User engagement is critical to the success of mobile health (mHealth) interventions. Text messaging interventions tend to have higher engagement than other forms of mHealth; however, the primary way users are engaged is via interactive, two-way texts that request a response. We developed an mHealth intervention for adults with type 2 diabetes which involved sending users text messages for 9 months including two-way texts and texts with embedded web links. As part of a former analysis, we learned response rate to the two-way texts was notably high (82%). For this study, we examined engagement with the web-linked texts including overall click rates and participant characteristics associated with odds of clicking the link.

Participants received two types of texts that included web links. Each time they had their A1c tested as part of the study (up to 5 times), they received a text with a link to access their A1c results. In addition, each participant received up to 5 text messages with links to educational content on managing diabetes. We defined engagement as clicking on the links in the texts. We used logistic regression with robust standard errors to examine independent associations between participant characteristics and likelihood of clicking the links.

Participants (N=152) were, on average, aged 55.7±11 years, 51% male, and 64% non-Hispanic White; average years of education was 15.4±2.7 and 18.4% had annual incomes less than US $35K. All included participants owned a smartphone. The overall click rate for the A1c links was 73%, whereas as the click rate for the educational links was only 24%. Both diabetes duration and prior health app use were significantly associated with likelihood of clicking the educational links. Each additional year of having diabetes was associated with a 4.0% [0.01%, 7.6%] lower odds of clicking the link. Participants who had never used a health app, had 55.3% [8.8%, 78.1%] lower odds of clicking the link.

Text messaging interventions tend to have high engagement; however, engagement with text messages offering access to online diabetes resources was low despite evidence that participants had the capabilities to click links (as evidenced by the high click rate for A1c links). Inclusion of online resources may only benefit a small proportion of users in a texting intervention, although presenting in alternate ways (e.g., pairing links with A1c values) could help boost engagement.