I have written about a wide variety of conventional and alternative therapies for arthritis pain in dogs. A recurring issue in evaluating these therapies is how we know whether or not interventions designed to reduce pain in animals are effective. While I do not think dogs have beliefs or expectations about their treatment, and so do not experience the kind of direct placebo effect people can, they can manifest differences in both the objective features of their illness and in the symptoms they exhibit when given placebo therapies, due to a variety of effects others have described (e.g. classical conditioning, human interaction, etc.). The indirect placebo effect of a therapy on the vets and owners involved in studying a pain control treatment for dogs are well-documented, and these frequently lead us to believe such therapies are effective when they aren’t. So finding a reliable, repeatable, and accurate measure of the effect of a pain medication is critical.
The gold standard for testing arthritis medications in dogs is force plate analysis. This tool involves having a dog walk across a plate that records the amount of weight carried on each limb. As pain is decreased with treatment, lameness improves and the amount of weight the dog can carry on the affected limb increases. This is a technically challenging tool, but the most accurate and objective way we have of knowing if we have successfully diminished pain in a dog’s limb.
A subjective measure of some kind is often used in arthritis treatment studies instead of force plate analysis because such tools are easier and less expensive to use. With proper blinding and controls, subjective evaluations by owners and veterinarians can have some value in measuring response to pain treatment. However, the results of such evaluations are often inconsistent, and the risk of uncontrolled placebo effects is high.
A new study has attempted to validate a subjective owner rating tool by comparing the results of this survey to force plate analysis in dogs with newly diagnosed arthritis given a NSAID or a placebo.
Brown DC, Boston RC, Farrar JT.Comparison of force plate gait analysis and owner assessment of pain using the canine brief pain inventory in dogs with osteoarthritis. J Vet Intern Med. 2013 Jan;27(1):22-30.
The dogs in this study were assessed by force plate gait analysis the first day, and the owners completed a previous validated questionnaire about pain and the degree to which it interfered with their pets’ activities. The dogs were then randomly assigned to either a placebo or NSAID treatment and re-assessed in 2 weeks.
As expected, the dogs on NSAID therapy showed a significantly greater increase in the weight placed on their arthritic limbs than those placed on the placebo. This objectively showed that the medication improved their lameness, and presumably their pain. The two measures of effect in the client questionnaire also showed significant improvement in dogs on NSAID compared to dogs taking placebo. This would seem to show that the questionnaire also accurately identified the improvement in pain on the medication..
However, the authors also found that there was no correlation or concordance between the survey measures and the force plate analysis. In other words, while overall the survey scores changed in the expected way, the survey scores and force plate analysis measurements for individual dogs didn’t seem to be related or to change together.
What does this mean? Well, the authors argue that both the survey and force plate analysis are measuring real improvements in pain, but the force plate specifically measures the change in lameness associated with treating pain while owners are measuring more general signs of comfort and normal activity at home. This is certainly a plausible explanation, but it doesn’t entirely make sense.
If the drug reduces pain, and pain is causing the lameness, then as pain decreases the lameness should decrease. The force plate results show that this is what is happening. However, if the surveys are also measuring accurately the reduction in pain, even if they are not measuring lameness, then the decrease in pain and the improvement in lameness should occur together and the two measures should correlate.
Think of it this way. Let’s say we want to measure the effect of a drug in reducing nausea and vomiting. If we count the number of episodes of vomiting, and the drug clearly reduces how often a person throws up, then that is an objective, though indirect, measure of nausea (analogous to force late analysis evaluating pain by measuring lameness). And if we ask people about their nausea, and if they say they are less nauseated when taking the drug, that is another way to measure nausea and the effect of the drug (analogous to asking clients about their dogs’ pain). But if both tests are measuring the same thing, then the results should correlate. People who say they feel less nauseated should vomit less than people who say they don’t feel any better.
But what happens if we give the drug and on average the group reports they feel less nauseated and overall they vomit less, but the individual responses on the survey don’t match the frequency of vomiting? People who say the feel better might or might not vomit less. And people might show less vomiting even when they say they don’t feel any less nauseated. Are the survey and the frequency of vomiting both measuring nausea? And which is more important? Do we primarily want people to report feeling better but still vomit just as much, in which case the survey might an appropriate tool? Or is it more important to reduce the amount of vomiting even if some people who vomit less say they don’t feel better?
In the case of arthritis pain in dogs, it is difficult to picture a situation in which reducing the signs of pain perceptible by the owner is more important than reducing the lameness caused by the pain. This raises the question not only of what the owner survey is actually measuring but also the question of when would it be an appropriate tool to use to measure the efficacy of an analgesic therapy. If animal owners report their pets have less pain but we don’t have an objective measure such as force plate analysis, it seems likely that the intervention is changing something about the dogs’ behavior which the owners are detecting, but it isn’t clear that something is pain.