Clinically relevant change and clinically relevant difference in knee osteoarthritis

Abstract

As clinical trialists in osteoarthritis, we were repeatedly met with the objection whether the difference between two treatments (as measured by group difference in change of disease status) was clinically relevant. To address this unsettled problem, we report post hoc analyses of three placebo- controlled clinical trials in knee osteoarthritis where Lequesne's 24-point pain and function index (ISK) and a 100-mm visual analogue pain scale (VAS) to measure global pain during the past week were used as endpoints. The aim of this evaluation was to establish change scores which are 'minimal clinically relevant' as distinguished from clinically unimportant. We postulate that a difference between two treatments, one generating relevant and the other unimportant change score is also of clinical relevance. Two groups were extracted from the three trials using all patients with data available: patients who subjectively felt their complaints were reduced after about half a year of follow-up and those who expressed unaltered complaints at that time point. Change score was calculated as absolute change of ISK and VAS from baseline to half-year follow-up in each patient. The difference in mean reduction of ISK was 0.7. It was 8.4 mm regarding mean improvement on VAS. Results were in accordance with cut off points determined by linear discriminant analysis: 0.4 (cut-off for change of ISK) and 8.4 mm (cut off for change of VAS). These data suggest which difference in change is relevant to patients and may therefore be considered as a difference which is relevant for treatments.