A review of records from 4,000 users found glucose readings within the target range for nearly three-quarters of the time. However, relatively few participants reported type 2 diabetes, and the observational analysis cannot establish comparative benefit.
A real-world analysis of 4,000 people using the twiist automated insulin delivery system found that average glucose levels were within the recommended range for much of the day, with results broadly similar among participants who identified as having type 1 or type 2 diabetes. The findings, published in Diabetes Technology & Therapeutics, add evidence on the system’s use outside a clinical trial, although relatively few participants reported having type 2 diabetes.
Participants used the system for at least 14 days between April 2025 and April 2026. Across the group, glucose readings were within the 70–180 mg/dL target range 73.5% of the time on average, rising to 77.9% overnight. Nearly two-thirds spent more than 70% of the time in range; 86% kept readings below 70 mg/dL to less than 4% of the time, and 97% spent less than 1% of the time below 54 mg/dL. The 2026 American Diabetes Association Standards of Care set those same thresholds as goals for many adults, while noting that targets should be tailored to individual circumstances.
The results were observational: researchers analysed de-identified device and glucose-monitoring records rather than assigning participants to treatments or comparing them with a control group. Although reported time in range was similar for type 1 and type 2 diabetes, only 151 participants identified as having type 2 diabetes, compared with 2,797 who reported type 1. Diabetes type was not reported for more than a quarter of the full group, limiting how confidently the subgroup comparison can be applied to the wider population.
Separate evidence offers a more focused look at type 2 diabetes. In a 13-week, multicentre trial of 307 adults using insulin, average HbA1c declined from 8.1% to 7.4%, while time in the target glucose range increased from 57% to 73%. As that trial was also single-arm, it did not compare the system with a concurrent control group. Together, the studies support further consideration of automated insulin delivery for people with either diabetes type who use insulin, while leaving questions about comparative benefit and outcomes across broader patient groups.
Disclaimer: This content is for informational purposes only and is not intended to be a substitute for professional medical judgment, advice, diagnosis, or treatment.
