Mealtime insulin, yes. Basal insulin is a separate question with a separate answer. Seven randomized trials pooled 1,332 adults already on basal-bolus insulin or multiple daily injections. Taking the mealtime doses out and putting a GLP-1 in left HbA1c flat at −0.08 points. It removed 33.32 units of insulin a day [1]. Body weight fell 4.94 kg with it.
That is a trade, not a win. Nothing about glucose control got better. What changed is everything around it, and the bill. Insulin is one of the cheapest drugs in this market. A GLP-1 is one of the dearest. That is the half of the comparison the economic studies keep modeling rather than measuring.
The outcome that did not move
HbA1c was the primary endpoint. The difference was −0.08 percentage points, interval −0.28 to 0.12.
That interval sits across zero. It is a null, and here a null is the objective. The point of the strategy is to take mealtime insulin away without losing control of blood sugar.
A trial showing something is better is a different object from one showing it is no worse. And the second gets read as the first constantly.
What the swap bought
Weight fell 4.94 kg, interval 7.37 to 2.51. Total daily insulin fell 33.32 units, interval 52.23 to 14.41.
Thirty-three units a day is a large reduction. So is five kilograms. For somebody on four injections a day, fewer of them is its own benefit that no endpoint captures. Weight removed this way is still weight that comes back when the drug stops.
Adding one to insulin is not the same trade
The trials above remove insulin. A second set of trials adds a drug on top of it, and the numbers look nothing alike.
REIMAGINE 3 put cagrilintide-semaglutide on top of basal insulin in 274 adults whose diabetes was not controlled [2]. HbA1c fell 2.33 and 2.10 points at the two dose levels, against 0.66 on placebo. Mean baseline was 8.8%, so these are people well above target. The comparator was placebo rather than semaglutide alone, which is the comparison the trial cannot make.
A third route is one injection carrying both. IcoSema combines weekly insulin icodec with semaglutide, and the pooled COMBINE trials report 98% heterogeneity on the glucose outcome [3]. No cardiovascular event was counted in either trial. Every benefit there is a surrogate, as the pooled analysis itself concedes.
What it costs
A GLP-1 costs many times what mealtime insulin costs. None of these three analyses prices anything.
So the trade in money runs opposite to the trade in convenience. Fewer injections, less weight, a much larger bill. Whether that is worth it depends on what coverage does with it. That is the question that decides most of these decisions before any seller quotes a figure.
Nothing on this page is a reason to change an insulin regimen. Reducing insulin without a prescriber directing it is dangerous. This is a clinical decision, not a comparison-shopping one.
How long it holds
Unknown. The trials are short and the authors say so.
Insulin requirements move over years. A strategy that holds HbA1c for months is not proof of one that holds it for a decade. The same gap sits under the case for stacking two drug classes, where the best-ranked combination was never randomized at all.