Afterword Five steps to increase the payoff of chronic pain trials

Congratulations to the editors and authors for an educational gem, the first multiauthor treatise on analgesic clinical trial design to appear in many years. The essays in this issue focus on the parallel-group repeat-dose design, the most common approach in current studies of chronic pain. Previous works1–3 had focused on single-dose studies. Taken together, the authors’ arguments suggest to me five steps that will greatly increase the public health benefits of pain research: To understand how great efficiency can quickly be gained in analgesic development, consider the general form of sample size formulas for clinical trials: ![Formula][1] When one discovers a factor that reliably accounts for more than a few percent of the variance, one can remove that proportion of the variance from the numerator of the sample size equation and either decrease the N by this proportion or decrease the size of the effect one can detect. For example, a collaborator (Robert Edwards et al, unpublished data) recently reanalyzed a published trial4 and found that age, sex, and baseline pain explained 10% to 20% of the variance in response to the drug, depending on the particular pain outcome examined. Nathaniel Katz, writing in the current collection, also suggests the value of incorporating baseline variables in covariate analyses. However, Katz could apparently find only one example of this approach to pain data, an article in which methodologist Andrew Vickers5 reanalyzed four acupuncture trials and showed that this analysis turned an equivocal study into a clear positive. I am unaware of any other chronic pain studies in which this has been examined. This is not rocket science. How could we have overlooked this simple step, standard practice in many areas of therapeutic research? Dworkin, Katz, and Gitlin offer an historical explanation. Until 1998, most chronic pain researchers relied on … [1]: /embed/graphic-1.gif