Background: Pain is an important outcome in a large number of clinical trials and systematic reviews. Trials investigating interventions for painful conditions frequently present their findings as the difference between the mean pain score. However, it is unclear when such a difference is large enough to be clinically relevant,making interpretation of such measures challenging. An attempt to solve this problem is to empirically determine a minimal clinically important difference (MCID) i.e. the smallest difference in a score which patients perceive to be beneficial or harmful and which would lead the clinician to consider a change in treatment. The MCID is also important in order to estimate adequate sample sizes in clinical trials. Empirical studies aimed at determining MCIDs report differing results.

Objectives: To assess the typical minimal clinically important difference in chronic pain and to examine reasons for its variability.

Methods: We searched PubMed, EMBASE and Cochrane Library with no language restrictions. We included empirical studies examining MCID in chronic pain on one-dimensional pain scales (e.g. visual analogue scale or numeric rating scales). Studies using retrospective pain assessment were excluded. Two independent observers extracted data. We plan to estimate the average MCID with random effects meta-analyses, and to perform subgroup analyses to examine the effects of co-factors (condition, baseline pain, gender and age). If studies are considered too heterogeneous for a meaningful quantitative summary, we will summarise results qualitatively.

Results: We have preliminary included 34 studies and analyses are ongoing.

Conclusions: Pending.