An observed statistically significant difference between two interventions does not necessarily imply that this difference is clinically important for patients.
We aimed to interpret the results of published trials in terms of clinical relevance and statistical significance. In particular, we assessed whether randomized controlled trials (RCTs) achieved the minimal important difference (MID) planned in the sample size calculation.
We performed a cross-sectional study of all RCTs included in Cochrane Reviews published up to April 2017, focusing on the efficacy of rehabilitation interventions for low back pain. We considered RCTs with a sample size calculation and a planned MID. Using an ad hoc checklist, two independent investigators collected the study characteristics, the mean differences and their confidence intervals for the primary continuous outcomes and the planned MID. The primary aim was to identify the number of RCTs classified as statistically significant and clinically relevant, statistically significant but not clinically relevant, clinically but not statistically significant, and neither statistically nor clinically significant. The secondary aim was to carry out the same categorization for all compared interventions.
From 20 eligible systematic reviews including 101 RCTs, we identified 42 RCTs (41.6%) involving 4966 patients and 81 intervention comparisons. Overall, we found that more than one-third of RCTs (37.5%, n = 15) were both statistically and clinically significant, whereas few (23.8 %, n = 10) were statistically significant but not clinically relevant. Taking into account all comparisons in multiple-arm trials the scenario was slightly different: less than one-fifth of the compared interventions were both statistically and clinically significant (24.7%, n = 20/81). Among these, 70% (n = 14/20) considered placebo or usual care while only 30% (n = 6/20) explored head to head comparisons.
Despite the amount of effort involved in conducting RCTs in low back pain rehabilitation, few of them obtained both statistically significant and clinically relevant results and the majority of these were comparisons against open-and-shut interventions.
It is essential to consider clinical relevance in order to involve the patient perspective and transfer research results into clinical practice.