Data investigation
Who Was Actually in the Trials
Every result comes with an unstated question: does it apply to you? A review of 246 trials and 139,566 participants shows who was studied, and the mismatch has a consistent shape.
Every result on this register comes with an unstated question: does it apply to you? A systematic review answered it directly, examining the baseline characteristics of 246 randomized trials covering 139,566 participants across twelve obesity medications and three decades.[1] The people studied are not the people affected, and the mismatch has a consistent shape.
Who was over-recruited, and who was missing
| Over-recruited | Under-recruited |
|---|---|
| White participants | Non-White participants |
| Female participants | Male participants |
| Aged 40 or over | Older participants |
| Class 1 obesity (BMI 30.0–34.9) and class 2 (35.0–39.9) | Class 3 obesity (BMI 40 or above) |
The people with the most severe obesity are the least represented in the evidence for treating it.
That last row deserves the most attention, because it is backwards. People with class 3 obesity have the most to gain from an effective treatment, are the most likely to be prescribed one, and are the group a clinician is most confident about recommending it to. They are also the group the trials studied least.
What this does and does not mean
Uncertainty is not bad news, but it still matters. A confidence interval reported for a trial population applies to that population. Extending it to a group who made up a small fraction of participants is an act of judgment, usually a reasonable one, and always a step beyond what was measured. Some groups were hardly enrolled at all: for adults with a chronic spinal cord injury, the only semaglutide study we found followed five people.
There is one relevant piece of reassurance. Where a rare genetic subtype of obesity has been examined inside a trial, response was comparable to everyone else (see genetics caused it and the drug still worked). That is a different question from demographic representation, and it points the encouraging way.
A second check points the same encouraging way. Adults aged 40 or under are underrepresented in type 2 diabetes trials, so a post hoc analysis of the SUSTAIN and PIONEER programs went back and examined them separately. It found semaglutide reducing HbA1c and body weight across every age subgroup, particularly in young adults, with comparable rates of serious adverse events.[2]
That is the useful response to this problem: go back to the data and check, instead of assuming an underrepresented group responds the same or differently. Both times anyone has checked here, the answer has been reassuring, and so was a tirzepatide analysis of people whose obesity began before age 25.
Where this shows up in specific trials
Once you know the pattern, you will see it in almost every trial this register covers. These are not criticisms of individual studies; they are the same systematic skew, one paper at a time.
- 82.4% women in the genome-wide study of who responds, which was also restricted to European ancestry (the genetics are real, and small).
- 92% women, mean age 64 in the bone density study, which shapes everything it found (bone density and fracture risk).
- 75% men, median age 68 in the peripheral artery disease trial. That is the opposite skew, for a condition that itself skews that way (how far you can walk).
- 85% White in the retatrutide body composition substudy.
- Chinese BMI thresholds and a mean age of 33 in a phase 2b trial whose results do not transfer straightforwardly (the drugs that are not here yet).
The BMI question has a second dimension the review does not address. Obesity is defined at a BMI of 25 rather than 30 in many Asian populations because risk rises at lower body mass. A trial recruiting at class 1 and class 2 by Western thresholds is therefore recruiting a different metabolic population than the same numbers would describe elsewhere. We cover that in the BMI threshold is not universal.
What to do with it
- Check the baseline table, not just the result. Every trial publishes who was in it, and it takes ten seconds to see whether that includes people like you.
- Treat effect sizes as estimates from a described population, which is what they are, rather than as universal constants.
- Do not read under-representation as evidence of worse response. It is evidence of not knowing.
- Expect this to improve slowly. The review’s own recommendation is that future trials recruit the groups they have been missing, which is the right conclusion and a slow one.
For what these numbers look like when applied outside a trial altogether, see what these drugs do outside a trial and who responds in the real world.
Frequently Asked Questions
References
- 1.Alsaqaaby MS, Cooney S, le Roux CW, Pournaras DJ. Sex, race, and BMI in clinical trials of medications for obesity over the past three decades: a systematic review The Lancet Diabetes & Endocrinology. 2024. PMID: 38723646.
- 2.Zaccardi F, Aroda VR, Arslan EG, et al. Semaglutide Treatment in Young Adults Living With Type 2 Diabetes: A Post Hoc Analysis From the SUSTAIN and PIONEER Clinical Trials Diabetes, Obesity and Metabolism. 2026. PMID: 41994903.
Compare the sellers
Related research
Comparing Trials That Never Met
We refused a cross-trial comparison in an earlier article. This article runs the methodologically proper version of it, and explains why it appears to contradict a head-to-head trial without actually doing so.
7 min read
Another Molecule You Cannot Buy
A dual agonist reported 14.7% weight loss in thirteen weeks. The drug is approved nowhere, and a thirteen-week figure sits on the steepest part of a curve that flattens.
6 min read
A Composite Is Only Its Components
A reanalysis found tirzepatide getting far more patients to three or more treatment targets. The comparator was semaglutide at 1 mg, and the gap was carried by two of the four targets.
7 min read
Real, and Below the Threshold
Semaglutide improves patient-reported physical functioning by 1.71 points, with an interval excluding zero. The threshold at which a patient notices a change is 3.7 points. Both are true at once.
6 min read
When Pooling Studies Hides the Answer
A meta-analysis of 323,439 patients found no significant cardiovascular advantage for tirzepatide. The more important number sits beside it: the studies being pooled disagree with each other almost completely.
7 min read
The Mental Health Score That Did Not Move
In the head-to-head trial, physical quality-of-life scores improved clearly over 72 weeks. The Mental Component Summary did not significantly improve in either arm, a result the paper's own conclusion omits.
6 min read
Where to get GLP-1 online, safely: sellers our editors have checked
These are telehealth sellers our editors have checked. For each one we hold a price, the form the drug comes in, and the states it reaches.
No insurance needed · vetted by our editors
Some of the links on this page earn us money. If you sign up with a provider after following one, that provider may pay GLP Watchdog a commission. Learn more
WellMedRx
Comparing compounded and brand GLP-1 plans with a published state list
From $49/mo
Get started →altRX
Compounded semaglutide from $89/month at the same price at every dose
From $89/mo
Get started →CoreAge Rx
Flat-priced compounded semaglutide at $79 a month, any dose
From $79/mo
Get started →