Peptide basics
STEP 4 enrolled 902 people and randomised 803. The missing 99 matter
On this page
Some trials run a preliminary period before randomly assigning anyone to a group. This is called a run-in.
STEP 4's registry record shows 902 enrolled and 803 randomised — a gap of 99 people.
Those 99 left during the run-in, typically because they could not tolerate the treatment or did not stay with the protocol.
The randomised group is therefore already selected for people who got on with it, which makes the eventual result look better than it would in an unselected group.
This page explains a common trial design feature and how it affects the interpretation of results. It uses published registry records as examples and contains no dose, schedule or route.
The two numbers
STEP 4's registry record posts an enrolment of 902. Its baseline characteristics table covers 803 participants, because those are the ones who reached randomisation. The difference is 99 people who took part in the study and appear in none of its comparative results.1
This is not concealment. Both figures are on the public record, and the design is described in the protocol. It is simply easy to miss, and the larger number is the one that tends to get quoted.1
What a run-in is for
A run-in period gives everyone the treatment, or a placebo, before the randomisation happens. Participants who cannot tolerate it, do not follow the protocol, or change their minds leave at that point.3
The stated rationale is efficiency: a trial that randomises only people likely to complete it needs fewer participants to answer its question, and fewer dropouts mean cleaner data.3
What it does to the result
It changes who the result is about. The randomised population is not a sample of people starting treatment; it is a sample of people who already started, tolerated it, and stayed.13
So a tolerability figure from a run-in trial understates what an unselected person faces, because the people who did badly left before the count started. The effectiveness figure is affected in the same direction for the same reason.3
How to spot one
Compare the enrolment figure on the record with the number in the baseline characteristics table. If they differ, something happened between the two, and the design section will say what.12
Common questions
Is a run-in a way of hiding dropouts?
The numbers are public, so nothing is hidden. It does move some dropouts outside the analysed population, which is why both figures are worth reading.
Do all GLP-1 trials use run-ins?
No. Designs vary, and the registry record for each trial states what it did. STEP 4's structure differs from STEP 1's.
Which number should I quote?
The randomised number, when discussing results, because that is the population the results describe. Say which one you mean.
Sources
- Primary studySTEP 4 (NCT03548987) — enrolment and randomised populationClinicalTrials.gov, U.S. National Library of Medicine, 2021clinicaltrials.gov/study/NCT03548987 ↗↩ Back to text
- Primary studySTEP 1 (NCT03548935) — enrolment and designClinicalTrials.gov, U.S. National Library of Medicine, 2021clinicaltrials.gov/study/NCT03548935 ↗↩ Back to text
- RegulatoryClinicalTrials.gov — glossary of common study termsU.S. National Library of Medicine, 2026clinicaltrials.gov/study-basics/glossary ↗↩ Back to text