Detecting Alcohol Consumption Among Homeless Individuals Using Ecological Momentary Assessment, Transdermal Sensors, And Timeline Follow Back Methods

Purpose: The present study examined the extent to which self-reported measures of alcohol use from ecological momentary assessment (EMA) among homeless drinkers corresponded with estimates from a transdermal alcohol sensor (SCRAM) and self-reported timeline follow-back (TLFB) recall measures. Methods: Participants were 63 homeless adults who were receiving services at a homeless shelter in Dallas, TX. Participants' alcohol consumption data were collected via EMA, SCRAM sensor, and a TLFB recall measure at the 4-week follow-up. For each assessment approach, we created two daily alcohol use variables: any use (1= alcohol use positive or 0 = alcohol use negative) and alcohol use quantity. We analyzed data using multilevel models, calculated intraclass correlation coefficients for inter-rater agreement, and estimated pairwise correlations and means across all three assessment methods. Results: Across the three assessment methods, the intraclass correlation coefficient for inter-rater agreement was 0.81 for any alcohol use and 0.76 for alcohol use quantity, indicating excellent agreement. Furthermore, the EMA assessed the quantity of alcohol used was highly correlated with SCRAM peak transdermal alcohol concentration estimate, whereas TLFB had low to modest correlations with EMA and SCRAM measures of alcohol use quantity. Conclusions: Compared with a transdermal alcohol measure, EMA is a valid measure of alcohol use among homeless drinkers. Given the substantial day-to-day variation in alcohol consumption and the ease of EMA compared to biological measures, EMA-based measures of alcohol consumption may be an important tool for clinical research, especially among underserved populations.