Opportunity

More than 38 million U.S. adults have diabetes. Care teams use A1c tests that measure average blood sugar levels over time to diagnose prediabetes and diabetes and help patients properly manage diabetes to avoid complications.

Proactive outreach to engage patients in A1c testing is necessary. However, traditional outreach methods such as phone calls, mailers, and messaging via patient portals can be costly, labor-intensive, time-consuming, and yield relatively low results.

Intervention 

A team of clinicians at Penn Medicine is leveraging Way to Health to test the efficacy of Diabetes Outreach by Text bot (DOTbot), a bi-directional texting program designed to engage patients with uncontrolled diabetes to adhere to A1c testing standards.

For the initial pilot in 2022, clinicians identified 544 patients who had uncontrolled diabetes with outstanding A1c labs. A non-clinical population health staff member used DOTbot to sent algorithmic text messages to these patients, notifying them that they were due for A1c testing and advising them on the next steps. Patients were given the option to walk into any Penn Medicine lab to be tested or have the order mailed to them to fill at a lab of their choosing.

After this pilot, the DOTbot team expanded testing to more than 1,128 patients at 20 urban and suburban practices.

Impact

With DOTbot in use, lab testing and office visit rates rose significantly among patients with uncontrolled diabetes, and the program lowered staff time spent on outreach by nearly 90 percent.

The fraction of patients completing labs was higher for both the pilot and scaled groups – 11 percent and 21 percent, respectively, compared to 10 percent in the manual outreach group (608 patients). The same was true of office visit rates: These were 18 percent and 24 percent, respectively, compared to 7 percent for those receiving manual outreach.

Whereas patients who received manual outreach had a 31 percent response rate, 45 percent of patients in the DOTbot urban pilot and 55 percent in the scaled intervention responded to texts. This is notable because engagement was the greatest predictor of lab completion and office visits when the researchers examined factors including patient demographics and median income based on zip codes.