In STEP 2, semaglutide 2.4 mg produced an estimated 9.6% mean weight reduction at 68 weeks versus 3.4% with placebo in adults with overweight or obesity and type 2 diabetes. That is a population average, not a target every patient must reach.
Start with the study that matches the condition
STEP 2 studied adults with overweight or obesity and type 2 diabetes. The estimated mean weight change at week 68 was −9.6% with semaglutide 2.4 mg and −3.4% with placebo, a reported treatment difference of −6.2 percentage points. Novo Nordisk funded the trial.[1]
The useful answer is therefore more specific than “semaglutide causes fifteen percent weight loss.” That familiar headline is often borrowed from a different study population. A trial involving people with diabetes provides a more relevant reference point for this particular question.
| At week 68 | Estimated mean reduction |
|---|---|
| Semaglutide 2.4 mg | 9.6% |
| Placebo | 3.4% |
| Between-group difference | 6.2 percentage points |
A percentage of starting body weight is different from a percentage-point gap between groups. It is also different from the proportion of people who reached a threshold. Advertisements sometimes place these kinds of percentages beside one another without identifying the denominator.
Why a lower average is not a failed treatment
A person with diabetes may care about weight, glucose management, treatment burden and other health outcomes at the same time. Judging treatment solely against a weight-loss number from a different population can obscure meaningful benefit. It can also encourage an unnecessary search for a service that promises a larger number.
The trial average includes a range of experiences. It does not establish the right personal target, how long an individual should continue or whether a particular response warrants changing treatment. Those decisions require the broader clinical picture.
A useful progress conversation identifies which outcomes were intended when treatment began and how each will be assessed. That is more informative than treating one published percentage as a pass-or-fail score for everyone who receives a prescription.
Do not turn STEP 1 versus STEP 2 into a causal experiment
STEP 1 enrolled adults without diabetes, while STEP 2 addressed adults with type 2 diabetes.[2] [1] The studies offer useful context, but subtracting their means does not isolate diabetes as the sole cause of a difference. Participants were not randomized to have or not have diabetes.
Other features of the populations and treatment context can differ. This is the same reason that comparing two unrelated medicine trials is weaker than a direct head-to-head trial. The result may prompt a research question without answering it by itself.
Our cross-trial comparison guide explains this principle in another setting. The rule is not to ignore separate trials; it is to describe their differences before turning the numbers into a ranking.
The care model matters when medicines interact
A diabetes-focused prescribing decision involves the existing treatment list and the clinician managing glucose. A new remote provider should have a clear process for reviewing that information and communicating changes. Convenience is valuable, but duplicated or disconnected care can make the overall arrangement harder to manage.
The Virta review and Teladoc review describe programs whose access and scope can depend on health-plan or employer eligibility. The provider shortlist helps distinguish these from direct-pay weight programs.
When comparing cost, include existing covered care before adding another membership. Ask which visits would be new, which services overlap and whether medication is covered separately. The first-year cost comparison keeps recurring care charges distinct from the pharmacy bill.
A realistic expectation is not a pessimistic one. Using the diabetes-specific evidence gives the reader a better starting point for discussing benefit, while avoiding the expense and frustration of chasing an outcome borrowed from a different trial population.