Selection, Confounding, and Attrition Biases in Randomized Controlled Trials of Rehabilitation Interventions