Evaluating Imaging Inclusion Criteria for Stroke Clinical Trials: Appropriate Enrichment Versus Cherry Picking

• Activity Number: 664 - Three Pillars of Successful Health Studies: Inclusion, Design, and Measurement
• Type: Contributed
• Date/Time: Thursday, August 3, 2017 : 10:30 AM to 12:20 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Robyn L Ball* and Bin Jiang and Manisha Desai and Max Wintermark
• Companies: Stanford University and Stanford University and Stanford University and Stanford University
• Keywords: clinical trials ; stroke ; epidemiology ; study design

Abstract

We compared the imaging inclusion criteria of MRCLEAN, EXTEND-IA, ESCAPE, and SWIFT PRIME by utilizing an independent registry and assessed whether the inclusion criteria resulted in appropriate enrichment or suffered from selection bias (cherry picking). The registry includes imaging biomarkers, demographics, treatment course, and clinical outcomes of 612 stroke patients. Using bootstrap methods and published trial results, we compared the number needed to treat to achieve a success (NNT) and the number needed to screen to achieve a success (NNS) and assessed differences among patients that may have been included in trials with stricter inclusion criteria (strict trials) and patients that may have been included in trials with more relaxed inclusion criteria (relaxed trials). Strict trials resulted in a lower NNT but although there were significant differences in imaging biomarkers among patients in strict trials compared to patients included in relaxed trials, these differences did not translate to significant differences in treatment course or clinical outcomes. This innovative method can be applied to a larger set of pooled trial data, if made available, to inform future trials.