Determining a Welfare Prioritization for Horses Using a Delphi Method.

Equine welfare issues are receiving increasing attention in the UK, but welfare problems can arise from a wide range of causes. In order to identify the most important welfare concerns for horses, we used a Delphi method with 19 equine welfare experts. An initial list of 84 equine welfare issues was generated using an online discussion board and NVivo thematic analysis. Subsequently, experts ranked these welfare issues for perceived prevalence, severity and duration of suffering associated with each issue on a 6-point Likert scale. All issues with a mean score of 3 or above (n = 37) were included in subsequent rounds. Finally, a subset of experts attended a two-day workshop to determine the final priority list of welfare issues. The welfare issues perceived to be most prevalent were lack of biosecurity, delayed euthanasia, lack of owner knowledge of equine welfare needs, fear and stress from use, and obesity. The issues considered to cause greatest suffering for individual horses were delayed euthanasia, lack of recognition by owners of pain behaviour, large worm burdens, obesity and being fed unsuitable diets for equine feeding behaviour. These outcomes can help to focus research and education interventions on the most pressing welfare issues for horses.


Introduction
Survey studies and estimates suggest that there are around 7 million horses in the EU [1], with approximately one million horses in the UK (850,000 horses in England and Wales [2], 100,000 in Scotland [3] and approximately 160,000 in Ireland [4]. Although the majority of horses are kept as

Materials and Methods
All research generated from this study was approved by the University of Edinburgh's Human Ethics Review Committee (HERC). The study formed part of a larger study to identify and prioritise welfare issues for a range of different species. Detailed methods were as previously described [21] and are given in brief here.

Recruitment of Experts
The aim was to recruit between 12 and 20 horse welfare experts to the study and to recruit a broad range of stakeholders, including practising veterinarians, academics, trainers, charity sector employees and equine industry representatives. Experts were defined as having worked in their field for at least 3 years and being based in the UK. Experts were recruited using a snowball sampling technique and were sent a consent form to sign in accordance with HERC guidelines before commencing the study.

Horse Welfare Issues
A list of horse welfare issues was developed through the use of a Delphi Conference procedure using an online discussion board. An initial list of horse welfare issues (derived from a list generated by the British Veterinary Association) was provided and experts were then able to add to and comment on issues anonymously for two weeks. All comments and discussion from the online platform were collated, and a detailed thematic analysis was conducted using NVivo 11 Pro. Using an emergent coding process, each comment from the board was categorized into themes, reaching saturation at 12 themes, which were later combined into 9 categories. Duplicate welfare issues were deleted, and the final list was checked by two independent assessors to remove redundancies or to add omissions.

Questionnaires
Two rounds of anonymous surveys were conducted using the Online Survey tool (formerly Bristol Online Survey, JISC, Bristol, UK). In both rounds, demographic data were collected from the experts including age, gender, expertise (experts could indicate expertise in more than one category; for example, they may be both veterinarians and academic researchers) and highest level of educational attainment.
For the first survey, participants were asked to score each of the potential welfare issues derived from the thematic analysis for severity (the intensity of suffering likely to be a consequence of the welfare issue in their opinion), duration (the estimated time period over which an animal was likely to experience the welfare issue in their opinion) and perceived prevalence (the likely proportion of the UK horse population they considered to be affected by the welfare issue). Each issue was scored on a 6-point Likert scale for each factor, where 1 = never/none, and 6 = always/high. An even-numbered scale was chosen as this forced the experts to make a choice (prioritised or not).
The results of the first survey were reviewed and mean scores for each welfare issue and each factor calculated. Those issues that scored at least 3.0 or above (i.e., were considered by the experts to be at least somewhat important for any of the three welfare factors) were included in the second survey as a ranked list. In the second round, experts were asked whether they agreed or not with the ranking position, and whether, in their opinion, the issue should be ranked higher or lower. Agreement between experts was assessed by calculating Fleiss' kappa statistics and by assessing the proportion of experts that agreed with each rank position.

Workshop
The final stage of the process was an expert workshop conducted in Edinburgh over 2 days. This involved two horse welfare experts drawn from the experts who completed the earlier rounds, as well as 19 other experts from the wider study who had broad animal welfare expertise in other species, including farm and companion animals (overall workshop participants consisted of academics = 6, veterinarians = 5, industry representatives = 4, charity/NGOs = 6). Over the two days, experts participated in small-group (species-specific) and larger-group exercises to achieve a final rank order, working from the ranked prioritized lists of horse welfare issues that had been generated in the earlier rounds. In a few places, experts also combined areas which they felt to be too similar to separate. Consensus at the workshop was considered to be unanimous where a final list was produced with the agreement of all participants.

Horse Welfare Issues
Horse welfare experts generated a comprehensive list of 84 welfare concerns from the analysis of the discussion board. These were finally classified into 9 themes in preparation for the first online survey (Table 1).

Category of Concern Specific Welfare Issue
Nutrition and management practices Obesity, e.g., due to lack of work, unrestricted grazing Inappropriate diet, in particular, low-fibre, high energy concentrates, restricted access to high-fibre forages Rise in the blanket use of supplements Undernourished (hunger) Lack of access to fresh, clean water (thirst) Feeding methods of stabled horsed contrary to their evolution (for low-energy forage) Turned out 24/7 on green grass (e.g., relative high energy, risk of obesity/laminitis) Poor field management, including ragwort, grazing too rich/poor, mud Unsuitable living environment e.g., unsafe fencing

Expert Rankings
The outcome of the first ranking process conducted by experts is shown in Table 2. Of the initial 84 welfare issues shown in Table 1, 34, 37 and 35 issues were retained into round 2 for perceived prevalence, severity and duration of the welfare issue respectively.
Horse experts had an overall low level of agreement with the rank order generated in the first round survey (Fleiss' kappa = 0.259, 0.227 and 0.243 for perceived prevalence, severity and duration respectively), and generally less than 75% of experts agreed on the top 10 highest-ranking welfare issue for perceived prevalence, severity or duration, although there was better agreement (70%-85%) on the five lowest ranking positions for all welfare issues. Experts considered that lack of understanding of horse welfare needs by owners, over-rugging, inappropriate training and handling, and delayed euthanasia were more prevalent than as scored in the first round. In addition, hunger and obesity were considered less severe, but lack of recognition of pain behaviour, delayed euthanasia and inability to perform normal social behaviours were more severe (see Supplementary Material for details). Table 2. Mean scores (with SD) and rank order for horse welfare issues that scored at least mean 3.0 or above for perceived prevalence, severity and duration of the welfare issue (issues scored using a Likert scale from 1-6, where 1 = never/none, and 6 = always/high).

Workshop Rankings
The workshop was more successful than the anonymous online surveys in achieving a consensus (unanimity was achieved), and over two days of discussion, a final welfare ranking on which the experts agreed was achieved (Table 3). Table 3. Ranked prioritized welfare issues for horses derived from the workshop for individual horses (combined severity and duration of the welfare issue in the experts' opinion) and for the UK horse population (perceived prevalence).

Rank Order
Horse Population (Perceived Prevalence) Individual Horses (Severity + Duration)

Discussion
Although the horse experts in the initial rounds of the Delphi did not achieve very high levels of agreement, a better consensus was achieved during the face-to-face discussions in the workshop, and many of the areas considered to be important in the survey were also ranked highly in the workshop. Overall concerns reflected all aspects of horse management and use, from health (biosecurity, worm burdens and delayed euthanasia decisions), owner knowledge (lack of recognition of pain behavior and lack of understanding of equine welfare needs), management (diets, weaning and social groups) and use of horses (fear and frustration, overwork, poorly fitting tack and inappropriate rider weight). The initial rounds of the Delphi were anonymous and had the advantage that individual respondents were not influenced by the decisions of others. However, a disadvantage of the Delhpi method is that it prevents live discussion where individual ideas and perceptions can be broken down, discussed and reassessed. In our study, we found that considered discussion, in a structured way in the workshop, allowed a better consensus to be achieved, building on the preliminary analyses derived in the anonymous process.
Other qualitative studies that have considered horse welfare have used varying techniques, and not all studies yielded outcomes that could be directly compared to our prioritization. However, a UK study using interviewing and focus groups of industry professionals [18] found that horses being underweight or overweight, poor foot care, internal worms and laminitis were the most frequently mentioned horse health issues. Large worm burdens, obesity and hunger all also appeared in the most important issues in our study for individual horses. Likewise, prolonged stabling, under-or over-feeding, inappropriate rugging and social isolation were identified in previous work [18] and were also considered important in our study for individual horses, and feeding of unsuitable diets was perceived to be prevalent. Use of poorly fitting tack was identified in both studies as a significant cause for welfare concern. In the study of Horseman et al. [18], the most prevalent welfare issues (identified as the perception by stakeholders to be the situations in which welfare was most likely to be compromised) were keeping horses in unsuitable environments, inappropriate use, misunderstanding behaviour, disruption of social routines (e.g., through moving yards), abandonment, transport and delayed euthanasia. Of these, issues with delayed euthanasia, lack of owner understanding, issues with use and instability of social routines were also considered by the experts in our study to be prevalent with horses in the UK. A Delphi study conducted in Canada identified lack of owner knowledge, delayed euthanasia decisions, inappropriate training or use, inappropriate feeding, lack of turnout and lack of social companions as important welfare issues at the individual horse level and issues with breeding, delayed euthanasia, biosecurity and owner ignorance as important welfare issues at the population level [17]. Finally, a study using the Five Domains Framework to assess horse welfare issues within different categories [22] concluded that the most severe impacts on horse welfare were abrupt individual weaning, feeding 100% low-energy concentrate, use of indoor tie-stalls without social contact, forced flexion (sometimes called 'Rollkur' or hyperflexion [15]), restrictive nosebands, ear twitches and transport, either individually or in a group with unfamiliar companions. A number of veterinary procedures or surgical interventions were also identified as causing severe welfare impacts [22]. However, this study did not assess the likelihood or prevalence of these issues, only the most severe within a set of specific categories.
Despite use of different qualitative methods, different selection of experts or industry professionals, and different ways of prioritization, there are a number of similar issues that have arisen in UK and Canada that may reflect general areas of concern for the welfare of horses. In our study, we specifically asked experts to address welfare by focusing on the amount of suffering caused by the welfare issue. The study of McGreevy et al. [22] used the Five Domains model, which asked experts to focus on the impact on the mental state of the horses in determining the welfare impact. Although other studies did not appear to use a specific definition of welfare to direct respondents, participants may have utilized similar thinking about the impact on equine mental state and suffering of the welfare issues that were mentioned frequently. Other studies have also identified racehorses or competition horses [17,18] and horses owned by travellers [18] as areas of particular welfare concern. In our study, although these areas were mentioned in the initial rounds of developing the lists of horse welfare concerns, they were not included in the final prioritization. Recent studies of racehorse welfare [16] and perceptions of welfare in travellers [23] have addressed these issues and suggest that some concerns may be misplaced, which perhaps reflects the lower concern for these areas in our study.

Owner Knowledge and Understanding
Of the issues that arose in all studies, owners' lack of knowledge, particularly in relation to horse behaviour, was clearly highlighted. In our study, this was identified as a specific theme in the initial development of the list of welfare concerns for horses and was also included as one of the most important issues in the final rankings for recognition of pain behaviour and lack of knowledge of horse welfare needs. Lack of owner knowledge may also contribute to other welfare issues, such as inappropriate feeding and training methods and use of restrictive or poorly-fitting tack. In their review of recreational horse management, Hemsworth et al. [24] suggest that a number of horse owner attributes might contribute to poor horse welfare, including commitment to horse ownership, income, education, knowledge of horse husbandry and attitude to horse management. Of these, levels of education and income, as well as lack of knowledge, have been shown to be associated with horses having welfare problems [24]. Hartmann et al. [25] found high agreement amongst Scandinavian horse owners that horses should be kept in groups but also observed that many horses were not kept with social companions, suggesting that factors other than just lack of knowledge, such as opportunity or financial ability to manage horses differently, may also contribute.
A number of studies have suggested that the vast majority of owners believe horses to be sentient and capable of specific emotions such as pain, fear or joy [26,27]. However, DuBois et al. [27] suggest that belief in sentience did not appear to reflect understanding of welfare issues. In a recent survey [28], horse owners could recognize at least some behaviours associated with distress but missed or incorrectly classified behavioural indicators of negative affective states as positive. Further, a number of participants in the survey indicated that they would be happy for their horse to be treated by methods they had already identified as causing distress [28], suggesting that knowledge was not the only factor involved in poor horse welfare.
The impact of owner knowledge or management responses has also been shown to be an issue with health practices as well as behaviour. Although lack of knowledge or understanding was not specifically assessed, Thompson et al. [29] found that 19% of Australian horse owners did not vaccinate their horses against tetanus, and 11% did not ensure regular dental care. In addition, McGowan et al. [30] demonstrated that owners could not identify all clinical signs of ill health in their aged equines and did not seek veterinary advice in all cases when it was warranted. These data support the view that lack of sufficient owner knowledge can contribute to poor health and welfare in horses. However, other barriers to improving horse welfare may be related to financial constraints, opportunity to change practices, habit and cultural norms around horse management.

Delayed Euthanasia Decisions
The consequences of delaying euthanasia decisions, and so increasing the risk of animals suffering uncontrollable pain or disease, was considered by experts in this study to be one of the most prevalent horse welfare issues and to cause the greatest suffering to individual horses. Delayed euthanasia was also identified as important in two other studies that have attempted to prioritize horse welfare issues [17,18], suggesting that this is widely considered an important issue for horses. This may be attributable to owners' lack of knowledge or ability to correctly identify and understand the consequences of clinical symptoms as discussed above, and may also be related to issues with income or ability to pay for horses to be euthanized. Horse slaughter at abattoirs is permitted in the UK, which may alleviate some of the financial impact of euthanasia, and has been suggested to improve horse welfare [31]. However, other issues such as emotional attachment, peer pressure and negative attitudes to death can also play a role [18]. Currently, there is little published research to allow quantification of the impact of delayed euthanasia on horse welfare, although many veterinary clinics and charities do provide guidance on recognizing a deteriorating quality of life (e.g., www.bluecross.org.uk and www.bhs.org.uk) or emergency conditions where euthanasia would be recommended.

Impact of Inappropriate Training or Use of Horses
Horses are unusual amongst other domesticated species in the degree of use and training that they usually receive, often by owners who themselves have had no specific education in learning theory or training practices. This lack of owner knowledge of equine learning or training methods was a significant concern in our study and in others [17,18,22]. There has been a recent increase in scientific studies assessing the impact of training, tack use and competition on equine welfare. For example, studies have shown significant lesions in the mouths of Finnish trotters and Icelandic horses when used in competition from the action of the bit [7,32]. The use of coercive hyperflexion of the head and neck or Rollkur, for example in dressage horses, has been shown to cause adverse behavioural [15] and physiological indicators [33], suggesting increased stress in these animals compared to working in a less flexed position. Application of the Five Domains model also identified working in Rollkur as one of the most adverse welfare impacts in foundation training [22]. As already described, a recent study [28] has shown that conflict, stress, fear and frustration were identified as present in a variety of different training and riding disciplines, including natural horsemanship and bridleless riding, and that not all horse owners perceived this to be inappropriate for horse handling. In our study, the misclassification of pain-related behavioural responses as 'naughty' was considered to be a significant source of equine suffering and suggests that owner knowledge about equine behaviour, attitudes and understanding of animal pain and beliefs about particular rising practices may need to be tackled to improve equine welfare.

Inappropriate Feeding
Concerns about feeding practices in our study were either about the quantity of feed given (i.e., that horses were fed too much and were overweight, or not enough and thus were thin or hungry) or that the method of feeding management, such as restricted forage access, might cause adverse behavioural consequences for horses.
Studies in the UK, Sweden, Italy, Germany and the USA suggest that between 25% and 45% of horses and ponies are overweight or obese [13,[34][35][36][37]. These studies confirm that this is indeed a prevalent condition as identified by experts in our study. Obesity is associated with insulin resistance, equine metabolic syndrome [38] and increased likelihood of laminitis (a painful condition of the feet that can result in euthanasia), as well as issues with heat stress, reproductive and inflammatory responses [39]. Thus, obesity is a significant risk factor for poor welfare from the painful and debilitating conditions that are predisposed. By contrast, for horses that were not fed enough food in our study, it was the negative feelings associated with lack of food, i.e., hunger, that were identified as the welfare issue, rather than specifically the consequences of being underweight. In survey studies, only 2% of horses were considered in poor body condition in Scotland [34] and 4.5% in older horses (>15 years) in England and Wales [40]. Thus, it seems that hunger from chronic lack of food does not occur at high prevalence in the UK, although the severity of the condition was the predominant issue for experts in this study. Hunger can occur because horses might be left without food for prolonged periods, even if the intention is to reduce weight in obese animals, such as managing overweight horses on a drylot [36]. Bedding investigation and bedding eating have been suggested to be indicators of feeding frustration which may be associated with hunger [41], and increased motivation to access feed is usually considered an indicator of hunger in other species [42]. Periods of food deprivation have also been linked to an increase in gastric ulceration [43] and are a risk factor for the development of stereotypic behaviour [44].
Actions to manage the weight of horses can induce hunger and/or frustration in horses, as described above. However, these emotional states may also occur when horses are fed in ways that are very different from their evolved feeding responses. In our study, feeding animals unsuitable diets for equine feeding behaviour was considered by experts to be both an important cause of horses suffering and to be prevalent. Feral horses spend 12-18 hours a day grazing [45], and their digestive system is adapted to a continuous intake of relatively poor quality forage. Feeding management of stabled horses may often utilize feed restriction (as described above), and feeding in 'meals' rather than continuous forage. A survey of UK horse owners found that 4% were continuously housed, and 92% had pasture access for an average of between 70 and 126 hours per week [46], suggesting that horses were spending between 25% and 58% of their time stabled. The majority (82%) of horses received additional forage feeds (such as hay) and concentrate feeds (86%), although the frequency of feeds was not reported [46]. The study of McGreevy et al. [22] concluded that feeding a 100% concentrate diet, even if this met the horses' metabolic needs, had the most adverse impact on welfare through its impact on behaviour and that pasture choice and feeding forage were the least likely to impact welfare. Rochais et al. [10] showed that feeding systems that encouraged a longer period of feeding were also associated with reduced stereotypic responses and more positive behavioural change in horses.

Inappropriate Environments and Social Behaviour
In our study, issues with social behaviours were considered both to be prevalent (instability of social groups) and to cause significant suffering to individual horses (inability to perform normal social interactions). Prolonged stabling and management in inappropriate environments were identified as welfare issues in the study of Horseman et al. [18]. In our study, it was the impact of housing/management on social contact that was considered the greatest issue, although these issues are often experienced together. The study of McGreevy et al. [22] also concluded that worse welfare was experienced by animals kept either indoors or restricted outdoors (tethered) without social contact and that best welfare was achieved by living outdoors with full social contact. Horses are social animals, and keeping them in groups is widely considered to be important for good welfare [47]. For example, horses have been shown to have a high motivation to maintain full, or partial social contact with another horse [48], and stereotypic behaviour is linked to single housing [49].
Various studies suggest that 83%-92% of horses are continuously managed at pasture in groups or have pasture access [25,46,50] but 4%-10% of horses may be continuously housed with little or no exercise opportunities [13,25,46]. This figure was higher for entire male horses, where nearly 40% were found to be kept alone in a Nordic study [25]. Of the stabled horses in a study of Italian and German horse welfare [13], 22% of horses had no physical or visual contact with other horses. These data suggest that a significant percentage of horses are kept in environments that may prevent social contact for at least some of the day and that a small percentage of horses may have no access to other horses for prolonged periods of time. Nearly 20% of stabled horses have been shown to display stereotypic behaviours [13], suggesting that welfare is compromised in these animals.

Conclusions
By comparing the outcomes from our data with those of other studies that have attempted similar types of assessments, we have found that there are common welfare concerns that are raised frequently with different groups, which lends greater weight to arguments that these are important issues for horse welfare. In particular, lack of owner knowledge, or application of knowledge to the management of horses, is an important welfare concern, which has also been seen in other species [21,51]. In addition, many of the other issues that have been highlighted may also stem from poor owner knowledge or the application of traditional or culturally mediated methods of managing or using horses. These include poor biosecurity practices, not recognizing pain behaviour, use of poorly fitting or restrictive tack, inappropriate training practices and keeping horses in environments that do not meet their physical, nutritional or behavioural welfare needs. Increasing numbers of studies have demonstrated that these practices can result in fear, frustration or distress in horses, and an effective means to transfer this knowledge to owners and an assessment of the barriers to implementing changes are required.
One of the limitations of a Delphi procedure, and for other techniques that rely on expert opinion, is that the outcomes reflect only the knowledge and understanding of the experts [52]. However, it is a useful technique when empirical data are not available, or when comparisons of, for example, chronic versus acute conditions are attempted. Although it uses a consensus approach and seeking experts from a range of different backgrounds [53] can improve the issue, it is still a possibility that this does not reflect reality and a different group of experts might achieve a different outcome. However, by comparing our outcomes with other studies that have used similar approaches, we have been able to overcome this issue and hence conclude that a consensus on priority horse welfare issues has been achieved.
Supplementary Materials: The following are available online at http://www.mdpi.com/2076-2615/10/4/647/s1, Table S1: Percentage agreement with the ranking of horse issues from first online survey scored by experts in the second online survey.