乌鸦传媒

An open-source automated insulin delivery (AID) system -- also known as a do-it-yourself system -- was both safe and effective for patients with type 1 diabetes, according to the CREATE trial.

Over 24 weeks, users of the AID system spent more time in target glucose range (70 to 180 mg/dL) -- an average of 14% longer -- than those who were using sensor augmented pump therapy without automation, reported Martin de Bock, PhD, of the University of Otago in Christchurch, New Zealand, during a presentation at the American Diabetes Association (ADA) annual meeting.

The open-source AID system consists of the OpenAPS algorithm from a version of AndroidAPS implemented in a smartphone, paired with the DANA-i insulin pump and Dexcom G6 continuous glucose monitor. The researchers previously published additional information on the ins and outs of the algorithm in the .

"Open-source AID, despite not being [FDA] regulated ... is safe and efficacious in children and adults with type 1 diabetes compared to sensor augmented pump therapy," de Bock noted. "I've got no commercial gains from doing this study and it gets me hoping that in some groups where you can't get commercial automation, this might provide an avenue for people who deserve to be getting the glucose benefits of automated insulin delivery."

By the end of the trial, adults using the AID system spent 74.5% (±11.9%) of time in target glucose range -- increasing 9.6% (±11.8%) since the run-in phase. As for children, they spent 67.5% (±11.5%) in range -- seeing an increase of 9.9% (±14.9%) since run-in.

The AID system was most effective at night, allowing children to spend nearly 20% longer time in range and adults more than 30% longer time in range. Overall, the AID was active 94% of the time throughout the study.

Users of the system immediately saw a greater time spent in range, as early as the first 3 weeks of use, which was sustained throughout the trial for both children and adults. "[This was] super pleasing," de Bock said. "There was a worry that the technical burden of open-source might be hard and you'd get some drop-out, but we didn't see that."

In contrast, users of sensor augmented pump therapy saw no improvement in time spent in range. At week 24, time spent in target range was 56.5% (±14.2%) and 52.5% (±17.5%) for adults and children, respectively.

In addition, 60% of users of the open-source AID system achieved the time-in-range target of 70% or more, according to current clinical practice guidelines. In comparison, only 15% of sensor augmented pump therapy users hit this mark.

Both systems were deemed safe, with no events of severe hypoglycemia or diabetic ketoacidosis in either group.

There were some adverse device events, however, split fairly evenly between the AID and control groups, most commonly hyperglycemia, site reaction or infection, and urticaria.

For the trial, a total of 97 patients -- 48 children ages 7-15 and 49 adults ages 16-70 -- were randomized, all of whom had type 1 diabetes and were established on insulin pump therapy for at least 6 months prior to randomization. At baseline, all participants had a mean HbA1c below 10.5%.

Exclusion criteria included comorbidities or psychological issues that would make participants "unsuitable," non-English language, plans to become pregnant, alcohol or drug dependence, or intolerance to NovoRapid insulin.

Patients underwent a 4-week run-in phase followed by the 6-month randomized phase. They're currently being followed for another 6 months in a continuation phase where all patients are using the AID system. Data will be presented later this year at the European Association for the Study of Diabetes meeting, said de Bock.

Comment