Eighteen published reviews say that adding a GLP-1 drug to insulin helps people with type 1 diabetes. A new analysis shows most of them are reading the same few trials, and the outcome you would most want the drug to improve, the share of the day spent in a safe blood-sugar range, did not move.
The work, an "overlap-informed umbrella review" published September 10 in Diabetes, Obesity and Metabolism ↗, is a review of reviews. The authors pulled 18 systematic reviews covering 56 unique studies (35 randomized trials plus 21 observational ones), then measured how much those reviews overlap. The overlap was very high. A metric called corrected covered area came in at 19.6 percent, meaning the same primary trials get counted again and again across supposedly separate reviews. Two trials, ADJUNCT ONE and ADJUNCT TWO (both testing liraglutide), drive most of it. So "18 reviews agree" is closer to "a few trials, re-pooled many times."
Type 1 diabetes is the autoimmune kind. The pancreas stops making insulin, so patients inject it for life. GLP-1 drugs like liraglutide ↗ (sold as Victoza and Saxenda) and semaglutide ↗ (Ozempic, Wegovy) are built and approved for type 2 diabetes and obesity, where the body still makes insulin but uses it poorly. Using them in type 1 is off-label, and the appeal is obvious. Many adults with type 1 are also overweight and take large insulin doses, and GLP-1 drugs are the best weight tool medicine has.
On that front, the numbers held up. Pooling the trials fresh, the authors found adjunct GLP-1 therapy cut body weight by about 3.9 kilograms (roughly 9 pounds) and total daily insulin by about 5.7 units, both rated moderate-to-low certainty. HbA1c, the three-month average blood-sugar measure, fell by 0.23 percentage points. That is a real change, and a small one.
The catch is what did not change. Time in range, the fraction of the day a continuous glucose monitor reads a safe level, is the number type 1 patients and their doctors actually steer by. It did not improve. The gain was about 2 percent with a confidence interval running from worse to better, rated very low certainty. A drug that trims your weight and your insulin dose but leaves your glucose swings where they were is doing something, just not the thing type 1 diabetes is about.
And the most certain findings in the whole review were the harms. Nausea was nearly three times as common (relative risk 2.88), vomiting more than three times (3.11), and dropping out of a trial because of side effects about twice as likely (2.10). All three were graded high certainty, the strongest evidence tier in the analysis. The reassuring numbers, by contrast, were soft. Severe low-blood-sugar episodes and diabetic ketoacidosis, the two dangers that would make GLP-1 use in type 1 genuinely risky, showed no clear increase, but the evidence was too thin to rule one out.
That asymmetry is the story. The things the review is surest about are the side effects. The safety questions that matter most are the ones it cannot answer.
The authors land where the data pushes them. This "does not support routine use" but points to "a potential role in selected individuals for whom weight reduction and lower insulin requirements are therapeutic priorities." In plain terms, a heavy adult with type 1 diabetes on a big insulin dose might reasonably try it, eyes open about the nausea, and without expecting steadier glucose.
It also fits a pattern in how GLP-1 drugs behave differently across the two diabetes types. A TriNetX study this year found autoimmune thyroid disease moved in opposite directions ↗ in type 1 versus type 2 patients on GLP-1 drugs. The receptor is the same. The disease it lands in is not.
Liraglutide, semaglutide, and exenatide ↗ all act on the GLP-1 receptor ↗. In type 2 diabetes that receptor has a large, repeatedly confirmed payoff. In type 1, this review suggests, the payoff is real but narrower than the stack of reviews makes it look.