Tirzepatide users lost about 13 percent of their body weight in a year. Semaglutide users lost about 6 percent. The three older GLP-1 shots landed near 4 percent. That is the ranking that fell out of medical records for 14,046 adults with obesity, and the order is the same one the drug trials produced.
The study, published online July 14 in Drug Design, Development and Therapy ↗, pulled its patients from the National Institutes of Health All of Us Research Program, a large and deliberately diverse database of everyday American patients rather than trial volunteers. The authors ran it as a new-user, active-comparator cohort, which means they only counted people starting one of the drugs for the first time and compared them against each other rather than against nobody. Five agonists were in the race: exenatide ↗, liraglutide ↗, dulaglutide ↗, semaglutide ↗, and tirzepatide ↗. All of them work on the GLP-1 receptor ↗; tirzepatide also hits the GIP receptor ↗, a second gut-hormone switch, which is the usual explanation for why it pulls ahead.
The order the trials predicted
After adjusting for baseline differences between the groups, tirzepatide was more than five times as likely as the older drugs to get a patient to a 15 percent weight loss (adjusted hazard ratio 5.57, 95 percent confidence interval 3.66 to 8.49). Semaglutide was about 1.8 times as likely (adjusted hazard ratio 1.77, 1.56 to 2.01). Both cleared the older agents by a wide margin. Exenatide, dulaglutide, and liraglutide finished bunched together at roughly 4 percent of body weight, close enough that the study could not reliably separate them.
That is the number worth keeping. The head-to-head trials that got these drugs approved were run in controlled populations, with monitored dosing and motivated participants, and skeptics have long argued that the neat ordering they produced might not survive contact with routine care. Here it survived. The relative standings in a messy real-world cohort matched the standings in the trials, which is a point in favor of the trial numbers meaning what people think they mean.
Where the confidence runs thin
The authors are unusually frank about the soft spot. The tirzepatide group was only 386 patients, a sliver of the cohort, and follow-up was short. So while the direction of tirzepatide's lead is clear, the exact size of it, that 13 percent figure and the fivefold hazard ratio, rests on a small sample and could move with more data. An observational study also cannot rule out that the people prescribed the newest, most expensive drug differed from the rest in ways no adjustment fully captures. The authors call the design's limits out directly: residual confounding cannot be excluded.
One nuance held across the board. The ordering was the same whether or not patients had type 2 diabetes, though semaglutide produced more weight loss in people without diabetes while tirzepatide worked about equally well either way.
The ranking itself is not news to anyone who read the trial readouts. What is new is watching it reproduce outside the trial, in a database built to look like the actual patient population. That external check is the whole value of a study like this, and it is why the small tirzepatide arm is the part to watch rather than the part to quote.
The size of the gap between these drugs is not an academic point. It is the subject of an active lawsuit: Novo Nordisk sued Eli Lilly over weight-loss advertising ↗, and the fight turns on how the two companies compare their drugs' effects at different doses. A real-world ranking that puts tirzepatide first and semaglutide second, with the exact margin still fuzzy, is the kind of evidence that argument will keep circling.