SEARCH

SEARCH BY CITATION

Keywords:

  • horse;
  • demography;
  • National Equine Database;
  • infectious disease;
  • risk modelling

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors' declaration of interests
  8. Sources of funding
  9. Acknowledgements
  10. Authorship
  11. References

Reasons for performing the study

The National Equine Database (NED) contains information on the size and distribution of the horse population, but the data quality remains unknown. These data could assist with surveillance, research and contingency planning for equine infectious disease outbreaks.

Objectives

1) To assess the extent of obsolete and missing data from NED, 2) evaluate the extent of spatial separation between horse and owner location and 3) identify relationships between spatial separation and land use.

Methods

Two questionnaires were used to assess data accuracy in NED utilising local authority passport inspections and distribution of questionnaires to 11,000 horse owners. A subset of 1010 questionnaires was used to assess horse–owner geographic separation.

Results

During 2005–2010, 17,048 passports were checked through local authority inspections. Of these, 1558 passports (9.1%; 95% confidence interval [CI] 8.7–9.5%) were noncompliant, with 963 (5.6%; 95% CI 5.3–6.0%) containing inaccurate information and 595 (3.5%; 95% CI 3.2–3.8%) classified as missing. Of 1382 questionnaires completed by horse owners, 380 passports were obsolete (27.5%; 95% CI 25.2–29.9%), with 162 (11.7%; 95% CI 10.0–13.4%) being retained for deceased horses and 218 (15.8%; 95% CI 13.9–17.7%) having incorrect ownership details. Fifty-three per cent (95% CI 49.9–56.1%) of owners kept their horse(s) at home and 92% (95% CI 90.3–93.7%) of horses resided within 10 km of their owners.

Conclusions and potential relevance

Data from a small sample survey suggest the majority of data on NED are accurate but a proportion of inaccuracies exist that may cause delay in locating horses and contacting owners during a disease outbreak. The probability that horses are located in the same postcode sector as the owner's home address is larger in rural areas. Appropriate adjustment for population size, horse–owner spatial separation and land usage would facilitate meaningful use of the national horse population derived from NED for risk modelling of incursions of equine diseases into Great Britain.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors' declaration of interests
  8. Sources of funding
  9. Acknowledgements
  10. Authorship
  11. References

Reliable data for susceptible host populations are key for meaningful surveillance, epidemiological research and contingency planning for dealing with incursions of equine diseases. In situations where the population is highly susceptible and little detailed data on its demography are available the consequences may be costly, particularly where the epidemic is extensive and potentially rapidly spreading. This was particularly highlighted in an outbreak of equine influenza in Australia in 2007 in which approximately 4500 infected premises, in an area of 278,000 square kilometres were affected in less than 2 months [1, 2] and measures to effect subsequent eradication required considerable efforts to locate large numbers of horses [3]. It has been hypothesised that an outbreak of equine infectious disease in the United Kingdom (UK), such as African horse sickness (AHS) virus, could have a similarly devastating impact on the UK equine industry, particularly if horses and their owners could not be rapidly located, hence the need for accurate denominator data on the horse population. A nonspatial, temperature-dependent epidemiological model has been developed to assess the risk of bluetongue to UK livestock [4] and a spatial-dependent extension of the model could be employed to study the risk of an outbreak in the UK of similar midge-borne equine disease, such as AHS, if the data are available.

Risk of infectious disease transmission can be evaluated using the basic reproduction number (R0) of the infection, which is defined as the expected number of secondary cases caused by one infectious horse, introduced into a completely susceptible population. If, on average, R0 is <1, the infection will eventually die out but if, on average, R0, is >1 then the infection will continue to spread within the host population [5, 6]. A key conclusion of the aforementioned nonspatial, temperature-dependent epidemiological model [4] was that the ratio of Culicoides midges to host and the ambient temperature are the most sensitive parameters for the variation of R0, i.e. even small changes in vector-to-host ratio and temperature are likely to have a significant influence on the resulting value of R0. Another conclusion of the work was that the value of R0 is affected by the close proximity of other host species. This is particularly important for AHS in horses, as cattle, sheep and pigs are all common targets of biting Culicoides midges, the vector of AHS virus. Although they are not susceptible to AHS virus, their close proximity to horses, such as through co-grazing the same pasture, would have a potentially large impact on the feeding patterns of the infected midges and ultimately on the spread of AHS virus infection. Consequently a realistic assessment of the risk of AHS in the UK requires accurate knowledge of 1) ambient temperatures, 2) numbers and distribution of horses, 3) vector abundance and 4) numbers of other hosts in close proximity to horses. Inaccurate data for the numbers and distribution of horses and their proximity to other hosts might lead to erroneous associations with localised parameters resulting in incorrect estimates of the magnitude of R0 and not simply a spatial redistribution of its values. Estimates on the size of the horse population have ranged from 840,000 [7] to 1.2 million [8] horses, based on various different sampling methods, such as telephone surveys [8] and using data from equine organisations such as charities as well as agricultural census data from the Department for Environment Food and Rural Affairs (DEFRA) and the National Equine Database (NED), which collates horse passport data [7]. It is plausible that NED should contain the most accurate data on the horse population as it is a legal requirement for all horses in the UK to have a passport and data from passports are centrally stored on NED. Horse Passport Regulations (2004) came into force on the 10 June 2004 in England [9] and are now covered by Horse Passports Regulation (2009) and European Commission Regulation 504/2008/EC [10], which both cover the identification of horses for breeding and food production purposes. It is the owner's responsibility when they purchase a horse to notify the passport-issuing organisation that issued the passport of the change of owner details and submit their name and address and the name and identification of the horse. Owners must also notify the passport-issuing organisation that an animal is no longer alive within 30 days of it dying or being subjected to euthanasia. Failure to fulfil these requirements not only contributes to inaccurate horse passport data but is also illegal under the UK's Horse Passports Regulations (2009) [10]. As data from all horse passports in the UK are centrally stored on NED, it has been suggested that NED could be a useful resource for rapidly acquiring equine demographic data in the event of a significant disease outbreak, however the data on NED remain unvalidated [11]. It has previously been noted that while NED holds the addresses of horse owners' residence when passports are issued, it does not contain details of where horses are actually located [11, 12], or the number of horses per premises, as equine premises are not registered to the same extent that agricultural premises are. Previous UK surveys have suggested that 90% of horses are located within 10 km of their owners and around 60% of horses are located at the owner address [11]. There may also be a proportion of horses that do not have passports (these data will be missing from NED), and a number of owners that do not inform NED when their horse dies (these data will be obsolete on NED). There may also be a proportion of owner details on NED that are out of date, if owners have failed to inform the relevant organisation of a change of address, or new owners have failed to inform the passport-issuing organisation when they have bought a new horse.

Our aims were to: 1) assess the extent of both obsolete and missing data from NED, 2) evaluate the extent of spatial separation between the NED-registered addresses of horse owners and the location where their horse(s) are kept and 3) identify a potential relationship between spatial separation and land use. This information could be applied to develop a credible national distribution of horses in Great Britain that will be suitable for use in future risk modelling of equine infectious disease incursions.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors' declaration of interests
  8. Sources of funding
  9. Acknowledgements
  10. Authorship
  11. References

Assessment of the accuracy of data held on NED

The extent of data missing from NED (passport compliance/noncompliance)

As local authority trading standards officers are responsible for enforcing horse passport regulations, in order to assess the extent of data missing from NED, a survey was conducted of local authority trading standards database records. A Freedom of Information request was submitted to all local authority trading standards offices in Great Britain (n = 202), requesting information on annual data for the number of passports checked and the number of compliant, missing and inaccurate passports from 2005 to 2010, or overall totals if data were not available annually. Noncompliant passports were defined as either ‘missing’, where horses were found, but no passport was available or ‘inaccurate’, where passports did not have the correct owner details, horse details or had section IX (declaration of intention of horse for human consumption or not) missing or incomplete. Descriptions of compliance/noncompliance rates were taken from the data and ArcGIS (ArcMap version 10.0, ESRI USAa) mapping software was used to generate county-level passport checking and proportion of compliance/noncompliance data for Britain based on aggregated data for 2005–2010.

The extent of obsolete data

A data protection-compliant, questionnaire-based survey of 11,000 NED-registered horse owners was conducted with the assistance of 7 participating passport-issuing organisations, whose data are part of NED (Fig 1). In order to ensure compliance with data protection protocols, participating passport-issuing organisations electronically issued data relating to the randomly selected 11,000 owner and horse details to a legally approved, commercial mailing house that merged the data with predesigned Animal Health Trust covering letters and questionnaires. Horse owners were asked to verify if they still owned the horse currently registered to them with the passport-issuing organisation and NED database and if so, asked to confirm the accuracy of the data relating to their horse's passport. To help facilitate this, horse owners were provided with the mandatory descriptive data stored about their horse on the respective passport-issuing organisation database. Data included Unique Equine Life Number (UELN), horse name, date or year of birth, gender, breed and colour.

figure

Figure 1. Flow of data between the passport-issuing organisations (PIOs), mailing house and Animal Health Trust (AHT), demonstrating how questionnaires were printed and mailed to horse owners while remaining confidential. UELN stands for Unique Equine Life Number and is used as the unique passport number.

Download figure to PowerPoint

Assessment of the extent of spatial separation between NED-registered horse owner addresses and location of their horses

As previously described [11], complete postcode records available for 1010 owners and their horses from the NED-registered owner survey were used to estimate the distribution of the distances between the 2. Postcodes for horse and corresponding owner locations were converted to grid references using GeoConvert online software (http://geoconvert.mimas.ac.uk). Based on Pythagoras' theorem, the easting and northing values in grid references were used to establish the Euclidean (straight line) distance (measured in kilometres) between the 2 postcodes.

Identification of a relationship between owner–horse spatial separation and land use

The final key objective of the current work was to identify a possible relationship between the owner–horse spatial separation and land use. It was hypothesised that horses are kept in premises located in rural areas although their owners might reside in urbanised areas. Consequently, the 1010 responses from the owner survey that had complete records for owner address and corresponding horse location were used to calculate the proportion of horses located at the owner's postcode and these data were used in conjunction with the value of urban coverage quantified at the owner's address (data from the Centre for Ecology and Hydrology, http://www.ceh.ac.uk/).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors' declaration of interests
  8. Sources of funding
  9. Acknowledgements
  10. Authorship
  11. References

Passport compliance and accuracy monitoring by local authorities

Responses were received from all 202 of the local authority trading standards offices contacted. Of these, 100 (49.5%) checked horse passports on a regular or occasional basis and recorded the data, 40 (19.8%) checked passports but did not store the data or have it in an accessible format and 62 (30.7%) did not check horse passports so no data were available. Of the 62 local authorities that did not check horse passports, 25 (40%) were in Greater London and the remainder were predominately urban authorities.

In Great Britain between 2005 and 2010, the number of passport inspections carried out each year increased annually with the exception of a small decrease in 2007 (Fig 2). A total of 17,048 individual passports (approximately 1.4% of the total number of horses registered on NED) were checked by 64 local authorities, excluding authorities that did not record the number of individual passports checked annually. Of these, 15,490 (90.9%; 95% confidence interval [CI] 90.5–91.3%) passports were found to be compliant and 1558 (9.1%; 95% CI 8.7–9.5%) were noncompliant. Missing passports constituted 595 (3.5%; 95% CI 3.2–3.8%) of the noncompliant passports, with the remaining 963 (5.6%; 95% CI 5.3–6.0%) containing obviously inaccurate information. The extent of inaccurate information was lower in 2005–2007 than in later years. County-level spatial differences in compliance rates are illustrated in Figure 3.

figure

Figure 2. Number of horse passports checked each year from 2005 to 2010 by local authorities, including the number of compliant, missing and inaccurate passports.

Download figure to PowerPoint

figure

Figure 3. Spatial differences in a) the total number of passports checked, b) the % compliant, c) % missing and d) % inaccurate passports in each county in Great Britain.

Download figure to PowerPoint

The extent of obsolete data retained within the sample passports

Of the 11,000 surveys mailed to horse owners, 1787 (16.2%) were returned either completed by the horse owner (n = 1400, 12.7%) or as undelivered (n = 387, 3.5%), with data from undelivered surveys (34%; 95% CI 31.7–36.1%) not being included in subsequent analyses. Of the 1382 owners that responded to the question regarding ownership, 1002 owners (72.5%; 95% CI 70.2–74.9%) stated they currently owned the horse registered to them on the database. In total, 380 passports (27.5%; 95% CI 25.2–29.9%), were classified as obsolete as the horse was either deceased and the passport not been returned to the passport-issuing organisation (n = 162; 11.7%; 95% CI 10.0–13.4%) or had been sold and either the passport had not been passed on to the new owner, or the new owner had not informed the passport-issuing organisation of the new ownership (n = 218; 15.8%; 95% CI 13.9–17.7%). There were some differences in data accuracy between the 7 individual passport-issuing organisations used for the survey (Fig 4).

figure

Figure 4. Variation in data accuracy between passport-issuing organisations (PIOs) with numbers of returned passports listed above each bar in brackets. A–D are breed societies and E–G are ID-only PIOs.

Download figure to PowerPoint

The majority of the mandatory horse data on the passport-issuing organisation databases were accurate, with over 95% of data being accurate for all fields except breed, which was 94.9%. There were 24 passports (2.5%; 95% CI 1.5–3.5%) that had an incorrect year or date of birth recorded as verified by the current owner and a further 221 passports (15.8%; 95% CI 13.9–17.7%) had no year or date of birth recorded.

Assessment of the extent of spatial separation between NED-registered horse owner addresses and the location of their horses

Of the 1010 owners that provided a complete postcode for both horse and owner location, it was calculated that 53% (95% CI 49.9–56.1%) of horses lived at the owner's address, while 73% (95% CI 70.2–75.7%) of horses resided within 3 km of their owners, 81% (95% CI 78.58–83.42%) within 5 km of their owners and 92% (95% CI 90.3–93.7%) within 10 km of their owners. These data are comparable with a previous pre-existing database of 1440 horse and owner addresses at the Animal Health Trust [11], where 61% (95% CI 58.4–63.5%) of horses lived at the same address as their owner, 73% (95% CI 70.7–75.2%) resided within 3 km, 80% (95% CI 77.9–82.0%) within 5 km and 90% (95% CI 88.45–91.55%) within 10 km of their owners.

Relationship between owner–horse spatial separation and land use as an informative tool to develop a credible national distribution of horses

The spatial distribution of horse population density, based on owner addresses in NED (the number of horses per postcode sector, divided by its area, to provide anonymity for the horse owner), illustrated in Figure 5a, appeared to mirror urban coverage (Fig 5b). In particular, 2 highly urbanised locations (City of Westminster and one of the Greater London boroughs), had exceptionally high spatial distribution (>2000 horses per km2). These patterns suggest that using owner addresses as a proxy for horse location instead of actual horse locations might lead to an unrealistic equine population map. Such unrealistic effect can be further amplified in a disease risk map. This is the scenario when the presence of a nonsusceptible host, such as ruminants, reduces the transmission of the disease and thus has a dilution effect. As the number of ruminants is negligible in urban areas (see Fig 5c) the basic reproduction number can be unrealistically overestimated in urban settlements. Preliminary analysis (not shown) have confirmed that in such scenario highest values of R0 occurs in the most urban areas.

figure

Figure 5. a) Distribution of horses' location based on owner address in National Equine Database. b) Percentage of built-up coverage (Centre for Ecology and Hydrology data). Horses are expected to be located in rural areas. c) Distribution of other nonsusceptible host (cattle) based on Rapid Analysis and Detection of Animal Related Risk (RADAR)–Department for Environment Food and Rural Affairs data. R0 is affected by the presence of other hosts at the horse's position.

Download figure to PowerPoint

To detect a potential association of owner–horse separation with land use, the proportion of horses kept at owner location vs. the percentage of urban coverage was plotted. Figure 6 shows that proportion of horses located at the same postcode as the owner in general decreases with increases in the urban coverage; this implies that mapping owner addresses as a simple proxy for horse location introduces an important source of error resulting in a spatial distribution biased towards large urban areas.

figure

Figure 6. Proportion of horses located at owners' postal codes vs. urban coverage. Each bar represents the proportion of horses located at the same postal code as the owners, thus sharing the same average value of urban coverage.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors' declaration of interests
  8. Sources of funding
  9. Acknowledgements
  10. Authorship
  11. References

Based on information derived from several publicly available equine industry reports published between 2004 and 2009, evidence of wide variation and uncertainty in estimates of the overall size of the horse population in Great Britain was provided [11], ranging from 988,000 [13] to 1.2 million [8]. Recently, a more conservative estimate of 840,000 horses [7] has been published. Furthermore, among the 80 passport-issuing organisations in Great Britain, fewer than 25% requested keeper address details in addition to the mandatory owner address details [11], thereby ensuring that only a small proportion of NED-registered horses were likely to have information available with which to locate the keeper, as a proxy for horse location. However, based on previously collated data from a nonpassport-issuing organisation syndromic surveillance database containing 1440 linked horse and owner locations, there was evidence that 90% of horses in Great Britain were actually located within 10 km of their owner's address [11].

In the current study the accuracy of some of the data held on NED was assessed via Trading Standards' records and owner questionnaires, thereby allowing further evaluation of NED's suitability as a resource for establishing the size and distribution of the horse population in Great Britain. Such knowledge about susceptible animal populations is widely accepted as being important for conduct of meaningful disease surveillance and optimising contingency planning for effective disease control. Based on 2 separate assessments conducted in this study, some inaccuracies were shown to exist in the sample of NED's information on the horse population in Great Britain. Although the majority of data in the sampled data were accurate and up to date, there was both evidence of data that were missing (i.e. horses without passports not recorded by passport-issuing organisations and NED) and evidence of retention of obsolete records (i.e. data retained by passport-issuing organisations and NED for dead horses). Missing and obsolete data will act to cancel each other out to some extent when evaluating the size of the population at risk; however, they would affect NED's efficiency in rapidly and accurately locating horses and horse owners in the event of a major disease outbreak affecting horses in Great Britain. This was particularly highlighted in this study with the observation that of 1787 surveys returned completed or undelivered from passport-issuing organisation/NED-registered owners, overall 605 had the wrong owner address recorded.

Regarding the assessment of the extent of missing data from NED, some caution is required in interpreting findings derived from local authority trading standards inspections, as there was evidence within these data of variation in the quality of data recorded and made available by different local authorities. However, use of Freedom of Information requests ensured a 100% response rate from all 202 local authorities contacted and although not all authorities provided the data requested this was still useful information for the investigation as it not only provided information about how active authorities were regarding passport inspections but also provided more of a national picture of noncompliance, rather than just focusing on the more proactive local authorities. Furthermore, the authors did consider other methods for making such assessments but after due consideration it was felt most unlikely, without the legal powers afforded to local authority trading standards officers, that horse owners would readily volunteer, even anonymously, to admit to not possessing equine passports, even to nonDEFRA affiliated researchers.

In 2006 and 2007 there were higher proportions of missing passports, compared with passports containing inaccurate data, possibly because horse owners took time to comply with the regulations in applying for a passport relatively soon after they became mandatory. From 2008, the proportion of passports with inaccurate data was greater than the proportion of missing passports, particularly in 2008 which also recorded the highest noncompliance rate of the past 6 years. It is speculated that the higher rate of passports with inaccurate data since 2008 has been due to increased awareness among local authority personnel over time as to what constituted an inaccurate passport. It is also possible that as horse passport legislation was in place for a longer period of time, there was increased opportunity for the passports not to be updated when horses died or changed ownership or owners changed address.

The assessment of the extent of obsolete data retained within a sample from NED was limited due to the small sample size. Approximately 0.1% of the horse-owning population was sampled, which is not an ideal sample size; however, due to data protection restrictions, a larger sample was not available through NED. When dealing with whole population data, small sample sizes, as long as they are truly randomly selected can be adequate representations of the population. Small samples are frequently used in exit poll surveys, where only a very small percentage of the total population are surveyed, and these data are then extrapolated to the whole population. For example, in UK election exit polls an approximate sample of 16,500 is used, which is only 0.02% of the total population [14]. These data are extrapolated up to an estimated population size of 62.3 million [15] and can produce accurate results, for example the prediction by some polls of the hung parliament in the 2010 UK general election [16]. The sample of horse owners in the current study was randomly selected by horse UELN, therefore despite being a small sample of the total population, it should be representative. It was likely that the survey of NED-registered owners was subject to responder bias, as owners with accurate and up-to-date passport records were probably more likely to provide a response, thereby leading to an underestimation from this survey of the true proportion of records containing obsolete or inaccurate data. However, it was also possible that a proportion of records estimated as containing obsolete records were in fact in the process of being appropriately updated with the passport-issuing organisation and NED around the time of the survey but these had not yet been reconciled with the passport-issuing organisation records.

Based on a calculated 1.2 million live horses on NED and estimates from the studies described here of both the extent of data missing from NED (3.5%) and the extent of obsolete records retained in NED (11.7%), the overall UK horse population can be estimated to be 1.10 million animals, within 95% confidence limits of 1.08–1.13 million horses. This is between the estimate of 1.2 million horses published in the British Equestrian Trade Association (BETA) National Equestrian Survey of 2006, which was based on a telephone survey of 5087 households extrapolated to a national population of 58.45 million people [8], the 2011 BETA survey that predicted a total of 988,000 horses [13] and a recently published paper that estimated a population of 840,000 based on data from NED and ‘stakeholder’ datasets from the agricultural census and equine organisations such as the British Horseracing Authority, Weatherbys, British Eventing and equine charities as a best estimate of the total population size [7].

Not only are numbers of susceptible hosts important in accurately predicting transmission dynamics for emerging vector-borne diseases but their spatial distribution and relative proximity to vectors and important other hosts species is also significant. Epidemiological models for an outbreak of bluetongue virus in cattle and sheep [4] emphasise the importance of having an accurate map of host distribution, as the most sensitive parameters depended on the location of the host. This would also be particularly relevant to the equine, midge vector-borne disease of AHS. In this and a previous study [11], it has been shown that although the majority of horses are located within 10 km of their owners, a small proportion (8–10%) is not. The illustration in the uncorrected distribution map (Fig 5a) that areas with the largest equine populations coincided with large urban settlements is considered a consequence of postcode densities increasing in more urban areas as well as NED recording owner postcodes rather than actual horse location. One solution to this problem might have been to interpolate the distribution of horses across the entire surface of Great Britain however this technique would not have taken into account the urban coverage. It was shown that the proportion of horses kept at the same location as the owner was strongly inversely related to the level of urbanisation. This empirical dependence would be a necessary tool to implement a correction algorithm to infer a more realistic distribution of the host equine population in Great Britain. It is anticipated that combining the distributions of host location, with existing national data on ambient temperature at different times of year and other midge vector host species (cattle, sheep and pigs) will facilitate meaningful future risk modelling of equine exotic disease incursions into Great Britain. These data could also be used to develop risk models for other, nonexotic equine diseases such as equine influenza although more data would be needed on immune status and horse movement to ensure the models were accurate.

Despite provisions being put in place to enforce horse passport legislation, it is clear that these measures need to be more strictly adhered to. Without the cooperation of horse owners and local authority trading standards inspections and other industry officials, NED has a very difficult task to ensure the records are accurate and up to date, as it relies on these individuals to inform NED of any changes to circumstances regarding the passported horse. While it is possible that including keeper, or even horse location on the passport, would improve the accuracy of data on the spatial distribution of the horse population, logistically and financially it is probably not feasible to do this in real time, particularly with competition horses that spend the majority of their time travelling both nationally and internationally. Currently, using owner address as a proxy for horse address, while not infallible, is the most accurate method for locating horses in Great Britain.

Sources of funding

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors' declaration of interests
  8. Sources of funding
  9. Acknowledgements
  10. Authorship
  11. References

The work presented here was generously funded by the Horserace Betting Levy Board (ref PRJ/754).

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors' declaration of interests
  8. Sources of funding
  9. Acknowledgements
  10. Authorship
  11. References

The authors would like to acknowledge the generous support received for their work in this area from the Horserace Betting Levy Board. James Wood is supported by the Alborada Trust and the RAPIDD program of the Science and Technology Directorate, US Department of Homeland Security and the Fogarty International Center. Richard Newton is supported by contributions to the Animal Health Trust's Equine Infectious Disease Service from the Horserace Betting Levy Board, Racehorse Owners Association and Thoroughbred Breeders' Association. Simon Gubbins is supported by the Biotechnology and Biological Sciences Research Council (BBSRC). Gianni Lo Iacono is funded by the ESPA award Dynamic Drivers of Disease in Africa Consortium.

Authorship

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors' declaration of interests
  8. Sources of funding
  9. Acknowledgements
  10. Authorship
  11. References

All authors participated in the design of the study. C.A.R. and J.R.N. collected the data and performed the statistical analysis on the missing and obsolete data in NED and horse–owner spatial separation. G.L.I., J.L.N.W. and S.G. undertook the land-usage analysis. C.A.R. and J.R.N. helped to draft the manuscript. All authors read and approved the final manuscript.

Manufacturers' address
  1. aESRI Headquarters, 380 New York Street, Redlands, California, USA.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Authors' declaration of interests
  8. Sources of funding
  9. Acknowledgements
  10. Authorship
  11. References