Before starting the drug, these patients were asked to fill in some questionnaires. Nothing made them. Some did, some did not, and the ones who did were still taking semaglutide a year later at a noticeably higher rate. [1]
That is a useful finding and it is not the finding a digital health company would write on a slide.
What the gradient shows
Persistence at 48 weeks went 59.5% among those who never opened the platform, 71.9% among partial users and 79.5% among those who completed the forms. Each step up was associated with higher odds of staying, adjusted odds ratio 1.62 with an interval of 1.05 to 2.50.
A clean monotonic gradient across three levels is the kind of pattern that usually means something real. What it means here is that people who answer optional questionnaires are different from people who do not, in ways that also predict sticking with a hard treatment — the same selection that makes a digital clinic's twelve-month figure describe the people who stayed to be measured.
What the mediation analysis says
Persistence accounted for roughly 69% of the association between engagement and weight loss, and the remaining direct effect was not significant.
Read plainly: engaged people stayed on the drug, and staying on the drug is what took the weight off. There is no separate benefit of engagement once you account for still taking it.
The number a buyer should take
Weight change ran 3.9% among those who discontinued and 15.3% among those who reached the 2.4 mg dose. Mean loss was 11.8% across everybody and 14.6% among those who persisted.
That spread is the whole economics of this purchase. It is also confounded by construction — reaching the top dose requires staying long enough to titrate to it — so the two are not independent. What it does confirm is the pattern behind what stopping costs and why a cheap rate that assumes twelve months is priced against a thing most people do not do.
Why 74% is not the national figure
Overall persistence here was 74.3%, and 87.3% of those still taking it had reached the full dose. This paper’s own background quotes 32% to 50% at twelve months in the real world, and fewer than 15% reaching 2.4 mg.
One French center, an early-access program, consecutive enrollment and baseline questionnaires. Every one of those selects, and together they explain most of the gap against what people say when they quit. Take the gradient from this study and leave the level.
One behavioral scale survived correction out of seven tested, the Intuitive Eating Scale-2, in the 91 participants who completed everything. One result from seven questionnaires in ninety-one people is a hypothesis.