An online cross-sectional questionnaire was distributed to veterinary professionals who handle dogs routinely at veterinary practices across Canada and the United States. Inclusion criteria for participants required individuals be 18 years of age or older, a current resident of Canada or the United States, and an actively working licensed veterinarian, or veterinary staff member (either licensed or unlicensed) who handles dogs during routine health examinations. We used convenience, snowball sampling via advertisements distributed on social media. This sampling technique has been demonstrated to reach targeted populations who are challenging to access (45), such as veterinary professionals. Email invitations were also sent to various Canadian and American veterinary organizations (e.g., Canadian and American Veterinary Medical Associations). The questionnaire was advertised and available from October 2023 to February 2024. Participation was voluntary and anonymous, and participants were provided with informed consent before accessing the questionnaire. The questionnaire took approximately 15 min to complete. Participant responses were not connected to any identifying information to minimize potential social desirability bias.

The online questionnaire was created using Qualtrics survey software (Qualtrics LLC, Provo, UT, United States). The questionnaire was comprised of 26 questions, categorized into five sections: (1) participant information and clinic experience: gender, age, graduation year, Ten Item Personality Index, staff role (veterinarian, non-veterinarian), practice type (small animal, mixed animal, emergency clinic), dog bite history; (2) participant professional quality of life: using the ProQOL scale; (3) general examination practices: provision of treats, stress-reducing handling certification, exam location, dog approach and handling upon entering the examination room, and response to dog struggle during restraint, (4) perception of whether certain factors influence handling of dogs during a routine examination and ranked importance of certain examination factors; and (5) frequency of using 14 dog handling techniques and tools. For the full list of questions provided in the questionnaire, see the Supplementary material.

The Professional Quality of Life (ProQOL) scale was used to measure participants’ subscale scores of compassion satisfaction (CS), burnout (BO), and secondary traumatic stress (STS). Each subscale includes 10 items with a 5-point Likert scale (1 = never; 5 = very often; 40). The scale was adapted to reflect the work of veterinary professionals. For example, “I find it difficult to separate my personal life from my life as a helper” was adapted to “I find it difficult to separate my personal life from my life as a care provider.” The ProQOL scale is a widely used validated measure of the positive and negative aspects of working with people who have experienced stressful or traumatic events (38), such as veterinary professionals. Personality traits of participants were measured using the Ten Item Personality Inventory (TIPI). This is a 10-item measure of the Big Five personality dimensions (Extraversion, Agreeableness, Conscientiousness, Emotional Stability, and Openness), including 2-items for each dimension, and each item is assessed on a 7-point scale (1 = disagree, 7 = strongly agree; 48). Participants were also asked to rate their level of agreement on whether the following factors influenced their handling of dogs during a routine examination: veterinarian instruction, patient behavioral history, dog breed, size, and age, owner presence, staff comfort level, and time to complete the appointment. They were also asked to rank the following five factors in order of importance (1 = most important, 5 = least important) when conducting a routine dog examination: completing the appointment on time, client satisfaction, staff safety, minimizing stress for dog patients, and completing all examination components.

The frequency of using 14 dog-handling techniques were assessed for small (under 35 lbs) and large (35 lbs. or greater) dogs using a Likert-scale (always, often, sometimes, rarely, or never). Participants were asked to report their frequency of using each handling technique if the dog was calm (patient is relaxed, no signs of aggression or fear-related behaviors), fearful (patient shows fear-related behaviors, such as lowered posture, ears back, tail tucked, whimpering or whining, shaking or trembling, attempts to hide or escape), and aggressive (patient showing fear-related behaviors while showing aggression, such as baring teeth, attempting to bite, growling, lunging). Participants were asked to separately report their responses for small and large dogs, as dog size was hypothesized to influence the choice of handling techniques used and some restraints include minor adjustments to accommodate dog size. For example, the image depicting full-body restraint included two handlers, compared to one handler for a small dog (21). The 14 handling techniques (including the written descriptions and images) used in this questionnaire were based on previous research by Carroll et al. (21) exploring dog owner perceptions of handling techniques used during routine examinations. The types of restraints assessed range from full-body restraint (i.e., placing the dog in lateral recumbency with the front and hind legs held), secure restraint (i.e., holding the dog’s abdomen securely with one hand and neck with the other hand), minimal restraint (i.e., hands are placed on each side of a dog’s shoulder allowing some movement of the dog’s body and limbs), and chemical restraint (i.e., sedative or anesthesia). The use of different restraint equipment was also assessed, such as basket and soft muzzles, dog masks (i.e., mask placed over the dog’s eyes and clipped behind the head to reduce visual stimulation), and Elizabethan collars (i.e., cone shaped collar that reduces head movement and mouth access). For full details on the 14 different handling techniques assessed for each dog demeanor, refer to the questionnaire found in Supplementary material.

All analyses were performed with Stata Statistical Software v.15.1 (StataCorp, College Station, TX, United States).

A total of 31 variables were used for analysis. Variables were related to participant information and clinic experience (12 variables), ProQOL (3 variables), general examination practices (3 variables), and perceptions and ranking of importance of certain examination factors (13 variables). Referent categories for categorical variables were selected based on the most common response, or biological plausibility specific to each variable. For instance, for staff role, the biological plausible referent chosen was non-veterinarians, which was more common than veterinarians. To reduce the number of variables tested for analysis, related variables were collapsed into composite variables. For example, the variable “non-veterinarian” was created from the following variables: veterinary technician (licensed), veterinary technician (non-licensed), and veterinary assistant. As these variables were all directly linked to one survey question regarding staff role, these variables were highly conceptually related and therefore a reliability analysis was not deemed necessary. Variables assessed with a Likert-scale of agreeance (strongly agree, agree, disagree, strongly disagree) were collapsed into categories of “agree” and “disagree” for ease of interpretation. Though dichotimzation of this variable can reduce the sensitivy of models to detect variation in frequency of handling technique use, this approach aligned with our aim to evaluate factors associated with the use (vs. non-use) of minimal and full-body restraint and has been applied in similar cross-sectional studies (36). During data cleaning, questions with “other” options were cross-referenced with existing options to reduce misclassification bias. Responses suspected of being from automated bots were identified and removed based on speed of survey completion and open-text anomalies, such as non-sensical text answers (46, 47). Further, in efforts to reduce the risk of over-fitting, categories with very low response counts were collapsed or removed. Despite these efforts, these analyses should be considered exploratory, and the findings will require validation in future studies with larger samples.

The frequency of using minimal (lowest level of restraint) and full-body (highest degree of physical restraint and aversive for dogs) were selected as they represent disparate degrees of restraint (29, 32). To appropriately conduct logistic regression, the frequency of using minimal and full-body restraint were consolidated to create a binary outcome with two levels, “used” (always, often, sometimes) and “not used” (rarely, never). Participants reported their frequency of using each technique separately for small and large dogs, so models were created for each dog size. This resulted in a total of eight logistic regression models to assess the impact of the explanatory variables on the following outcomes: (1) use of minimal restraint during examination of fearful dogs, (2) use of minimal restraint during examination of aggressive dogs, (3) use of full-body restraint during examination of fearful dogs, and (4) use of full-body restraint during examination of aggressive dogs.

Linear relationships between continuous independent variables (CS, BO, and STS scores, and TIPI scores) and the outcome variables (minimal and full-body restraint use) were assessed using locally weighted regression curves and testing significance of the quadratic term (48). The relationship between the continuous explanatory variables and dependent variables were neither linear nor quadratic, and therefore, these variables were categorized (48). ProQOL scores were categorized based on established scale cut-points by Stamm (38), as low (range cutpoint), medium (range cutpoint), or high (range cutpoint), and TIPI scores were categorized as either below or above population norms (Extraversion: 4.44, Agreeableness: 3.23; Conscientiousness: 5.40, Emotional Stability: 4.83, and Openness: 5.38; 48), as recommended by scale developers (49). Correlations between all explanatory variables were tested, and none were found to have a correlation coefficient of >70%. Univariate analysis was performed to test the association of each independent variable against each outcome variable. Independent variables with a liberal p-value of p ≤ 0.20 were retained in the model to allow potentially meaningful variables within the model (48). Each multivariable model was then built using a forward stepwise selection process where only significant explanatory variables (p < 0.05) were retained in the full model. Confounders were based on biological plausibility and were identified as an explanatory variable that caused >30% change in the coefficient of another variable in the model. This threshold was chosen as a conservative criterion for identifying covariates with substantial influence on other variable coefficients (48). Biological plausibility was assessed on potential influence on handling decisions, informed by previous literature on individual characteristics and animal handling (36, 42). When confounders were identified they were included in the model to account for confounding bias. To achieve parsimony, only significant and biologically meaningful interactions were included if they resulted in lower Bayesian information criterion (BIC) values. BIC was used to determine the appropriate model for the given dataset. Results were reported as the odds ratios (OR), where OR greater than one indicates higher odds (possible causal effect) and OR less than one indicates lower odds (possible protective effect). With a pre-determined Type I error rate set at 5%, 95% confidence intervals were also reported for each finding. Given the cross-sectional design of this study, the term “risk factor” is used descriptively to refer to variables that are statistically associated with the use of minimal or full-body restraint. These associations do not imply causality.

A total of 920 responses were collected from the questionnaire, and 691 complete responses were retained for analysis after removing 229 suspected bot responses (46) and non-qualifying participants based on eligibility criteria. Most participants resided in the United States (78.3%, 541/691), identified as female (62.2%, 429/690), and were 25–44 years of age (77.4%, 531/686). Participants were non-veterinarians (67.7%, 454/671) and veterinarians (32.3%, 217/671) and work at small animal practices (52.4%, 361/689), mixed animal practices (30.3%, 209/689), emergency clinics (15.2%, 105/689), and/or other, such as shelter medicine (2.0%, 14/689). Most participants reported having a stress-reducing certification (86.9%, 544/626), which included certifications (86.9%, 544/626), which included certifications such as Fear Free Certified Veterinary Professional, Sophia Yin Low Stress Handling^® Silver Certified, and Karen Pryor Better Veterinary Visits. For full details on participants refer to Supplementary Table 1.

The frequencies of handling methods did not vary substantially by dog size. Therefore, only the frequencies for large dogs are reported, as this pattern was consistent across all explanatory variables in the descriptive data. Participants reported using minimal restraint most frequently for calm dogs (82.7%, 566/684), followed by fearful (73.1%, 499/683), and aggressive dogs (51.9%, 352/678; Figure 1). In contrast, full-body restraint was most commonly used for aggressive dogs (68.6%, 465/678), followed by fearful dogs (63.9%, 434/679), and calm dogs (58.5%, 400/684).

The general handling of dogs during routine veterinary examinations varied among participants (Figure 1). Minimal restraints were commonly reported for use on calm dogs; however, it was not uncommon for restrictive techniques (e.g., sitting full-body restraint) to also be reported. Fearful dogs were frequently reported to be handled using minimal and secure restraints, and for aggressive dogs, full-body and secure restraints were mostly used (Figure 1). Most participants use chemical sedation to restrain fearful (72.8%, 477/655) and aggressive (84.9%, 556/655) dogs. Participants indicated using various types of tools to restrain dogs of all demeanors, including dog masks (calm: 42.8%; aggressive: 55.4%) soft (calm: 48.8%; aggressive: 86.5%) and basket muzzles (calm: 48.5%; aggressive: 80.1%), and Elizabethan collars (calm: 54.9%; aggressive: 68.5%); however, all tools assessed were more frequently reported handling fearful and aggressive dogs and less for calm dogs. Complete data on the frequency of using the 14 handling methods on calm, fearful, and aggressive dogs is presented in Figure 1. When first entering an examination room, most participants reported that they allow the dog to explore the examination room before initiating handling (80.3%, 553/689). In addition, the majority indicated that they approach the dog indirectly (e.g., by crouching or kneeling and not facing the dog directly) rather than using a direct approach (e.g., standing or walking directly towards the dog; 19.0%, 130/685). Responses to a dog struggling during restraint varied. However, most participants reported that they release the restraint and allow the dog to calm down before reattempting (68.8%, 471/685). Less commonly, participants report either applying a more restrictive restraint or tool (15.6%, 107/685) or tightening their grip on the current restraint applied (15.6%, 107/685).

Factors associated with the use of minimal and full-body restraint on small and large dogs during routine examinations were identified from logistic regression analyses and are reported in Tables 1–4.

Handling decisions for aggressive dogs appear to be influenced by veterinary staff’s perceptions of safety and past experiences, including previous bite injuries. Participants who prioritized staff safety or had been bitten were less likely to use full-body restraint on fearful or aggressive dogs, and those unsure about having been bitten were more likely to use minimal restraint on aggressive dogs. In situations where the risk of injury is high, such as when dogs are aggressive, minimal restraint is advisable to avoid increasing arousal and aggression (24), thereby reducing the likelihood of injury. Although no research to date has directly examined the relationship between restraint level and injury risk, existing recommendations are primarily based on expert opinion and anecdotal evidence. Our findings suggest that prior experiences with injury may heighten risk perception among staff, leading them to adopt a more cautious approach, such as minimizing physical contact through the use of minimal restraint. While full-body restraint was frequently reported and may be perceived to enhance safety through increased control of movement, our results suggest that prior injuries may alter this perception and practice. Full-body restraint is associated with a greater likelihood of dog fear escalating to aggression (17, 22, 24, 25), and potentially, also higher risk of bite injury (23, 26). Experiencing a dog bite may lead veterinary professionals to recognize the association between high restraint and dog aggression, potentially motivating them to seek training in, or more consistently apply stress-reducing handling techniques. Further research should explore how factors, such as bite severity, frequency, antecedents, and the context of the injury shape risk perception, examination priorities (e.g., staff safety, appointment efficiency), and handling decisions.

Participant perceptions of examination factors, such as completing all examination components and managing available time, also influenced dog handling practices. Specifically, staff who prioritized completing all examination components were more likely to use full-body restraint on large aggressive dogs but less likely to do so for small fearful and aggressive dogs. These findings may reflect the belief that full-body restraint is more appropriate or practical for handling aggressive large dogs due to their size, whereas less restrictive restraint may be perceived as sufficient for small dogs. Also, those who did not perceive available time as an important factor influencing handling decisions, were less likely to use full-body restraint on fearful dogs. This may suggest that without time constraints there is more flexibility in using less invasive techniques with fearful patients, whereas staff who feel rushed may rely on higher levels of restraint to adhere to a schedule and may avoid stress-reducing strategies they perceive to be inefficient and/or prolong appointments. This aligns with previous research that identified time constraints and high workloads as barriers to implementing stress-reducing practices in veterinary settings (4, 35). However, preliminary evidence suggests that these practices (e.g., minimal or no restraint) may not prolong appointments (20) and can improve ease of examination and promote positive experiences during visits (3, 5). These findings may not generalize to all practice settings and patients, and further research is needed to evaluate the time efficiency of stress-reducing approaches in a variety of real-world contexts. Moreover, while appointment structure and staffing support are important factors, we recognize that these strategies alone may not address the needs of all patients, particularly those with behavioral concerns or histories of aggression, where behavioral pharmacology and referral to behaviorists may be needed. Further, this warrants a need to explore how appointment structure and volume influences handling decisions. Education and training on effective and practical ways to implement stress-reducing strategies, catered to practice types (e.g., emergency, general practice) may also support the provision of comprehensive, efficient, safe, and lower stress veterinary care. Despite most participants holding stress-reducing certifications, barriers such as appointment structure, high workloads, and perceptions of efficiency may hinder the use of minimal handling in clinical practice, in some cases.

Owner presence during an examination also influenced handling decisions. Veterinary staff who felt that an owner’s presence did not impact their handling decisions were less likely to use full-body restraint on fearful and aggressive dogs, suggesting that they may feel less pressured by clients and more inclined to use less restrictive techniques and focus on other factors, such as the dog’s behavior. In contrast, those who report that owner presence impacts their handling decisions may use greater physical control of the dog (e.g., through full-body restraint) to ensure safety and comprehensive examinations; however, this approach conflicts with clients’ preferences. For instance, research shows that dog owners prefer minimal over full-body restraint, particularly if their dog is fearful (21, 52, 53), and they perceive full-body restraint as excessive, inappropriate, and stressful (52). Additionally, some research suggests that owner presence reduces behavioral signs of dog stress during examinations (18, 54), while others suggest owner presence may increase dog stress if owners have negative interactions with their dog during the appointment (e.g., verbal punishment or aversive handling), or if the owner is highly stressed during the appointment (17, 55, 56). It is possible, that veterinary staff who perceive or anticipate dog behavior to worsen with owner presence, may rely on higher levels of restraint. Future research is needed to continue exploring how factors like owner presence, staff priorities, appointment pressures, and safety concerns influence handling decisions to identify approaches that optimize safety, minimize dog stress, and align with client preferences while enabling comprehensive care.

Having a stress-reducing certification and using recommended strategies for decreasing dog fear (e.g., provision of treats) were associated with increased use of minimal restraint on fearful and aggressive dogs, reflecting the application of stress-reducing principles (16, 17, 22, 24). A generational trend was also observed, with younger participants being more likely to use minimal restraint on fearful dogs, and those who graduated during the rise of Low-Stress Handling^® (22) being less likely to use full-body restraint on fearful dogs. This trend may reflect differences in curricula, as recent graduates or younger veterinary professionals may have had more education on stress-reducing handling principles and its benefits for staff and patients.

Participant gender also influenced handling decisions, where men were more likely to use minimal restraint on aggressive dogs. While our study did not assess underlying psychosocial variables, one possible explanation is that gender-related differences in risk perception may influence handling choices. Previous research has identified gender differences among veterinarians in areas such as client communication (57), and exposure to occupational stressors, such as high workloads (58). It is possible that women may perceive a higher risk of injury and opt for more restrictive handling when encountering dog aggression. However, this remains speculative and further research should directly examine how psychosocial factors, such as perceived risk, workload demands, performance pressures, work-life balance, and other systemic challenges shape handling decisions across genders.

Staff role and practice type influenced the use of full-body restraint, with increased use reported by support staff (e.g., technician or assistant). Veterinary technicians and assistants may use more restrictive handling potentially due to limited exposure to stress-reducing practices during their education/training or as a result of veterinarians delegating certain handling techniques to support staff. Regarding practice type, participants working in emergency clinics were less likely to use minimal restraint on aggressive large dogs. Emergency staff may be less likely to implement minimal restraint due to frequent exposure to highly distressed and possibly aggressive dogs and thus may opt for restraints that allow for more control of the dog’s movement to apply necessary treatment. However, emergency staff were also less likely to use full-body restraint on aggressive small dogs, which may suggest that handling techniques used in emergency settings may depend on dog size, potentially due to perceived risk differences (35). We did not examine differences in the implementation of stress-reducing techniques across practice types. Thus, future research should explore how contextual factors shape handling decisions in different clinical environments.

Personality traits and professional quality of life (ProQOL) influenced the use of full-body restraint. Specifically, the use of full-body restraint on fearful dogs was more common among participants with low to moderate secondary traumatic stress (STS), than those with high STS. Individuals with high STS are typically overwhelmed by a negative experience at work, characterized by fear, and may experience symptoms such as exhaustion, sleep disturbances, and avoiding activities that are traumatic triggers (38). Thus, veterinary professionals who experience high STS may be more cautious about exposing themselves to potentially stressful situations or witnessing their patients in distress, such as dog fear or aggression that can result from full-body restraint (32), which is anecdotally physically and emotionally demanding on handlers. This aligns with recommendations for countering STS, which includes changing caseload, work environments, or introducing other safety measures (38). Thus, these individuals may avoid practices that induce fear for themselves or their dog patients as a way to cope or protect against further stress. In contrast, those with low or moderate STS may not feel the need to avoid such stressors when making handling decisions regardless of the potential stress elicited. Burnout and compassion satisfaction were not associated with handling practices, possibly due to limited variability in ProQOL scores. Alternatively, handling practices may remain unaffected despite experiencing BO or STS, due to unmeasured factors such as resiliency. or the ability to cope with stress (59). Resilience may enable professionals to manage their responses in stressful situations, such as when handling an aggressive patient. Similarly, Perret et al. (39) found no relationship between poor veterinarian mental health and low client satisfaction, suggesting client interactions may remain consistent even under psychological strain. Compassion satisfaction can help meditate compassion fatigue (60); thus, any effects that compassion fatigue (i.e., BO and/or STS) may have on patient interactions may be mitigated by having high satisfaction from their work. Categorizing veterinary professionals based on combinations of their ProQOL scores and examining these scores with handling practices (e.g., frequency of using minimal and full-body restraint) may capture other trends in animal handling.

Regarding the personality traits assessed, participants with lower extraversion scores were less likely to use full-body restraint on aggressive dogs. As extraversion reflects an individual’s sociability and positive emotionality, such as enthusiasm (61), results may suggest that less extraverted individuals may experience greater discomfort when applying higher levels of physical restraint. The relationship between personality, well-being, and patient handling is likely complex, and though the current study provides preliminary insights, more research is needed to further explore the role of these factors in shaping handling decisions.