A Digital Cognitive-Behavioral Intervention for Depression and Anxiety Among Adolescents and Young Adults

Overall, 506 adolescents and young adults received an RxWell prescription. Of the 278 who enrolled, 217 (78%) were female, 240 (86%) were White, and 184 (66%) had commercial insurance; the mean±SD age was 19.0±2.6 years, and 181 (65%) were receiving therapy or antidepressants from their pediatric practice. Of 275 patients with EHR data before an RxWell order, 39 (14%) were on an antidepressant only, 74 (27%) were receiving therapy only, 66 (24%) received both an antidepressant and therapy, and 96 (35%) had received none of the two treatments. At baseline, 82% (N=228) of patients’ GAD-7 and PHQ-8 scores were consistent with an anxiety or depression diagnosis (scores >8), but for many participants, no formal diagnosis was assigned on the basis of structured diagnostic criteria during the index visit because of busy workflows at the clinic. For a subset for whom RxWell was an adjunct to behavioral treatment, the vast majority were diagnosed as having anxiety or depressive disorders.

In total, 67% (N=339) of RxWell prescriptions were from pediatricians and the remainder from embedded behavioral therapists. In total, 220 (79%) of the patients completed at least one RxWell technique, 161 (58%) engaged in the app, and 176 (63%) messaged their coach. The median number of techniques used was four; mean length of app use was 25 days, which were not necessarily consecutive. Of patients receiving behavioral health treatment before downloading the app, 81% (N=114 of 140) continued to see a therapist with RxWell supplementation.

RxWell use was associated with a significant decrease in anxiety and depression scores between baseline and 1 month and up to 3 months (Table 1). Subgroup analyses revealed that patients receiving dCBI alone had a similar magnitude of reduction in anxiety (mean±SD change=3.6±4.5; from baseline to the last score over the 3-month period) compared with those for whom dCBI supplemented antidepressants alone (3.8±4.8), therapy alone (3.2±4.8), or both (2.9±5.3). We observed 44 risk escalations involving 38 patients, with the majority (84%, N=37) deemed being at low risk, and no serious adverse events (e.g., suicide attempts) occurred.

Interviews about app use were conducted with 24 randomly selected participants who enrolled in RxWell and 10 who did not enroll. Enrolled patients provided two main reasons for app use: current mental health symptoms and free time. Most described using the app when they were anxious or in stressful situations. Common reasons for not enrolling in the app included time constraints, other support, disliking therapy, or using other apps. Some stated that more information about the app from providers would have been beneficial.

Enrolled patients discussed the benefits of a “real person” (i.e., the coach) who provided accountability, asked how content was helpful, sent reminders, and provided reassurance. Reasons for no interaction with the coach were limited app use and lack of awareness that the coach was a real person. App users described how techniques and goal setting supported their mental health. Many shared how app content and their coach promoted their ability to stay grounded during challenging times.

When asked about barriers that RxWell helped overcome, behavioral health providers endorsed that RxWell offered support when providers were unavailable and when patients were ambivalent about therapy. The providers also believed that RxWell improved the efficiency of health care delivery.

The results of this study indicate that it is feasible to embed a dCBI for emotional distress within outpatient pediatric clinics, with app enrollment rates being comparable to published rates of follow-through for in-person mental health services (8). The rate of clinical improvement in the severity of emotional distress observed in this study from baseline to ≥1 month is consistent with those reported for Internet- and computer-based therapies for adolescents and young adults (9). However, because our study design did not include a control group, we could not conclude that these improvements were due to the dCBI or determine how much dCBI enhanced outcomes over concurrent mental health treatment. Overall, enrolled patients who completed the qualitative interview shared positive experiences and believed that they had benefited from having the dCBI during a time of distress and from communicating with a real-life coach. The availability of a coach-enhanced dCBI was particularly impactful during the COVID-19 period when emotional distress among adolescents and young adults was elevated and incorporation of health technology into routine medical care became the new status quo. The use of a new digital tool to provide access to behavioral care or enhance the benefit of existing treatment for individuals making little or no treatment progress represents added value.

Strengths of this study included that the dCBI was nested within pediatric primary care, that pediatricians directly ordered the app through the EHR, and that the project included a sophisticated risk escalation protocol. We had a large sample and an empirical process for learning while RxWell was incorporated into routine care. Further studies are needed to determine a standard evidence basis for evaluating digital interventions (10), but our findings are consistent with a positive effect of the app alone and as a supplement to behavioral treatment.

Limitations included that this was an open-trial quality improvement project. An internal review board–approved, randomized controlled trial to evaluate efficacy of dCBI compared with a control group will be conducted to further validate the impact of the dCBI. This future study will recruit participants from clinics with a large minority and Medicaid population. Examination of the relationship between user engagement and outcomes would further strengthen links between the intervention and clinical improvement. It would also be of benefit to better delineate baseline characteristics of patients to determine whether those meeting diagnostic criteria for anxiety or depressive disorder would benefit from this type of dCBI as a first-line intervention in a stepped behavioral intervention or whether the dCBI would be better suited for preventing progression of mild or moderate emotional distress. Furthermore, most app users came from oversampled groups (i.e., White and female), and patients who did not enroll were more likely male and Black. The study highlights the need for steps to improve engagement for individuals from minority populations, which could include training coaches in cultural sensitivity. Additionally, we will provide culturally competent language for clinicians to use when discussing the app with patients. Some attrition occurred because self-reported assessments were captured only within the app.

As a quality improvement project, implementation changes were initiated at midpoint on the basis of lessons learned, including enhanced prescriber education, improved care coordination with therapists, and changes in messaging that emphasized that the coach was a real person. The resulting dCBI received high approval ratings by patients and was seen by prescribers as improving care.

With the next phase of planned validation and efficacy, another barrier to address will be cost reimbursement. UPMC offers RxWell as a health care app to employers, health care organizations, and health plans, enabling organizations to provide an evidence-based dCBI to their employees, members, and patients free of charge, while the organization absorbs a minimal per-use or per-eligible-member fee. Inclusion of digital tools in shared savings models or reimbursable CPT codes could provide sustainable scalable integration into clinical care. Our results indicate that it is feasible to foster engagement with digital behavioral interventions offered as part of routine pediatric primary care and that dCBI is associated with a reduction in the severity of anxiety and depression among adolescents and young adults.

The authors thank Kate Saucier, Meredith Strassburger, Nicole Kelly, and Karen Depperman for ensuring the success of this project by working with the donors or clinicians and patients. They also thank YourMomCares for providing a gift in kind to Children’s Hospital of Pittsburgh Foundation for funding of this project.