Selection, Confounding, and Attrition Biases in Randomized Controlled Trials of Rehabilitation Interventions What Are They and How Can They Affect Randomized Controlled Trials Results? Basic Information for Junior Researchers and Clinicians

A thorough knowledge of biases in intervention studies and how they influence study results is essential for the practice of evidence-based medicine. The objective of this review was to provide a basic knowledge and understanding of the concept of biases and associated influence of these biases on treatment effects, focusing on the area of rehabilitation research. This article provides a description of selection biases, confounding, and attrition biases. In addition, useful recommendations are provided to identify, avoid, or control these biases when designing and conducting rehabilitation trials. The literature selected for this review was obtained mainly by compiling the information from several reviews looking at biases in rehabilitation. In addition, separate searches by biases and looking at reference lists of selected studies as well as using Scopus forward citation for relevant references were used. If not addressed appropriately, biases related to intervention research are a threat to internal validity and consequently to external validity. By addressing these biases, ensuring appropriate randomization, allocation concealment, appropriate retention techniques to avoid dropouts, appropriate study design and statistical analysis, among others, will generate more accurate treatment effects. Based on their impact on clinical results, a proper understanding of these concepts is central for researchers, rehabilitation clinicians, and other stakeholders working on this field.