ELSEVIlER Preventive Veterinary Medicine 30 (I 997) 49-59

PREVENTIVE VETERINARY

MEDICINE

A risk-assessment model for foot and mouth disease ( FMD) virus introduction through deboned beef

importation

Peter Yu a, * , Tsegaye Habtemariam a, Saul Wilson b, David Oryang a, David Nganwa a, Mike Obasa a,

Vinaida Robnett a .a Center ji)r Compututionul Epidemiology, School c)j’Veterinury Medicine, Tuskearr University, Tuskegee. AL

36088. USA

b Interrufionul Center jiv Tropicd Animal He&h, School of Veterinary Medicine, Tuskegee Unioersity.

Tuskegee, AL 36088, USA

Accepted 3 June 1996

Abstract

We present a risk-assessment model to assess the risk of introduction of foot and mouth disease (FMD) virus associated with deboned beef importation. The model was developed in accordance with the risk-reduction procedures proposed by the European Community for meat importation. The risk reduction procedures include farm-level inspection, ante-mortem inspection, post-mortem inspection, chilling and deboning. The risk assessment was based on the prevalence of FMD-infected cattle in herds as well as the prevalence of infected herds in the exporting country. Computer simulations were carried out to evaluate the probability of FMD virus introduction by importing 100 tons of deboned beef in relation to FMD prevalence, number of cattle selected from each herd, and sample sizes in ante-mortem and post-mortem inspections. The effects 01’ the risk-reduction procedures on the probability of FMD virus introduction were examined. 0 1997 Elsevier Science B.V.

Keyword.~: Risk assessment; Foot and mouth disease; Epidemiology

* Corresponding author

0167-5877/97/$15.00 Copyright 0 1997 Elsevier Science B.V. All rights reserved.

PI/ SO167-5877(96)OlOSS-9

50 P. Yu et al./ Preventive Vrtrrincrry Medicine 30 (1997) 49-59

1. Introduction

Foot and mouth disease (FMD) is an acute highly-contagious viral infection of all cloven-hoofed animals. It is a major threat to the health of livestock around the world.

Krystynak and Charlebois (1987) studied the potential economic impact of an outbreak of FMD in Canada. Their results indicated that even a small outbreak of FMD would

have serious economic consequences for the livestock sector with farm cash receipts

declining by $2 billion. In the study by Berentsen et al. (1992), a variety of preventive and control strategies with respect to FMD in Netherlands were examined using

cost-benefit analysis. Davies (1993) conducted an analysis of the risks attached to vaccination and non-vaccination strategies against FMD in the European Community.

The United States has been free of FMD since 1929. Enforcement of strict import regulations on animals and animal products has prevented introduction of FMD so far.

However, the United States Department of Agriculture, Animal and Plant Health Inspection Service (APHIS, 1991) pointed out that the introduction of FMD into the U.S. must be considered possible due to the increase in the volume of traffic and

economic pressures to reduce trade barriers between countries. APHIS provided State and Federal Animal Health officials with general guidelines on emergency operations

for control and eradication of FMD. The international animal health communities advocate a more-widespread use of

quantitative risk assessment in making international-trade decisions (Miller et al., 1993). In order to prevent the introduction of foreign animal diseases, regulatory officials are required to make decisions in the face of uncertainties. With rapid changes in world trade, current regulations related to animal and animal-product import are being chal- lenged based on the risk of introduction of foreign animal diseases. Regulatory decisions must have a scientific basis which is consistent and defensible. Growing support for using risk-assessment methods can be seen in the North American Free Trade Agree- ment (NAFTA) and the General Agreement on Tariffs and Trade (GATT). Arbitrary plant and animal health-restrictions without a scientific basis in trade regulations could result in international sanctions under the GATT rules.

The member states of the European Community have common policies for import controls of animals and animal products (Davies, 1993). The European Community accepts beef from FMD countries under conditions which reduce the risk of FMD transmission by such products. These measures require specific inspections and slaugh- tering processes. Recently, officials of Caribbean countries requested an assessment of the risk of introduction of FMD into the Caribbean countries by deboned beef importa- tion based on the European Community directives (The Commission of the European Communities, 1993, Wilson, 1994). APHIS has been exploring methods of quantitative risk assessment to support decision making related to animal and animal-product importation into the USA (McElvaine et al., 1993). Garner and Lack (1995) evaluate four alternate control strategies for FMD in three different regions of Australia. Their results show that slaughter of dangerous contact herds as well as infected herds is most effective under conditions where FMD is likely to spread rapidly.

Quantitative risk-assessment methods are an extension of standard statistical and epidemiological methods which enable one to evaluate the likelihood and consequences

P. Yu et al./ Preventive Veterinury Medicine 30 11997) 49-59 51

of an adverse event occurring. These methods are especially useful when the information

is incomplete or uncertain. Miller et al. (1993) outlined a process with key steps for performing a quantitative risk assessment, and described a method for quantifying the uncertainty associated with results of risk assessment. Morley (1993) presented a model

for the assessment of the animal-disease risks associated with the importation of animals and animal products. His model uses prevalence of infection in the entire animal

population of the exporting country. We developed a model to assess the risk of introduction of FMD virus by deboned

beef importation from countries where FMD exists. The risk assessment is based on the prevalence of infected cattle in herds as well as the prevalence of infected herds in the

exporting country. Computer simulations are carried out to evaluate the probability of FMD virus introduction by importing 100 tons of deboned beef in relation to FMD prevalence, number of cattle selected from each herd, and sample sizes in ante-mortem

and post-mortem inspections. The effect of risk-reduction procedures on the probability of FMD virus introduction is examined. In this paper the term deboned beef refers to

fresh meat of cattle origin where bones are removed to reduce the chance of virus

survival and eventual transmission.

2. Materials and methods

Quantitative risk assessment provides a method for measuring risk and providing decision makers with the information obtained. The quantitative risk assessment process involves: (a> identifying the hazard; (b) developing a scenario tree which outlines the pathway of expected events and all the failures which could occur, culminating in the occurrence of the identified hazard; (c) gathering and documenting the evidence; (d) developing equations or functions; (e) performing calculations to summarize the likeli- hood of the hazard occurring; (f) considering risk management options; (g) preparing a

written report (see McElvaine et al., 1993, Miller et al., 1993 for a complete discussion of the risk assessment process).

A scenario tree for the risk of FMD virus introduction associated with deboned beef importation is shown in Fig. 1. The scenario tree consists of a sequence of specific events, from the point of origin of the product to its destination. For each point or event in the tree, a specific question related to the risk of FMD virus introduction is asked. The accumulation of the answers to those questions determines the final risk related to the deboned beef importation.

Let ~1~ denote the probability that a randomly-selected exporting herd in an exporting country is infected with FMD, pA the probability that a randomly-selected slaughter-aged animal from an infected herd is infected with FMD. Suppose m herds are selected, then nk cattle are selected from herd k, k = 1, 2,. , . , m, to produce deboned beef for export. The process for selecting nk cattle is equivalent to conducting nk independent trials. Each trial has a probability of pA that the animal is infected with FMD. Let Y, be the number of FMD-infected cattle selected from herd k, then Y, has a binomial distribution with parameters ( nk, pA) (Larsen and Marx, 1990), and we have

P(Y,=i)= ‘I;” ~;(1-~*)n~-i,i=0,1,2 ,...) nk ( 1

52 P. Yu et al./ Preventive Veterinary Medicine 30 (1997) 49-59

0 cl

0 1

0 2

0 3

0 4

0 ; El El

lnitlating event of beef exportation

Is a randomly-selected exporting herd infected with FMD?

Is FMD detected in an infected herd in the farm by the Animal Health

Attention System?

Is a randomly-selected animal from an exporting herd infected with FMD?

Is FMD detected In an infected animal during ante-mortem inspection?

Is FMD detected in an infected carcass during post-mortem inspection?

Does FMD virus in an infected carcass survive chilling treatment?

FMD virus contaminated beef introduced to an importing country

As planned scenario FMD virus not introduced to an importing country

Fig. 1. Scenario pathway for the risk of FMD virus introduction associated with deboned beef importation

where i is the number of FMD-infected cattle selected from herd k. The total number of cattle selected for export is given by

m n= En,

k= I (2)

Let R, denote the event that at least one of the II selected cattle is infected with

FMD, then R, can be represented as follows:

M

R , = U (herd k is infected] rl {at least one animal selected from herd k is infected} k= I

(3)

where notation U and n are the union and intersection of sets respectively. Note that FMD virus introduction can happen if one or more FMD-infected carcasses are imported. However, if FMD virus is introduced to an importing country, a large number of FMD-infected carcasses may do more harm than one FMD-infected carcass in terms of disease-spreading. The risk assessment in this study is focused on the risk of virus introduction. One FMD-infected carcass has the same risk as a large number of FMD-infected carcasses in terms of virus introduction. Further study should be con-

P. Yu et al./ Prevenrive Vererimry Medicine 30 (1997) 49-59 53

ducted to integrate the risk of FMD virus introduction with the disease-spread dynamics in an importing country where intensity is important.

Note that P(herd k is infected} = pH, and P{at least one animal selected from herd k is infected} = P(Y, > 0). Using the complement property of probability functions, we

have

m P(R,) = I- np -PHwpo)l

k- 1

where notation ny=, flk) is the product of fll>, f(2), . . . , and flm>, and PfY, > 0) is given by

P(Y,>O)=l-(l-p,)“” (5)

The probability given in Eq. (4) is called the risk of FMD virus introduction into an

importing country without any risk-reduction measures. In this study, we examine the following risk-reduction procedures in the process of

beef importation: (I) herds for exportation are inspected in the farms by the Animal Health Attention System; (II) cattle are inspected in the slaughterhouse pens during ante-mortem examination; (III) carcasses are inspected for FMD infection in the slaughter house during post-mortem examination; and (IV) carcasses are chilled and

deboned (bones are removed to reduce the chance of virus survival and eventual transmission). Note that standard sanitary procedures are carried out for transportation and lairage of animals for slaughter. These include routine disinfection of transport vehicles and lairage facilities. The abattoirs used to slaughter cattle for export are different from those used for local consumption. The overall risk of infection after the animals leave the farm is very small, therefore is not considered in this study. Letp, = the probability that FMD is not detected in an infected herd by the Animal Health Attention System,p, = the probability that FMD is not detected on an infected animal during ante-mortem inspection,p, = the probability that FMD is not detected on an infected carcass during post-mortem inspection, andp, = the probability that FMD virus in deboned beef of an infected carcass survives the chilling and deboning treatment.

Suppose a total of nk cattle are selected from herd k to produce deboned beef for export, and i out of the nk cattle are FMD infected. A random sample of size sk is

drawn from the nk cattle for inspection in the slaughterhouse pens during ante-mortem examination. Let X, denote the number of infected cattle in the antemortem sample, then X, has a hyper geometric distribution with

P(X,=j)= ,j=o,1,2 , . . . , min( sk ,i) (6)

where j is the number of FMD-infected cattle in the antemortem sample, min(s, ,i) is the minirnum of sk and i.

In accordance with the European Community directives (The Commission of the European Communities, 19931, if FMD is detected in any animal from a given herd

54 P. Yu et ul./ Preventive Veterinury Medicine 30 (1997) 49-59

during ante-mortem inspection, the export of deboned beef from that farm will be canceled. Given that there are j FMD-infected cattle in a sample of size sk, the probability that FMD is not detected during ante-mortem inspection is given by pi. Given that i FMD-infected cattle are selected from herd k, the probability that FMD is

not detected in the consignment from herd k during ante-mortem inspection, P(Uii), is

Sk

P(U,) = CP( x,=j>p; (7) j=O

Note that P( X, = j> is dependent on i, the number of FMD-infected cattle selected from herd k.

For post-mortem inspection, suppose there are i out of the nk carcasses that are FMD infected. A random sample of size rk is drawn from the nk carcasses for inspection. Let Z, denote the number of infected carcasses in the sample, then Z, has a similar distribution to X,. Given that i FMD-infected carcasses are selected from nk carcasses,

the probability that FMD is not detected during post-mortem inspection is

P(V,) = ;P(Z, = j)p{ (8) j=O

Note that Eqs. (7) and (8) are based on the assumption that the sensitivity and specificity of ante- and post-mortem inspections is 100%. That is, we do not consider any false results in the inspections in this study. However, it should be interesting to conduct further research on the effect of false positives and negatives on the risk assessment (Martin et al., 1992).

The probability that FMD virus in deboned beef of at least one of i infected carcasses survives the chilling and deboning treatment, P(Wi), is given by

P(W;) = 1 - (1 -p4)i (9)

Noticing that

P(Y,>O)= &(Yk=i) (‘0) i= I

in order to incorporate the risk-reduction factors, we rewrite Eq. (4) as follows:

P(R,)=l-fi I-pHfJP(Yk=i) k= I [ i= I 1

(‘1)

Incorporating the risk-reduction factors Eqs. (7)-(9) into Eq. (111, we have P(R,), the probability that FMD virus from at least one infected animal passes the risk-reduc- tion pathway:

P(R,) = 1 -J.j 1 -~H~,&(Y~=i)P(Ui)P(Vi)P(Wi) k= I i= I 1 (12)

where R, is the event that FMD virus from at least one infected animal passes the risk-reduction pathway. This probability is called the risk of FMD virus introduction

P. Yu et ul./Preuentioe Veterinary Medicine 30 (1997) 49-59 55

associated with importing nw kg of deboned beef in accordance to the risk reduction

procedures of the European Community directives, where n is the total number of cattle selected for export, and w is the average deboned beef weight (kg) per animal.

The Tarameter values in the model were estimated based on the available information obtained from South American Common Market countries (Wilson, 1994), and erred on the side of caution. The values of the probabilities were estimated as follows: pH = 0.005, p* = 0.166, p, = 0.2, p2 = 0.03, pj = 0.01, and p4 = 0.1. These probability estimates

were the basis for the simulations in this study. In the simulations, each of the

probabilities changes in a given range (see the figures in the next section) while the

others were set at their basic values. The ranges of the probabilities were determined based on the information from South American Common Market countries or the natural domain (all possible values). Other parameter estimates used in the simulations are:

w = 19 kg, nk = 80, and m = 66. The value of m (number of herds) was derived from

the values of w and nk to ensure that 100 tons of deboned beef was produced. Through computer simulations, we evaluated the probability of FMD virus introduction by

importing 100 tons of deboned beef in relation to FMD prevalence, number of cattle selected from each herd, and sample sizes in ante-mortem and post-mortem inspections. We examined the effect of the risk-reduction procedures on the probability of FMD virus introduction.

3. Results and discussion

Computer simulations were carried out on a NeXT work station with C as the programming language. Fig. 2 shows the probability of FMD virus introduction by

importing 100 tons of deboned beef as a function of both the prevalence of infected herds (/in > and the prevalence of infected cattle within a herd ( pA >. The risk of FMD

,004

PH

Fig. 2. Probability of FMD virus introduction by importin, D 100 tons of deboned beef as a function of the

prevalence of infected herds ( pH) and the prevalence of infected cattle in a herd ( pA ).

56 P. Yu et al./ Preventive Veterinary Medicine 30 (1997) 49-59

virus introduction increases as the prevalence of infected herds increases. We were interested to notice that the risk first increases rapidly as the prevalence of infected cattle within a herd increases from 0 to 0.0264, then decreases after that. This is possibly due to some type of balance between the prevalence of infected cattle within a herd and the

chance to detect FMD in ante-mortem and post-mortem inspections. On one hand, low prevalence means low risk; on the other hand, low prevalence of infected cattle within a herd would decrease the chance of detecting at least one FMD-infected animal in

ante-mortem and post-mortem inspections, thus the risk of FMD virus introduction would be increased. The results indicate that the risk of FMD virus introduction is quite

high (> lo-‘) when the prevalence of infected cattle within a herd is relatively low.

Therefore, the early stage (lower prevalence) of an FMD outbreak may impose a high risk of FMD virus introduction to an importing country. We should note that the results were obtained without considering the variations in sensitivity/specificity of diagnostic procedures. See Martin et al. (1992) for some interesting discussions on the sensitivity/specificity of diagnostic procedures.

The effect of risk-reduction procedures on the probability of FMD virus introduction by importing deboned beef is shown in Fig. 3. The risk of FMD virus introduction is almost linearly related to the probability of failure in detecting FMD in an infected herd by the Animal Health Attention System (Fig. 3a). The risk increases as the probability of

(b)

0 0.2 0.4 0.6 0.8 1 0 02 0.4 0.6 0.8 I

P3 P4

Fig. 3. Probability of FMD virus introduction as a function of: (a) probability of failure in detecting FMD virus

in an infected herd by the Animal Health Attention System; (b) probability of failure in detecting FMD virus

during ante-mortem inspection; (c) probability of failure in detectin g FMD virus during post-mortem

inspection; Cd) probability of FMD virus surviving chilling and deboning.

P. Yu et al./ Preuentiue Veterinary Medicine 30 (1997) 49-59 57

failure irr detecting FMD during ante-mortem or post-mortem inspection increases (Fig.

3b,Fig. 3~). The increase is slower for smaller values of p2 and ps than that for larger values. The risk of FMD virus introduction increases rapidly as the surviving probability

of FMD virus increases from 0 to 0.2, then slowly from 0.2 to 1.0 (Fig. 3d). The probability of FMD virus introduction by importing 100 tons of deboned beef is

a decreasing function of the number of cattle selected for exportation from each herd

(Fig. 4). Note that given the total number of cattle needed to produce 100 tons of deboned beef for exportation, the number of herds needed decreases as the number of

cattle selected from each herd increases. Fig. 5 shows the probability of FMD virus introduction associated with deboned beef

importation in relation to the sample sizes in ante-mortem and post-mortem inspections. It is clear that the risk of FMD virus introduction is very high for smaller sample sizes

(lo-*). The risk can be reduced to a level of 10m6 by increasing the sample sizes. The

simulation results appear to indicate that it is not necessary to check every animal and carcass during ante-mortem examination and post-mortem examination. Further increase in sample size beyond a critical value may not importantly reduce the risk of FMD virus

introduction. For instance, if 80 cattle are selected for exportation from each herd, then it is good enough to check 50 cattle during ante-mortem examination and post-mortem examination. Note that these findings are valid only for the given range of assumptions

tested in this study. In order to prevent the introduction of animal diseases, animal-regulatory officials are

required to make decisions under uncertainty. Quantitative risk assessment allows one to evaluate the risk of an adverse event occurring even when the information is incomplete or uncertain. Uncertainty in parameter values has a great impact on the outcomes of risk assessment. Nonetheless, decisions have to be made using whatever data are available. This model is developed to assess the risk of introduction of FMD virus associated with

0.0006 -

60 64 69 75 82 91 101 114

Number of animals selected I herd

Fig. 4. Probability of FMD virus introduction by importin, 0 100 tons of deboned beef as a function of the

number of cattle selected from each herd.

58 P. Yu et aI./ Prevrnrioe Vererinury Medicine 30 (1997) 49-59

1 c .g

:

z b 0.01 c

: B 0.0001

,z 5

d ; 0.000001

post-mortem inspection

Fig. 5. Probability of FMD virus introduction by importin g 100 tons of deboned beef as a function of sample

sizes in ante-mortem inspection and post-mortem inspection.

deboned beef importation. The results in this study are dependent on the risk reduction procedures for deboned beef importation. However, the methodology may be used to assess the risk of introduction of other animal diseases associated with the importation of animals and animal products.

The risk assessment in this study is based on the prevalence of infected cattle in herds as well as the prevalence of infected herds in the exporting country. The prevalences are constant in this model, but would change over time in reality. In the future, this model may be integrated with a dynamic model of FMD transmission to study the evolution of the risk over time, and to evaluate control strategies if FMD is introduced to an importing country.

Acknowledgements

This work was partially supported by grant No. SG12 RR03059-07 from the National Institute of Health (NIH/RCMI), and grant No. RII-9005621 from the National Science Foundation (NSF/RIMI). The authors thank M.D. McElvaine, R.M. McDowell and R.W. Fite from USDA-APHIS for helpful discussions on the risk assessment methodol- ogy, P. Sutmoller and J. Toussiant for providing information on the procedures of deboned beef importation from South American Common Market countries (Mercosul countries).

References

Animal and Plant Health Inspection Service (APHIS), 1991. Foot-and-mouth disease emergency disease

guidelines. USDA, Hyattsville, USA, pp. 144.

P. Yu et ol./ Preventive Veterinary Medicine 30 (1997) 49-59 59

Berentsen, P.B.M., Dijkhuizen, A.A. and Oskam, A.J., 1992. A dynamic model for cost-benefit analyses of

foot-and-mouth disease control strategies. Prev. Vet. Med., 12: 229-243.

Davies, G., 1993. Risk assessment in practice: a foot and mouth disease control strategy for the European

Community. Rev. Sci. Tech. Off. Int. Epiz., 2: I 109-I I 19.

Garner, M G. and Lack, M.B., 1995. An evaluation of alternate control strategies for foot-and-mouth disease

in Australia: a regional approach. Prev. Vet. Med., 23: 9-32.

Krystynak, R.H.E. and Charlebois, P.A., 1987. The potential economic impact of an outbreak of foot-and-mouth

disease in Canada. Can. Vet. J., 28: 525-527.

Larsen, R.J. and Marx, L.M., 1990. Statistics. Prentice-Hall, Englewood Cliffs, NJ, USA, pp. 829.

Martin, S.W., Shoukri, M. and Thorbum, M.A., 1992. Evaluating the health status of herds based on tests

applied to individuals. Prev. Vet. Med., 14: 33-43.

McElvaine, M.D., McDowell, R.M., Fite, R.W. and Miller, L., 1993. An assessment of the risk of foreign

animal disease introduction into the United States of America through garbage from Alaskan cruise ships.

Rev. .S:i. Tech. Off. Int. Epiz., 12: 1165-l 174.

Miller, L., McElvaine, M.D., McDowell, R.M. and Ahl, A.S., 1993. Developing a quantitative risk assessment

process. Rev. Sci. Tech. Off. int. Epiz., 12: 1153-l 169.

Morley, R.S., 1993. A model for the assessment of the animal disease risks associated with the importation of

animal and animal products. Rev. Sci. Tech. Off. Int. Epiz. 12: 1055-1092.

The Commission of the European Communities, 1993. Commission decision concerning animal health

conditions and veterinary certification for imports of fresh meat from South American countries. Off. J.

European Communities. No. L 179: I l-22.

Wilson, S., 1994. Quantitative assessment of the risk of foot and mouth disease introduction into the Caricom

region by the importation of deboned beef from mercosul countries. Presented to the Chief Veterinary

OfficeIs of the Caribbean Countries.