In a randomized experiment, they rated the drug below losing nothing. Willingness to associate ran 0.52 points lower for a person described as using a GLP-1 than for one losing weight through diet and exercise [1]. The same description also scored 0.26 points below a person described as not having lost weight at all. That is the finding, and it is the one most likely to be quoted out of shape.
What was measured
Two randomized experiments gave participants a short description of a fictional person. The weight history in that description was varied and everything else was held constant. The first experiment ran with 607 participants, the second with 706.
Willingness to associate was the measure that separated. The GLP-1 condition ran below the diet-and-exercise condition at 0.52 points, 95% CI 0.27 to 0.76. It ran below the no-weight-loss condition at 0.26 points, 95% CI 0.02 to 0.51.
Regain is judged separately
The second experiment tested what happens after the weight comes back. The GLP-1 and diet-and-exercise descriptions were rated similarly once regain was described. Both were rated below a description of somebody who maintained the loss.
So the judgment attached to regain is not specific to the drug. It attaches to the regain, whatever produced it. How much weight comes back, and how fast, is set out in what regain actually measures.
What patients reported from inside it
Thirty adults across 15 US states were interviewed about taking these drugs, 23 of them still on one [2]. Stigma came up as varying by what the drug was prescribed for. People treated for diabetes described a different reception from people treated for weight.
Cost and access came up as prohibitive in the same interviews. Neither appears in any efficacy figure, and both shape whether somebody stays on treatment. The rest of what those participants described is collected in what patients said when asked.
The number behind the second judgment
A meta-analysis pooled 17 studies covering 3,793 participants in cessation arms [3]. Mean regain after stopping came out at 7.20%, 95% CI 5.93 to 8.48. By drug it ran 4.83% for liraglutide 3.0 mg, 7.19% for semaglutide 2.4 mg and 13.04% for tirzepatide.
Heterogeneity was 97.6%, which means the included studies produced very different answers. The ordering between drugs is the signal and the exact percentages are soft. In the three randomized trials with a withdrawal phase, cessation arms gained 14.26 percentage points more than continuation arms.
That ordering matters to a purchase as well as to a reception. The drug that takes the most off gives the most back, so a course that ends early costs more than its monthly price implied. What an early stop does to a prepaid term is worked out in the prepay arithmetic.
What this evidence does not cover
Participants rated a paragraph, not a colleague. A vignette experiment measures a stated attitude toward a described scenario. That is a real thing to measure, and it is not the same as how anybody behaves toward a person they know.
The authors frame their conclusion as a case for stigma-reduction work rather than a measurement of lived treatment. For somebody weighing whether to tell people, the usable content is narrow. The concern is not imagined, and stopping carries a separate judgment that lands on everyone who regains. Why people stop is counted in why people stop taking a GLP-1. What a month costs before any of that is in what a GLP-1 actually costs per month.