A GLP-1 clinical trial result needs more context than a percentage in a headline. Check who was studied, what the comparison was, when the outcome was measured and how the analysis handled people who stopped treatment. This guide helps you turn those details into a clear reading note.
Start with the study context
The methods section describes the question the trial was designed to answer. FDA’s overview of clinical research explains how a protocol specifies the participants, duration, comparison, treatment and assessments.
| Detail | What to write down | Why it matters when reading |
|---|---|---|
| Participants | Eligibility criteria, starting characteristics and number enrolled. | The result describes a defined study population. |
| Intervention | The exact medicine, preparation and study regimen. | A shared ingredient name does not describe the whole intervention. |
| Comparator | Placebo, another treatment or another comparison. | A within-group change and a difference between groups answer different questions. |
| Timing | The measurement week and any earlier treatment or run-in period. | Results measured at different stages need that context. |
| Other care | Any dietary guidance, activity support or other care provided to each group. | These are part of the study conditions being described. |
Randomization means assignment by chance. It does not mean the volunteers represent every person who might later receive a medicine. The NIH glossary explains randomization and eligibility criteria.
Name the number you are reading
Percentage weight change relates a person’s change to their starting weight. A change from 100 kg to 88 kg is a 12% reduction: 12 divided by 100, multiplied by 100. A trial’s reported mean summarizes a group; it is not an individual promise.
A responder percentage counts people who meet a specified threshold. If 60 of 100 people meet a trial’s threshold, the response rate is 60%. That does not mean their average weight reduction was 60%. Always keep the threshold, time point and denominator with the number.
FDA’s January 2025 draft guidance on weight-reduction trials (PDF, pages 8–9) distinguishes mean percentage weight change from responder analyses and warns that threshold-based results can exaggerate the apparent effect. It remains draft guidance, not a final rule.
Work through a fictional comparison
Invented numbers for arithmetic practice — no medicine or real trial is represented.
Imagine two groups assessed at the same time point with the same analysis approach. The report gives these mean reductions from starting weight:
The study group changed by 12% on average. Its reduction exceeded the comparison group’s by 8 percentage points. Calling this simply “8% more weight loss” would hide the distinction between a percentage and a percentage-point difference.
This arithmetic alone cannot tell you the result’s uncertainty, how many people contributed data, what adverse events occurred or whether the finding applies to someone outside the study.
Separate stopping treatment from missing data
A participant may stop the study medicine and still attend follow-up visits. Someone else may leave the study, leaving a planned measurement unavailable. Those situations are different.
An estimand is the precise treatment-effect question an analysis aims to answer. For example, an analysis may consider outcomes regardless of treatment discontinuation, or estimate outcomes under a defined hypothetical scenario. Read the paper’s explanation instead of assuming every reported number answers the same question.
Also check how many measurements are missing, the assumptions used to handle them and whether alternative assumptions changed the conclusion. FDA’s ICH E9(R1) guidance (PDF) explains these distinctions. “Completed treatment,” “completed follow-up” and “included in the analysis” should not be treated as interchangeable labels.
Read uncertainty and harms alongside benefit
A confidence interval describes the precision of an estimate. A wider interval indicates less precision. It is not the range of results every individual participant experienced. The NCI-hosted CDISC glossary defines confidence intervals.
Statistical significance also differs from practical importance. Ask how large the difference is and what it means for the outcome being studied. NCCIH’s guide to study results explains this distinction and the value of checking financial relationships and repeated findings.
Read the adverse-event table and reasons for discontinuation with the same care. Record the number of people assessed and the observation period. A headline about benefit may leave those details out. NCCIH’s health-news checklist highlights study length, participant relevance and limitations.
Keep a useful reading note
Before repeating a trial result, fill in this short note:
In [population], [intervention] was compared with [comparator] over [time]. The reported outcome was [measure and result], using [analysis approach]. Important limits were [uncertainty, missing data and harms]. Source: [full paper and date].
If a detail is unavailable, mark it as unknown. Avoid ranking medicines by headline percentages from separate trials without examining their populations, methods and study conditions. A precise account of one study is more useful than a comparison its design cannot support.
This article explains research-reading methods using the linked primary sources. The arithmetic example and reading-note template are editorial teaching tools. They do not assess any particular treatment or person’s care.