JMIR Publications

ABSTRACT

Background:

The COVID-19 pandemic has accelerated the adoption of digital tools to support individuals struggling with their mental health. The use of a digital intervention plus human coaching (“dual” intervention) is gaining momentum to increase overall engagement in digital cognitive behavioral interventions (dCBIs). However, there is limited insight about the methodologies and coaching models used by those deploying dual interventions. To achieve a deeper understanding, we need to identify and promote effective engagement that leads to clinical outcomes, versus simply monitoring engagement metrics. Motivational Interviewing (MI) is a collaborative, goal-oriented communication approach with particular attention to the language of change and an effective engagement approach for helping people manage mental health issues. However, this approach has been traditionally used in in-person or telephonic interventions and less is known about the application of MI to digital interventions.

Objective:

We sought to provide a dual intervention approach and address multiple factors across two levels to operationalize a dCBI that combined: (1) Cognitive behavioral therapy (CBT)-based techniques; and (2) MI-based interactions between the digital health coach (DHC) and user.

Methods:

We reviewed hundreds of digital exchanges between DHCs and users to identify and improve training and quality assurance activities for digital interventions.

Results:

We tested five hypotheses and found that: (1) users of a dual digital behavioral health intervention had greater engagement levels than users of a non-coached intervention (p<0.0001); (2) DHCs with a demonstrated competency in applying MI to digital messages had more engaged users, as measured by DHC-to-user message exchange ratio (p<0.001); (3) DHC-to-user message exchange ratio was correlated with more engagement in app activities (r = 0.28 (95% CI [0.23,0.33])); (4) DHCs with demonstrated MI proficiency elicited greater amount of “change talk” from users than did DHCs without MI proficiency (H = 25.12, p<0.0001); and (5) users who were engaged by DHCs with MI proficiency had better clinical outcomes compared to users engaged by DHCs without MI proficiency (p= 0.0151).

Conclusions:

This data indicates potential and need for further research in establishing coaching models in dCBIs that incorporate MI to promote effective engagement and optimize positive behavioral outcomes.