Spine (Phila Pa 1976). 2004 (Nov 1); 29 (21): 2410-2417
Jennifer E. Bolton
Anglo-European College of Chiropractic,
STUDY DESIGN: Prospective, single-cohort study.
OBJECTIVE: To determine the relative sensitivity of a range of outcome measures used in evaluating treatment interventions in patients with neck pain and the magnitude of change scores on the neck Bournemouth Questionnaire and the cutoff score on the Patients' Global Impression of Change scale associated with clinically significant improvement.
SUMMARY OF BACKGROUND DATA: Traditionally, evaluation of treatment interventions using subjective outcome measures has been based on the statistical significance of the difference between group mean values. To be clinically meaningful, however, information is required from sensitive outcome measures on the proportion of patients undergoing a clinically important improvement and from this, the number needed to treat for a single patient to benefit.
METHODS: Patients with nonspecific neck pain from a number of treatment centers completed a battery of self-report questionnaires, including the Bournemouth Questionnaire, before and after chiropractic treatment. After treatment, patients also completed a retrospective analysis of their overall improvement since the start of treatment (Patients' Global Impression of Change). The a priori definition of clinically significant improvement was defined as a Reliable Change Index on the Bournemouth Questionnaire of >1.96.
RESULTS: The best cutoffs with a balance between the highest sensitivity and highest specificity in detecting clinical improvement were a score of 2 or less on the Patients' Global Impression of Change (11-point Numerical Rating Scale: 0 = much better, 5 = no change, and 10 = much worse) and a raw change score of three or more points on each of the seven 11-point Numerical Rating Scale subscales of the Bournemouth Questionnaire. For the total score of the Bournemouth Questionnaire, raw change scores of 13 or more points, percentage change scores of 36% or more, and individual effect sizes of 1.0 or more were all associated with clinically significant improvement. The sensitivity of the Bournemouth Questionnaire in terms of its effect size was comparable with that of pain intensity scales and the Neck Disability Index.
CONCLUSIONS: The present findings will assist in the choice of outcome measures in trials on neck pain. The study also illustrates a methodologic framework for interpreting change scores in terms of clinical improvement, facilitating the process of making sense of research data in the clinical setting.