Written by Kevin Liu
Spoon Feed
In a 13-week randomized controlled trial (RCT) involving adults with insulin-treated type 2 diabetes (T2DM), automated insulin delivery (AID) was associated with a greater reduction in hemoglobin A1c (HbA1c) than intermittent insulin delivery.
Moving insulin treatment in T2DM to an automated world
This trial offers strong evidence supporting AID in adults with insulin-treated type 2 diabetes. In this 13-week, multicenter study, both groups received continuous glucose monitoring (CGM), with 319 patients randomized either to AID or to continue their pretrial insulin regimens. The AID group saw a greater reduction in HbA1c (0.9% vs 0.3%, 95%CI –0.8 to –0.4; P<0.001) and more time in the target glucose range (+14%, 95%CI 11 to 17; P<0.001). Hypoglycemia remained low in both groups, with only 1 severe case in the AID arm.
Current guidelines favor intermittent insulin therapy in hospitalized patients and caution against continuing home oral diabetes medications (metformin, sulfonylureas, thiazolidinediones, glucagon-like peptide-1 receptor agonists, sodium-glucose cotransporter-2 inhibitors) due to safety concerns. Dipeptidyl peptidase-4 inhibitors (DPP-4s) are safe to continue inpatient due to evidence showing comparable glycemic control to insulin without more hypoglycemia.
How does this change my practice?
As a hospitalist, a recent inpatient practice change involves the continuation of automated insulin delivery systems instead of transitioning patients to hospital-provided intermittent insulin delivery. Additionally, we have also continued certain oral agents (DPP-4s) safely while inpatient as well. This trial adds to those endocrinologist-led practice changes. Anecdotally, I have seen less benefit with AID in patients requiring large daily totals of insulin, and the trial notes that the additional HbA1c reduction appears to be attenuated in patients who are already at an HbA1c goal < 7.0% and who take > 200 units of insulin daily.
Source
A Randomized Trial of Automated Insulin Delivery in Type 2 Diabetes. N Engl J Med. 2025 May 8;392(18):1801-1812. doi: 10.1056/NEJMoa2415948. Epub 2025 Mar 19. PMID: 40105270
