| Objective: The results of observational studies can not only inform the clinician concerning clinical outcomes but are often more appropriate or practical than randomised controlled trials. Furthermore, the reporting of statistically significant changes in continuous outcome measures in comparison to clinically significant change using categorical outcomes can be markedly different. The objective of this study therefore was to compare the statistical and clinical significance of outcomes reported by low back pain patients undergoing treatment. Design: This study observed a single cohort of low back pain patients undergoing chiropractic treatment in a chiropractic outpatient clinic. Methods: Patients completed the Bournemouth Questionnaire (BQ) at baseline, the following four visits and at 6 months. Conventional approaches to outcome analysis were used to compute within subject changes in BQ scores. In addition to this, both reliable change and clinically significant change was calculated, which provided a method to categorise patients into one of four groups:
(i) clinically significant improvement;
(ii) reliable improvement that was not clinically significant;
(iii) no change; and
(iv) reliable deterioration. The results of both analysis approaches were compared. Results: Using simple change scores based on continuous data, analysis of the entire sample, and acute (<7 weeks) and chronic (>7 weeks) sub-groups, there was a highly statistically significant change in total BQ scores at all time intervals (p < 0.001). When analysed using cut-offs for reliable and clinically significant change however, the findings were less impressive. At visit 4, 61.5% of acute patients had clinically improved compared with 27.7% of chronic patients. At 6 months, these proportions were 64.4% and 51.0% respectively. Conclusions: Conventional approaches based only on statistical investigation of change scores may over-estimate what patients report during treatment and bear little resemblance to patient outcomes in terms of reliable and clinically significant improvement. This abstract is reproduced with the permission of the publisher; full text by subscription.
|
