GLP-1: reported results by outcome

What did these trials count as a result?

systematic review

A result in this literature is a number read on a date chosen before anybody enrolled. In the 72-week obesity trial of the oral molecule orforglipron, the primary end point was the percent change in body weight from baseline to week 721. A 44-week semaglutide trial counted two things at once, because its primary endpoints were percentage change in mean bodyweight and proportion of participants having reached a weight reduction of at least 5% of bodyweight from baseline to week 442. The heart trials counted something else again, pooling major adverse cardiovascular events (MACE; ie, cardiovascular death, stroke, or myocardial infarction)3 across seven trials in people with type 2 diabetes.

Each is a group figure with a comparison arm attached, read on a scheduled day. That is a different object from what most readers mean by results, which is what happened to one person across a year nobody scheduled.

Did everyone in the trials lose weight?

human RCT

No, and some of these trials counted how many did not. In a 36-week phase 2 trial of the oral molecule, a weight reduction of at least 10% by week 36 occurred in 46 to 75% of the participants who received orforglipron, as compared with 9% who received placebo4. On the assigned amount that moved the most participants, a quarter still finished below ten per cent, and on the weakest one slightly more than half did.

The comparison arm moves as well: in the 44-week semaglutide trial, estimated mean percentage change in bodyweight from baseline to week 44 was -12·1% (SE 0·5) with semaglutide 2·4 mg versus -3·6% (0·7) with placebo2. An average summarises a distribution rather than establishing a floor beneath it, and that distribution is where any individual actually lands.

Why am I not losing weight on a GLP-1?

review

Three things separate a published average from a stalled month, and each is documented rather than inferred. The first is how much is actually going in, and a 2025 review of ordinary clinical practice reports that real-world data shows that many patients use lower doses and do not stick to their treatment as strictly as participants in a controlled trial might, leading to less weight loss5. The second is whether the schedule is maintained, since the same review found that those who do follow their plans closely can achieve results similar to those in trials5. The third is simply where in the published distribution an individual falls, which the section above counts.

None of that supports a verdict about one reader. The amounts the trials assigned are laid out on the GLP-1 dosage page, the molecule behind the largest averages belongs to the GLP-1 benefits page, and the documented adverse effects are taken apart on the GLP-1 safety page.

How long would it take to lose 20 pounds?

systematic review

These trials cannot answer that in pounds, because none of them reports a result in pounds. A 2024 pooled analysis of 28 trials in adults with type 2 diabetes found that compared with placebo, tirzepatide was more efficacious than semaglutide for reducing body weight, with reductions ranging from 9.57 kg (tirzepatide 15 mg) to 5.27 kg (tirzepatide 5 mg)6. Those are gaps against a placebo arm built up over trials of at least twelve weeks, not a monthly rate anyone can multiply out. In the 72-week obesity trial, the mean change in body weight from baseline to week 72 was -7.5% on the lowest assigned amount1.

A percentage is a different quantity for every reader, because it is a share of what they weighed at the start. None of the trials covered here follows one person to a fixed number of pounds, so turning a group percentage into somebody's calendar is arithmetic this evidence does not carry.

What did the trials give people besides the molecule?

human RCT

A diet and an activity programme, in every arm including the placebo one. The 44-week semaglutide trial gave its weekly injection plus a diet and physical activity intervention2, and the 72-week tablet trial ran its assignment as an adjunct to healthy diet and physical activity for 72 weeks1.

One Copenhagen programme did the weight loss first. During an 8-week low-calorie diet (800 kcal/day), 195 adults with obesity and without diabetes lost 12% in body weight7, and only then were they randomly assigned for a year. Its exercise arm was supervised and logged, and participants randomized to exercise performed a median 2.65 session/week (116 min/week at 79% of maximum heart rate)8.

A published figure is therefore what a molecule added on top of a structured programme, measured against people following the same programme without it. No trial covered here was built to isolate what it does alone.

Why does one trial report two different numbers?

human RCT

Because the same data can be counted by two rules, and one phase 3 tablet trial published both. Two estimands addressed two efficacy-related questions: a treatment policy estimand (regardless of trial product discontinuation or rescue medication use) and a trial product estimand (on trial product without rescue medication use) in all randomized patients9. An estimand is simply the rule deciding whose numbers get counted at the end.

The two rules produced two sets of figures at week 26: treatment policy estimand, -0.6% [3 mg], -0.9% [7 mg], and -1.1% [14 mg]; trial product estimand, -0.7% [3 mg], -1.2% [7 mg], and -1.4% [14 mg]9. The second set is larger throughout, because it describes only the participants who stayed on what they were assigned. Same trial, same people, two totals. Newer obesity trials chose the wider rule, assessing weight change by a treatment-regimen estimand covering everyone randomised1, and which rule a headline came from is rarely printed beside it.

Who were the people behind these figures?

human RCT

Named and weighed before anything began, which is the part that rarely travels with the number. In a 40-week trial comparing two weekly injections, at baseline, the mean glycated hemoglobin level was 8.28%, the mean age 56.6 years, and the mean weight 93.7 kg10. Glycated haemoglobin is a blood test reflecting average blood sugar over roughly three months. In the first oral programme, adults with type 2 diabetes insufficiently controlled with diet and exercise were randomized9, while the 44-week weight trial recruited participants from 23 hospitals and trial centres in China, Hong Kong, Brazil, and South Korea2.

Every figure on this page belongs to a particular starting weight, a particular diagnosis and a particular set of countries. Who these trials would not enrol at all is a different question, and the exclusion criteria behind it are collected on the GLP-1 safety page.

Why are older people quitting GLP-1?

human RCT

The honest answer is that age is barely covered by these figures, and the discontinuation is counted without it. A 2025 review of ordinary practice reports that a major issue with GLP-1RAs is that many patients stop using them within the first year due to side effects or high costs of the medications, especially if not covered by insurance5. What that review does not do is separate that rate by how old anybody was.

The trials are little help either. The Copenhagen programme studied 193 adults with obesity (age 18-65 years, body mass index 32-43 kg/m2) without diabetes mellitus8, an upper limit that stops at retirement age, and the one large trial with a clearly older population was built for the heart rather than for weight: the mean age of randomized participants at baseline was 64.1 years, diabetes duration 14.7 years, HbA1c 8.4%, and BMI 32.6 kg/m211.

Why somebody over sixty stops and why somebody under forty stops are different questions, and none of the studies covered here separates them. The adverse effects behind that first reason are documented on the GLP-1 safety page; what a course costs is checked on the GLP-1 sourcing page.

How many people stayed in these trials to the end?

human RCT

Fewer than were randomised, and the drop-off is printed in the papers themselves. In the 40-week SURPASS-1 trial of a weekly injection, 66 (14%) participants discontinued the study drug and 50 (10%) discontinued the study prematurely12.

The Copenhagen year narrows in three steps. Of the 195 adults sent on from the diet, a total of 166 participants completed the trial7. The main result rests on a smaller group again, because statistical analysis was performed on 130 participants adherent to the study interventions (per-protocol population)7. A year after it all stopped, 109 participants attended the post-treatment study13.

So a number at week 52 is a number about the people still present at week 52, read against the comparison arm of that same trial. The ones who left took their outcomes with them, and outside a trial they are the larger group.

What do before-and-after pictures show?

human RCT

Much less than a trial readout, and the gap is the comparison group. The Copenhagen bone analysis printed all four arms beside each other: the total estimated mean change in weight losses during the study was 7.03 kg in the placebo group, 11.19 kg in the exercise group, 13.74 kg in the liraglutide group, and 16.88 kg in the combination group14. The placebo arm is the point, because people lost weight in it as well, so any single arm read alone overstates what a molecule added.

The scans go further than the scales. That same trial estimated abdominal obesity as android fat via dual-energy X-ray absorptiometry7, and its bone analysis read site-specific bone mineral density by the same scanner across everyone randomised14.

A photograph carries none of that: no comparison arm, no scan, no recorded starting weight, no account of the people who stopped posting. It describes an outcome rather than measuring one.

What result stands behind a GLP-1 patch?

human RCT

None that was measured, and the reason matters more than a verdict would. Every figure on this page arrived by one of two routes: a weekly shot, with participants assigned to receive a subcutaneous injection of either semaglutide 2·4 mg or placebo once a week2, or something swallowed, a tablet or a small molecule built to survive the gut.

Reaching that second route took a research programme of its own. A 2019 phase 3 trial tested the first oral glucagon-like peptide 1 (GLP-1) receptor agonist, oral semaglutide9, and was conducted in 93 sites in nine countries9. A 2022 review of diabetes medicines puts the engineering plainly, noting that while this has already been achieved for incretins, there are still some challenges for the oral application of insulin15, incretins being the gut hormones this class copies.

A route is a result in its own right, and the skin has produced none among the trials covered here. That is an absence of measurement rather than a finding of failure, and it is why a patch review describes a purchase rather than an outcome. The class and its two routes are described on the GLP-1 overview.

What is a GLP-1 booster review actually describing?

human RCT

A different kind of object from the ones measured here, and the difference lies in how each number was produced. The weight figures above emerged from designs built to remove the reader's judgement from the result: one was a 26-week, phase 3a, randomized, double-blind, placebo-controlled, parallel-group trial9, and in another, participants, investigators, and the trial sponsor were masked to treatment allocation until after database lock2.

Masking is what makes a placebo arm worth anything. In the 36-week tablet trial, nine per cent of the participants given placebo still crossed the ten per cent mark, without knowing which arm they had been allocated to.

A customer review carries no masking, no comparison arm and no record of who quit. It is an account of an experience, which is a reasonable thing to read and not the same thing as a measured outcome. Among the trials covered here, none tests a capsule, a drink or a blend sold over the counter.

What is still unmeasured about one person's result?

human RCT

Three gaps sit directly beneath the question most readers arrive with. The first is who responds: the trials counted how many participants crossed a threshold, but which participants crossed it, and what distinguished them from the rest, goes unreported.

The second is how long any of it holds, since the one programme covered here that followed people after treatment ended reassessed them a single year later, and its authors open by noting that long-term adherence in a real-world setting is challenging13.

The third is the reader. Whether a particular person lands at the top of a published distribution or the bottom is a different question from the one these trials asked, and whether any of it is worth doing is a judgement about a life rather than a reading of evidence. We do not make it. The documented harms are collected on the GLP-1 safety page, and the measured outcomes are gone through one by one on the GLP-1 benefits page.