A cross‐lagged analysis of measured and self‐estimated intoxication: Mobile health data from a large United States sample examined 2016 to 2022

BACKGROUND AND AIMS: Feedback from mobile health technologies can facilitate self-awareness of intoxication among substance users, thereby reducing risky decision-making and potentially moderating consumption; however, real-world effects of mobile device engagement among drinkers remain underexamined. The current study leveraged longitudinal data from mobile breathalyzers via cross-lagged analyses to examine reciprocal pathways between breath alcohol concentration (BrAC) levels and BrAC self-estimation error. DESIGN: Observational study analyzing BrAC data collected via mobile breathalyzers and shared by the device manufacturer. SETTING: Participants were mobile breathalyzer users located within the United States from January 1, 2016, to February 28, 2022, who voluntarily submitted data during active real-world drinking episodes. PARTICIPANTS: 18 800 unique user accounts provided a total of 387 643 breathalyzer readings. MEASUREMENTS: BrAC self-estimation error was calculated as the absolute difference between a measured BrAC level and the corresponding BrAC self-estimation. RESULTS: Results of cross-lagged models indicated an asymmetrical relationship between BrAC levels and self-estimation error. In moment-to-moment analyses, a 0.02 increase in BrAC level at the preceding time point predicted a 2.35% increase in self-estimation error at the following time point, b = 1.16, p < 0.001, 95% confidence interval (CI) [1.08, 1.25]. Conversely, after participants received BrAC feedback, a 0.02 increase in self-estimation error predicted a subsequent 1.15% decrease in BrAC level, b = -0.58, p < 0.001, 95% CI [-0.64, -0.52]. Similarly, a tendency to underestimate at earlier time points predicted later decreases in BrAC levels, b = -0.07, p < 0.001, 95% CI [-0.08, -0.07]. SUMMARY: There appears to be a reciprocal, albeit asymmetrical, relationship between objectively measured (via mobile breathalyzer) and subjectively self-estimated intoxication. When drinkers are provided breath alcohol concentration feedback, inaccuracies in self-estimated intoxication are followed by behavioral adjustments in real-world drinking contexts.