Optimising vaccination strategies in equine influenza
Introduction
Equine influenza (EI) is a highly contagious infectious disease of equidae, which in fully susceptible animals causes a high temperature, harsh cough, and mucopurulent or serous nasal discharges. Secondary bacterial infections cause significant problems [1] and broncho-pneumonia occurs in a proportion of cases. In partially immune animals, the signs of disease are moderated and may just consist of a mild cough or mucopurulent nasal discharge [2].
Vaccination against equine influenza has been practised since the 1960s but although vaccines have improved considerably since then, there are continued problems with failure of efficacy under field conditions. Most products available internationally consist of whole killed virus, or sub-unit vaccines. The datasheets for most licensed equine influenza vaccines in Europe recommend that an annual booster dose of vaccine be given after an initial course of three doses.
In this paper we construct and parameterise a stochastic model of equine influenza to assess the risk of an outbreak in a flat race training yard under this recommended dosing schedule (which also represents the minimum vaccination policy under the Jockey Club rules in the UK). This model represents the next step forward from previous work which was based on simulating the management life cycle of a Thoroughbred population [3]. A stochastic model is essential when dealing with relatively small populations as chance events (such as failure of the infection to transmit) become important [4]. The model assumes that all horses in a yard are in one of four states: susceptible to infection (S), exposed to infection but not yet infectious (E), infectious (I) and resistant (R). Such ‘compartmental’ population models have been used successfully to study many infectious diseases including malaria [5] and measles [6]. The response of horses to administered vaccines and the relationship between the vaccine and infectious virus strains are critical factors in determining vaccine efficacy in the field [7]. In this paper, we only consider the situation where the strains are homologous.
The SEIR model is parameterised from several data sets. The latent and infectious periods are ascertained from a group of ponies which were vaccinated against, and subsequently challenged with, equine influenza.
The transmission rate has previously been estimated for unvaccinated animals [8] and we assume that in vaccinated animals it is less than or equal to this rate. A key component of the model is an empirical relationship between pre-challenge antibody and probability of becoming infectious (given exposure) which was derived from quantitative evaluation of data from equine challenge experiments performed at the Animal Health Trust. By using this relationship in conjunction with the model it is possible to simulate epidemic development in a yard provided that antibody levels of all the horses in the yard are known (other complicating factors such as horse age, gender and vaccine history need not be considered).
A realistic yard population structure, which takes account of population dynamics over the course of a year in a flat race training yard (e.g. sale of older horses and purchase of yearlings), is incorporated into the model which is then used to assess the risk of an outbreak of equine influenza in the yard under the current minimum policy in accordance with Jockey Club rules. A key preliminary finding of the models was that small epidemics are far more likely than large epidemics [8] and these small outbreaks could be responsible for maintaining equine influenza in the population at large. Consequently, our definition of risk includes small outbreaks and throughout the paper we ask: If equine influenza were introduced to the yard (from an external contact), what are the probabilities of epidemics affecting 3 and 10% of the yard population?
The model is then adapted to consider an alternative vaccination strategy (where the frequency of vaccination of older horses is increased) and the probabilities of both small and large epidemics are again estimated, providing a quantitative comparison between the current minimum policy and a plausible alternative. Finally, a two-yard model is implemented to address the question of risk of transmission between yards. This can occur locally at shared training areas such as gallops and nationally at race meetings. This two-yard model is the beginning of a more complex model which will look at large spatial scales, up to the national level.
Section snippets
Transmission parameters
The stochastic SEIR model uses the three-key epidemiological parameters for equine influenza: latent period (1/a), infectious period (1/g) and transmission rate (β). The first two rates (a and g) have been estimated from clinical observation of 27 homologously vaccinated ponies that were subsequently challenged with influenza and went on to show symptoms (Table 1). These data clearly show the benefits of vaccination against equine influenza when compared with a control group (Fig. 1).
Results
The following results address the question: If equine influenza were to enter the yard (from an external contact) in a given week of the year, how likely is it that there will be a small epidemic or a large epidemic? Probabilities of an outbreak affecting 3 and 10% of the yard population under the current minimum policy are presented as three-dimensional surface plots (Fig. 7, Fig. 8, respectively). The x-axes represent the transmission rate for equine influenza in vaccinated animals and the y
Discussion
We have constructed and parameterised a model that can predict the likelihood of an outbreak of equine influenza in a training yard of Thoroughbred racehorses. A stochastic formulation of the model is used to capture the inherent variability in epidemic development. The latent period and infectious period for equine influenza in a vaccinated population are established from clinical observations (Table 1, Fig. 1). The transmission rate is more difficult to estimate. However, it is known for an
Acknowledgements
We are very grateful to the Horserace Betting Levy Board who funded much of this work through their support of the Animal Health Trust’s diagnostic and surveillance services. We are also grateful to Stephen Cornell and Matt Keeling for helpful discussions. The manuscript was much improved following comments from our anonymous referee, to whom we are very grateful. B.T. Grenfell was supported by the Wellcome Trust.
References (18)
- et al.
Risk factors for equine influenza serum antibody titres in young Thoroughbred racehorses given an inactivated vaccine
Prev. Vet. Med.
(2000) - et al.
Modelling the spread of a viral infection in equine populations managed in Thoroughbred racehorse training yards
Prev. Vet. Med.
(2000) - et al.
Secondary bacterial infections following an outbreak of equine influenza
Vet. Rec.
(1992) - Grenfell BT, Dobson AP, editors. Ecology of infectious diseases in natural populations. Cambridge: Cambridge University...
- et al.
A contribution to the mathematical theory of epidemics
Proc. R. Soc. London A
(1927) - et al.
Chaos and biological complexity in measles dynamics
Proc. R. Soc. London B
(1993) - Mumford JA. Control of influenza from an international perspective. In: Proceedings of the Eighth International...
- et al.
Modelling equine influenza 1: a stochastic model of within-yard epidemics
Epidemiol. Infect.
(2002) - et al.
Experimental quantification of vaccine-induced reduction in virus transmission
Vaccine
(1994)
Cited by (36)
Diseases of the Respiratory System
2019, Large Animal Internal MedicineEstimating the potential for disease spread in horses associated with an equestrian show in Ontario, Canada using an agent-based model
2018, Preventive Veterinary MedicineCitation Excerpt :Further analysis is required to explore the effects of targeting facilities based on these potential risk factors, and to determine whether it is more effective to target facilities based on these risk factors compared to randomly targeting a large proportion of facilities in the population. Previous models used to investigate equine influenza outbreaks have included stochastic, differential equation models (Glass et al., 2002; Park et al., 2003; Baguelin et al., 2010), which have limitations regarding the inclusion of horse characteristics and detailed contact patterns. Stochastic, state-transition simulation models have also been used to evaluate disease control strategies between facilities, while incorporating spatial and temporal elements (Garner et al., 2011a; Rosanowski et al., 2016).
Equine influenza virus
2014, Veterinary Clinics of North America - Equine PracticeCitation Excerpt :This hiatus results in the third dose of the initial series being administered when the antibody response to the second vaccine dose has waned, and the amplitude of the antibody response to the third dose is consequently much greater.101 For high-risk populations (show horses, race horses and so forth), booster vaccinations should be given at 6-month intervals.59,99,102 Additional booster doses may be administered 1 to 2 weeks before potential exposure if a higher risk of infection is anticipated.
Expression of the hemagglutinin HA1 subunit of the equine influenza virus using a baculovirus expression system
2013, Revista Argentina de MicrobiologiaCitation Excerpt :Virions are harvested from egg allantoic fluid or culture medium and are inactivated or treated with detergent. A basic disadvantage these systems present is their variability and the heterogeneity of their cellular constituents9,13. Therefore, other methods for equine influenza vaccine production were developed.
Equine influenza-A global perspective
2013, Veterinary MicrobiologyCitation Excerpt :Antibody response to vaccines in the field is short-lived and intervals between 4 and 6 months are considered necessary to maintain protection in young horses (Newton et al., 2000b). Mathematical models based on assumptions and data derived from field as well as experimental challenge observations show that six monthly rather than annual boosters reduce the risk of influenza infection in racehorses aged 2 years and upwards (Park et al., 2003). Annual vaccination of older horses with several years’ history of vaccination may be adequate or in some cases excessive, as durability of immune responses increases with serial vaccination and/or natural exposure to virus.