CATEGORY: METABOLIC & BLOOD SUGAR
An insulin pump that decides its own doses has now been put through a proper test in type 2 diabetes, and it won. Over 13 weeks, adults who switched to automated insulin delivery saw their HbA1c fall from 8.2% to 7.3%. HbA1c is a lab value that boils about three months of blood sugar down to one number, and most treatment aims for 7% or below. People who stayed with their usual routine, nearly all of them daily injections, went from 8.1% to 7.7%.1 Everyone wore a glucose sensor. The pump made the difference.
A Pump, a Sensor, and Software Doing the Math
These systems pair a small wearable pump with a continuous glucose monitor, a sensor worn on the skin that reads blood sugar around the clock. Software connects the two, nudging insulin up or down every few minutes without being asked. That technology is already familiar to many people with type 1 diabetes. Whether it helps in type 2, where insulin needs behave differently, hadn’t been settled in a trial this size, and settling it is why the researchers ran one. They enrolled 319 adults with insulin-treated type 2 diabetes at centers across the United States. A computer drew the assignments, sending two of every three to the pump system and leaving the rest on whatever method they’d been using. Before the trial, 96% were taking several shots a day, and three quarters used fixed mealtime doses that never adjusted to the moment. The study ran on the Control-IQ+ system and was funded by Tandem Diabetes Care, the company that makes it.
Three Extra Hours a Day in the Target Range
Doctors track how many hours a day blood sugar sits in the recommended band of 70 to 180 mg/dL. Pump users went from 48% of the day in that band to 64%. The usual-care group barely moved, 51% to 52%. The gap between groups, 14 percentage points, works out to more than three added hours in range every day.1 The improvement reached the people who needed it most. Among those starting with an HbA1c of 9% or higher, the pump group dropped from 10.3% to 7.9%, while usual care went from 9.7% to 8.6%. It also held in the 44% of participants who were already taking a GLP-1 drug such as semaglutide, which the study’s senior author called an encouraging sign. Severe low blood sugar, the risk that makes insulin dosing nerve-wracking, stayed rare. One participant in the pump group had a severe episode. That was the only one.
Thirteen Weeks Is a Start, Not a Verdict
This trial was short. It can say what happens to blood sugar in three months, and nothing about heart disease, kidney damage, or the other complications that good control is meant to prevent over years. The maker’s funding is worth keeping in view too, even with the trial run by independent academic centers. And the comparison group still got real-time glucose sensors, so the result isn’t pump versus nothing. It’s the software’s dosing decisions versus a person’s. For the millions of adults with type 2 diabetes who take insulin, that’s exactly the question that matters, and for 13 weeks at least, the software answered it.
Key Takeaways
- Adults with insulin-treated type 2 diabetes who switched to an automated insulin pump reached an average HbA1c of 7.3%, versus 7.7% with their usual method. Both groups wore glucose sensors.
- Pump users spent about three more hours a day with blood sugar in the recommended range, and the benefit was largest in people starting above 9%.
- The trial lasted 13 weeks and was funded by the pump’s manufacturer. It measured blood sugar control, not long-term complications.
- The benefit held in people already taking a GLP-1 drug alongside insulin.
References
- Kudva YC, Raghinaru D, Lum JW, et al. A Randomized Trial of Automated Insulin Delivery in Type 2 Diabetes. New England Journal of Medicine. 2025;392(18):1801–1812. link