Investigating risk factors that predict a dog’s fear during veterinary visits

Attending the veterinary clinic is an integral part of the physical welfare of every companion dog. However, some dogs experience their veterinary visits negatively, which poses a risk of injury to the veterinary staff, their guardian (owner) and themselves during veterinary examinations. It may also influence the regularity of non-urgent veterinary appointments. To date there has been conflicting data on the proportion of dogs that are fearful during their veterinary visits. In this study, we explored the risk factors associated with fear during veterinary examination and in novel situations (including first time at the veterinary clinic) from 26,555 responses in the Canine Behavioral Assessment and Research Questionnaire database. According to their guardians, over half (55%) of companion dogs displayed some form of fearful behaviour (mild-extreme) when examined by a veterinarian, while 14% of dogs exhibited severe or extreme fear. A similar trend was observed with dogs responding fearfully in unfamiliar situations, including the dog’s first time at the veterinary clinic. Chi-squared tests showed every bivariate relationship was significant (p < 0.05). The most important predictors of fear in a veterinary examination were, in order: the dog’s breed group (27.1%), their history of roles or activities (16.7%), where they were sourced (15.2%), their weight (12%), the age of other dogs in the household (9.5%) and dog owner experience (6.3%). However, these risk factors accumulate to explain a total of 7% of variance of fear observed during veterinary examination. Results demonstrate that fear of veterinary visits is common in dogs, but that other factors (including the environment or human-animal interactions) are likely to contribute more to prevalence and severity of this problem than the demographic factors measured here. Finally, we highlight opportunities for future research aimed at facilitating less stressful veterinary visits for dogs and their guardians.


Introduction
Visits to veterinary clinics are integral to maintaining and improving the health and welfare of domestic dogs. However, the veterinary experience can also be stressful for them. It is currently estimated that between 10% and 78.5% of dogs become stressed or fearful in the veterinary clinic [1][2][3][4][5][6][7][8][9][10][11][12][13]. While Doring et al. [4] identified 13% of dogs who refused to enter the veterinary clinic, likely representative of severe fear or distress, Stanford [1] reported that 70% of dogs were unwilling to enter. Further, Mariti et al. [9] observed 29% of dogs displaying 'extreme' stress in the waiting room. Such disparity in prevalence is likely a reflection of the methodology employed. For example, variation in the behavioural and physiological measures used to assess stress or fear; the person taking the measurement (e.g., investigator, guardian, veterinary nurse, veterinarian); the locations within the veterinary clinic where stress is measured (e.g., waiting room, examination room or kennels and cages); and the context (e.g. owner present/ absent, mock/real examination). This makes an accurate estimate of the prevalence of stress or fear in dogs visiting veterinary clinics difficult to ascertain.
Negative veterinary experiences can have long-term impacts for the dog, pet owner (guardian) and veterinary staff. A North American study found that the very idea of taking a dog to the veterinary clinic can cause guardians to become stressed (26%). In fact, many guardians (38%) believe that their dog 'hates' going to the veterinarian [7]. As such, guardians want to see their veterinarian interact compassionately with their dog [14], especially when certain methods of handling and restraint can be stressful for animals [15][16][17][18]. These attitudes and experiences affect guardian decisions about which veterinarian they see and how often they attend [7]. Further, the behavioural or physiological signs of fear and distress can mirror those of pain, illness and some neurological conditions [19], making accurate diagnoses difficult. Not only can stressed or fearful dogs at the veterinary clinic injure themselves, but they also pose a risk of injury to the veterinary staff and their guardians [19][20][21]. Addressing fear at the veterinary clinic and promoting pet-friendly practice is integral to the continual improvement of companion animal welfare.
With the exception of the dog's sex [4], size [3], and the benefit of supportive guardian presence [11], our understanding of dog and guardian characteristics that may exacerbate or ameliorate a dog's fear or stress response at the veterinary clinic is limited. In order to address this shortfall, we explored the proportion of dogs that are fearful during veterinary visits from a large sample of companion dogs. To do this, we analysed two fear-related, veterinary specific questions and their corresponding behavioural subscales from the Canine Behavioral Assessment and Research Questionnaire (C-BARQ). The C-BARQ is a validated research questionnaire available online to dog guardians [20]. It has been used extensively to investigate factors that influence dog behaviour, personality and temperament in general [21][22][23][24][25][26], as well as to explore how domestication has influenced behaviour more specifically [27].
The C-BARQ has also been used to measure the behavioural effects of neutering in dogs [28,29], and to investigate the factors associated with aggression [30][31][32], trainability [33], boldness traits [34], and how training can impact dog intelligence [35], and to explore how dog behaviour or temperament can influence their health and lifespan [36], and the relationship between dogs and their owners [37]. As such, this previously collected, extensive dataset of dogs provides an opportunity to build on our understanding of how dogs experience their veterinary care.

Method C-BARQ
C-BARQ contains 100 items (questions) that have been validated into 14 subscales (factors) of dog behaviour [20]. Guardians respond on a 5-point Likert scale for how serious that behaviour is for their dog or how often it is performed, with '0' being 'none/ never' and '4' being 'extreme/ always'. C-BARQ provides guardians with clear examples of what mild to moderate fear may look like in their dog: "avoiding eye contact, avoidance of the feared object; crouching or cringing with tail lowered or tucked between the legs; whimpering or whining, freezing and shaking or trembling". Similarly, extreme fear is described as: "exaggerated cowering, and/or vigorous attempts to escape, retreat or hide from the feared object, person or situation". In this study, C-BARQ responses to two items (questions) related to fear in a veterinary context were analysed to explore the relationship between dog experience at the veterinary clinic and dog and guardian factors (see Table 1 [20]. Non-social fear contained 6 items (Questions 38, 41, 42, 44, 47 and 48], and refers to dogs that are fearful or wary of sudden or loud noises (e.g. thunder), traffic, and unfamiliar objects and situations [20]. In this study, the predictive value of factors on fear responses in the veterinary context is explored via the two veterinary specific items (Q43 and Q47) in conjunction with the greater insight provided by the two validated behavioural subscales (touch sensitivity and non-social fear).

Dog sample
Retrospective data was collected from guardians completing C-BARQ online (https://vetapps.vet.upenn.edu/cbarq/) between 2005 and 2016. Only dog breeds that were clearly identifiable and represented in at least 50 responses were included. Responses that did not include answers to the vital items -Question 43 (fear of veterinary examination) and Question 47 (fear of unfamiliar situations) -were excluded. Breeds were then categorised into breed groups via the Australian National Kennel Council (ANKC) breed list. In the event that a breed was unrecognised by the ANKC, it was categorised in accordance with the American Kennel Club (AKC) (e.g. American Eskimo dog, rat terrier, great Pyrenees, chinook, Spanish water dog). Some breeds were identified in ANKC by a different name (e.g. Belgian sheepdog as groenendael; English bulldog as British bulldog). Breeds not recognised by the ANKC, AKC or the Fédération Cynologique Internationale (FCI) breed lists (English shepherd and American pit bull terrier), were removed. Dog age, neuter age, and age of acquisition were converted into years, and dog weight converted to metric (lbs to kg). C-BARQ is available for guardians with dogs six months and over, and as such, 'puppies (<6months)' in this study refer only to dogs that were six months of age at the time of evaluation (N = 583).

Statistical Analysis
Cross-tabulations were performed between all independent variables and the outcome measures, fear of veterinary examination and fear of unfamiliar. Chi-square tests revealed that every bivariate relationship was significant, due to the large sample size. Accordingly, our analyses focused on determining the magnitude of effect sizes, and assessing the relative importance of each variable in predicting fear of veterinary examination and fear of unfamiliar, in both a univariate and multivariate context. In the first step, we assessed relative importance using two metrics within a framework called dominance analysis [38,39]. The first metric is calculated by entering each variable in isolation (e.g. simple regression), and expressing each model R 2 relatively: e.g. as a percentage of the sum of R 2 explained by all models. The second metric, lmg [40], considers the R 2 over all possible combinations of predictors: e.g. given predictors X1, X2, X3, models y ~ X1, y ~ X2, y ~ X3, y ~ X1 + X2, y ~ X2 + X3, y ~ X1 + X3, y ~ X1 + X2+ X3 are considered. Thus, a single predictor's relative contribution is assessed in terms of the drop in R 2 when it is removed, in all possible multivariate contexts. In the second step, two multiple regression models were fitted for fear of veterinary examination, fear of unfamiliar, touch sensitivity and non-social fear. The first model included all predictors. The second model included only those predictors that contributed more than 5% of the explained variance according to the lmg importance metric.
Though 5% is an arbitrary threshold for inclusion, in contrast to stepwise methods the lmg criteria is a robust variable selection method, because the entire space of possible regression models is evaluated.

Results
The sample of 26,555 valid responses was evenly distributed by dog sex (51.79% male; Table   1], and the mean dog weight was 22.78kg (median 21.60kg). The mean dog age was 4.52 years (median 4.00 years), mean neuter age was 0.84 years (median 0.48 years) and mean age acquired was 0.77 years (median 0.21 years). The majority of dogs were healthy (84.63%), neutered (75.67%), and purchased as companions (70.97%), with no specific sporting or working role. The most common reasons for neutering were birth control (25.85%), and required by breeder (23.72%). Dogs were most commonly acquired from a breeder (41.38%) and shelter or rescue (31.03%), while the least common source for dogs was a pet store (3.47%), followed by those bred by their guardians (4.08%). The majority of guardians were experienced dog owners, having had dogs previously as adults (81.82%) and/or as children (80.11%). Mixed breeds or dogs of unknown breed were the most commonly reported (27.75%), followed by working breeds (16.41%) and gundogs (15.77%).
Of the 26,555 dogs, over half (55.25%) exhibited some form of fearful behaviour (mildextreme) when examined by a veterinarian, and 14.23% of guardians reported that their dog showed severe or extreme fear during veterinary examination. Similarly, over half the dogs (57.70%) showed at least some signs of fear in new situations, including potentially the first visit to the veterinary clinic, while 11.02% of dogs displayed severe or extreme fear in unfamiliar situations. In contrast, just over a third of dogs displayed mild-extreme fear for the corresponding subscales of touch sensitivity and non-social fear (35.01% and 37.00% respectively). The mean score of fear for both items (fear of veterinary examination and fear of unfamiliar) and both subscales (touch sensitivity and non-social fear) for each of the independent variables are displayed in Table 1.

Multivariate regression of fear response
Linear regression models were used to explore the predictive importance of dog and guardian factors in determining fear of veterinary examination and fear of unfamiliar.    All factors dummy coded (0,1); beta coefficients reflect mean differences from the reference category. All dependent variables scaled from 0 (not fear) to 4 (extreme fear)

Relative importance of factors in explaining variation of fear
The relative importance of each of the factors in predicting fear of veterinary examination, fear of unfamiliar, touch sensitivity and non-social fear are shown in Table 3

Breed group
Breed group was the largest predictor of fearful behaviour at the veterinary clinic (Table 2)

Other contributing factors
The reason a dog was neutered also contributed to the variance of fear observed across the majority of variables, but did not contribute over 5% of variance toward fear of veterinary examination. Dogs neutered in order to correct behaviour problems exhibited the highest That is, while the subscales include items referring to a dog's veterinary experience, the scales also contain other items that do not. As such, the higher prevalence of fear observed in the individual items reflects the many factors within the veterinary context that may be the cause or catalyst of that fear.
The prevalence of fear in a veterinary context may be influenced by a dog's genetic predisposition to fear [41]. Godbout et al. [3] identified a small proportion of puppies (10%) that displayed extreme avoidance behaviours during a mock examination. They suggest this likely reflects a proportion of dogs that exhibit anxious behaviours through to adulthood, as a result of a genetic predisposition to an anxious temperament. This is an important area for future research and our results indicate a similar proportion of dogs with severe fear.
As a group, all predictors explained between 5 and 7% of variance in fear of unfamiliar and fear of veterinary examination. The most important predictors were, in order, the dog's ANKC breed group, the dog's employment or activity history, where they were sourced, their weight, the age of other dogs in the household, reason for neutering and guardian's level of experience of dog ownership. As such, each risk factor accumulates to set the foundation of a dog's predisposition to fearful experience in the veterinary clinic. This low effect size (e.g. 7% for fear of veterinary examination) also highlights the integral role other factors (e.g. environmental, previous experience or human-animal interactions) play in determining the severity of the fear response. This mirrors findings reported in previous studies investigating neuter age and stranger-directed aggression [28] or fear-related behaviours [29], source of acquisition and non-social or stranger-directed fear [42], and litter size and personality [43].
Indeed, Casey et al. [44] suggest the factors associated with human-directed aggression explain a similar amount of variance (<10%), and emphasise that individual experience is likely of much greater importance in determining behaviour.
A dog's breed group was the best predictor across all the dependent variables. Toy dogs, hounds and mixed breeds were most likely to have the highest scores of fear during veterinary examination in comparison to other breed groups, whereas mixed breeds and hounds predicted higher fear of unfamiliar situations. The same trend was observed in touch sensitivity and non-social fear. Further, according to their guardians, mixed breeds were generally more likely to be fearful of noises in comparison to other breed groups [45]. In contrast, in a cross sectional study, dogs in utility and hound groups were more aggressive to family members than mixed breeds [44]. Although this may simply highlight the difference between aggression and fear, it suggests context is important in determining behaviour -the same dog may react fearfully to an unexpected noise but aggressively toward an unfamiliar person. Additionally, some individual breeds (e.g. dachshunds, Chihuahuas or Jack Russell terriers) have been associated with an increased likelihood of showing aggression toward their guardians and strangers [30]. Considering a dog's ANKC breed group explained the largest proportion of variance of fear observed, providing additional time within the veterinary consult or extra support to guardians of specific breeds may be valuable in reducing fear in the veterinary clinic.
In the present study, dogs previously employed in working, breeding and showing roles had lower fear of the veterinarian and fear of unfamiliar, while companion animals were most likely to show high levels of fear in the same contexts. The same pattern was observed in touch sensitivity and non-social fear. This is supported by dogs in these roles also having the lowest touch sensitivity scores. The roles that dogs are employed in can influence aspects of their personality and behaviour. Indeed, Lofgren et al. [25] suggest that Labrador retrievers purchased as companions or employed in a gundog role showed higher human and object fear than Labradors that were show dogs, while companion Labradors exhibited greater noise fear than those that were gundogs or show dogs. The reduced risk of fear of veterinary exam and unfamiliar in dogs employed in breeding or working roles may reflect an increased familiarity with procedures associated with veterinary care (i.e. grooming, handling or restraint). As such, we suggest that appropriate handling and grooming practice for companion dogs is equally as important as basic manners training and socialisation in reducing fear in the veterinary context.
The source of acquisition of the dog was also a predictor of fear response in a veterinary context. Dogs acquired from friends or relatives, pet stores shelters and rescues, or as strays were most likely to be fearful during veterinary visits or have high touch sensitivity and nonsocial fear. In contrast, those bred by their guardian exhibited less fearful behaviour. This reflects a similar finding by Blackwell, Bradshaw and Casey [45], that dogs bred by their guardians less likely to show fear responses to noises than dogs from other sources. Further, puppies from pet stores had increased risk of behavioural issues in comparison to puppies purchased from breeders [42,46]. Indeed the quality of maternal care can have long term behavioural fallout and alter the physiological responses to stress [47]. Therefore, the puppy's experience for the first several weeks of life requires careful consideration when investigating the causes of fear during veterinary visits.
Dog size also influences the overall variation of fear observed during veterinary examinations, with lighter (and therefore generally smaller) dogs (<22kg) predicting higher fear scores than heavier dogs (>22kg). Smaller dogs have also been found to be more vocal in the veterinary clinic than larger dogs [3], and are associated with aggressive and excitable, and anxious and fearful behaviour in comparison to larger dogs [48]. However, we highlight that while both breed group and dog weight were predictors of fearful behaviour, the statistical models estimated the effect for breed when controlling for size and vice versa. As such, the extent to which each of the factors contribute to fear of veterinary examination and fear of unfamiliar individually is unknown, and likely inseparable, considering artificial selection for breed phenotype includes size.
The ages of other dogs in the household also contributed to fear responses. Dogs living without conspecifics were most likely to exhibit higher fear in every dependent variable.
Lower fear scores for dogs in multi-dog households may reflect social learning. As with the benefit of a positive guardian presence in reducing fear [11], dogs that attend the veterinarian with another familiar, confident dog may take their social cues from that dog and be less fearful. However, whether all dogs in the home attend the veterinarian together, or whether a reduced risk of fear at the veterinary clinic results from some social interaction that occurs at home is unclear. Further, Meyer and Forkman [49] report a link between high social fear in dogs and high guardian emotional closeness with their dog, and so perhaps beneficial guardian or conspecific presence is conditional on the type of attachment. As such, the positive and supportive presence of familiar conspecifics, or a guardian, may help reduce fear during veterinary visits, although further investigation is required.
Guardian level of experience was another contributing factor for fear of veterinary examination and touch sensitivity. Guardians that had never owned a dog previously were more likely to have dogs that exhibited higher fear responses. While over 25% of guardians are able to identify obvious signs of stress in dogs [50], Flint et al. [51] suggest a lack of experience in dog behaviour or attendance at dog training classes is associated with guardians being less likely to identify fear correctly. This suggests a potential for fear to be underreported in dogs with inexperienced guardians. It also constitutes a significant risk to companion animal welfare as accurately recognising fear is essential in reducing fear in the veterinary context [52][53][54][55][56]. Overall, the predictive weight of guardian experience relative to all other demographic factors is small, although it highlights the importance of guardian education focusing on dog training, behaviour and communication.
Guardians that neutered their dog in order to correct behaviour problems had dogs that were more fearful in new situations, and had higher touch sensitivity and non-social fear. This is supported by Lind et al. [13] who found that dogs with guardian-reported behaviour problems were also rated as more stressed during veterinary visits by the guardian, and the (blinded) test leader. However, as the study design is cross-sectional it may be that dogs that already have behavioural problems, and hence are neutered, are then more likely to be touch sensitive and have non-social fear. Further longitudinal studies are necessary to investigate the influence of neutering on dog behaviour.
While the C-BARQ is a validated questionnaire that clearly describes the behaviour of interest, it is still limited by measurement errors, including: conservative reporting (if guardians are predisposed to report on items like problem behaviours optimistically); guardian interpretation of the items or behaviours; guardians not noticing behaviour during previous veterinary visits, and; time since last veterinary visit. Further, the ability of guardians to accurately identify fear in their own dogs is questionable [9,50,51]. Indeed, while Flint et al. [51] found training in recognising fear in dogs resulted in guardians being more likely to correctly identify mild/ moderate and high/extreme fear, they observed no corresponding change in reporting on the guardian's rating of their own dogs. Conversely, while the current study's sample size is large, it may reflect responses from guardians that actively seek to know more about their dog's behaviour, and so, may represent responses from those who are more aware of their dog's fear or body language. Further, we suggest that the proportion of dogs exhibiting fearful behaviour in the context of unfamiliar situations may be over-representative of experience in the veterinary clinic, as guardians may be reporting on fearful behaviour that occurs in unfamiliar circumstances outside of the veterinary clinic. Future research into dog experience in the veterinary context should corroborate C-BARQ responses for dogs who have recently visited a veterinarian with physiological measures of fear or distress and objective observations.
Overall, it is important to emphasise that the proportion of dogs negatively experiencing their veterinary visits is under-represented by C-BARQ respondents. The items (fear of veterinary examination and fear of unfamiliar situations) within C-BARQ explicitly reflect fear responses only, with no corresponding items for aggression. While several aggression items do refer to grooming or handling by an unfamiliar person (Q14, Q21), they do not expressly mention the veterinarian or veterinary clinic and so were not included in analysis in this study. Aggressive behaviour in the veterinary clinic is also a very real risk for dogs distressed during their veterinary care [55]. As such, it is highly likely guardians with dogs that behave aggressively at the veterinary clinic are not represented in the proportion of dogs experiencing distress during veterinary examination or in unfamiliar situations.

Conclusion
The results from the present study indicate that around half of companion dogs are experiencing some level of fear or stress when receiving veterinary care. Together, these factors predict approximately 7% of the variation of fear observed during veterinary examinations, and 5% of fear of unfamiliar situations. That is, while they play an important role in determining dog experience in the veterinary context, other factors (e.g. environmental set up or human-animal interactions) are much stronger predictors of a dog's fear in the veterinary clinic. The cause of the fear in dogs visiting veterinary clinics is still yet to be fully explored. For example, is fear a response to a previous negative experience in a clinic, or are