This one is a bit technical, and intended for those who are looking at and interpreting primary source papers – just some of the things I try to think about as I read them – do you agree?
The expansion of studies evaluating treatment of pain in animals is a significant achievement for veterinary medicine and animal welfare, but science isn’t always clear-cut. Sometimes results are unexpected and differ from previous findings. Abstracts are necessarily concise, and may not reflect all of the considerations pertaining to patients.
How, then, should we assess scientific investigations into putative analgesics to treat our clinical cases?
Our first considerations are related to be the study design. Is this a retrospective case series without a control group? in which case any treatment effect may be confounded by care-giver placebo effects and regression to the mean (spontaneous improvements in disease state). Is this a placebo-controlled cross over study, in which owner and investigator are unaware of treatment allocation (‘double-masked’)? If the order of treatment and placebo administration blocks is randomised, then this study design enables each animal to act as its own control, and can be a powerful study design to detect treatment effects. A placebo-controlled double masked grouped study can provide evidence of treatment effect if the control group is appropriately matched to the treatment group.
In pain medicine we have very few magic bullets – it is unlikely that a single intervention will provide total relief, and so investigators need to make a decision whether to test the new intervention is isolation (although the treatment effect will need to be very large, or a large sample size may be necessary to demonstrate a more modest effect), or to evaluate the treatment in addition to current care (which most patients will already have established).
The sample size of the study needs to be large enough so that clinically relevant differences would be identified with statistical testing – this requires investigators to make assumptions about the size of treatment effect they expect, and also regarding the natural variance within the population – if these assumptions are incorrect then the results may need careful evaluation.
Persistent pain impacts many areas, or domains, of a patient’s life. Although these may be interlinked, it is likely that treatments will benefit some areas more than others. Therefore, the particular outcome measures which are being investigated by a study may be more or less relevant to our patients. Examples of outcome measures might include objective measures of limb use in appendicular osteoarthritis, subjective evaluations of function using client specific outcome measures, total activity or activity patterns over the day using collar mounted accelerometers, skin or muscle sensitivity using sensory testing measures, clinical metrology instrument (CMI) scores to evaluate pain and quality of life, and dose of ‘rescue’ analgesics administered. Different outcome measures should ideally be validated – for example many CMIs are validated only in certain painful conditions. Side effects of drugs used in interventions should also be considered – for example increased locomotor activity may be associated with pain free movement, but can be associated with the use of opioid drugs.
It is possible that some interventions may improve pain relief and quality of life without necessarily increasing activity, or changing distribution of weight.
We should also carefully consider the attributes of the patients in the study and how closely these relate to our patients. Although osteoarthritis patients all present with a degree of synovitis and cartilage damage, the severity of these is variable and the secondary changes (myofascial pain, neuroplasticity, muscle atrophy) may further increase the variability between patients. Different patients with differing degrees of secondary changes may respond differently to analgesic interventions – therefore defining these characteristics in study populations is important. Less common pain conditions are likely to be even more variable, and difficult to perform clinical studies.
In the future pharmacogenomics may provide information on how individual patients process pain signals and respond to analgesics, and could be an exciting area for future developments.