Ecological Momentary Assessment Methods in Relapse Reporting

Introduction

It can be expensive or infeasible to collect and evaluate biological measures of substance use. Thus, several studies have examined concordance between recall-based self-reported use versus biological measures such as urine analysis. These studies have found a moderate level of concordance, which varies by factors such as: substance of use, age, and race/ethnicity. In recent years, ecological momentary assessment (EMA) has gained popularity as a real-time data collection tool in studies of substance use. As this method does not suffer from recall bias, EMA might be a more reliable self-report method for assessing substance use than recall-based reporting. Few studies have assessed the concordance between self-reported use collected by EMA and biological measures of use, particularly for illicit substance use. In this report, we examine the level of concordance between self-report using EMA methodology and biological measures of use among a diverse sample of licit and illicit substance users in treatment.

Methods

Participants completed three week-long EMA measurement bursts, with 6 weeks between each burst. During each burst, participants received four brief EMA surveys per day that assessed substance use. At the end of each burst, participants completed an exit survey in which they were asked to again report on past week substance use. Participants were asked to allow our study to access their clinic data, which included urinalysis test results and electronic medical records. We used survival analysis to compare time to relapse by self-report and by clinic data for the subset of 71 participants for whom we had access to clinical data. Relapse was defined as use of any substance that the participant was receiving treatment for, or use of any other illicit drug.

Results

Through combined self-report and clinical data, we estimated that 83% of the participants in this analysis experienced relapse over the course of the study: 55% self-reported relapse and clinic data indicated relapse for 65%. The level of agreement between self-report and clinic data was very poor (Cohen’s Kappa=.04; % Agreement = 54%).

Discussion

In some cases, participants did not self-report use by EMA or recall when clinic data showed use. This type of disagreement is expected, as participants are reporting on stigmatized or illegal behavior. However, we also observed cases in which participants self-reported by EMA or recall, but clinic data did not show use. This could occur for participants with infrequent urine tests, gaps in care, or for those who did not seek medical treatment for substance-related reasons. We are in the process of conducting additional analyses to determine which methods are more reliable for certain substances of use, or for certain demographics.