Peptide basics

Why 'studied in 2,000 people' tells you almost nothing

This page contains no dose, schedule or route and makes no claim about any compound. It explains what a sample size does and does not establish, using the posted baseline table of one large trial as the worked example.

What a sample size buys

A larger sample narrows the uncertainty around an estimate. That is the entire benefit, and it is a real one: a trial of 2,000 people produces a tighter interval around its average than a trial of 200.¹

What it does not do is widen the range of people the estimate applies to. Two thousand participants recruited under one set of eligibility criteria describe one kind of person very precisely. They do not describe a second kind of person at all.¹²

The worked example

STEP 1 is the trial behind most semaglutide weight-loss claims and is usually introduced by its size. Its posted baseline characteristics table describes those 1,961 people in some detail.¹

• Sex: 1,453 female (74.1%) and 508 male (25.9%).
• Mean age: 46 years.
• Race as recorded: 1,472 white (75.1%), 261 Asian (13.3%), 111 Black or African American (5.7%), 27 American Indian or Alaska Native (1.4%), 2 Native Hawaiian or Other Pacific Islander (0.1%).
• Ethnicity as recorded: 236 Hispanic or Latino (12.0%), 1,669 not Hispanic or Latino (85.1%).¹

Read that table and the sentence changes. The trial did not study adults; it studied a group that was three-quarters female, three-quarters white, and centred on age 46, under eligibility criteria that excluded people meeting a diabetes threshold.¹

SURMOUNT-1 posted a comparable picture from a different sponsor: 1,714 female of 2,539 (67.5%), a mean age of 44.9 years, and 1,792 of 2,539 recorded as white (70.6%).⁵

Race and ethnicity are recorded categories, not facts about biology

The categories in that table come from a standard set. FDA guidance to sponsors on collecting race and ethnicity data in clinical trials sets out the recommended categories and the way they should be collected, which is why trials from different sponsors report comparable buckets.³

Those categories describe how participants were classified in a study record. They are useful for checking whether a trial recruited broadly, and they are not a statement about how any compound behaves in any group.³

Where the regulator agrees

FDA guidance on eligibility criteria, enrolment practices and trial design recommends that sponsors broaden eligibility where criteria are not needed for participant safety or scientific validity, precisely because unnecessarily narrow criteria limit what can be learned about the people who will use a product.²

ICH E7 makes the same argument for age, stating that protocols should not ordinarily include arbitrary upper age cutoffs and that geriatric participants should be present in meaningful numbers.⁶

Approved labeling reflects the same discipline. It describes what was studied rather than generalising it, which is why the geriatric use section of a label prints a headcount rather than a reassurance.⁴

The one-minute check

1. Find the NCT number of the trial being cited. It is usually in the paper's abstract or methods.
2. Open the ClinicalTrials.gov page for it and go to the posted results, then baseline characteristics.
3. Read the sex, age and race rows. Those four lines are the population the headline actually describes.
4. Read the exclusion criteria. If a group you care about is on that list, the trial says nothing about them.¹

STEP 1's size, 1,961 randomised participants, is the number most often quoted about semaglutide. The four rows below it in the same table are the ones that determine what the number means.

Common questions