Cross-sectional evaluation of outcome comparison methodology in dermatologic clinical trials and a simulation-based discussion of why outcome comparison technique matters-Reference-Cited by-同舟云学术

Cross-sectional evaluation of outcome comparison methodology in dermatologic clinical trials and a simulation-based discussion of why outcome comparison technique matters

Published:2023-02-10 Issue: Volume: Page:
ISSN:
Container-title:
language:
Short-container-title:

Author:

Hopkins Zachary H.¹,Swink J. Michael¹,Kaprive Jessica Forbes²,Carlisle Ryan¹,Olayinka Jadesola³,Moreno Christopher⁴,Frost Zachary⁵,Secrest Aaron M.¹

Affiliation:

1. University of Utah

2. HCA Lewisgale Montgomery, Virginia College of Medicine

3. Icahn School of Medicine at Mount Sinai

4. University of Arizona

5. Noorda College of Osteopathic Medicine

Abstract

Abstract Clinical trial outcome comparison methodology can affect trial efficiency and interpretation. Our goal was to describe current trends in clinical trial outcome comparison methods among high-impact dermatology trials and perform simulations comparing direct comparison (DC), change-score based (CS), and baseline-adjusted analyses (BAA) to demonstrate the more subtle differences in these methods. We performed a cross-sectional examination of top-cited dermatology clinical trials over five years. Trials were retrieved from journal websites and extracted from PubMed. All clinical trials published in The Journal of the American Academy of Dermatology (JAAD), JAMA Dermatology, Journal of Investigative Dermatology (JID), The British Journal of Dermatology (BJD), and The Journal of the European Academy of Dermatology and Venereology (JEADV) from 2015–2019. Clinical trial analysis type including DC, CS, BAA, percent change from baseline (PoB), and responder analysis (RA). We evaluated the proportion of trials reporting each endpoint were reported and simulations assessing impact of analysis type on outcomes. Among 252 eligible trials, outcome comparison techniques included DC (75/252), CS (40/252), PoB (36/252), RA (98/252), and BAA (3/252). Among trials using CS, PoB, or RA, 25/174 adjusted for baseline score. No trials discussed relationship between baseline and final values; 85/252 selected patients based on baseline score; 8/85 gathered a post-randomization baseline score; 20/85 used a pre-randomization run-in period. Simulations demonstrated that BAA maximized trial efficiency for continuous outcomes. CS, PoB, and RA were common; however, when trials have continuous or ordinal outcomes, BAA is most efficient. PoB and RA are particularly inefficient. RA can promote misleading inference. Use of BAA optimizes inference, trial efficiency, and resource utilization.

Publisher

Research Square Platform LLC

Reference26 articles.

1. 1. S. Senn, Statistical issues in drug development, Third edition, John Wiley and Sons, Ltd, Hoboken, NJ, USA, 2021.

2. 2. F. Harrell, Responder Analysis: Loser x 4, Data Methods. (2019). https://discourse.datamethods.org/t/responder-analysis-loser-x-4/1262 (accessed April 20, 2021).

3. 3. J.M. Bland, D.G. Altman, Comparisons against baseline within randomised groups are often used and can be highly misleading, Trials. 12 (2011) 264. https://doi.org/10.1186/1745-6215-12-264.

4. 4. A.J. Vickers, The use of percentage change from baseline as an outcome in a controlled trial is statistically inefficient: a simulation study, BMC Med. Res. Methodol. 1 (2001) 6. https://doi.org/10.1186/1471-2288-1-6.

5. 5. J.M. Bland, D.G. Altman, Best (but oft forgotten) practices: testing for treatment effects in randomized trials by separate analyses of changes from baseline in each group is a misleading approach, Am. J. Clin. Nutr. 102 (2015) 991–994. https://doi.org/10.3945/ajcn.115.119768.