Objectives: To determine the minimal clinically important difference (MCID) of changes in chronic musculoskeletal pain intensity that is most closely associated with improvement on the commonly used and validated measure of the patient's global impression of change (PGIC), and to estimate the dependency of the MCID on the baseline pain scores.

Methods: This was a prospective cohort study assessing patient's pain intensity by the numerical rating scale (NRS) at baseline and at the 3 month follow-up, and by a PGIC questionnaire. A one unit difference at the lowest end of the PGIC ("slightly better") was used to define MCID as it reflects the minimum and lowest degree of improvement that could be detected. In addition we also calculated the NRS changes best associated with "much better" (two units). In order to characterize the association between specific NRS change scores (raw or percent) and clinically important improvement, the sensitivity and specificity were calculated by the receiver operating characteristic (ROC) method. PGIC was used as an external criterion to distinguish between improved or non-improved patients.

Results: 825 patients with chronic musculoskeletal pain (233 with osteoarthritis of the knee, 86 with osteoarthritis of the hip, 133 with osteoarthritis of the hand, 290 with rheumatoid arthritis and 83 with ankylosing spondylitis) were followed up. A consistent relationship between the change in NRS and the PGIC was observed. On average, a reduction of one point or a reduction of 15.0% in the NRS represented a MCID for the patient. A NRS change score of -2.0 and a percent change score of -33.0% were best associated with the concept of "much better" improvement. For this reason these values can be considered as appropriate cut-off points for this measure. The clinically significant changes in pain are non-uniform along the entire NRS. Patients with a high baseline level of pain on the NRS (score of >7 cm), who experienced either a slight improvement or a higher level of response, had absolute raw and percent changes greater that did patients in the lower cohort (score of less than 4 cm).

Conclusions: These results are consistent with the recently published findings generated by different methods and support the use of a "much better" improvement on the pain relief as a clinically important outcome. A further confirmation in other patient populations and different chronic pain syndromes will be needed.