Eli Lilly ran its own weight-loss drug through an insurance database and reported that 22,512 Americans started tirzepatide for obesity in an eight-month stretch, and that two-thirds of them were still filling it six months later. The headline weight-loss number, a 10.5 percent drop, came from 808 of them.
That gap between the two counts is the story.
The study, published online August 10 in Diabetes, Obesity and Metabolism ↗, used the Healthcare Integrated Research Database, a large pool of US commercial insurance claims and linked health records run by Carelon Research, the analytics arm of the insurer Elevance. The authors pulled adults who filled at least one prescription for tirzepatide for obesity or overweight, without a type 2 diabetes diagnosis, between November 2023 and June 2024, and who had continuous coverage for a year before and at least six months after. That left 22,512 people, average age 46, nearly three-quarters women. Ninety-one percent already carried an obesity-related condition, most often high cholesterol, high blood pressure, or anxiety.
Tirzepatide, sold as Mounjaro and Zepbound, is a once-weekly shot that activates two gut-hormone receptors at once, GIP and GLP-1, which together blunt appetite and slow the stomach. It is the most potent of the current weight drugs by trial data. The question this study asked was not how well it works in a trial but whether people in the real world actually stay on it, and what the ones who stay get.
The persistence numbers are the real find
On staying power, the results were better than the field's track record. Fifty-five percent of initiators were adherent, meaning they had the drug on hand at least 80 percent of the days measured, and 68 percent were persistent, meaning they had not stopped for a long gap by the six-month mark. Real-world persistence has been the soft spot for injectable weight drugs. Earlier commercial-claims work put six-month persistence for the GLP-1 class closer to 46 percent, so a tirzepatide figure near 69 percent is a genuine data point, not a rounding difference.
That matters because a weight drug only works while you take it. The trial-quality weight losses that make headlines assume people keep injecting for a year or more, and the real world keeps proving that a large share do not. A drug that holds two-thirds of its starters through month six is doing better on that axis than its predecessors did.
Then the denominator collapses
The effectiveness half of the paper is where the reading has to slow down. The 10.5 percent average weight loss, and the 12.0 percent seen in people new to GLP-1 drugs, was measured only among persistent users who happened to have both a before and an after weight recorded in the database. That subgroup was 808 people. Out of 22,512 initiators, roughly one in twenty-eight.
Insurance claims are built to track billing, not body weight, so a recorded weigh-in before and after treatment is not random. It skews toward people who stayed in regular care, saw a clinician who logged the number, and by implication were engaged enough to keep coming back. Those are exactly the patients most likely to be losing weight. The 10.5 percent is a best-case slice, not the average experience of the 22,512, and the paper's own framing (effectiveness "among persistent individuals with available measurements") says so in the fine print. What the average initiator lost, including the third who quit and the many with no recorded weight, this design cannot say.
Who wrote it
Four of the eight authors are employees of Eli Lilly, which makes tirzepatide, and the paper carries a Lilly copyright. The other four work for Carelon Research, which owns the database. None of that makes the numbers wrong. The persistence rates are the persistence rates. But it shapes the adjectives. A "favourable persistence profile" is the sponsor's gloss on a finding that also means one in three people stopped a drug they had started for a serious condition within half a year, and the paper does not dwell on why they left.
This is the second real-world read on tirzepatide we have covered in as many weeks. A late-July All-of-Us cohort ranked five GLP-1 shots on weight loss ↗ and put tirzepatide ↗ first. That study asked which drug takes off the most weight. This one asks whether people stay on the drug at all, and answers it with better numbers than the class has posted before, wrapped around an effectiveness figure that rests on 808 selected people. Both questions matter. A sponsor-independent database that logs weight for everyone, not just the engaged minority, is what it would take to turn the second one into an average rather than a best case.