Why Clinical-Trial Results and Real-World Results Differ on GLP-1s

• written by Perjan Duro

You have seen the numbers.

Around 15% average weight loss on semaglutide. Up to 22.5% on tirzepatide.

What usually gets left out is the rest of the protocol those numbers came from: a dietitian every four weeks, a daily food and activity record, and a trial population that mostly stayed on treatment for well over a year.

That context is not a footnote. It is a large part of why headline trial results and real-world results look different.

Quick answer

The pivotal GLP-1 trials tested medication plus a structured lifestyle programme, in both the drug and placebo arms. Real-world users get the medication, and often none of the surrounding structure.

Real-world studies also report that many people stop within the first year, and that reported weight loss scales with how long people stay on treatment.

So the gap between "trial results" and "your results" is not only about the drug. It is about the routine wrapped around it, and how long that routine runs.

What the trials actually involved

STEP 1 (semaglutide 2.4 mg)

Wilding and colleagues randomised 1,961 adults and reported a mean change of −14.9% body weight at 68 weeks with semaglutide 2.4 mg, versus −2.4% with placebo.

Both groups — drug and placebo — also received a lifestyle intervention. As reported in the trial, participants had:

  • counselling sessions with a dietitian roughly every 4 weeks;
  • a target 500 kcal/day energy deficit;
  • a target of 150 minutes/week of physical activity;
  • daily recording of food intake and activity, via a diary or an app.

That is not a small add-on. It is a year and a half of structured, supervised routine running underneath the injection.

STEP 3 (semaglutide added to intensive behavioural therapy)

Wadden and colleagues studied 611 adults and layered semaglutide on top of an even more intensive programme.

Both arms received 30 counselling visits across the trial plus a low-calorie diet run-in period. This trial was specifically designed to ask what the medication adds when the behavioural support is already maximal — not to test the drug alone.

SURMOUNT-1 (tirzepatide)

Jastreboff and colleagues reported mean weight reduction of up to −22.5% at 72 weeks on tirzepatide.

Here too, the medication was studied as an adjunct to lifestyle counselling, provided in both the drug and placebo arms.

A neutral note on provenance: STEP 1 and STEP 3 were funded by Novo Nordisk, and SURMOUNT-1 by Eli Lilly — the manufacturers of the medications studied. Industry funding is normal for pivotal drug trials and does not by itself invalidate the results; it is simply part of reading them accurately.

The pattern

Three trials, one design choice in common: every reported number — STEP 1's −14.9%, STEP 3's drug-on-top-of-therapy result, SURMOUNT-1's −22.5% — was produced with structured lifestyle support running in both arms, from dietitian check-ins to daily logging.

None of these were "take the medication and see what happens" studies.

What real-world data looks like

The picture outside a trial is different in one specific way: how long people stay on treatment.

Persistence. Marshall and colleagues analysed 33,607 initiators of high-potency, weight-loss-indicated GLP-1 medications. One-year persistence was 33% among 2021 initiators, rising to roughly 61% for people starting in the first half of 2024. Persistence was higher for tirzepatide than for semaglutide.

Two things are true there at once. A third of people continuing at one year is a long way from a trial population running 68 to 72 weeks. And the trend has been improving substantially.

Outcomes tracked with persistence. Samuels and colleagues followed 2,306 patients in an academic obesity clinic. About 50% had discontinued by 12 months. Reported total body weight loss scaled with how long people stayed on:

  • 5.5% among those persisting to 3 months;
  • 14.4% among those persisting to 12 months.

Read that carefully. This is an observational association, not proof of cause. People who feel well, tolerate the medication, and can afford it are also more likely to stay on it — so persistence and outcome influence each other in both directions. But the association is consistent and it is large.

Why people stop. Van Laren and colleagues characterised 1,374 real-world patients and reported a 31.7% discontinuation rate by 6 months. The leading reasons given were:

  • adverse effects — 26.8%
  • cost — 14.4%
  • non-adherence — 11.2%

Notice that the top two are largely outside the reach of a better routine. Side effects and cost are clinical and financial problems, and they deserve a real conversation with a clinician or pharmacist rather than more willpower.

What you can actually replicate

You cannot give yourself a dietitian every four weeks for free, and you should not try to self-manage side effects or costs by pushing through. Those belong with your care team.

But a meaningful share of the trial scaffolding is structural, not clinical — and it is available to anyone:

1) A daily record

STEP 1 asked participants to log food and activity every day, by diary or app. That was part of the protocol, in both arms.

You can do that. It requires no prescription and no appointment.

2) A fixed review cadence

The trials did not review progress continuously. They reviewed it on a schedule — every four weeks in STEP 1.

A recurring review beats constant checking. It also gives daily noise time to average out into something readable.

3) A defined routine, written down

Trial participants were given explicit, specific targets rather than a vague intention to eat better. The specificity is the point: a defined routine can be followed, missed, and resumed. A vague one can only be felt good or bad about.

Your specific targets are a conversation for you and your clinician. Having them written down at all is the replicable part.

4) A long enough horizon

The headline numbers are 68 and 72-week numbers. Not 8-week numbers.

If you judge your plan on a fortnight, you are reading a different study than the one the number came from.

5) Continuity as its own goal

Given what the real-world data reports about persistence, "am I still on track with my plan and my appointments?" is at least as useful a question as "how much did I lose this week?"

If side effects, cost, supply, or life logistics are threatening continuity, that is the thing to raise at your next appointment — early, with specifics, before it becomes a gap.

The honest summary

The trial numbers are real. They were also produced by a package: medication, structured routine, daily recording, regular professional contact, and 15+ months of continuity.

If you only have part of the package, expecting the full headline result is comparing yourself to a study you are not enrolled in. That is not a reason to be discouraged — it is a reason to be precise about which parts you can add back.

Velto implementation

The daily record and the fixed review cadence are the two parts of the trial routine you can run on your own.

That is exactly what Velto is for: logging treatments, symptoms, appetite, weight, and habits in one timeline, so a monthly review shows you the pattern instead of asking you to remember it — and so your next appointment starts from a record rather than a guess.

Velto is a tracking companion. It does not treat, diagnose, or make dosing decisions, and it cannot promise you a trial result.

References

Related reading

Medical Disclaimer: Educational content only. Not medical advice. For treatment decisions, consult a licensed clinician.
Back to Blog

Articles you might like