Asked to choose between hypothetical weight-loss drugs, 800 American adults ranked how a drug is taken above how much weight it takes off. In a discrete-choice survey published on 17 September, route of administration was the single most important attribute driving choice, ahead of average weight loss, cardiovascular risk reduction and the share of people reaching 20% weight loss. The study was designed and co-authored by employees of Novo Nordisk, and the profile that won its head-to-head comparison was the one built to resemble Novo Nordisk’s oral drug.
What the survey asked
The OPTIC study surveyed 800 US adults who had overweight or obesity plus at least one obesity-related complication, online, in October and November 2025. The sample was assembled in balanced halves: 400 had taken an obesity medication before and 400 had not, 400 had used an injectable and 400 had not.
In the main exercise, each participant repeatedly chose between two hypothetical drug profiles that varied on efficacy, cardiovascular risk reduction, side effects, route and frequency of administration, and dosing-instruction requirements. Relative preference weights were estimated with a random-parameters logit model. Route of administration came out most important, followed by average weight-loss percentage, cardiovascular risk reduction, and the proportion of people achieving at least 20% weight loss. Dosing-instruction requirements ranked last of the set. Only 23.4% of participants said that taking a tablet on an empty stomach and waiting half an hour before eating would be disruptive to their lives, and 73.3% of those who had never taken an obesity medication said they were open to an oral one.
The paper then put a separate fixed question to the same participants: a straight choice between two predefined oral profiles, one constructed to resemble oral semaglutide and one constructed to resemble orforglipron. The semaglutide-like profile was chosen by 84.2% in the choice experiment and 90.0% in the fixed question.
Who ran it, and why that matters
Four of the eight authors are employees of Novo Nordisk Inc in Plainsboro, New Jersey. Two are at RTI Health Solutions, a contract research organisation. The remaining two are academics, at UT Southwestern Medical Center and Weill Cornell Medicine. Novo Nordisk makes oral semaglutide. Orforglipron is Eli Lilly’s oral GLP-1 drug, and Lilly is Novo Nordisk’s closest competitor in this market.
So the headline result is that people preferred a profile resembling the sponsor’s product to one resembling its main rival, by better than five to one. The attribute ranking deserves the same reading. Oral semaglutide carries a fasting requirement that orforglipron does not, and dosing instructions is exactly the attribute this study found people care about least. A survey that ranks the sponsor’s main practical disadvantage as the least important thing on the list, and places the category where its product competes at the top, is not evidence of bad faith — but it is the shape a favourable result takes, and it is worth naming before these numbers turn up in marketing.
None of that makes the arithmetic wrong, and the broad signal is plausible on its own terms: a lot of people would rather swallow a tablet than inject themselves, and that preference is commercially consequential no matter who measures it. What the conflict does mean is that the head-to-head comparison is the part to treat most carefully, because it is the part that was constructed rather than observed.
Nobody in this study took anything. Participants chose between descriptions of drugs that do not exist exactly as described, and stated preference is a weak predictor of what people do when facing an actual prescription, an actual price and actual side effects. Cost was not among the attributes participants traded off at all, which in the US market is a substantial omission — price is frequently the thing that decides which of these drugs a person ends up taking, or whether they take one. The two head-to-head profiles were hypothetical by construction, so the 84% is a comparison of descriptions rather than of experience. And a sample recruited online and split into equal halves by prior medication and injection experience is engineered for statistical balance, not built to represent the US population.
Sources: Almandoz JP, Tchang BG, Myers K, Traina A, Bhavsar J, Divino V, Kent C, Bell ST. “Preferences for Obesity Medications Among People With Overweight or Obesity in the United States: A Discrete-Choice Experiment (OPTIC).” Diabetes, Obesity and Metabolism, published online 17 September 2026. doi:10.1111/dom.71318 · PubMed 42755136.
