The Number of Ratings Needed for Valid Pain Assessment in Clinical Trials: Replication and Extension

Objectives To provide additional empirical findings regarding the number of pain ratings needed to obtain valid measures for assessing outcomes in pain clinical trials. Design Secondary analyses of data from a clinical study examining the effects of psychological treatments on pain. Eleven adults with multiple sclerosis and chronic pain reported on four domains of pain intensity (current pain and 24‐hour recalled worst, least, and average pain) on four occasions before and after receiving 16 sessions of psychological pain treatments. We evaluated the reliability and validity of four single ratings and 16 different composite scores. Results Many of the single pain ratings were inadequately reliable while almost all of the composite scores, including the scores created from two ratings, evidenced adequate to excellent reliability. There was a noticeable increase in validity (ability to detect treatment effects) as the number of ratings used increased from one to two. However, there was little change in the validity as the number of items used to create composite scores increased from 2 to 3 or more. The findings also indicated that the scores assessing recalled worst pain were more valid than the scores assessing any of the other pain intensity domains. Conclusions Composite pain intensity scores created from two individual ratings of recalled pain appear to be adequately valid for detecting treatment effects. Moreover, the findings indicate that the selection of the pain intensity domain to use as a primary outcome variable may play a more important role than increasing reliability by obtaining more assessments; specifically, ratings of recalled worst pain may be more valid for detecting treatment effects than ratings of average pain.