A STEPP Forward in Tailoring Treatment: New Research on the STEPP Methodology

• Activity Number: 124
• Type: Contributed
• Date/Time: Monday, August 1, 2016 : 8:30 AM to 10:20 AM
• Sponsor: Section on Risk Analysis
• Author(s): Wai-Ki Yip*
• Companies: Dana-Farber Cancer Institute
• Keywords: subgroup analysis ; subpopulation treatment effect pattern plot (STEPP) ; randomized clinical trial (RCT) ; meta-analysis ; generalized linear model (GLM)

Abstract

STEPP (Subpopulation Treatment Effect Pattern Plot) is a non-parametric approach to perform subgroup analysis in clinical trial setting. The original method is developed for time-to-event censored data. We have extended the STEPP methodology beyond the original setting: 1. A new window type based on events instead of the number of patients can be used in STEPP. 2. The STEPP methodology has been extended to analyze clinical trials with continuous, binomial and count end points via Generalized Linear Models. 3. It is common to pool results from several clinical trials in a meta-analysis to combine evidence and to draw a stronger conclusion. We extend STEPP to combine evidence from many related clinical trials. This new technique is called meta-STEPP. 4. Finally, software extension to R stepp software for the new window type and GLM stepp to show patterns for more than two treatments on the same plot. Availability of these capabilities in software helps make subgroup analyses easier.