There is an excellent article in the current issue of the Journal of the American Veterinary Medical Association which I wanted to draw attention to, both for its conttent and the design of the study.
Brown,D.C.; Boston,R.C.; Farrar,J.T. Use of an activity monitor to detect response to treatment in dogs with osteoarthritis. J.Am.Vet.Med.Assoc. 2010, 237, 1, 66-70.
A perennial problem in the evaluation of any therapy for arthritis in animals is the difficulty in assessing effect. Most studies use subjective evaluations by investigators or owners, which are subject to many kinds of bias and almost always show a benefit of every therapy for at least some of the multiple measures of function evaluated. A few studies use force plate analysis, where the dog walks over a device that measures how much weight they place on an arthritic limb. But this is a complex and often difficult measurement to take. The recent article describes a study using a small accelerometer to measure activity at home over a prolonged period. This is potentially a very useful tool in evaluating the efficacy of arthritis treatments in dogs.
The study was very well designed, and the authors clearly recognized the importance of controlling for investigator and owner bias even when using an objective measurement tool. The patients were randomized to treatment either with an agent already clearly established as effective (the NSAID Rimadyl) or an identical placebo, and both investigators and owners were blinded to the group assignment. Confounding factors were controlled for by selecting a study population newly diagnosed and not already treated, and by using statistical techniques to account for factors such as age, weight, and so on.
The results showed an expected increase in activity for the patients given Rimadyl (about 20% greater activity than before treatment) and no change from baseline for the placebo group. Interestingly, some of the dogs in the placebo group did show an increase in activity (20% of the dogs in this group had an increase of 10-30% in activity from the baseline period). As the authors explained, this sort of change likely represents the phenomenon of regression to the mean. For many chronic diseases, including arthritis, the symptoms wax and wane, getting worse and better alternately around an average, or mean, degree of severity. People tend to seek medical treatment, for themselves and their pets, when their symptoms are especially bad, and so the natural course of the disease tends to lead to an improvement in symptoms following an intervention regardless of whether the intervention is actually doing anything. Add to this the tendency for people to behave differently when involved in a medical study, usually exhibiting better health habits than they usually do, and it is routine for studies to show improvement with placebo therapies that aren’t actually doing anything. This is a significant contributor to the so-called “placebo effect.” It also explains why adequate controls are critical to showing real treatment effects, and why poorly controlled studies tend to find effects where they don’t really exist.
I am hopeful that the monitoring device discussed in this article will make it possible to more easily and inexpensively evaluate the effectiveness of the various purported osteoarthritis treatments currently in use as well as new treatments. I am less hopeful that such future studies will be as carefully designed and controlled as this one. Objective measures and all the statistics in the world don’t make experimental results real or useful unless human bias is adequately compensated for through techniques such as random allocation of subjects, blinding of investigators and owners, and placebo controls, or ideally a combination of active and placebo controls as well as no treatment at all. This study is a nice example of how such techniques should be used, as well as a demonstration of why they are needed.