Background: Automated SMS text messaging as an adjunct to psychotherapy treatment for depression may help patients feel more supported in between therapy sessions, thus decreasing dropout in live intervention. However, more research is needed into how to design and tailor text messaging content and delivery to increase user engagement with text messages, with the goal of improving therapy attendance and treatment adherence.
Objective: This study aimed to explore user engagement by assessing 1) response patterns over time to automated daily mood monitoring messages and cognitive behavioral therapy (CBT) messages; 2) types of text messages most likely to elicit a response; and 3) feedback related to perceived benefits of a text messaging adjunct.
Methods: We included English- or Spanish-speaking primary care patients who attended outpatient group therapy for depression and participated in an automated text-messaging adjunct. We examined differential response rates to message types and explored common themes and key words among messages with the highest response rates. Quantitative and qualitative data were compiled and integrated to provide recommendations for improving user engagement.
Results: Survey data and text message responses from 63 patients were analyzed. Average response rates decreased over time. Response rates varied significantly across message types: the average response rate across mood monitoring messages was 56.0%; for CBT messages, 32.2%; and for feedback messages, 43.1%. Messages containing a closed-ended question tended to have higher response rates than messages containing an open-ended question, tip, suggestion, or positive reinforcement. Qualitative data from feedback questions suggested that participants found the text messages to be useful and engaging, and they felt more supported and connected throughout the intervention due to this adjunct. Lack of response to text messages was often due to inconvenient timing of messages.
Conclusions: Overall, the text messaging adjunct was positively received, and patients were engaged. Messages should include a combination of questions, tips, suggestions, and personalized feedback for maximum impact. Machine learning and AI approaches may be a useful addition to maximize the relevance of messages and optimize timing of message delivery to participants. Findings from this study contribute to our understanding of how to design text messages to supplement standard therapy and increase engagement.