Missing Data and Other Problems Inherent in Analyzing Clinical Data Obtained from Programmable Medical Devices

• Activity Number: 509
• Type: Topic Contributed
• Date/Time: Thursday, August 11, 2005 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Andrew Mugglin*+
• Companies: Medtronic, Inc.
• Keywords: missing data ; clinical trial ; memory ; medical device ; bias ; pacemakers

Abstract

Many medical devices, such as cardiac pacemakers, are really implantable computers with ever-growing programmability and memory storage capabilities. Thus, they serve not only as medical treatment, but as continuously functioning monitors and recorders of biologic phenomena. Increasingly, such device-stored data are becoming endpoints in primary and secondary objectives of clinical trials. But this raises a number of challenges, including the specter of missing data. How do we know what the device has failed to record? What effect do memory limitations have on clinical endpoints? What if memory is limited differently for different types of biologic phenomena device models used in the same clinical trial? What if a comparison of treatment strategies involves separate programming regimens that differentially influence the kinds of endpoint events logged in memory? These and similar questions will be explored in the context of detection of arrhythmias in clinical trials involving cardiac pacemakers and implantable defibrillators.