A tale of confusion from overlapping confidence intervals: brief report

In clinical research presentations, study results are commonly reported in the form of p-values and confidence intervals as an estimate of association and treatment effect. The interpretation of confidence intervals that overlap can be confusing and difficult for the reader to draw clinically meaningful conclusions. In this brief report, we describe the basics of confidence intervals and present an example from a recently published randomized control trial (RCT) to illustrate a common confusion that overlapping confidence intervals between the means of two independent groups may not necessarily reject the true significant difference of effect. It is recommended that investigators use the direct difference of means between groups for confidence interval estimation to reduce Type II errors. Clinicians should interpret overlapping confidence intervals with caution, and avoid the assumption that overlapping confidence intervals always implies a lack of difference of treatment effect to decide application of treatment. Corresponding Author: Dinesh Kumbhare MD, PhD, FRCPC, FAAPMR, Department of Medicine, Division of Physical Medicine and Rehabilitation, University of Toronto, dinesh.kumbhare@uhn.ca, Telephone: 416-597-3422 x 4612 Author Disclosures: All the authors report no conflict of interest for the completion of this manuscript. Also, no funding was received. Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.

https://ift.tt/2BlVacm