In T1DM, open-source automated insulin delivery increased glucose time in target vs. sensor-augmented pumps

A challenge to wider use of the open-source systems to manage type 1 diabetes (T1DM) is that they are not approved by the FDA and thus require clinicians to be willing to learn from and support their patients, said an ACP Journal Club commentary.

Patients with type 1 diabetes (T1DM) successfully increased their time in target glucose range using an open-source automated insulin delivery (AID) system, a recent trial found. A total of 97 adults and children were randomized to either the AID system or sensor-augmented insulin pump therapy (SAPT). At 24 weeks, the mean time in the range of 70 to 180 mg/dL (3.9 to 10.0 mmol/L) increased from 61.2% to 71.2% in the former group and decreased from 57.7% to 54.5% in the latter group.

The study was published by the New England Journal of Medicine on Sept. 8, 2022, and was summarized in the September 2022 ACP Diabetes Monthly. The following commentary by Michael Lockhart, MD, and Sean F. Dinneen, MD, was published in the ACP Journal Club section of Annals of Internal Medicine on Jan. 3.

Technology for management of T1DM has advanced at a phenomenal pace over the past 2 decades. The holy grail of an artificial pancreas has been all but realized, with hybrid closed-loop systems allowing real-time adjustment of insulin doses. Some patients are unwilling to wait for device companies to market their hybrid closed-loop systems solutions and instead have created their own do-it-yourself systems through open-source algorithms and compatible pumps and sensors. Evidence of benefit from these open-source AID systems has come mainly from anecdotal and real-world evidence instead of clinical trials.

CREATE by Burnside and colleagues is the first high-quality RCT to show that an AID system increased time in range vs. SAPT. The trial has many strengths, including recruitment of both children and adults, a strong comparator, and outcomes that reflect both the reality of day-to-day glucose variability (sensor glucometrics) and a single integrated biochemical measure (i.e., HbA1c). The reported improvement in time in range is similar to that seen with commercially available AID systems. Issues with devices, episodes of ketoacidosis, and severe hypoglycemia were rare or nonexistent.

What is the future of open-source AID systems, and do patients who use these devices need input from clinicians at all? Does confirmation of efficacy with one open-source AID system mean that all other systems can be endorsed? The answer to these questions will depend on how market forces, guideline writers, and online diabetes communities interact in an ever-changing, complex health care environment. According to the American Diabetes Association 2022 Clinical Practice Guidelines, open-source AID systems are not approved by the U.S. Food and Drug Administration. Therefore, they remain outside of the usual regulatory sphere. This can be problematic for clinicians and calls for a willingness to learn from, and support, their patients. At a societal level, it is important that equity of access is addressed in a way that is fair to all patients with T1DM, not just those fortunate to have access to sophisticated devices.