Glucose Sensors Cut A1c 0.9 Points More Than Routine Care in Type 2 Diabetes Without Insulin. The Trial Lasted 26 Weeks.

Continuous glucose monitors already have strong randomized evidence behind them for people with diabetes who use insulin. For the much larger group with type 2 diabetes who do not, the proof came mostly from small trials. A study published October 14 in the New England Journal of Medicine adds a large randomized one, and its headline number is a 0.9 percentage point gap in A1c.

The trial is called CONNECT. Researchers enrolled 283 adults at 22 primary care practices across the United States, all with type 2 diabetes, none treated with insulin, and all with an A1c of 7.5 percent or higher at the start. Average A1c at entry was 8.8 percent, and 31 percent of participants started at 9 percent or above. About half were assigned to wear a Dexcom G7 sensor for 26 weeks. The rest got routine care with a blood glucose meter. Everyone received diet and exercise education and kept their usual glucose-lowering drugs. Of the 283, 265 finished.

Time spent in the target range of 70 to 180 mg/dL averaged 62 percent with the sensor and 41 percent with routine care, a difference of about five hours a day, according to the Cleveland Clinic Journal of Medicine's summary of the results. In participants who entered with an A1c above 10 percent, A1c fell 3.1 points with the sensor, 2.1 points more than routine care. The authors reported the benefit held in every subgroup, including people already taking GLP-1 drugs (40 percent of the trial) or SGLT-2 inhibitors (37 percent).

Who ran the trial and who paid for it

Dexcom sponsored CONNECT, and the device tested was Dexcom's own. The journal's conference coverage says two of the authors, Roy Beck of the Jaeb Center for Health Research and Thomas Martens of the International Diabetes Center in Minneapolis, reported consulting or research support. Martens presented the first results at the American Diabetes Association's June 6 meeting. Beck, the senior author, said Level A evidence "has historically driven meaningful changes in standards of care."

Tamara Oser, the first author, is a professor of family medicine at the University of Colorado Anschutz and runs its Primary Care Diabetes Lab. She estimated that about 20 million U.S. adults have type 2 diabetes and do not take insulin. Her team hopes the result will "strengthen national diabetes care guidelines and expand insurance coverage," according to the university's release.

The university's release says the American Diabetes Association rated the evidence for sensors in people not on insulin at Level C, the lower tier. The Cleveland Clinic Journal of Medicine notes that the association's 2026 standards already recommend sensors for adults with type 2 diabetes on any regimen where they help. On those two accounts, the recommendation moved ahead of the evidence grade, and CONNECT supplies the stronger evidence.

What 26 weeks cannot show

The trial measured A1c and time in range. It did not measure heart disease, kidney disease, or vision loss, the complications Oser cited as the reason glucose control matters. Participants also knew which group they were in, which is unavoidable with a wearable device. The sensor group saw continuous readings while the comparison group checked with a meter. The Alberta authors write that in this population the benefit may be driven more by lifestyle and behavior change than by medication adjustment.

Safety results, including how often blood sugar dropped too low, were not in the summaries reviewed for this report, and the full New England Journal of Medicine paper could not be loaded for this article. The account here rests on two secondary summaries that trace to the same trial.

A smaller trial saw the gap close

A separate group looked at what happens after the sensor comes off. Darren Lau and colleagues at the University of Alberta had randomized 105 adults with type 2 diabetes, none on insulin, to a six-week course of FreeStyle Libre 2 sensors with two virtual educator visits, or to the two visits alone. At three months the sensor group was ahead on A1c. They then followed 81 of the completers through Alberta's provincial medical records and reported the result in a letter in Diabetes, Obesity and Metabolism.

The three-month gap, adjusted for baseline, was 0.62 points in favor of the sensor. At six months it was 0.03 points, with a confidence interval running from 0.59 points in favor of the sensor to 0.64 points against it. At 12 months it was 0.14 points in favor of the sensor, with an interval that also crossed zero. Fifteen of the 43 people in the sensor arm used sensors again off protocol.

The Alberta authors said their sample was too small to conclude that sensors stopped working. They concluded only that the three-month difference was no longer observable. The two studies also ask different questions. CONNECT participants wore sensors throughout, with a median wear time of 97 percent. In Alberta, most stopped after six weeks.