pat_auckland said:

The dose-response is real but shallow at the top.

This is exactly what I could not find anywhere else.

Clinical perspective, offered as context rather than as advice.

jason_sac26 said:

...but the FDA says semaglutide...

Interesting point. I want to add some regulatory nuance: the FDA labeling reflects the specific clinical trial data submitted for approval. Real-world clinical practice often extends beyond the FDA label based on emerging evidence and clinical judgment.

Example: semaglutide was first approved for diabetes (Ozempic), then obesity (Wegovy). The molecule didn't change — our understanding of its applications expanded. Similarly, semaglutide may evolve as more data accumulates.

roxy_nash said:

Steady state is the thing most people miss.

This is where I part company with the consensus forming above. The trial means are being read too generously in this thread. STEP populations were selected, supported, and titrated by protocol, and the real-world curves are consistently a few points worse. That difference is not noise, it is what happens when you remove the study infrastructure.

Ask again with the specifics and you will get a better answer than this one.

One concrete data point for the thread. One habit that pays for itself: post the method alongside the number. A figure without its method cannot be checked, and an unchecked figure is how this community accumulates folklore.

Moderator note: good thread. Keeping it here rather than moving it, because the question is general enough to be useful. Thread quality here is what the rules are for. Keep it up.