The study was conducted in 17 adult-serving EDs in the northeastern US urban and suburban locations; it is part of a larger project examining SMS text messaging interventions for UAU [29].

This study was approved by the institutional review board at the Feinstein Institutes for Medical Research (approval number 21-0756). All participants consented to study procedures using verbal consent processes and consent information sheets. All qualitative data were deidentified prior to analysis; survey data were anonymous. Patients received a US $15 gift card incentive; staff participants were not compensated. Generative artificial intelligence was not used in any part of the manuscript writing.

We used an explanatory, sequential mixed methods approach (quantitative → qualitative) [30]. Data were integrated at several points. Quantitative EHR and survey data classified sites on low or high implementation potential; classifications were used for data collection and site selection for qualitative interviews and analysis. Data were integrated during result interpretation. Reporting is in accordance with the COREQ (Consolidated Criteria for Reporting Qualitative Research) and CHERRIES (Checklist for Reporting Results of Internet e-Surveys) checklists for qualitative health research involving interviews and internet e-surveys.

The 3-item AUDIT-C (Alcohol Use Disorders Identification Test-Consumption) [31] detects UAU (≥3 for women and ≥4 for men) and was built into all participating ED EHRs; it provided information on each ED’s capacity to address UAU (eg, low screening completion rates may indicate a lack of time). Data, extracted from 2021, included percentage of patients screened (number of AUDIT-C completions divided by the total patient census).

ED physician chairpersons from the 17 participating EDs received an email from the study team inviting them to complete a 13-item web-based survey of three validated 4-item scales assessing conceptually distinct, but interrelated constructs often used in formative research to examine implementation barriers: feasibility (eg, “A text messaging intervention for UAU seems possible in my ED”), acceptability (eg, “I would welcome our clinical team offering a text messaging intervention for UAU to patients during their ED visit”), and appropriateness (eg, “A text messaging intervention for UAU seems fitting for patients in my ED”) [32]. An additional confidence item (“I am confident that our clinical ED team would offer a text messaging intervention for UAU to patients during their ED visit”) was included. All items were rated on a 5-point scale (1=completely disagree to 5=completely agree). To complete site classification, we calculated each site’s survey totals (range 4‐65; higher values indicate positive implementation outcomes). All invited EDs participated; in cases where both primary and secondary chairs completed the survey, scores were averaged.

We classified low versus high implementation potential based on AUDIT-C completion rates and survey scores (low: ≤median on both measures; high: >median on both measures), resulting in 5 low sites and 4 high. Sites with mixed classifications were not recruited. We considered several factors (eg, census and location) to select the 4 final EDs for interviews (2 low and 2 high) using stratified purposeful sampling [33,34].

In the 4 final EDs, chairpersons were interviewed and then each identified staff. They were invited via emails from study staff, resulting in 18 (82%) out of 22 invited interviewies. Interviews lasted ~30 minutes, were conducted over Zoom, recorded, and cleaned for analysis. Interview guides covered views on alcohol screening, SMS text messaging interventions and how to best offer them, barriers and facilitators to intervention implementation, and general practice change.

Patients were recruited from 6 EDs; sites were selected based on location and census to gain perspectives from different populations and ED sizes (eg, race or ethnicity and urban or suburban). Patients were approached during their ED visit if they screened positive on the AUDIT-C. Of the 30 patients invited, 21 (70%) were interviewed. Interviews lasted ~20 minutes and were recorded and transcribed. The interview guide was adapted from previous ED studies (eg, mobile device use and interest in receiving supportive messages on UAU reduction) [35,36].

We used rapid qualitative analysis [37] to analyze interview data following procedures previously outlined [37,38]. Two researchers (MO and AS) summarized 3 transcripts from patient and staff interviews to ensure consistent summaries. One researcher (AS) completed the remaining summaries with the second researcher (MO) reviewing for accuracy and alignment; discrepancies were resolved via discussion. For patients, we created comparison groups between AUDIT-C scores indicative of moderate to high risk (score 4‐7) versus severe risk (score 8‐12) to examine response differences; for staff, we created comparison groups by low versus high implementation potential sites. See Multimedia Appendix 1 for demographic information for chair surveys (n=26), staff (n=18), and patient interviews (n=21).

Average AUDIT-C completion rates (73%) were relatively high but ranged greatly (35%‐93%). Survey implementation scores indicated a relatively high level of confidence, acceptability, and appropriateness regarding the UAU texting intervention; feasibility had the lowest mean, indicating logistical implementation concerns. See Multimedia Appendix 2 for absolute values on AUDIT-C completion rates and implementation scores.

Table 1 illustrates implementation barriers and facilitators. In comparing staff in low (n=7) versus high (n=11) implementation sites, low sites more frequently indicated addressing alcohol use was not a high priority. All staff discussed challenges to include information about the intervention in discharge papers, noting the paperwork is bulky and patients rarely read it. Facilitators were similar across high and low sites. Low sites discussed the difficulty of making changes, and that change takes time and requires strong justification. When asked about recent practice changes, low sites had fewer examples of what worked well and more frequently mentioned low staff morale, high turnover, and burnout.

Patient results indicated strong interest and potential benefit from a UAU text intervention with limited concerns (Table 2). When comparing patients who screened moderate to high (n=6) versus severe (n=15) on the AUDIT-C, only patients at moderate to high risk said they had a goal to use the intervention to prevent future heavy drinking, while both groups mentioned abstinence and moderation. Perceived benefits also varied. Patients at severe risk thought the program could help stop the progression of alcohol use disorder symptoms, change their lifestyle, and be a distraction from drinking; patients at moderate to high risk thought the program could improve their job roles and family relationships, and provide information about drinking and change their thoughts about drinking. Both groups thought the intervention could provide support and improve their health and had similar content preferences. Patients at severe risk additionally mentioned using the text intervention to track drinking, set and complete goals, engage family support, avoid people who trigger their use, and reduce heavy drinking during drinking episodes.