Accepted answer
Read the 1 mg row, not the pooled one. A programme that randomised more than one dose level reports each arm separately, and the figure that circulates afterwards is usually either the top-dose arm or an average across arms nobody was randomised to. If STEP 4 ran a 1 mg arm, that row carries its own sample size and its own confidence interval, and both are narrower than the trial-level ones by roughly the square root of however many arms there were. Take the primary publication rather than the press release: one reports by arm, the other reports whichever number is largest. A dose level inside a trial is a protocol decision made under supervision, not a recommendation, and nothing here is medical advice.
A trial establishes what happened to a defined group under a defined protocol. Extending it beyond that group is inference, and inference is allowed as long as it is labelled.
Trial populations are selected. Exclusion criteria in this class routinely remove people with significant renal impairment, prior pancreatitis and unstable psychiatric illness, which is exactly the population the results are then quoted for.
Open-label extensions are not the same evidence as the randomised phase. Once everyone knows what they are taking, the reported outcomes acquire a bias that no analysis fully removes.
Where a result is quoted from a conference abstract rather than a peer-reviewed publication, the numbers routinely move between the two. It is worth checking which one you are reading.
One qualification: absence of a signal in a trial of this size is not evidence of absence for a rare event. It is evidence that the event is rarer than the trial could detect.
When two sources disagree, the answer is almost always in the methods section of the one you have not read.