Efficacy of EMLA™ cream for reducing pain associated with venepuncture in felines

[...]

Appraisal, application and reflection

Behaviour as a measure of pain

In all four papers, the primary outcome measured was the response of the subjects to restraint and venepuncture, which the investigators attempted to quantify using behavioural descriptors and numerical rating scales (NRS). Therefore, the results of each study must be interpreted in the context of the inherent challenges associated not only with the accurate assessment of behaviour, but also with the use of behaviour as a measurement of pain.

The correlation between behaviour and pain is not necessarily proportional, nor consistent^11-13. Pain involves both the sensation of a noxious stimuli and the provocation of an aversive emotional response^14,15. Although domestic animals are known to possess the neuroanatomical apparatus and physiological components required to detect tissue damage in a way analogous to human beings, their inability to self-report means that the affective aspect of pain is very challenging to scrutinise⁶. This is further complicated by the myriad of confounding factors that may influence how and to what degree pain is expressed through behaviour, including species, age, health status, and other emotions such as fear¹⁶. The usefulness of behaviour as a definitive measure of pain is greatly challenged by the fact that many behaviours used as proxies for pain are also induced by emotions such as fear, even in the absence of actual tissue damage^6,17.

Feline-specific pain behaviours

Given that behaviour is strongly influenced by the evolutionary history of the species, it is essential to assess and interpret behaviours in a species-specific context^17,19,20. The study by Flecknell et al.² aimed to measure pain in multiple species, including dogs, cats, rabbits, and rats. The NRS designed by Flecknell et al.², therefore, used very basic behavioural responses that are conserved across species despite having divergent ethograms. Consequently, it is possible that more nuanced signs of discomfort expressed specifically by cats, such as purring, growling, posturing, ear positioning, pupil dilation, and facial tension^17,20, were overlooked, resulting in erroneously low discomfort scores²¹.

A similar scoring system was adopted by Oliveira et al.³, however, the implication of this is unique because all subjects were sedated at the time of venepuncture. Assuming that a sedated cat is more likely to demonstrate simple and reflex-like responses to pain, such as limb withdrawal, rather than complex responses, such as hissing²², it could be purported that the use of a simple NRS is not detrimental. However, it can also be argued that the analgesic, sedative and muscle-relaxing effects of the opioids and α₂-agonists used in the sedation protocols may have impacted the ability of the subjects to physically react to any noxious stimuli detected^3,22. It should be noted that the use of sedation does not invalidate the significance of the outcomes measured by Oliveira et al.³ because both the treatment and control groups were reportedly sedated to an equal degree, and yet there was still a significant difference in the reactivity to venepuncture. A study in children likewise found that although pain caused by venepuncture was reduced with the use of systemic opioids, the application of EMLA™ cream at the site of venepuncture still produced a measurable and significant difference in the degree of discomfort experienced²³.

The NRS used by Wagner et al.⁴ and Crisi et al.¹ was adapted from the Modified Behavioural Pain Scale (MBPS)^10,24. This was originally developed by paediatric clinicians to evaluate postoperative pain²⁴ and was later used to assess pain during venepuncture in children^25,26. Although it featured a comparatively more representative sample of feline-specific behaviours related to discomfort, such as attempting to scratch or bite, ear positioning, growling or hissing¹⁶, no literature or methodology was cited to explain how or why certain behaviours were selected. Furthermore, these revised versions of the MBPS have not been validated as an accurate measure of pain in cats.

A systematic review by Merola & Mills¹⁶ described an ideal NRS as one that is highly sensitive and has minimal variability across observers, conditions, and subjects in which pain is being assessed. While the majority of studies appraised by the review utilised an NRS to measure feline pain, only the UNESP-Botucatu Multidimensional Composition Pain Scale was found to demonstrate good validity, reliability and sensitivity⁶. All four papers, thus, could have benefited from incorporating all or part of the UNESP-Botucatu Multidimensional Composition Pain Scale into their methodology^16,21,27-29.

Age as a confounding factor

In cats, age has been implicated as a factor that may influence the behavioural expression of pain¹⁸. The study designs by Crisi et al.¹ and Oliveira et al.³ attempted to control this confounder by evidencing that age did not differ significantly between the randomised treatment groups. In comparison, Wagner et al.⁴ and Flecknell et al.² failed to demonstrate the absence, or presence, of any significant difference in age between treatment groups by way of statistical analysis. Moreover, the study by Flecknell et al.² did not report any randomisation of subjects for allocation to treatment groups, which reduces the strength of evidence by permitting confounding bias³⁰.

Performing a paired analysis whereby each subject was tested both with and without the EMLA™ cream on different occasions and comparing their reactions may have achieved better control over innate inter-subject behavioural differences³⁰. However, such a design has its own limitations with regard to minimising other confounding factors in between procedures³¹, and would have decreased the degree of study design refinement. In all studies, except Crisi et al.¹, patients were only enrolled if they had a non-related indication for venepuncture, which would be much more challenging to achieve for a paired analysis design, potentially generating exclusion bias³⁰.

Fear as a confounding factor

In a conscious animal, a noxious stimulus may not be the only factor contributing to the expression of pain-associated behaviours³². Indeed, many of the behaviours used by these studies as criteria for quantifying pain are also recognised as fear responses in cats³³. It is well established that many feline patients become readily fearful during veterinary visitation, even when pain-free^17,33. Often, this is triggered by prior negative experiences, physical restraint, and/or the presence of frightening sensorium, such as unfamiliar sounds, sights and smells^20,32,33. As such, these stimuli must be considered confounding factors but are highly impractical for clinical researchers to exclude.

Both Crisi et al.¹ and Oliveira et al.³ attempted to mitigate the impact of inherent fearfulness as a confounding factor by excluding subjects that were resistant to handling during clinical examination, or were deemed “aggressive” by the clinician, respectively. While the method utilised by Crisi et al.¹ for identifying and rejecting reactive subjects was well-defined and repeatable, the impact of excluding ‘feral’ and ‘aggressive’ subjects in the paper by Oliveira et al.³ is unknown as neither of the terms were specifically defined and could have various interpretations.

It should be noted, however, that the behavioural threshold employed by Crisi et al.¹ is quite intolerant, requiring patients to demonstrate minimal to no fear associated with clinical examination for inclusion. While this improves the internal validity of the study by creating a narrow behavioural baseline amongst its study population, it simultaneously decreases its external validity. That is, the results cannot be well extrapolated to a population in which most feline patients that would typically demonstrate even a modest degree of fear (i.e., score of 2 or greater using the system¹ employed by Crisi et al.¹) when presented to a veterinary clinic.

The validity of all studies could be improved by firstly performing a pilot study to characterise the range and distribution of temperaments in the target population, and subsequently enrolling a large and representative sample.

Given that the studies by Wagner et al.⁴ and Flecknell et al.² did not exclude subjects based on temperament, it is possible that their results are more indicative of the efficacy of EMLA™ cream in fearful or aggressive patients. However, this is impossible to verify as neither study reported nor integrated analysis of pre-intervention temperament into their design, which is a significant limitation.

Furthermore, the study by Flecknell et al.² is greatly limited by its lack of pre-intervention temperament scoring because its subjects were laboratory animals acquired from specific breeders. As such, it would be valuable to know if these felines were more or less fractious than the general population.

In humans, it has been shown that the relationship between degree of tissue damage and the experience of pain is not linear and may be impacted by the emotional context^12,15. That is, even a procedure that creates a consistent degree of tissue trauma may induce different degrees of pain in a patient depending on their emotional state at the time^12,15. This implies that the emotional domain of pain should not be ignored when attempting to improve patient welfare⁶ and lends credence to the use of techniques that make procedures inherently less stressful for patients³³.

Only the study by Crisi et al.¹ reported attempts to minimise iatrogenic fear, anxiety or distress by keeping subjects in separate and quiet waiting areas and allowing subjects to rest undisturbed between treatment application and venepuncture^32,33. None of the studies, however, described in detail the techniques used for the handling and restraint of subjects during patient preparation, treatment application, and venepuncture. This is significant because poor handling techniques may have contributed to patient stress, which could have increased the likelihood of resistance, vocalisation, or other behaviours that were being assessed as measures of pain^16,18,32,33. In the study by Oliveira et al.³, the impact of stress or fear on the demonstration of pain is less likely to be a confounding factor because patients were sedated during all procedures, and so the reaction to venepuncture is more likely to represent the sensation and reflex response to pain^22,23.

[…]