JMIR Publications

ABSTRACT

Background:

Transition to adult healthcare for young people and young adults (YP/YA) with Type 1 Diabetes Mellitus (T1DM) starts around 11 years-of-age, but transition services may not always meet their needs. Post-transition many YAs are reluctant to request diabetes self-management support, this can lead to increased anxiety and deterioration in diabetes control and quality of life. Chatbots increasingly offer social support in daily life so could also provide diabetes self-management support, but there is a lack of co-developed, evidence-based, developmentally appropriate chatbots. Therefore, the DigiBete Chatbot, the first user-led, clinically approved transition chatbot for YP/YA with T1DM was recently developed

Objective:

Study objectives were to (1) evaluate the DigiBete Chatbot in four English diabetes services for YP/YA, and (2) assess the feasibility of a future full trial of the chatbot

Methods:

YP/YA received password-protected access to the chatbot and were encouraged to use it ad libitum. YP/YA completed web-based questionnaires (the Mobile App. Rating Scale (MARS), the Hospital Anxiety and Depression Scale (HADS), the Short-Form 36 (SF36), and the System Usability Scale (SUS) at baseline, 2-weeks and 6-weeks to evaluate quality of life, anxiety and depression and chatbot usability and acceptability. After 6-weeks, YP/YA and parents participated in individual qualitative interviews to ascertain their views on the chatbot’s design, content and usability. Parents also discussed the chatbot’s potential to enable their child to become autonomous in T1DM self-management. Healthcare professionals (HCPs) participated in focus groups to determine their views on the chatbot and their perceived role in supporting its use if it became standard practice. Data were analyzed using descriptive statistics and Framework Analysis

Results:

Eighteen YP/YA enrolled. YP/YAs' SUS and uMARS scores demonstrated improvement from baseline to the second timepoint in perceived ease of use, functionality and confidence, and most viewed the chatbot as highly usable with over 50 percent finding it easy to navigate. Some modifications were recommended to increase accessibility. HADS anxiety scores were higher than the depression scores and overall, there was no significant change across the 3 time points for all SF36 domains. Four parents, 24 HCPs and 12/18 YP/YA completed qualitative interviews. Questionnaire uptake/outputs and the emergent qualitative themes: Living with T1DM, Using the Chatbot and Refining the Chatbot, indicated that the study measures are feasible to use, the chatbot is acceptable and functional, and with refinements incorporating our results, could beneficially support YP/YA during transition. Users scored the chatbot as ‘good’ to ‘excellent’ for being engaging, informative and aesthetically pleasing, and would use it again.

Conclusions:

Study results specify minor refinement of the chatbot and further investigation in a full cohort study prior to it's wider clinical use. Our research design and methodology could also be transferred to the development, evaluation and use of chatbots for other chronic conditions. Clinical Trial: Not relevant