Robust methods to correct for measurement error when evaluating a surrogate marker

Concordance Between Point-of-Care Urine Ethyl Glucuronide Alcohol Tests and Self-Reported Alcohol Use in Persons with HIV in Uganda

Screening and assessing alcohol use accurately to maximize positive treatment outcomes remain problematic in regions with high rates of alcohol use and HIV and TB infections. In this study, we examined the concordance between self-reported measures of alcohol use and point-of-care (POC) urine ethyl glucuronide (uEtG) test results among persons with HIV (PWH) in Uganda who reported drinking in the prior 3 months. For analyses, we used the screening data of a trial designed to examine the use of incentives to reduce alcohol consumption and increase medication adherence to examine the concordance between POC uEtG (300 ng/mL cutoff) and six measures of self-reported alcohol use. Of the 2136 participants who completed the alcohol screening, 1080 (50.6%) tested positive in the POC uEtG test, and 1756 (82.2%) self-reported using alcohol during the prior 72 h. Seventy-two percent of those who reported drinking during the prior 24 h had a uEtG positive test, with lower proportions testing uEtG positive when drinking occurred 24-48 h (64.7%) or 48-72 h (28.6%) prior to sample collection. In multivariate models, recency of drinking, number of drinks at last alcohol use, and Alcohol Use Disorders Identification Test - Consumption (AUDIT-C) score were associated with uEtG positivity. The highest area under the curve (AUC) for a uEtG positive test was for recency of drinking. Overall, we concluded that several measures of drinking were associated with POC uEtG positivity, with recency of drinking, particularly drinking within the past 24 h, being the strongest predictor of uEtG positivity.