The surgery types performed at this health system are RYGB and SG. We plan to enroll 100 participants who are between 6 months and 5 years after MBS. This will allow us to capture a range of drinking behavior, from reinitiation (to minimize recall bias) to continued use, by enrolling those who may have begun to escalate their frequency and amount of alcohol use [4,31]. Patients who underwent MBS at our health system will be randomly selected in batches and contacted via email with a brief description of the study and a link to a brief online eligibility survey through REDCap (Research Electronic Data Capture; Vanderbilt University) software, a secure and Health Insurance Portability and Accountability Act (HIPAA)–compliant web-based program. Patients will be asked if they have consumed alcohol since surgery. If so, they will report the frequency of use over the previous month (ie, not at all, once per month, 2-3 times per month, once per week, 2-3 times per week, 4-6 times per week, or every day). Patients will be eligible if they have consumed alcohol at least 2 to 3 times in the previous month, ensuring enrollment of those with current alcohol use. This criterion was also selected, given our goal to capture at least 2 drinking episodes during the EMA period. Once eligibility is determined, a research assistant will contact eligible patients via email or phone to schedule a time to complete informed consent and the baseline assessment. After obtaining participant consent to participate in the study, the research assistant will email a link to the baseline assessment, which will be completed through REDCap. Participants will also complete the assessment battery through REDCap at 6- and 12-week follow-ups (Figure 1). Participants will receive up to 5 reminders via email or phone if they have not completed each assessment. In addition, after completing the baseline assessment, each participant will complete 3 weeks of EMA. The EMA will include brief assessments (eg, 2-3 min) delivered twice per day to each participant’s smartphone. We propose to collect data each morning (eg, 8 AM) and in the evening before bedtime (eg, 8 PM); administering 2 surveys will reduce recall bias while also being cognizant of participant burden. Morning and evening assessments are frequently used in other EMA studies that have used a time-based approach [30]. An automatic push notification from the EMA app will be sent to the participant to signal when the items can be completed. Participants will be informed that the survey will be available on their dashboard 2 hours before the push notification is scheduled to allow for flexibility in the event that the participant wishes to complete it before receiving the push notification. Automatic reminders will be sent at 1 and 2 hours after the push notification if the survey has not yet been completed. Each signal will expire 4 hours after the push notification is sent. During the consent appointment, the research assistant will instruct participants on how to download and use the EMA app. If a participant misses multiple EMA signals in a row, the research assistant will attempt to contact the participant via email or phone.

Measures for the distal factors will be administered at baseline and at 6 and 12 weeks unless otherwise noted (Table 1).

Proximal factors are measured during the EMA period. Table 2 presents the measures to be used in the daily EMA surveys. Completion of the complete set of items is expected to take less than 3 minutes.

This study is approved by the Henry Ford Health Institutional Review Board, where recruitment will take place (17061). Institutional review board approval was obtained in March 2024. To protect privacy, participants will be asked whether they consent to share their data in a repository. Participants who do not consent to this will not have their data shared. No identifying information of any participant will be shared or published. Participants will be compensated for their participation. They will receive US $50 for completion of each of the assessments at the baseline, 6-, and 12-week follow-ups. During the EMA, they will receive US $2 for each signal completed (up to US $4 per day) for a total of up to US $84 for the 21 days. Participants who complete at least 80% of the EMA signals will also receive a US $50 bonus.

This study received funding from the National Institute on Alcohol Abuse and Alcoholism (R21 AA029423) in May 2023. The primary goal of this study is to evaluate the feasibility of the study design. We will calculate the enrollment rate (the number of those who complete the baseline assessment of those eligible) and retention rates at the 6- and 12-week follow-ups. We anticipate that at least 50% of the eligible individuals will enroll and that at least 80% will complete the 6- and 12-week follow-ups. We will also calculate adherence to the morning and evening assessments during the EMA period. We anticipate that 75% of the participants or more will have at least a 75% response rate to the EMA assessments across the full 3-week EMA period.

The second goal of this study is to determine distal and proximal predictors of alcohol use. We plan to measure alcohol use in several ways. Our primary outcome is alcohol use on a given day during the EMA period (yes or no), and the secondary outcomes include the number of drinks and hazardous alcohol use. We will also explore whether there are differences in predictors for planned versus unplanned drinking. Data collection began in November 2024 and will conclude in December 2025.

We will conduct multiple regression analyses predicting alcohol outcomes at 6 and 12 weeks. Predictors will include baseline measures of depression, anxiety, emotional eating scores, number of food addiction symptoms, frequency of disordered eating behaviors (binge eating, loss of control eating, large portions, grazing, and unhealthy food choices), and drinking motives; we will control for sex and examine sex differences by testing interactions with sex. We will control for time since drinking reinitiation, given that there will be variability in timing. We hypothesize that negative emotionality will be associated with increased drinking in lieu of disordered eating. We will also produce path models that include data from all 3 time points (baseline and 6 and 12 weeks) in a single model within Mplus (Muthén & Muthén).

We will use continuous time survival analysis [49-52] to investigate initiation of alcohol use following MBS (time 0). We can predict time to first drink from time-invariant variables, such as sex and history of preoperative alcohol use.

To examine proximal predictors of drinking, we will use multilevel modeling [53,54], which accounts for clustering (clustering for days within each person) and permits varying numbers of observations to account for missing reports. Level 1 (L1) effects correspond to within-person effects (eg, proximal behaviors at the level of the day), and level 2 effects indicate between-person effects (eg, distal predictors such as sex or L1 variables such as eating behaviors or depression aggregated over time). Consistent with recommendations, level 2 effects will be grand mean centered, and L1 effects will be person mean centered, such that the L1 effect reflects a deviation from one’s own typical behavior [55,56]. Outcomes will include any drinking and amount of drinking as well as our measure of unplanned drinking; predictors will include same-day mood and eating behaviors. Moreover, we can use the survey data to code whether drinking precedes mood, follows mood, or occurs contemporaneously to examine directionality within a day; these categorical variables can be used in subsequent analyses. In addition, daily-level data can be lagged to examine whether there are mood effects on next-day alcohol use and vice versa. We will also examine daily associations among the mood and eating variables. We hypothesize that alcohol use will occur on days characterized by greater negative mood is present and in lieu of disordered eating, given that we expect maladaptive coping strategies (ie, alcohol use or disordered eating in response to negative mood) to be used at different times, and we can code directionality among mood and drinking and eating behaviors for use in analyses.