Can dog appeasing pheromone ameliorate stress behaviours associated with anxiety in mature domestic dogs?

[...]

Appraisal, application and reflection

Anxiety can be defined as anticipation of future danger, whether real or imagined, often despite the absence of any specific object of threat (Tiira et al., 2016). Dogs suffering from anxiety can have a negative emotional response to many innocuous stimuli in their environment and those experiencing a continual state of anxiety have compromised welfare (Salonen et al., 2020; and Beata et al., 2007). Canine anxiety can strain the human-animal bond and is a major reason for animal surrender (Miller et al., 1996). Dog Appeasing Pheromone (DAP), a synthetic version of Canine Appeasing Pheromone marketed as Adaptil® (Ceva Santé Animale) is a popular adjunctive therapy for canine anxiety. Endogenous canine appeasing pheromone is secreted by the bitch shortly after parturition and is detected by the vomeronasal organ (Pageat & Gaultier, 2003). Interestingly, it is reported to have calming effects on both young and adult dogs (Pageat & Gaultier, 2003).

Frank et al. (2010) conducted a systematic review of prospective studies published between January 1998 and December 2008 to assess the evidence for treating undesirable canine behaviours with DAP. This systematic review determined that six studies yielded insufficient evidence for using DAP to treat undesirable behaviours, whilst the seventh provided sufficient evidence that DAP collar helped reduce fear or anxiety during training in puppies 12–15 weeks of age. However, several relevant studies have since been published and to the authors’ knowledge, these have not been appraised in a systematic manner. This Knowledge Summary was limited to randomised and non-randomised controlled trials as these study designs provide the most reliable evidence for assessing efficacy of interventions (O'Connor et al., 2014).

This PICO question was focused on dogs older than 6 months of age because although canine appeasing pheromone is endogenously released by a lactating bitch for puppies, synthetic DAP is still reportedly indicated for use in adolescent and adult dogs. From a behavioural development perspective, the juvenile period ends at 6 months of age, and this usually also coincides with sexual maturity (Scott & Fuller, 1965). Additionally, the dog has already surpassed both the early and late socialisation period by this age (Dietz et al., 2018).

Three studies investigated DAP spray, but the method of application differed. One study applied the spray on a bandana (Amaya et al., 2020) whereas the other two applied it to the corners of the kennel (Hermiston et al., 2018) or cage floor (Siracusa et al., 2010). The quantity of DAP used varied from 3–10 pumps for each treatment. Concentration of DAP used also varied between unreported (Hermiston et al., 2018), to 2% (Siracusa et al., 2010) to 15.72 mg/ml (Amaya et al., 2020). The findings between these three studies were inconsistent. In the study by Amaya et al. (2020), a short length of each observation and large gaps between each observation resulted in inadequate data collection to statistically analyse licking nose/lip behaviours which are well-known displacement behaviours (Lund & Jørgensen, 1999).  However, DAP-treated dogs spent more time lying down as well as less time panting, vocalising and moving the tail. In combination, these behaviours are suggestive of increased relaxation in this study. In contrast, Siracusa et al. (2010) found that DAP spray increased the duration of alertness (‘eyes kept open’) and visual exploration, which are behaviours classically associated with increased arousal rather than relaxation (Hammerle et al., 2015; and Overall, 2013). The significance of this finding is ambiguous and might simply reflect exploratory behaviour secondary to the presence of DAP in the environment. The third study by Hermiston et al. (2018) found that exposure to DAP was associated with reduced barking intensity only. This finding may or may not reflect a change in the animal’s emotional state and is difficult to interpret as all other indicators of stress were not significantly affected. In addition, there are many causes of barking and this behaviour is not necessarily specific to anxiety, stress or fear (Hermiston et al., 2018). The discrepancy in results and methodology between these three studies is problematic, but the design of Siracusa et al. (2010) was least prone to bias whereas Hermiston et al. (2018) was most prone to bias. The reasons for this have already been described in ‘The Evidence’ section of this Knowledge Summary.

For the three studies that evaluated DAP collar, only one (Broach & Dunham, 2016) reported the concentration of DAP impregnated within the collar. Landsberg et al. (2015) was the only study that had significant findings (P < 0.05). In this trial, DAP collar was reported to significantly reduce the frequency and intensity of ‘active’ behavioural signs of fear and anxiety in middle-aged to geriatric Beagles during and immediately after stimulation with thunderstorm audio. This study carries moderate evidence but was funded by the manufacturer of the product and at least one statistician involved was also employed by the manufacturer at the time. As there was no formal conflict of interest declaration, these are perceived conflicts of interest that should be noted. Allocation of subjects into treatment and placebo groups was performed by systematic alternation rather than true randomisation, which could lead to unequal distribution of confounding factors between groups. If any allocation was unmasked, then internal validity of the study would also be compromised (Berger & Grant, 2014). The study by Broach & Dunham (2016) found no significant effects of DAP vs placebo collar in any behaviour category including ‘distress’ and ‘aggression/affability’ (although not all canine aggression is necessarily attributed to anxiety (Sueda & Malamed, 2014)). In this study, it is assumed that the outcome ‘mean score’ for each behaviour category was an average of all three behavioural assessments as this was not clarified in the publication. However, including the results from all three behavioural assessments in the calculation of ‘mean score’ for a given behaviour category would potentially obscure any true differences between the treatment and control groups. This is because subjects did not wear collars in the first (baseline) behavioural assessment and had stopped wearing collars for 1 week by the time of the last (third) behavioural assessment. Furthermore, the outcome ‘change in score’ for each behaviour category was calculated by subtracting the final behavioural assessment score from the initial behavioural assessment score. Unless DAP collar has a residual effect of ≥ 1 week, this approach would clearly also be unlikely to detect any treatment effects.

The two studies that investigated DAP diffuser (Taylor et al., 2020; and Kim et al., 2010) yielded inconsistent findings and only one reported the concentration of DAP (Taylor et al., 2020). In both studies, it was unclear how long the DAP diffuser was switched on for before subjects were exposed. The manufacturer recommends allowing at least 24 hours for the pheromone to diffuse into a room before exposure to subjects (Ceva Animal Health Pty Ltd, n.d.). The first study found that a DAP diffuser did not influence behaviours (some of which were considered anxiety-related) in dogs separated from their owners in a laboratory environment (Taylor et al., 2020). The trial was placebo-controlled and double-blinded, but order and time effects may have biased the results. The effect size may also have been reduced due to sensitisation from learnt association of the study site with owner separation. Furthermore, results for yawning and lip-licking (which may be considered anxiety-related behaviours [Hammerle et al., 2015]) were not described. The second study (Kim et al., 2010) found that the intervention significantly (P < 0.05) ameliorated ‘excessive licking’ and ‘pacing’ behaviours. Excessive licking and pacing are commonly associated with stress and anxiety (Hammerle et al., 2015), so this finding could reflect a beneficial treatment effect. However, there were many potential confounders including level of pain, noise of other dogs and medical diagnoses that were not controlled for. Furthermore, subjects were hospitalised for a multitude of disease diagnoses that necessitated intrinsic differences in caregiving. In addition, the allocation of subjects was unclear, and the description of methodology and statistical analysis was ambiguous. As a result, it was unclear how some values presented in the results section were obtained (see ‘Summary of Evidence’ section).

Overall, no study reported a formal sample size calculation. Failure to perform this calculation could reduce the power of statistical analyses if sample sizes were too small. However, it is likely that sample size calculation is difficult for the outcome variables in behavioural studies (Taborsky, 2010). One study (Broach & Dunham, 2016) stated that a sample size calculation was not performed due to a lack of baseline data.

In this Knowledge Summary, all studies except for two (Taylor et al., 2020; and Siracusa et al., 2010) either lacked or did not report a statistical analysis of intra and/or inter-observer agreement for behavioural observations. These metrics are important to reveal any potential impact of rater scoring drift and observer bias (Girard & Cohn, 2016). This is especially important in behaviour studies where outcome variables are often subjective. The importance and prevalence of this issue in animal behaviour research have been discussed in detail in the literature (Burghardt et al., 2012; and Kaufman & Rosenthal, 2009).

The systematic review by Frank et al. (2010) cites some flaws in the design validity of reviewed articles published between January 1998 and December 2008. Some limitations included the introduction of confounding bias from concurrent treatments or environmental changes, and failure of intention-to-treat analysis when there was loss to follow up. This is important because failure of either can introduce selection bias and/or increase the likelihood of non-comparability between treatment groups with respect to potential confounders (O'Connor et al., 2014; and Fives et al., 2013). In this Knowledge Summary, the relevant controlled trials published between 2009 and June 2021 continue to be affected by similar limitations to those that were reviewed by Frank et al. (2010). All four randomised controlled trials included in this Knowledge Summary still failed to report the method of generating randomisation sequences (Amaya et al., 2020; Broach & Dunham, 2016; Grigg & Piehler, 2015; and Siracusa et al., 2010) and three failed to report concealment of allocation into treatment groups (Amaya et al., 2020; Broach & Dunham, 2016; and Grigg & Piehler, 2015). Additionally, blinding of assessors and/or caregivers in some of the studies appraised in this Knowledge Summary was either incomplete (Hermiston et al., 2018), or unclear (Amaya et al., 2020; Grigg & Piehler, 2015; and Siracusa et al., 2010).

Overall, the level of evidence from controlled trials published since the systematic review by Frank et al. (2010) remains of low quality and/or with moderate risk of bias. This was true for all studies regardless of their findings. There were inconsistencies in the effects of DAP on behavioural outcomes; flaws in experimental designs; differences in the concentration of DAP between and within formulations; heterogeneity in methodology between studies; and potential conflicts of interest sometimes declared or undeclared.

There are several limitations to this knowledge summary. Firstly, reviewing a dog’s body language is inevitably subjective and context-specific (Paul et al., 2005). Measures of behavioural signs between observers are not always reliable and this is a limitation of drawing conclusions from multiple studies (Bradshaw & Casey, 2007; and Wemelsfelder, 2001). Secondly, veterinary practitioners should be aware that a reduction in non-specific stress behaviours does not necessarily equate to successful treatment. This is because the behaviour displayed by an animal is only a surrogate measure of anxiety and is not a direct reflection of its emotional state (Mills, 2017). Thirdly, some behaviours occur due to a variety of motivational reasons (Paul et al., 2005) and aetiologies. These factors can result in individual variation between subjects and are problematic to control in experimental designs of behaviour studies. Therefore, control to limit variability between treatments, the requirement of a placebo, and well-defined hypotheses that can facilitate calculation of adequate sample sizes for each treatment group are paramount for study implementation. Additionally, concurrent interpretation of behaviour with physiological stress parameters may provide a more accurate assessment for canine anxiety (Taylor et al., 2020).Fourthly, this Knowledge Summary has highlighted the inconsistency in findings between eligible studies. This is likely at least partially due to the broad PICO question resulting in the inclusion of many studies with heterogenous study populations that assessed a breadth of indications. Fifthly, other study designs such as well-designed observational studies may have contributed more evidence for the PICO question but were excluded from appraisal in this Knowledge Summary. Lastly, this study focused on the population of patients older than 6 months of age but learning and occasionally cognitive development continues to progress beyond this. The timing of behavioural maturity in domestic dogs is a complicated and under-researched topic (Harvey, 2021). The possible effect of ongoing cognitive development and behavioural immaturity on an individual dog’s emotional state and behaviour after being exposed to DAP is unclear.

In conclusion, there is a continued knowledge gap concerning the effectiveness of DAP in treating non-specific stress behaviours associated with anxiety. Testing within the home environment should be a future consideration for further research, although controlling for differences in the home environment and caregiving would be problematic.

[…]