Evaluating a drink-counting and a breathalyzer-coupled app for monitoring alcohol use: A comparison with timeline followback and peth biomarker

This is the authors' abstract. We don't add key points for this paper.

• Alcohol use reported via app and breathalyzer was compared with traditional reports. • App and breathalyzer reports were also compared with an objective biomarker. • The app showed strong agreement with retrospective self-reports and objective data. • Breathalyzer data weakly aligned with retrospective and objective reports. • The app appears valid for tracking alcohol use, whereas the breathalyzer does not. Accurately measuring alcohol consumption remains a challenge. This study aimed to evaluate two app-based methods, one using drink-count logging and one using a breathalyzer, by comparing them to retrospective self-report (Timeline Followback, TLFB) and a biomarker of alcohol use (Phosphatidylethanol, PEth). Data were acquired from a randomized controlled trial involving alcohol-dependent adults (n = 110). Standard drinks, drinking days, and heavy drinking days reported via the drink-counting app or breathalyzer, were compared with TLFB data over the 12-week period using Lin’s concordance correlation coefficient (CCC). Correlation with PEth was assessed only at the 12-week mark, using Spearman’s rank correlation coefficient (rho) and Receiver Operating Characteristic (ROC) curves, reporting the area under the curve (AUC). Compared to app-based methods, TLFB consistently identified more drinking days and heavy drinking days. However, the drink-counting app’s estimates were still relatively close to TLFB and demonstrated strong agreement for drinking days across the different time intervals (CCC = 0.71–0.86). The drink-counting app also showed a strong correlation with PEth values for standard drinks and drinking days (rho = 0.74–0.78). In contrast, breathalyzer data generally showed weak agreement with both TLFB and PEth. Although TLFB yielded more drinking and heavy drinking days, the drink-counting app showed strong agreement with TLFB and correlated closely with PEth levels, indicating good validity. In contrast, breathalyzer data showed weaker agreement, likely due to lower usage during drinking episodes. These findings suggest that drink-counting apps could provide a reliable tool for monitoring alcohol use, offering advantages over both retrospective reports and breathalyzer measures.