Using Drug Screens to Inform Missing Data Mechanisms in Daily Drug Use and Mood Reports

• Activity Number: 62
• Type: Topic Contributed
• Date/Time: Sunday, August 3, 2014 : 4:00 PM to 5:50 PM
• Sponsor: Mental Health Statistics Section
• Author(s): Warren Comulada*+ and Dallas Swendeman and Robert E. Weiss
• Companies: University of California, Los Angeles Center for Community Health and University of California, Los Angeles and University of California, Los Angeles
• Keywords: Non-ignorable missing data ; ecological momentary assessment ; Bayesian model ; drug abuse treatment ; mood

Abstract

Monitoring relapse is a key element of drug treatment, yet ideal monitoring tools have yet to be found. Treatment programs typically rely on retrospective self-report that is subject to recall bias. In response, recent drug abuse studies have begun to incorporate two alternative data collection tools. Cell phone-based ecological momentary assessment (CEMA) enables participants to fill out daily reports and addresses shortcomings of traditional EMA, e.g., paper diary. Urine drug screenings provide more of a gold standard, but are collected less frequently, e.g. once a week. Both CEMA and screenings provide unique and valuable data on drug use patterns, but statistical approaches are needed to integrate them. We present such models in a Bayesian framework for daily drug use and mood ratings reported by CEMA. Missing data is a common problem and most likely informative, barring additional information. For example, individuals may be less likely to fill out CEMA on using days. Our models incorporate urine screenings to inform a missing data mechanism on drug use and mood. Motivating examples are provided by studies on HIV-positive adults and youth in drug abuse treatment.