Confounders in rehabilitation trials of task-oriented training: lessons from the designs of the EXCITE and SCILT multicenter trials.

Two multicenter randomized clinical trials (MRCT), the Extremity Constraint Induced Therapy Evaluation (EXCITE) to improve upper extremity function after stroke and the Spinal Cord Injury Locomotor Trial (SCILT) to enable functional walking after incomplete spinal cord injury, demonstrate that complex, task-oriented physical therapies can be studied using a scientific methodology during inpatient or outpatient rehabilitation. In the past, a new therapy's benefit may have been overestimated by comparing it to no treatment or to a conventional treatment at a low intensity of practice. Sample sizes were often too small and may have failed to detect the efficacy of a new intervention. In addition, whereas statistical significance in outcomes has been critical to understanding whether one treatment is better than another, the clinical significance of outcomes must also impact the interpretation of the results of a trial. MRCT designs will continue to improve through attention to the limitations of preclinical animal models that offer a conceptual basis for the treatment, [corrected] from enrichment strategies at every phase of trial development, [corrected] from more vigorous dose-response studies using adaptive methods, [corrected] by capturing interim measures of behavior and functional neurophysiologic adaptations during the treatment phase, [corrected] by aiming for a clinically meaningful control intervention, [corrected] and by including ratio or interval outcome measures when feasible that capture a target of the intervention in relation to gains in daily functioning and quality of life.