BACKGROUND AND AIMS: Several randomised controlled trials (RCTs) have demonstrated that delivering approach bias modification (ApBM) during residential alcohol use disorder (AUD) treatment helps prevent post-treatment relapse. However, few studies have examined ApBM's efficacy for AUD in outpatients. We trialled a personalised ApBM smartphone app in individuals receiving outpatient AUD treatment.

DESIGN: This double-blind RCT randomised participants to receive ApBM or sham training, adjunctive to treatment as usual.

SETTING: Participants were recruited from alcohol and other drug treatment services in Melbourne, Australia.

PARTICIPANTS: 79 participants (mean age 46.6 years; 45 male, 34 female) installed the app between May 2022 and January 2024.

INTERVENTION AND COMPARATOR: In the ApBM condition, the app delivered personalised, gamified ApBM. Notifications prompted participants (n = 39) to complete 2 ApBM sessions weekly for 4 weeks. The control version prompted participants (n = 40) to complete a weekly sham-training task for 4 weeks.

MEASUREMENTS: The primary outcome was number of standard drinks (10 g pure alcohol) consumed in week 4 of the intervention period, self-reported in the app. Secondary outcomes included past-week standard drinks at 8-week and 16-week follow-ups, past-week drinking days, past-week heavy drinking days (HDDs; days when ≥5 standard drinks were consumed) and questionnaire measures of AUD severity, quality of life and alcohol craving. Primary analyses followed an intention-to-treat (ITT) approach, with secondary complete-case sensitivity analyses also conducted for all outcomes.

FINDINGS: Groups did not statistically significantly differ in the primary outcome [values from ITT negative binomial model: ApBM = 75.49 standard drinks, control = 71.34 standard drinks, difference = 4.16, 95% confidence interval (CI) = -42.37 to 50.69, P = 0.859]. Most analyses of secondary outcomes showed statistically non-significant effects, with the only exception being past-week standard drinks at the 16-week follow-up, where ApBM participants showed statistically significantly larger decreases than controls in past-week standard drinks (reduction of 14.6, relative to baseline, versus 2.1 in controls; interaction β = -12.53, 95% CI = -23.85 to -1.22; P = 0.030). Time x group interaction effects were statistically non-significant for all other secondary outcomes (Ps > 0.069).

CONCLUSIONS: A smartphone app using approach bias modification showed no evidence for reducing alcohol use among alcohol use disorder outpatients after 4 weeks, or evidence for effects on most secondary outcomes, although 16-week follow-up results suggested that approach bias modification may have facilitated delayed/longer-term reductions in alcohol use.