WHAT IS A MINIMALLY IMPORTANT DIFFERENCE IN OUTCOME: EVIDENCE FROM THE UK BEAM TRIAL

Abstract

Background and Significance. The minimally important difference (MID) in outcome, at an individual patient or population level, is crucial to interpreting the outcome of randomised controlled trials (RCTs). We have used the UK BEAM dataset to define a minimally important difference in outcome for the Roland Morris Disability Questionnaire (RMDQ) for different baseline values of the RMDQ.

Method and Results: . We recruited 1,334 participants to UK BEAM. In addition to the RMDQ we included a specific health transition question on satisfaction with current health state that consists of a five point ordinal scale.. We used mean changes and ROC curves to identify the whole number of points change in RMDQ that represented a MID, defined as an improvement of one category on the satisfaction scale, using six bands of baseline RMDQ score 4–24, 4–7, 5–12, 9–16, 13–20 & 17–24, three months after randomization. Rounding up our data give values for the MID, for improvement, in each of these groups of 2,1,2,4,4,6 respectively

Conclusions: The MID, to an individual, is different at different baseline values of RMDQ. Interpretation of MID at high baseline values is difficult because of small numbers. Because of difference in methodological approach our analysis has produced smaller values than the previous study of MID. However, different values for MID should be used at different levels of baseline severity. The use of this approach would allow the number of individual gaining at least a minimally important benefit (or harm) from interventions tested in randomised controlled trials.