Statistical design considerations for pilot studies transitioning therapies from the bench to the bedside
- PDF / 178,428 Bytes
- 3 Pages / 610 x 792 pts Page_size
- 26 Downloads / 172 Views
BioMed Central
Open Access
Commentary
Statistical design considerations for pilot studies transitioning therapies from the bench to the bedside Rickey E Carter* and Robert F Woolson Address: Department of Biostatistics, Bioinformatics and Epidemiology, 135 Cannon Street, Suite 303, Medical University of South Carolina, Charleston, SC 29425, USA Email: Rickey E Carter* - [email protected]; Robert F Woolson - [email protected] * Corresponding author
Published: 28 October 2004 Journal of Translational Medicine 2004, 2:37
doi:10.1186/1479-5876-2-37
Received: 07 October 2004 Accepted: 28 October 2004
This article is available from: http://www.translational-medicine.com/content/2/1/37 © 2004 Carter and Woolson; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Abstract Pilot studies are often used to transition therapies developed using animal models to a clinical setting. Frequently, the focus of such trials is on estimating the safety in terms of the occurrence of certain adverse events. With relatively small sample sizes, the probability of observing even relatively common events is low; however, inference on the true underlying event rate is still necessary even when no events of interest are observed. The exact upper limit to the event rate is derived and illustrated graphically. In addition, the simple algebraic expression for the confidence bound is seen to be useful in the context of planning studies.
Introduction In the translational research setting, statisticians often assist in the planning and analysis of pilot studies. While pilot studies may vary in the fundamental objectives, many are designed to explore the safety profile of a drug or a procedure [1,2]. Often before applying a new therapy to large groups of patients, a small, non-comparative study is used to estimate the safety profile of the therapy using relatively few patients. This type of investigation is typically encountered in the authors' experiences as collaborating biostatisticians at our General Clinical Research Center as well as developing applications addressing the National Institutes on Health Roadmap Initiative http://nihroadmap.nih.gov/. In the context of pilot studies, traditional levels of α (the Type I error rate) and β (the Type II error rate) may be inappropriate since the objective of the research is not to provide definitive support for one treatment over another [3]. For example, the null hypothesis in a single arm pilot study might be that the tested intervention produces a
safety profile equal to a known standard therapy. A Type I error (rejecting the null hypothesis when it is false) in the context of this preliminary investigation would encourage additional examination of the treatment in a new clinical trial. This is in contrast to a Type I error in a Phase III/IV cl
Data Loading...
