For the people who stay on them, yes. A UK digital prescribing service recorded mean weight loss of 13.5% at six months and 17.4% at twelve [1]. The qualification is the whole article: 630,545 people started, the six-month figure was measured in 270,309 of them, and the twelve-month figure in 58,434. That last group is about nine in every hundred who began.
This is the largest published record of what happens when these drugs are sold the way sellers on this site sell them. Online, by subscription, without a clinic. Read each number with the group it describes, because a headline cohort is rarely the group a figure was measured in.
The numbers, with their denominators
630,545 eligible patients, 80.3% of them women, mean body mass index 36.1. At six months, mean weight loss among those with a recorded weight was 13.5%, 95% CI 13.51 to 13.56, in 270,309 people. At twelve months, 17.4%, 95% CI 17.3 to 17.5, in 58,434.
At six months, 89.9% had lost at least 5% and 72.1% at least 10%. Again, among those who reported a weight.
Self-reported, by people paying for it
Weights came from patients entering them into the service, not from a clinic scale.
That is how a digital service has to work, and it is a second selection on top of the first. Reporting a weight is itself a behavior, and it correlates with the thing being measured. The association between regular ordering and greater weight loss has the same shape. engagement predicting outcome describes who engages at least as much as what engaging does.
The clearest test of that: a form nobody had to fill in
A 48-week observational study followed semaglutide patients with severe obesity. Those who completed a voluntary digital questionnaire before starting were far more likely to still be on the drug at the end [2].
The form did not make them stay. It identified who would. That is the cleanest available demonstration that a digital program’s results partly measure who signs up for it. the predictors the study reported are worth reading beside any retention claim a seller publishes.
Does the support layer do anything?
One digital clinic added an AI support agent and adherence rose afterwards [3]. The regression inside the same paper, adjusting for everything else, found the opposite.
Cost predicted dropping out either way. Both findings are in one publication, which is the reason to quote neither alone. the adjusted model is the half that usually goes missing from a marketing page.
The finding that is about money
Patients from less deprived areas placed 26.1% more prescription orders than those from the most deprived areas.
Persistence decides what a course actually costs and whether any of the weight stays off. Here it tracks postcode. That is the same gradient behind most eligible patients declining on price and behind income appearing among the predictors in the persistence literature.
The dose line, read carefully
The authors report that clinically meaningful outcomes were frequently achieved without escalation beyond intermediate doses.
That is an observation about who stayed where, not a recommendation. People who stop escalating are disproportionately people for whom the current dose is working, and the comparison is not randomized. It does establish that the intermediate rungs are where a lot of this market sits, which matters when a seller’s price climbs with the dose.
What a buyer takes from it
That 13.5% at six months is a real result for people who kept ordering and kept weighing. The cohort it came from is four times larger than the group it was measured in.
Both halves belong to anyone deciding whether to prepay for a term. The patient-reported benefits are worth having too — 46.7% reported better mobility, 45.9% better mood and 26.7% drinking less — and they come from the same survivors.