System architecture for "Support through mobile messaging and digital health technology for diabetes" (SuMMiT-D): Design and performance in pilot and randomized controlled feasibility studies
Chi Y., Velardo C., Allen J., Robinson S., Riga E., Judge D., Tarassenko L., Farmer AJ.
Background: Diabetes is a highly prevalent long-term condition with high morbidity and mortality rates. People with diabetes commonly worry about their diabetes medicines and do not always take them regularly as prescribed. This can lead to poor diabetes control. The Support Through Mobile Messaging and Digital Health Technology for Diabetes (SuMMiT-D) study aims to deliver brief messages as tailored interventions to support people with type 2 diabetes in better use of their diabetes medicines and to improve treatment adherence and health outcomes. Objective: This paper describes the overall architecture of a tailored intervention delivery system used in the pilot and randomized controlled feasibility studies of SuMMiT-D and reports its performance. Methods: The SuMMiT-D system includes several platforms and resources to recruit participants and deliver messages as tailored interventions. Its core component is called the clinical system and is responsible for interacting with the participants by receiving and sending SMS text messages from and to them. The personalization and tailoring of brief messages for each participant is based on a list of built-in commands that they can use. Results: For the pilot study, a total of 48 participants were recruited; they had a median age of 64 years (first quartile, third quartile [Q1, Q3: 54.5, 69]). For the feasibility study, a total of 209 participants were recruited and randomly assigned to either the control or intervention group; they had a median age of 65 years (Q1, Q3: 56, 71), with 41.1% (86/209) being female. The participants used the SuMMiT-D system for up to 6 months (26 weeks) and had a wide range of different interactions with the SuMMiT-D system while tailored interventions were being delivered. For both studies, we had low withdrawal rates: only 4.2% and 5.3% for the pilot and feasibility studies, respectively. Conclusions: A system was developed to successfully deliver brief messages as tailored health interventions to more than 250 people with type 2 diabetes via SMS text messages. On the basis of the low withdrawal rates and positive feedback received, it can be inferred that the SuMMiT-D system is robust, user-friendly, useful, and positive for most participants. From the two studies, we found that online recruitment was more efficient than recruitment via postal mail; a regular SMS text reminder (eg, every 4 weeks) can potentially increase the participants’ interactions with the system.