(PS8-D96) Addressing Barriers to Care: Is a Live Video Session Required to Build a Working Alliance?

Although clinical science has identified several effective, evidence-based treatments for a variety of mental illnesses, widespread access to this care is still hampered by high costs of treatment and a persistent shortage of mental healthcare providers, the effect of which is particularly salient in rural communities.  This state persists despite evidence that cognitive behavioral treatment (CBT) is equally effective regardless of delivery medium. Digital messaging therapy is a relatively cost-effective and scalable ecological momentary intervention (EMI) which fills key access gaps in the present mental health treatment landscape. Yet, prompted by concerns over the establishment of a working alliance and the security and/or privacy of the patient, some US states established telemental health care laws requiring that synchronous audio- or video-based live sessions (LVS) be held at the outset of treatment. To date, it is unknown whether having these mandated synchronous sessions lead to better engagement and outcomes, but varying state-wide regulation provides an ideal context in which to empirically investigate this question.
In the present study, we examined the effect of an introductory LVS on user engagement for message-based care compared to message-based care without a LVS. Data was collected on 3502 users of a large digital mental health platform, including 1808 users from states mandating an introductory LVS (Arkansas, Idaho, Louisiana, Ohio, and Montana) and a matched sample of 1694 users in states not requiring an LVS.  All participants joined the platform in December 2021. There was no statistically significant difference in likelihood of cancelation at four weeks for LVS vs no LVS (OR 1.05, 97.5% CI [0.36, 3.15]). To analyze differences in outcomes, we also ran a pattern mixed model analysis controlling for cohort and dropouts while allowing correlation for intercept deviations due to individual user effects.  There was no statistically significant difference in GAD-7 scores as an effect of time (b = -2.77219, SE =0.05582, t = -49.665) or group (b = -0.44213, SE = 0.18962, t = -2.332).  Similarly, we found no statistical significance in PHQ-9 as an effect of time (b = -2.68431 , SE = 0.06697, t = -40.081) or group (b = -0.12381 , SE = 0.23478, t = -0.527).  In the context of a natural experiment, state-mandated introductory LVS were not shown to have an effect on user engagement and outcomes.