Use of scruffing and scruffing tools (eg, clipnosis clips) to immobilise cats is contentious, and cat handling guidelines vary in recommendations regarding these techniques. The current study examined whether cats show negative responses to the following restraint methods: (1) scruff (n=17), (2) clip application to the dorsal neck skin (n=16) and (3) full body (a known negative; n=19). Each cat was also handled with passive restraint (control) for comparison. During handling, cats were examined for behavioural (side/back ear positions, vocalisations, lip licking) and physiological (pupil dilation ratio, respiratory rate) responses. Full-body restrained cats showed more negative responses than passively restrained cats (respiratory rate: p=0.006, F3,37=4.31, p=0.01; ear p=0.002, F3,49=6.70, p=0.0007; pupil: p=0.007, F3,95=14.24, p=0.004; vocalisations: p=0.009, F3,49=4.85, p=0.005) and scruff-restrained cats (pupil: p=0.009; vocalisations: p=0.04). Clip restraint resulted in more negative responses than passive (pupil: p=0.01; vocalisations: p=0.007, ear p=0.02) and scruff restraint (pupil p=0.01; vocalisations: p=0.02). No differences were detected between full-body restraint, known to be aversive, and clip restraint. Full-body restraint and clip restraint resulted in the greatest number of negative responses, scruffing resulted in fewer negative responses and passive restraint showed the least number of responses. We therefore recommend against the use of full-body and clip restraint, and suggest that scruff restraint should be avoided when possible.
Statistics from Altmetric.com
Laboratory, shelter and owned cats are handled and restrained for many reasons, including routine health examinations, necessary diagnostic procedures, and medication administration. A recent study by Moody et al 1 suggests that scruffing is commonly used when handling cats for routine examinations and procedures in veterinary clinics throughout Canada and the USA. Scruffing involves grasping a cat’s skin and fur at the back of the neck using one hand held in a fist. Scruffing tools (eg, binder clips, ‘clipnosis’ clips) are argued to perform the same function as scruffing, bringing together the fur and skin on the back of a cat’s neck, resulting in a constant pressure. Clips have been recently introduced to veterinary clinics for use during handling, having previously been used in a variety of species (including cats), to induce an immobilisation response in a laboratory setting.2 Generally, this restraint method involves applying one or two specifically designed clips (eg, Cat scruffier clip; Campbell pet company, Washington, USA and Clipnosis gentle calming clip; Fairport Harbor, Ohio, USA) or 2-inch binder clips,3–5 to the loose skin at the back of the cat’s neck.
Applying pressure in the dorsal neck area has been used in laboratory species (mice, rats, rabbits, cats) to imitate both the maternal carrying of immature animals and physical restraint during predation. It is thought to induce an immobilisation response referred to as the dorsal immobility reflex.2 In nature, this antipredator behaviour is thought to be beneficial to young animals to facilitate carrying, and to prey species to help reduce the risk of continued attack 2 Cats naturally exhibit immobilisation when carried by their mother as kittens and during mating when the male grips the neck of the female with his mouth.6 The dorsal immobility reflex in mice is associated with a physiological response that includes both immobility and a calming effect,7 although this has not been examined in cats. Together, these natural, functional examples have been employed to justify the use of scruffing and clips during cat handling.
Overall, recommendations are divided around the use of manual scruffing and scruffing tools (eg, binder clips and clipnosis clips) when handling cats. Some guidelines recommend against use,8–10 while others suggest use on a case-by-case basis.3 11–13 To date, two studies have examined cat responses to clips or scruffing. These studies suggest that most cats show inhibition of behaviour (eg, spontaneous or manual acceptance of a lateral decubitus position, during application) with clip restraint.4 5 One study found that cats restrained by scruffing had a higher heart rate and pupil dilation than clip-restrained cats and that some cats displayed purring and kneading during restraint with clips5; based on these findings, the authors concluded that clips do not induce more stress than manual scruffing.
The current study used previously validated behavioural responses (back and side ear positions, lip licking) and physiological changes (pupil dilation, respiratory rate) responses indicative of negative affective states during handling14 to examine cat responses to: (1) scruffing, (2) clips and (3) full-body restraint. All cats were also assessed during a passive restraint technique, which was used as a neutral control for comparison. Manual full-body restraint is a known negative and psychological stressor: immobilisation by manual restraint evokes escape behaviour (struggling,8 15 sympathoadrenal activation)16 and is used to elicit a stress response in a variety of species,17 including cats.18–20 Negative vocalisations (growling, hissing, meowing) were also examined since increased vocalisations, along with fear and defensive responses, have been seen in cats during electrical stimulation of the amygdala and adjacent structures in the forebrain.21–24
Before testing, cats were categorised as friendly or unfriendly based on responses to a stranger. Stranger-directed responses were examined because cat responses to an unfamiliar person vary and are influenced by factors such as genetics, past handling experiences25–28 and socialisation experiences.29 30 We hypothesised that cats categorised as unfriendly would find all handling methods more aversive in comparison with friendly cats. We also hypothesised that handling methods involving increased restraint would be more aversive. This hypothesis predicts that cats undergoing scruff and clip restraint would show an increase in negative responses in comparison with passively restrained cats.
Materials and methods
The current study used 55 non-pedigree shelter cats (22 neutered males, 33 spayed females) from an animal shelter located in Ontario, Canada. At the time of testing, all cats had been housed in the shelter for at least 5 days, were singly housed and deemed healthy by a licensed veterinarian.
Of these 55 cats, two females displayed aggressive behaviour during restraint and were removed from the study; both had been assigned to the full-body treatment group. In addition, one male (clip treatment group) was discovered to have a urinary tract infection at the time of testing, and his data were removed from the study. Thus, 52 cats were included in the final data set: 21 males and 31 females, with an average age of 4.6 (SD ±2.4) and range of 1–10 years.
Experimental testing for each cat included the following two test procedures, which were performed consecutively: (1) an unfamiliar person test and (2) restraint. Testing for both procedures took place in the shelter’s clinic, located in a separate building from where the cats were housed. Testing took place during the light period between 09:00 and 18:00, over seven non-consecutive days.
Unfamiliar person test
An unfamiliar person test adapted from McCune31 was used to categorise cats as either friendly or unfriendly based on their responses to a stranger. Testing consisted of two phases: (1) willingness of the cat to approach a stranger and (2) responses to being approached and petted by a stranger. The experimental set up (figure 1) and procedure used are identical to that in Moody et al,14 section 2.2.1, except that testing took place in a room with a perimeter of 9.49 m. After this test was completed, the cat was placed into a holding cage, identical to their home cage, for 2 minutes before the restraint phase commenced.
Cats were restrained on a table (Lifetime; Utah, USA; 178 cm length×76.2 cm width×91.4 cm height), with rubber mats (43 cm×91.4 cm, Model 090500, Rubbermaid; Newell Rubbermaid, Ontario, Canada) to minimise slipping. Three research personnel were present during testing: the main investigator performed all handling and two research assistants moved cats through the different stages of testing, assisted with restraint when necessary, and collected video recordings and physiological measures.
All cats were restrained with passive restraint, in addition to one of the three other restraint techniques (full body, n=19; scruff, n=17; clip, n=16) for comparison (figure 2). Cats were randomly assigned to treatment, blocked for sex and age and order of restraint was counterbalanced between cats. The order in which the animals underwent testing was dependent on the animal shelter facility. Cats were restrained for 1 minute; as soon as the cat was securely restrained, respiratory rate (number of breaths seen in 15 seconds) was recorded, and then three consecutive pictures of each eye were taken using a video camera (Model HDR-CX330; Sony Corporation, China) for later assessment of pupil diameter. After 1 minute was up, the handler released the cat from restraint and took a step back from the table. The cat was left on the table for 30 seconds, for later analysis of posthandling behaviour. The cat then underwent restraint with a second handling technique using the same procedures.
A camcorder (Models HDR-CX220 and HDR-CX330; Sony Electronics, Corporation, China) was used to record all experimental procedures in high-definition recording mode for later behavioural scoring. Observers were blind to cat friendliness during all behaviour scoring and were also blind to treatment when scoring videos for posthandling responses. Observers were not blind to treatment when scoring responses to restraint since the technique was obvious; however, all observers were naive to the study hypotheses to reduce bias.
Videos from the unfamiliar person test were used to score cat behaviour. Friendly cats satisfied the following two conditions: (1) completely exited the carrier voluntarily within 2 minutes and passed within the 50 cm semicircle surrounding the chair where the unfamiliar person was sitting and (2) allowed the unfamiliar person to greet, approach and stroke the cat three consecutive times on the first attempt. Cats that did not satisfy at least one of these conditions were categorised as unfriendly.
Various cat behaviours (table 1)14 were scored from video using Noldus Observer XT (Noldus, The Netherlands). Scoring was completed during (1) restraint (starting when the cat was correctly placed into the assigned restraint and ending 1 minute later) and (2) posthandling (30-second period starting when the handler took both hands off the cat). During restraint, the presence or absence of side or back ear positions was recorded; this was scored at the start of restraint and every 5 seconds thereafter for 15 seconds, creating an ear score out of four for each cat. Number of vocalisations and number of lip licks were scored throughout the entire 1-minute duration of the restraint. Lip lick data and the amount of time the cat’s face was out of view were used to calculate the number of lip licks per minute of time the cat’s face was in view. On average (±SD), cats’ faces were out of view for 3.4±4.1 seconds (4.5 seconds for passive restraint, 1.9 seconds for full-body restraint, 1.3 seconds for scruff restraint and 3.9 seconds for clip restraint). To calculate pupil dilation ratio, the first clear eye image for each cat was chosen and imported into ImageJ software, V.1.50a (developed by Wayne Rasband, National Institute of Health, Bethesda, Maryland, USA; available at https://imagej.nih.gov/ij/index.html). The horizontal diameter of the pupil and iris were measured at the widest margins. For each image, this was performed three times, and the average iris measurement was divided by the average pupil measurement to calculate the pupil dilation ratio. Measurement of pupil dilation using ImageJ software has previously been used in human ophthalmology research.32–34 Cats undergoing clip restraint were also assessed for whether they spontaneously went into a hunched lateral decubitus position, as this has previously been used to examine effectiveness of the dorsal immobility reflex.5
For the 30-second posthandling phase, ear position was scored once, 2 seconds after the handler let go of the cat and vocalisations, walking or backing away from the handler and number of lip licks (taking into account the amount of time the cat’s face was in view of the camera) were assessed over the entire period.
All analyses were conducted using SAS V.9.3 or SAS University Edition 2016, with p<0.05 considered statistically significant. The individual cat was the experimental unit, and cat identification was included as a random effect in all models that included paired data. Final models were built using a backward elimination process whereby the covariates (sex, age, friendliness) and restraint type were included in the model and then removed one at a time if p>0.05. During preliminary analysis, no effects of restraint order on response to treatment were detected, so this variable was not assessed in the final models. Interaction variables based on all variables in the models were tested for inclusion. When analysing the number of lip licks out of the time the cat’s face was in view, a logit transformation was performed since the data were presented as a percentage of time.
Mixed linear regression was used to determine the associations between treatment and respiratory rate and pupil dilation ratio. Model fit was assessed using residual plots, histogram plots, Shapiro-Wilk normality test (alpha 0.05, p<0.05), the coefficient of determination and Akaike information criterion/Bayesian information criterion (with a lower value preferred).
Binomial logistic regression was used to assess the association between treatment and ear score (out of four) during handling, while binary logistic regression was used to assess the association between treatment and presence of a side or back ear position, vocalisations and whether or not the cat walked away from the handler towards the back of the compartment during the posthandling phase. Model fit was assessed by fitting all the interactions and terms available into the model to see if they were significant and explained the data.
Poisson regression was used to assess the number of vocalisations during handling and the number of lip licks (per minute of time the cat’s face was in view) during the handling and posthandling phases. Model fit was assessed by fitting all the interactions and terms available into the model to see if they were significant and explained the data, in addition to using the Pearson’s chi-squared statistic/degrees of freedom to assess over dispersion (>1).
A Pearson’s chi-square exact test was used to examine the association between treatment and vocalisations during the posthandling phase, due to a small number of positive responses in this data set. This test was used to examine if there was an overall difference in the proportion of cats vocalising between treatments, examining all treatment combinations separately.
Unfamiliar person test
Out of the 52 cats that were included in the full data set, 39 were categorised as friendly (n=14, full body; n=15, scruff; n=10, clip) and 13 as unfriendly (n=5, full body; n=2, scruff; n=6, clip). The average age of the friendly cats was 4.5 years (SD ±2.4; range 1–10 years), while unfriendly cats had an average age of 4.9 years (SD ±2.3; range 1–10 years). There were 16 males and 24 females included in the friendly category and seven males and six females included in the unfriendly cat category.
On average, cats undergoing full-body restraint showed more rapid respiration (mean number of breaths per minute 52.46, 95 per cent CI 46.7 to 58.9) than cats undergoing passive restraint (mean number of breaths per minute 44.77, 95 per cent CI 41.2 to 48.7; p=0.006). No other significant main or interaction treatment effects were seen when analyzing respiratory rate.
Cats undergoing passive restraint showed smaller pupil dilation ratios (F3,95 = 14.24; p=0.004; figure 3) than cats undergoing full body (p=0.007) and clip (p=0.01) restraint. In addition, cats undergoing scruff restraint showed smaller pupil dilation ratios than cats undergoing full body (p=0.009) and clip (p=0.01) restraint. A sex by friendliness interaction was also seen, with unfriendly female cats (mean 0.59, 95 per cent CI 0.53 to 0.65; F1,95 = 5.71; p=0.019) showing larger pupil dilation ratios than either friendly female cats (mean 0.46, 95 per cent CI 0.42 to 0.50; p<0.0001) or friendly and unfriendly male cats (mean 0.47, 95 per cent CI 0.43 to 0.51; p=0.0006 and mean 0.49, 95 per cent CI 0.44 to 0.55; p=0.01, respectively); no differences were detected between friendly and unfriendly male cats. No other significant main or interaction treatment effects were seen when analyzing pupil dilation.
In comparison with passively restrained cats, cats undergoing full-body restraint vocalised at a greater rate (rate ratio 3.38, 95 per cent CI 1.39 to 8.26, p=0.009; F3,49=4.85, p=0.005; figure 4) and the same held for clip restraint (rate ratio 4.45, 95 per cent CI 1.53 to 12.94, p=0.007). Cats held with scruff restraint showed a lower rate of vocalisations than full-body restrained cats (rate ratio 0.15, 95 per cent CI 0.025 to 0.88; p=0.04) and clip restrained cats (rate ratio 0.11, 95 per cent CI 0.018 to 0.72; p=0.02), while no differences were detected between scruff and passively restrained cats. There were no other significant main or interactiion treatment effects for vocalisation.
The odds of having a higher ear score (more side and back ear positions) was greater in cats restrained with a full body (OR 2.98, 95 per cent CI 1.51 to 5.88; p=0.002; figure 5), clip (OR 2.27, 95 per cent CI 1.13 to 4.55; p=0.02) or scruff (OR 2.89, 95 per cent CI 1.34 to 6.21; p=0.008) technique in comparison with being restrained with passive handling. No differences were detected between the other restraint techniques or there were no interactions with sex or temperament.
On average, the number of lip licks per minute was greater in passively restrained cats (mean: 1.13, 95 per cent CI 0.81 to 1.6; F3, 48= 3.29, p=0.03) than cats held with a scruff restraint (mean: 0.34, 95 per cent CI 0.15 to 0.82; p=0.009). In addition, female cats showed a greater number of lip licks per minute (mean: 1.0, 95 per cent CI 0.67 to 1.5; p=0.02) than males (mean: 0.47, 95 per cent CI 0.28 to 0.80), regardless of handling treatment. Temperament again did not interact with restraint technique.
The odds of showing lip licking was 2.36 (95 per cent CI 1.22 to 4.54, p=0.01; F1, 48=6.92, p=0.01) times greater in females in comparison with males (female mean: 2.3, 95 per cent CI 1.54 to 3.45 versus male mean: 0.98, 95 per cent CI 0.55 to 1.75). Age showed a quadratic two-way interaction (F1,49=6.51, p=0.01); as age increases so does the number of lip licks per minute until reaching an average maximum of 1.5 lip licks per minute at age 6, followed by a decline with increasing age. No treatment effects were detected when analysing lip licking posthandling in cats.
More cats vocalised after being handled with full body versus passive restraint (p=0.03). No other treatment effects were detected when analysing vocalisations, the presence of a side or back ear positions, or whether a cat walked away from the handler towards the back of the compartment.
As expected, cats restrained with the full-body technique showed more negative responses (respiratory rate, ear score, pupil dilation ratio, vocalisations) than passive-restrained cats. Clip-restrained cats showed similar responses to cats undergoing full-body restraint, with no significant differences detected between the behavioural and physiological parameters that were scored. Moreover, scruff-restrained cats showed fewer negative responses than both full-body and clip-restrained cats (pupil dilation ratio, vocalisations), suggesting this handling technique may be less negative than restraint with full body or clips. However, scruffed cats showed a larger ear score than passively restrained cats, suggesting that scruffing is more negative than passive restraint. Note that the technique that was used for scruffing did not result in the skin at the top of the head being pulled backwards (see figure 2), so ear position should not have been influenced by mechanical forces. Thus, overall, these data suggest that the negative welfare impact of these techniques are: full body = clip > scruff > passive.
In the current study, we found that clip restraint results in more negative responses (pupil dilation ratio, vocalisations) than scruff restraint, which conflicts with previous research suggesting that clip restraint is effective at inducing behavioural inhibition without a negative response for the majority of cats4 and that it is less stressful than scruffing.5 However, there were a number of methodological differences across the studies that might explain these conflicting results. Importantly, both of the previous studies involved the placement of two 2-inch binder clips on the dorsal neck skin, with one study assessing cat responses for 2–5 minutes during performance of a routine veterinary procedure,5 and the other rating the cats immediately after clip application.4 Pressure differences between Clipnosis clips used in the current study and 2-inch binder clips, as well as the effects of different durations of application are unknown. It is possible that the Clipnosis clips induced more pain or discomfort than the binder clips used in previous research, suggesting that further research is needed to compare these two methods of application.
In addition, the methods used to assess cat responses to handling differed between the current and previous studies. Pozza et al 4 found that the majority of cats went into a lateral decubitus position (with a hunched back and tail tucked) with clip placement and suggested that this was not a fear response because cats remained responsive and in an alert posture; however, the authors did not examine specific indicators of cat affective state to support this claim. Past literature suggests that a hunched posture is sometimes associated with fear in cats35 36 and that a tail wrapped close to the body is associated with a stress response in cats.31 Further, Nuti e t al concluded that clip restraint is less stressful than scruffing based on reduced heart rate and lower pupil dilation scores.5 The current study did not assess heart rate because earlier work validating cat responses to full-body versus passive restraint found that this measure was not a sensitive indicator of negative affect.14 However, detailed pupil assessment was completed in the current study, and our findings were opposite to this previous research and supported by a concomitant increase in vocalisations. In the current study, direct measurements were taken from photographs, whereas the previous study used subjective scores from live observer assessment; the latter may be biased by lack of blinding and the influence of other behavioural responses occurring at the time of scoring, which might account for the different results.
Cats in the current study were also assigned randomly to treatment and categorised as either friendly or unfriendly to minimise potential group differences, while Nuti et al 5 placed cats into the clip group only if the owner gave consent. If no consent was provided, the cat was placed into the scruff group, and this might have introduced bias. For example, if owners knew the cat had responded poorly to handling during previous appointments, they might have been more open to other handling suggestions as potential solutions to previous handling issues, or (perhaps more likely) they might have been more likely to decline the use of clips—so resulting in the clip group having calmer or less reactive cats than the scruff group.
Finally, there were two further experimental design differences between the current work and the previous study comparing scruff and clips, although we suggest these were less likely to contribute to the conflicting results. First, the current study used shelter cats, testing them within the familiar shelter facility where the cats were housed. In contrast, Nuti et al,5 conducted their study with owned cats in an unfamiliar veterinary setting. While baseline stress levels would be expected to be somewhat higher in the shelter cats, travel to the clinic and exposure to an unfamiliar environment likely would have caused acute stress elevations in cats assessed in the veterinary setting. Second, each cat in the current study completed an unfamiliar person test and was then handled with passive restraint plus one other restraint technique. Such repeated handling and interaction could potentially lead to an increase in response across successive exposures. However, the unfamiliar person test was relatively passive and also provided the cat with time to adapt to the room and the experimenters prior to the handling portion of the study; furthermore, handling order was counterbalanced to prevent systematic bias, and an assessment of order effects did not find any increase in response during the second handling exposure. Overall, however, these last differences in experimental design are unlikely to account for the different results observed across the two studies.
Turning to the cats’ negative responses to aversive handling, our own previous research assessed response differences between cats handled with full-body and passive restraint, and identified and validated the following negative responses during handling: more side/back ear positions, higher respiratory rate, larger pupil dilation ratio and greater number of lip licks.14 Most of these original response differences were replicated in the current study between full-body and passively restrained cats, except that cats did not show a significant difference in lip licking when undergoing full body versus passive restraint. In addition, cats in the current study showed an increase in vocalisations when handled with full-body restraint: a difference not observed in our original validation. Overall, these findings suggest that some of our response measures are more consistent and reliable than others. While pupil dilation, respiratory rate and ear position were consistent in both studies, lip licking and vocalisations appear to vary with the sample of animals. Thus, these measures might be influenced by other developmental or environmental variables that were not recorded, such as genetics or background fear and stress levels.
One unexpected finding was that the number of lip licks seen during handling was greater in passively restrained cats than in scruff-restrained cats. This finding is inconsistent with other treatment results, as passively restrained cats showed fewer negative ear positions than scruff-restrained cats. One possible explanation for this discrepancy is that scruffing might restrict the performance of lip licking in cats; the loose skin of the lower neck is pulled back during scruffing, possibly affecting opening of the mouth, and this is a hypothesis that should be tested in future work. Alternatively, since we did not distinguish between lip and nose licking, it could be that only nose licking was elevated by passive restraint. This is relevant because Bennett et al 37 suggest that nose licking in cats might be associated with frustration, rather than with the fear that the three other handling techniques may have caused (although this hypothesis has yet to be experimentally tested). Thus, the results of the current study suggest that lip licking and vocalisations should be interpreted with caution in future studies of handling stress, particularly if passive control and full-body negative treatments are not included as controls. In addition, perhaps more research is needed to disentangle the relative roles of fear versus frustration in different licking responses.
Overall, full-body restraint and clip restraint resulted in the highest number of negative responses, followed by fewer responses during scruff restraint, and the least number of negative responses in passively restrained cats. This suggests that handling with full body, clips and to a lesser degree scruffing, are negatively perceived by cats. Thus, it is recommended that the use of clip and scruff restraints are avoided when possible. However, when some level of cat immobilisation is necessary, manual scruffing may be preferable to clip and full-body restraint since its use resulted in fewer negative responses.
We want to thank Justine Antunes and Sophia Lee for help running the study, Quinn Rausch and Siobhan Mellors for help with behavioural scoring, William Sears for help with statistical analyses, and the animal shelter facility for allowing us to complete this research at their facility. Portions of Figure 2 were reprinted from 'Moody C, Mason G, Dewey C, Landsberg G, Niel L, Testing two behavioural paradigms for measuring post-handling cat aversion behaviour. Appl Animal Behav Sci, 2019, 210: 73-80', with permission from Elsevier.
Funding This study was supported by the Ontario Veterinary College Pet Trust Fund, and a Discovery Grant from the Natural Sciences and Engineering Research Council of Canada.
Competing interests LN reports grants from Natural Sciences and Engineering Research Council of Canada, and from Ontario Veterinary College Pet Trust Fund, during the conduct of the study.
Ethics approval All animal procedures were approved by the University of Guelph’s Animal Care Committee.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement No data are available.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.