Primary outcome measure use in back pain trials may need radical reassessment

BACKGROUND The answers to patient reported outcome measures and global transition questions for back pain can be discordant. For example, the most commonly used outcome measure in back pain trials, the Roland Morris Disability Questionnaire (RMDQ), can show improvement even though participants say that their back pain is worse. This gives cause for concern as transition questions are used as anchors to estimate minimally important change (MIC) thresholds on patient reported outcome measures such as the RMDQ. We aimed to explore and compare what people with back pain think when they respond to a transition question and when they complete the RMDQ. METHODS We purposively sampled people enrolled on a back pain randomised controlled trial who completed the RMDQ and two transition questions. One enquired about change in ability to perform tasks, the other about change in back pain. We sampled participants with discordance (in both directions), and participants with concordant scores. We explored participants' thought processes using in-depth interviews. RESULTS We completed 35 in-depth interviews. People with discordant RMDQ change and transition question responses attend to different factors when responding to transition questions compared to people with concordant scores. In particular, those for whom the RMDQ change indicated greater improvement than transition questions, prioritised their pain ahead of functional disability. When completing the RMDQ, participants' thought processes were comparatively more objective, and specific to each statement. CONCLUSION Approaches to primary outcome assessment in back pain needs re-assessment. The RMDQ may be unsuitable for use as a primary outcome measure since patients may not attend to thinking about their back pain when completing it: patients' abilities to cope with tasks can be independent of the change in their back pain. Some participants who improve on the RMDQ consider themselves globally worse. As transition questions can be driven by pain and other physical factors, transition questions should not be used to anchor minimally important change thresholds on the RMDQ.