Among 57,456 people taking a GLP-1 drug for weight loss, the ones on semaglutide were about 12 percent less likely to develop diabetes over the next year than the ones on liraglutide. The catch is when that gap showed up: nothing in the first six months, then a clear separation after.
Both drugs are approved for chronic weight management, and both beat placebo at holding off diabetes and heart disease. What nobody had pinned down was how they compare to each other. A study published July 16 in the British Journal of Clinical Pharmacology ↗ took the question to real-world data.
How you run a trial without running a trial
The authors used MarketScan insurance claims to build what epidemiologists call a target trial emulation. The idea is to mimic the design of a head-to-head randomized trial using records of people who were already prescribed one drug or the other. They took adults who started semaglutide ↗ or liraglutide ↗ between 2021 and 2023, none of whom had diabetes or cardiovascular disease at the outset, and matched them one to one on age, other conditions, and medication use. The matched group had a mean age of 45, and 83 percent were women, a profile that reflects who takes these drugs for weight rather than for blood sugar.
Both drugs target the GLP-1 receptor ↗, a gut-hormone switch that curbs appetite and improves how the body handles glucose. Semaglutide, the molecule behind Ozempic and Wegovy, is the more potent of the two and is dosed once a week. Liraglutide, sold as Victoza and Saxenda, is an older daily injection.
The number, and the timing
Over an average of one year, the cohort logged 1,104 new diabetes diagnoses. After adjustment, semaglutide users had a 12 percent lower risk than liraglutide users (hazard ratio 0.88, 95 percent confidence interval 0.78 to 0.99), a result that just clears the line of statistical significance.
But the risk did not fall evenly across the year. The statistical assumption that a treatment's effect stays constant over time was clearly violated, so the authors split the follow-up. In the first six months, semaglutide and liraglutide were indistinguishable on new diabetes (hazard ratio 0.99). After six months, semaglutide pulled ahead, with a 20 percent lower risk (hazard ratio 0.80, 95 percent confidence interval 0.68 to 0.94). The benefit is real but it is a slow build, not a day-one effect, which fits a drug whose metabolic advantage compounds as the weeks add up.
What the study cannot say
The heart-disease comparison is the honest weak spot. Over one year in a group this young and this healthy, only 57 cardiovascular events occurred, far too few to draw a conclusion from. The point estimate drifted the wrong way (hazard ratio 1.25) but the confidence interval ran from 0.74 to 2.11, wide enough to include a large benefit or a large harm. The right read is not that semaglutide raised heart risk. It is that this study had no power to measure it either way.
The rest of the usual claims-data caveats apply. Insurance records capture prescriptions filled, not pills swallowed. A one-year window is short for a disease that takes years to develop. And matching on measured variables cannot rule out the possibility that the people prescribed the newer, pricier drug differed in ways the records never saw.
Still, the shape of the result is useful. Two drugs that both look good against placebo are not interchangeable against each other, and the difference here is not a headline percentage so much as a clock. On peptidemodel, semaglutide and liraglutide sit as separate cards on the same receptor, close cousins with a widening gap between them. This study puts a number and a timeline on that gap: six months in, the newer molecule starts to win.