Saturday, October 9, 2010

When is Pain Relief “Meaningful”?

EBPM Logo Understanding Evidence-Based Pain Management (EBPM)

An important deficiency of many research reports is that they focus on pain relief of a therapy as a statistical phenomenon, rather than examining whether there were clinically meaningful benefits for patient care. A recent study investigated characteristics of “responders” to fibromyalgia therapy that can make a difference in achieving successful outcomes that are durable over time.

Clinical trials of drugs and other treatments for pain often demonstrate “statistically significant” pain reductions in patients without demonstrating any “clinically meaningful” reductions, wrote Gabriel Miller in the September edition of Pain Medicine News [Miller 2010]. The article was based on a presentation at the 2010 meeting of the American Pain Society (APS) in which researchers noted how, with an adequately large sample size, a one-point reduction on a visual analog scale (VAS) can be statistically significant, yet may do nothing to truly improve a patient’s pain or quality of life.

The researchers further reported that patients who seem to achieve less than certain pain threshold at 6 months post-treatment — specifically, a score of 4 on a 10-point VAS scale — generally continue to improve, whereas those who never get down to that threshold more often return to pretreatment pain levels. Their study involved 301 female patients with fibromyalgia who underwent a standardized course of rehabilitation (not further described). Overall, there was a statistically significant reduction in pain intensity from baseline in almost all patients (p<0.001), achieving an average score of 5 on the 10-point scale. However, at the 6-month followup only 73 patients reported pain levels below a score of 4, which was considered the threshold between “mild” and “moderate to severe” pain; whereas, the majority of patients reverted back to having debilitating pain.

The researchers observed that one way of resolving the conflict between statistically significant versus clinically relevant reductions in pain relies on responder analysis; that is, looking separately at patients who were either “responders” or “non-responders” to the intervention under investigation. In their study, the 73 patients who were responders had pain below a 4 on the VAS and continued to improve; eventually, this group’s pain score dropped more than 3 points from baseline, to a 2.2 on average.

Another group, of 140 patients, who also were hovering around a score of 5 on the VAS immediately after treatment turned out to be non-responders and went in the exact opposite direction; at 6 months this group’s average pain was higher than their baseline score. Hilary Wilson, PhD, from the University of Washington who participated as a researcher in the study, noted, “It really demonstrates that something happens when you can get a patient’s pain at a 4 or less. They are able to somehow improve their function, get back into their life, and continue to realize improvements. If you can get your patient in treatment below that threshold, they can continue to get better.”

COMMENT: Perspectives from this article shed more light on the typical flaws in research in the pain management field and what practitioners and their patients should expect from effective therapies. Research reports focusing on average, or mean, improvements in pain (as many do) are unhelpful. Readers do not know whether all patients improved to some degree or a only small segment of patients benefitted greatly, which skewed the overall average higher. Researchers need to present their data in ways that can help identify which patients, if any, can benefit most from a particular therapy.

One approach, as noted in the above article, is to identify the characteristics of “responders” who achieve a “minimal clinically important difference” (MCID), such as a predetermined improvement from baseline in self-reported VAS scores. We have previously described [here] that a 10% to 20% reduction in VAS score is generally considered as minimally acceptable, with 30% to 36% reductions being moderate, and a 50% or greater reduction deemed substantial.

One of the problems with data such as self-reported VAS scores is that there can be what is called a “regression toward the mean” over time. That is, regardless of whether the treatment was effective, patients may drift back toward their initial scores; hence, group averages may eventually reflect no meaningful differences in treatment effects. The research reported at APS suggests that a VAS score of 4 is somewhat of a critical threshold for identifying patients who have meaningfully benefitted from a therapy and will continue to show improvements, as compared with other patients who may have benefitted initially but may revert back to their initial pain status or worse. Of course, the interpretation of “clinically meaningful” is somewhat relative and depends on how high a level of pain the patient was experiencing before treatment; however, it is something to consider.

REFERENCE: Miller G. Studying the Difference Between ‘Statistical’ and ‘Clinically Meaningful.’ Pain Medicine News. 2010(Sep);8(9) [article available here].