Two results out of many came back significant. The authors say so plainly. That sentence is the most useful thing in the paper, and it is the part a headline drops.
The setup
Adults having a first hip replacement between 2003 and 2023 were pulled from a claims database. [1] Anyone with at least three GLP-1 prescriptions in the year before surgery counted as exposed. Matching on demographics, conditions and lab values left 1,262 patients per side.
Two ninety-day outcomes separated. Deep vein thrombosis ran 1.6% against 3.0%, a relative risk of 0.53, 95% CI 0.31 to 0.89, p = 0.014. Readmission ran 1.1% against 2.8%, relative risk 0.40, 95% CI 0.22 to 0.71, p = 0.001.
How big is that
Those rates on 1,262 people come to roughly twenty clots against thirty-eight. The paper does not print the counts; that is arithmetic from its own percentages.
Eighteen events is a thin margin. It clears significance and it would not survive much reshuffling. A relative risk of 0.53 sounds larger than the 1.4 points of absolute difference behind it, which is the usual gap between the two ways of saying it.
The flat results
Most of this study found nothing. Revision at five years was 94.6% against 95.5% survival, p = 0.674. Joint infection was 94.7% against 94.4%, p = 0.465. Other ninety-day outcomes did not separate either.
That matters. The reason people worry about a GLP-1 before surgery is delayed stomach emptying under anesthesia, and this study did not find worse outcomes. It found the same outcomes, with two exceptions.
Who is in the exposed group
Three filled prescriptions in a year. That is a filter, not a randomization. It selects people who tolerated the drug, afforded it, and kept going — a minority, since most people do not.
Whatever makes someone a persistent filler probably also makes them a better surgical patient. Matching on database codes cannot see it.
What a buyer takes from this
Nothing to act on. It is reassurance, not a reason. Nobody should start a GLP-1 to lower a clot risk, and nobody should read a surgical benefit into a price.
If a hip replacement is in the picture, the sequencing question is a clinical one and the order matters more than the drug. What a prevented event is worth is a separate calculation, covered in the cost-per-event piece.