Digital Mental Health Service for Non-Treatment Seeking Young Adults

The primary purpose of this trial is to test the feasibility of conducting a randomized controlled trial of digital mental health intervention (DMHI) for non-treatment seeking young adults with symptoms of depression or anxiety, comparing an 8-week adaptive (personalized) messaging intervention, an 8-week non-personalized digital mental health intervention, or an active control which will deliver psychoeducational content via clickable URLs. The adaptive intervention will be powered by reinforcement learning to deliver tailored SMS messages based on user profiles and the ways in which a user interacts with the intervention system, along with links to longer psychoeducational content. The non-personalized intervention arm will not personalize the messages, but will still be an active treatment. This study will examine the adaptive intervention's ability to engage users and effectiveness in decreasing psychological distress relative to a non-personalized intervention. Participants who become disengaged (5 continuous days of not clicking on content URL links, responding to messages, or message ratings) from the adaptive or static intervention arms at any point in the first two weeks of treatment will be randomized to receive low-intensity human coaching or continue without coaching for the remainder of the intervention period.

Recruitment will take place via Mental Health America's (MHA) website. MHA is a large mental health advocacy organization that hosts online screeners and has large number of visitors the majority of whom meet clinical cutoffs for depressive or anxiety disorder symptoms. This study will enroll individuals who have completed a depression or anxiety disorder screener on MHA's website, and meet clinical cut-offs on the Patient Health Questionnaire - 9 (PHQ-9) or Generalized Anxiety Disorder - 7 (GAD-7).

Initial randomization will be generated in permuted blocks of 5 using a computer program. To prevent allocation bias, randomization will be conducted by the biostatistician, who will not inform the study team of the treatment arms until the baseline assessment has been completed and the patient has been enrolled. Second randomization of disengaged participants will occur as disengagement is detected during the first two weeks. Half (50%) of disengaged individuals will be randomized to receive additional human coaching throughout the remainder of the study while remaining in either the adaptive or non-personalized treatment arms. The disengaged participants not randomized to receive additional human coaching will continue to receive either the non-personalized or adaptive automated treatment the participants were randomized to initially.

Coaching will consist of providing users with support and accountability via positive reinforcement, goal and expectation setting, and monitoring. Coaching outreach will focus on adherence to the treatment but will not provide treatment advice. Coaches will provide a brief (15-minute phone call or equivalent depending on the medium) engagement session over the phone, SMS, or email, depending user preference. Thereafter, coaches will check in with participants via messaging, phone call, or email, twice per week and respond to patient texts.

Participants will be assessed using self-report measures at baseline, weeks 4, 8, 12, and 16 weeks.