There is a good idea behind this and not enough patients to test it.
The idea
High blood sugar around heart surgery is linked to wound infection, kidney injury and death. The standard fix is intravenous insulin, which brings its own risk of driving glucose too low. [1]
A GLP-1 lowers glucose mainly when glucose is already elevated. In principle that separates the control from the hypoglycemia, which would be a real advance in a setting where both matter within hours. It is the same logic that makes swapping mealtime insulin worth testing outside a hospital.
What the numbers actually permit
Four trials, 446 patients, reported across seven papers. Count the papers and you count the same people more than once.
Thirty-day mortality came out at 0.42 with an interval from 0.06 to 2.81. That rests on one death in 161 against three in 160. Four deaths cannot establish anything about mortality.
Complications ran 0.92, cardiac events 1.08, hypoglycemia 0.85 with an interval to 2.13. Nausea and vomiting came out at 3.01, interval 0.26 to 35.27.
What would settle it
A trial large enough to count events. For 30-day mortality after cardiac surgery that means thousands of patients, not hundreds.
That is expensive and nobody has run it. The same absence sits behind two drug classes compared only by inference, and it is the ordinary state of most questions in this field rather than an unusual gap.
If you take one and have surgery coming
Tell the anesthetist what you are on. That is the whole of it, and it matters for a reason this paper does not cover — these drugs slow the stomach, and food can still be there after an overnight fast.
Nothing here is a reason to start, stop or change anything before an operation. That is a decision for the team doing it.