Objective: To compare methods commonly used to derive minimal important differences and recommend a rule for defining patients as clinically improved on the low back pain-specific Roland-Morris Disability Questionnaire (RMDQ).
Methods: 447 primary care low back pain consulters completed a questionnaire at consultation and 6 months. Patients were classified as having achieved an important change based on methods with the best theoretical qualities, that is, the standard error of measurement, reliability change index (RCI), and modified RCI (RC(indiv)), and using a 30% reduction in score from baseline. To assess clinical importance, improvements based on these methods were compared with improvements on other back pain-related measures.
Results: The percentage of patients rated as improved ranged from 14 to 51% by method. Using a simple rule it was possible to identify patients who had clinically important improvement (36%), patients not improved (53%), and a group of possible improvers (11%). Clinical improvement is shown if RMDQ score is reduced by 30% from baseline and back pain is rated as better on a global rating scale.
Conclusion: A minimal clinically important difference is derived that is clinically relevant, incorporates the measurement error of the RMDQ, and allows subjects with different grades of severity to improve.