FMD and called the official veterinarians43.
Bulgaria 1993	Cooperative farm44	Dairy cattle45	Contaminated trucks from Turkey suspected45	FMD suspected in 210 dairy cattle45.
Greece 1994	Unknown	Cattle47	Illegal importation of infected live sheep from Turkey to Lesbos as early as April, possibly to increase flock numbers in response to EU subsidies45	FMD became evident in Xanthi (NE Greece) when infected sheep imported from Lesbos spread infection to cattle47. Vesicles on sheep and cattle had been observed by owners on Lesbos and samples were submitted to the laboratory, but tested for blue-tongue virus and not FMD virus112. 32 infected holdings found at time of initial report45.
Greece 1996	Unknown	Sheep50	Spread from Turkish Thrace. Up to 7 different introductions occurred — 1 probably from illegal importation of live sheep and 5 due to illegal immigrants45	Unknown.
FRY 1996	Holding 500 m from Macedonian border113	Cattle113,114	Spread from Albania and/or former Yugoslav Republic of Macedonia115	FMD suspected in 74 heifers113,115.
Bulgaria 1996	Private farms in a village51	Cattle51	Unknown	FMD first suspected in cows and calves at pasture116. 11/47 cattle affected52.
Taiwan 1997	Farrow-to-finish farm117 with approximately 2000 hogs53	Pigs54	Pigs or other animal products illegally imported by fishing boats54,118	Owner noticed vesicular lesions on muzzle and feet of sow in farrowing crate in middle of farrowing barn117. 28 farms affected by time of first report119. Possible that earlier cases had been mis-diagnosed as SVD, as diagnosis of SVD routinely made clinically without laboratory confirmation. Serological evidence of SVDA virus detected at WRLB53.
Japan 2000	Farm with 10 Japanese black cattle57	Beef cattle57	Chinese wheat straw used on index farm57	Owner noticed cattle with pyrexia, anorexia, and coughing on 8 Mar/00. A private veterinarian visited the farm on 12 Mar/00 and noted anorexia and nasal and mouth erosions that were spreading to other cattle. Veterinarian reported to officials on 21 Mar/00, and they visited farm on the same day. Samples submitted on 22 Mar/00. All 10 cattle were affected. Veterinarian had visited 105 farms around 21 Mar/0057.
S. Korea 2000	Dairy farm with 12 cows and 3 calves60	Dairy cattle60	Unknown: Imported hay, international travellers, and wind-borne introduction by Asian (yellow) dust suggested120	Owner observed clinical signs in 1 lactating cow on 20 Mar/00. 10 animals were sick on the 22 Mar/00, and all 12 on 24 Mar/00, when he notified the authorities60. This or another farm probable true index case120.
Greece 2000	Free-grazing beef herd64	Beef cattle64	Animals from Turkey crossing the Evros river65	Mild clinical signs observed in 4–5/50 cattle121 during routine veterinary inspection prior to issuance of movement permit. Further investigation detected 1 herd in vicinity likely infected earlier64. By the time of detection, FMD had become widespread inside the area and escaped to a contiguous town65.
Argentina 2000	Communally owned establishment (22 owners)68	Cattle67	Illegal importation of cattle67	Routine surveillance measures detected FMD virus antibodies in 10 illegally imported cattle in August67.
Brazil 2000	Smallholding122	Dairy cattle123	Unknown	Detected by passive surveillance activities122 in traditional herd producing milk for local consumption123. Conflicting information regarding dates of incursion and first notification presented in OIE literature15,122,123.
S. Africa 2000	Pig farm in KwaZulu-Natal province58	Pigs74	Galley waste from a passing ship, obtained by farmer at Durban harbor and fed illegally as untreated swill124	Private veterinarian investigated deaths in pigs due to suspected pesticide toxicity from containers in which the swill was stored. Samples sent to private pathologist with negative results. Samples then submitted to virology laboratory at Onderstepoort Veterinary Institute, which suspected FMD and alerted the Exotic Diseases Division124C. 247/726 pigs affected74.
Uruguay 2000	Field shared by 3 owners77	Beef cattle77	Sow that had consumed feed of animal origin122	Infection spread from pigs to cattle in which FMD was detected38.
Swaziland 2000	Abattoir78	Beef78	Imported from cattle from S. African feedlot58	Mouth lesions noticed by inspector in abattoir after 44/110 in group were slaughtered. Lesions found in 2/66 remaining 78.
UK 2001	Abattoir70	Pigs70	Contaminated waste food fed to pigs81	Lameness noted in 27 sows in lairage of abattoir during ante-mortem inspection on Monday 19 Feb/01. 28/109 remaining pigs found to have vesicular lesions. Pigs had spent weekend in lairage125. At least 57 premises infected by 20 Feb/0183.
France 2001	Farm with 114 cattle84	Dairy cow84	Infected sheep imported from UK by neighbor of index case84	Farmer noticed lesion on udder of dairy cow at milking. By afternoon, 2 animals noted with lesions and farmer called his private veterinarian, who examined animals and informed the authorities84.
Netherlands 2001	Farm with 74 veal calves and 545 milking goats89	Milking goats	Irish veal calves imported on 24 Feb/01. They were infected at a resting point in Mayenne, France, by FMD-infected sheep from the UK. Calves went to farm in Oene where first suspicion was recorded88	Veterinarian reported clinical signs of FMD in goats on 15 Mar/01. ELISAD negative, but more animals with same signs on 16 Mar/01. Animals culled on 17 Mar/01. This case was confirmed on 22 Mar/01 after virus isolation from goats. No clinical signs noted in the calves, but 4/74 were seropositive. In the meantime, characteristic signs of FMD reported in adult cows at dairy farm on 20 Mar/01, and disease was laboratory confirmed on 21 Mar/0189. 9 herds believed infected by time of initial confirmation88.
Ireland 2001	Holding with 113 cattle and 447 sheep, 8 km S. of infected premises in N. Ireland92	Sheep90	‘Strong circumstantial evidence’ of spread by indirect contact with index case in N. Ireland via neighboring farm70	On 19 Mar/01, stock manager noted a ‘ewe with legs spread’ that he caught and then observed lesions in her mouth. As a veterinary inspector was coming the following day, he waited until then to report the suspicion90. Cattle on farm were unaffected70.
Uruguay 2001	Farm with cattle and sheep, near Argentinian border94	Cattle94	Spread from Argentina70	39/430 cattle showed clinical signs, no clinical signs observed any of the 640 sheep94.
Botswana 2002	Crush area with 700 cattle96	Cattle96	Spread from Zimbabwe, probably by smuggling due to price differentials126	Surveillance teams noticed clinical signs in cattle in a crush and immediately reported to the authorities76.
S. Korea 2002	Farm with 8302 pigs98	Pigs98	Virus similar to isolates from China and Mongolia in 2001. Most likely routes considered foreign workers on index farm or neighbouring farmers that had travelled overseas98	Pigs had decreased appetite and blisters on tongues and hooves since 20 Apr/02, local feedmaker official reported suspect disease to the National Veterinary Research and Quarantine Service on 2 May/02127. Discovered later that pig farmer may have buried > 200 dead pigs between 27 Apr and 2 May to hide disease128.
Paraguay 2002	Farm on border with Brazil101 with 2200 cattle103	Breeding/fattening cattle101	Unknown	Initial occurrence of vesicular disease investigated 23 Sept/02, reported to OIE as infectious bovine rhinotracheitis101. 4/196 cattle (males vaccinated against FMD virus) recorded as cases101. Argentina and Brazil closed borders early Oct/02. International pressure lead PAHOE to investigate, and virus was isolated on 4 Nov/02111.
Botswana 2003	Crush pen with 700 cattle104	Cattle104	Unknown	Animals seen limping and oral and foot lesions noted in 9/700 animals in crush pen104. Originally noted by farmer who immediately reported it to the veterinary assistant in charge of the area55,129.

^ASwine Vesicular Disease

^BWorld Reference Laboratory

^CVosloo W. Personal communication. 23/1/2003

^DEnzyme-Linked ImmunoSorbent Assay

^EPan American Health Organization

Stamping out and movement restrictions, with or without vaccination, were the control measures applied for all epidemics, except for those that occurred in Argentina in 2000 and Uruguay in 2001. Stamping out is the “slaughter of all sick and contaminated animals, with destruction of their carcases (by burying, incineration etc.), followed by cleansing and disinfection of the premises” (15). Stamping out was initially applied in Uruguay in 2001, but it was abandoned in favor of vaccination (without the subsequent destruction of vaccinates) when the extent of the epidemic became clear (16). Vaccination and movement restrictions were used to control the outbreak in Argentina (17).

The number of infected premises did not necessarily reflect the number of animals killed for disease control purposes (Figure 2). For example, because of the decision to “vaccinate to live,” relatively few animals were destroyed in Uruguay in 2001, despite 2057 recorded outbreaks. Conversely, some epidemics that consisted of relatively few infected premises resulted in many animals being slaughtered due to the type of control measures implemented, such as emergency vaccination with vaccinates targeted for subsequent destruction, as in S. Korea in 2000 and the Netherlands in 2001.

Figure 2.

Relationship between number of recorded infected premises (outbreaks) and number of animals destroyed for disease control purposes in epidemics of foot and mouth disease (FMD) between 1992–2003 in countries and zones considered FMD-free by the World Organisation for Animal Health (OIE). Outliers are labelled.

We estimated the date of incursion of the virus for 21 of the 24 epidemics based on the following: a published account of the descriptive epidemiology (8 epidemics), the known date of importation of infected animals (4 epidemics), the age of lesions of the index case (1 epidemic), or the Emergency Report to the OIE (8 epidemics). The time to detection ranged from 1 to 100 d. We did not find a direct relationship between the time to detection and either the total number of infected premises (Figure 3) or the numbers of animals killed for disease control purposes.

Figure 3.

Association between the time to detection and the number of infected premises (outbreaks) for epidemics of foot and mouth disease (FMD) between 1992–2003 in countries and zones considered FMD-free by the World Organization for Animal Health (OIE). Outliers are labelled.

The type of premises where disease was initially detected was recorded for all but 2 epidemics (Greece 1994 and 1996). In 2 epidemics, FMD was first detected in abattoirs (UK and Swaziland). Foot and mouth disease was first discovered in crush pens in the 2 epidemics in Botswana. In the remaining 18 epidemics, FMD was initially found on private or communal farms. Foot and mouth disease was most commonly detected initially in cattle (17 out of 24 epidemics). In at least 2 epidemics, pigs (Uruguay 2001) or sheep (Greece 1994) were infected first, but clinical signs went unnoticed until cattle became infected.

Detailed information about the circumstances leading to the detection of the 1st outbreak was available for 15 of the 24 epidemics. Two of these 15 (13%) were discovered during antemortem inspection at abattoirs, 4 (27%) were recognized by routine surveillance activities, and 1 (7%) was reported by a member of the public. The remaining 8 epidemics (53%) were discovered as a result of a farmer alerting a private veterinarian or the authorities to a problem with his/her animals. Reasons for delayed detection included initial misdiagnosis of disease, deliberate concealment of sick livestock by producers, mild clinical signs in small ruminants, and failure of the laboratory to isolate the virus.

Discussion

This retrospective study yielded 2 results of particular interest. The 1st was the presence of a striking dichotomy in the size of the FMD epidemics studied (Figure 1), and the 2nd was the lack of a clear relationship between the time to detection of the incursion of FMD virus and the magnitude of the epidemic (Figure 3). Both of these findings are explored in detail below.

In the past 10 y, while most epidemics consisted of fewer than 150 infected premises, there were 4 extremely large epidemics, each consisting of more than 2000 infected premises. No epidemic resulted in between 150 and 2000 infected premises. The same dichotomy is apparent if the severity of the outcome is measured by the number of animals destroyed instead of the number of infected premises (Figure 2), but there are only 2 exceptionally large epidemics (UK 2001 and Taiwan 1997).

It is possible to consider 3 distinct groupings of epidemics rather than 2: fewer than 5 infected premises (11 epidemics), between 15 and 150 infected premises (9 epidemics), and more than 2000 infected premises (4 epidemics). However, the division into 3 groups rather than 2 is less convincing because: 1) the same distinction cannot be applied if the number of animals destroyed is used as a measure of size in lieu of the number of infected premises, and 2) the definition of ‘outbreak’ varied from country to country. For example, in Bulgaria entire villages were considered single outbreaks. In France, premises on which seropositive sheep were found were not called outbreaks, whereas in the UK they were. Such inconsistencies would not make a difference in distinguishing between 150 and 2000 infected premises, but they could affect the distinction between 5 and 14 infected premises.

The reasons that the 4 epidemics were so much larger than the others warrant consideration. Livestock density, late detection, animal movement, and effectiveness of control measures have been identified as contributors to the magnitude of the epidemics in both the UK and Taiwan (3,18,19). Additionally, inadequacy of vaccine supply has been cited as an important factor in Taiwan (19). Movements of animals, agricultural machinery, and people have been implicated in the dissemination of the virus during the epidemics in Uruguay (20) and Argentina (21). Unfortunately, comparatively little has been published describing the epidemics in Uruguay and Argentina.

According to the published dates of incursion of the virus, the presence of the disease was not detected considerably later in the very large epidemics than the smaller ones (Figure 3). However, the date of incursion of the virus is only an estimate, and its true value is particularly uncertain in the epidemic in Taiwan, with estimates ranging from October 1996 (22), to mid-February 1997 (used in our analysis) (22), to March 10, 1997 (23). The different estimates result from speculation that the initial cases of FMD were misdiagnosed as swine vesicular disease, endemic in that country (2,4,22).

The epidemics in both the UK and Taiwan occurred in livestock-dense areas, which facilitated local spread of disease both before and after the detection of the viral incursion. In both of these epidemics, a large number of outbreaks were detected daily in the early stages of the epidemic, overwhelming the resources available to implement control measures (19,24). This led to delays in the slaughter and disposal of infected herds, thus prolonging the period during which infected herds shed virus, which further fuelled the spread of infection (25). However, the inadequate response observed in both the UK and Taiwan was a result of the unexpected scale of the early epidemic, not its cause.

We suspect that the movement of infected animals through markets was the most critical factor that contributed to the unusual magnitude of these epidemics. In the UK, sheep infected with the FMD virus passed through the largest sheep market in the country (Longtown Market) during a seasonal peak of sheep sales and movements. The first infected sheep were at the market prior to the initial detection of the virus, which resulted in the infection of approximately 35 premises before the first case of disease was confirmed (26). Different infected sheep passed through the market on another day prior to the nationwide imposition of animal movement restrictions.

In Taiwan, the inability of the government to shut down livestock markets has been implicated in the size of the epidemic (19). The majority of outbreaks occurred on farms located in the proximity of livestock auction markets or abattoirs, suggesting that movement of subclinically infected pigs was the major route of viral dissemination (27). Further, the disease occurred during the Chinese New Year, a period traditionally known for an increase in animal movements (4).

Cattle (presumably infected with FMD virus) moved through an auction a few days prior to the initial detection of the 2001 epidemic in Uruguay as well, disseminating the virus to other departments within the country (20). However, as little has been published describing this epidemic, it is difficult to assess the significance of this observation. We are unaware of markets playing a role in the epidemic in Argentina, but again, little has been published describing the early spread of disease in that epidemic.

While the contact structure of populations has long been recognized as important in the transmission of infectious disease, until recently it has been considered too complex to study quantitatively within large populations (28). This has changed with advances made in modeling networks in order to describe the elements or nodes of a system (which might be individuals, Web sites, or farms, for example) and the connections or links between them. Within several complex networks, each composed of a very large number of nodes, a few nodes have been found that possess a huge number of links (29). These particular nodes have been termed “hubs.” The World Wide Web is an example of a complex network, and Google^TM of a hub within it. If a virus, computer or biological, is introduced to a network containing hubs, at least one hub will tend to become infected, because hubs are connected to many other nodes. Once a hub has been infected, it will pass the virus to numerous other sites, eventually compromising other hubs, which will then spread the virus throughout the entire system (30).

Bigras-Poulin et al (9) used graph theory to describe the network of cattle and sheep movements in Denmark. Farms, markets, and abattoirs were considered nodes of the network, linked by animal movements. Most premises were associated with few movements, but a small proportion (5–10% of premises) were associated with many movements. These premises, most often animal markets and abattoirs, fit the description of a hub. The markets that received and sold animals infected with FMD virus in the UK and Taiwan were most likely hubs in the animal movement networks of the respective areas. This movement through a hub or hubs may be the key factor that distinguished the very large epidemics from the others. However, as data indicating whether or not infected animals passed through a hub are lacking from the other epidemics, we cannot test this hypothesis with the information available.

The 2nd striking finding from this study is the lack of an evident relationship between the time to detection of disease and the size of the resultant epidemic (Figure 3). There are 2, nonexclusive, explanations for this.

First, the dates of viral incursion used in the analysis are only estimates of variable, and often questionable, reliability. Because all but 1 of the estimates were less than 25 d, if the time to the incursion of the virus was underestimated by as little as a week, the analysis could be impacted. Such a difference between the estimated and actual date of incursion is very plausible. For example, in Taiwan, published estimates range from a few days to 6 mo (22,23). Therefore, we might not have detected a true relationship between time to detection and the size of the epidemic because of the poor quality of the data.

Second, while late detection may be a factor in the scale of an epidemic, it does not act in isolation. Rather than guaranteeing a large epidemic, late detection of FMD increases the probability of the occurrence of a large epidemic by extending the time period in which another event to augment the epidemic may occur. Such an event might be animals passing through a hub, the occurrence of climatic conditions suitable for windborne dispersal of the virus, or extensive local spread in an area of extremely high animal density.

Based on this study, we can present a number of recommendations for consideration when developing contingency plans for an outbreak of FMD or other contagious foreign animal diseases. First, premises that act as hubs with respect to normal animal movement should be identified prior to an outbreak, and this information should be kept updated as animal movement patterns change. If normal animal movement patterns are particularly amenable to the spread of infectious diseases, policies can be developed to change them, as occurred with the implementation of a 6-day movement standstill in the UK following the 2001 epidemic (31).

Second, as recommended by others (24,32,33), all animal movement should be immediately halted upon discovery of the viral incursion. Resources should be allocated to ascertain if infected animals have passed through a hub or hubs. If a hub is involved, a very large epidemic may be expected, and this should be reflected in the scale and range of deployment of disease control teams. Further research concerning the logistics of how to respond most effectively in these circumstances is needed.

Third, reasons for delayed diagnosis (misdiagnosis, either clinically or by the laboratory; deliberate concealment of sick animals; and subclinical infection of small ruminants) should be addressed. This might be done through programs to raise and maintain awareness about FMD among farmers and veterinarians, active surveillance programs for high-risk premises/ situations, high laboratory standards, and the provision of adequate compensation for destroyed infected stock.

Despite the valuable information gained from this study, the analysis has limitations and constraints. First, direct comparisons between epidemics in different countries may not always be appropriate. As described previously with respect to the FMD epidemics in France and Bulgaria, “infected premises” or “outbreak” might have a different definition and connotation in different locations. Second, we examined the number of infected premises and the number of animals destroyed in the disease control campaign to compare the severity of different epidemics. We did not consider the economic consequences of the epidemic, also an important indicator of severity.

The most serious constraint to this study was that data from the different epidemics were of variable quality and completeness. With notable exceptions (3,13,33–37), we found few papers in the scientific literature that outlined the descriptive epidemiology of the epidemics in our study. Thus, we relied heavily on Emergency Reports to the OIE published in Disease Information for data about the detection of outbreaks and dates of viral incursion. Some of this information, particularly the date of incursion of the virus, was found to be inaccurate in light of further investigation subsequent to publication. However, we found that this was rarely reported in published materials easily accessible to the public or research community. Further, there are limitations to the usefulness of the information provided to the OIE, which relies heavily on self-reporting by the relevant governments (38). This can lead to a conflict of interest, as the repercussions of reporting diseases such as FMD may involve heavy economic losses due to consequential trade restrictions.

These issues are illustrated by information published on the epidemic in Argentina in 2000–2001. Within the OIE information sources (World Animal Health, Scientific and Technical Review, Disease Information and Handistatus II), 3 different versions of the events of the year 2000 are described. According to Disease Information and World Animal Health (2000), FMD virus (type A) was isolated from illegally imported cattle in August 2000 in the course of routine surveillance measures (39). There was no evidence of clinical signs or spread of disease. An OIE mission visited Argentina and, on October 6, 2000, reported that their disease-free status should not be revoked (40).

However, in World Animal Health (2001), it is stated that there were animals infected with FMD virus on 124 premises in 2000 and that both virus types A and O were isolated. A 2nd epidemic was reported to have begun in March 2001 that resulted in 2394 infected premises by the end of the year. Finally, Handistatus II (accessed on June 23, 2004) indicated that there were outbreaks of FMD every month from July 2000 to January 2003. Similar discrepancies concerning this epidemic are to be found in other FAO and OIE reports summarizing the FMD situation in S. America and worldwide (41,42). Papers published more recently clarify the situation, indicating that the Argentine Animal Health Service (SENASA) reviewed the situation and confirmed that 1 large epidemic involving 2519 infected premises occurred between July 2000 and January 2002, involving virus types A and O (17,21). Until this review, the 268 outbreaks that occurred in January and February 2001 were not officially recognized (17).

Important lessons may be learned by studying historical epidemics, as many of the same mistakes and issues recur, irrespective of the country involved. By identifying common patterns of behavior and types of events that occur repeatedly, it should become possible to manage future epidemics of foreign animal diseases more efficiently. Factors that influence the final size of epidemics may be better defined and understood. In turn, this should guide policy development and the allocation of resources to issues most likely to minimize the impact of future epidemics. An objective source of outbreak information would both facilitate and validate future studies of this nature. Such a source could be compiled and published retrospectively, thus avoiding some of the concerns present during and just after the epidemic, such as trade issues. CVJ