The answer so far

A headline percentage is a group summary. To understand it, ask who was studied, what the comparison was, when the result was measured, how missing data were handled and what harms occurred alongside the average.

An average weight change in a trial describes the people and analysis specified by that study. It is not a promise that each participant, product or future population will have the same result. The comparator, time point, analysis method and withdrawals are part of the result.123

On this page
  1. Read the denominator first
  2. An average does not describe everyone
  3. Attach the comparator and time
  4. Put harms beside benefits
  5. Sources
A DESIGNED COMPARISONGroup AGroup BRandom allocation

In this schematic comparison, participants are allocated at random to group A or group B.

Conceptual schematic. Simplified and not to scale; not a scan, microscopy image or validated simulation.

Read the denominator first

Find the number randomised, the number included in the analysis and the time point. If participants stopped treatment or missed measurements, ask whether the analysis estimates an effect under continued treatment, treatment policy or another defined strategy.45

An average does not describe everyone

A mean or model-based estimate summarises a group. Individual changes can be smaller, larger, absent or in the opposite direction; measures of spread describe variation between people. A confidence interval around a group estimate expresses uncertainty about that estimate, not the range of individual results.13

Attach the comparator and time

A percentage compared with baseline answers a different question from a difference between a medicine and placebo. A result at 68 weeks is not interchangeable with one at 24 or 48 weeks, and a weight endpoint is not a cardiovascular endpoint.123

For example, STEP 1 assigned 1,961 adults with overweight or obesity but without diabetes to once-weekly subcutaneous semaglutide injections with a 2.4 mg target dose or placebo, alongside lifestyle support. At 68 weeks, the estimated mean weight change was −14.9% and −2.4% of starting weight, respectively. The reported estimated difference was −12.4 percentage points, with a 95% confidence interval from −13.4 to −11.5 points. That is a group comparison under the study’s analysis, not a prediction for one person. The trial was funded by Novo Nordisk. See the semaglutide product guide for study context and adverse effects.7

Put harms beside benefits

Gastrointestinal events, discontinuations, serious adverse events and other risks should be read beside the efficacy estimate. A short trial can show common events while remaining unable to estimate uncommon or long-term harms precisely.61

Further reading

Trusted external resources selected to help explain this page. They open in a new tab.

Sources, kept close.

Selected supporting sources; this is not a complete literature review.

  1. Tirzepatide Once Weekly for the Treatment of ObesityJastreboff et al. / New England Journal of Medicine · Peer-reviewed randomised trialAccess checked 2026-11-03 · Source date: 2022 · PubMed abstract and publisher-indexed report text were inspected, alongside separately listed sponsor protocol and analysis plan. Journal supplements have not been independently appraised.
  2. Tirzepatide versus Semaglutide Once Weekly in Patients with Type 2 DiabetesFrias et al. / New England Journal of Medicine · Peer-reviewed randomised open-label trialAccess checked 2026-11-03 · Source date: 2021 · PubMed abstract and publisher-indexed report text were inspected, alongside separately listed sponsor protocol and analysis plan. Journal supplements have not been independently appraised.
  3. Triple–Hormone-Receptor Agonist Retatrutide for Obesity — A Phase 2 TrialJastreboff et al. / New England Journal of Medicine · Peer-reviewed randomised clinical trial; primary reportAccess checked 2026-11-03 · Source date: 2023-06-26 · Publisher report text was retrieved through web source access; a local browser encountered a challenge. A sponsor-submitted protocol is available separately through ClinicalTrials.gov; the NEJM supplementary appendix remains unappraised.
  4. SURMOUNT-1 statistical analysis plan, version 2Eli Lilly / ClinicalTrials.gov · Sponsor-submitted trial protocol or statistical analysis planAccess checked 2026-11-03 · Source date: 2021-09-28 · Public sponsor-submitted document inspected for planned design and analyses; this does not independently establish how those analyses were implemented.
  5. Retatrutide GZBF phase 2 statistical analysis plan, version 4.0Sponsor-submitted document / ClinicalTrials.gov · Publicly posted study protocol or analysis planAccess checked 2026-11-03 · Source date: 2022-12-15 · Sponsor-submitted document inspected as a record of planned design or analysis; it does not itself demonstrate implementation or clinical benefit.
  6. Zepbound (tirzepatide) prescribing information, August 2026 revisionU.S. Food and Drug Administration · U.S. regulator product labelAccess checked 2026-11-04 · Source date: 2026-08 · August 2026 US label: section 16 describes a total room-temperature limit; single-dose pen instructions also say not to return to refrigeration. Check exact presentation and latest label. Section 5.2 says the product is not recommended in severe gastroparesis.
  7. Once-Weekly Semaglutide in Adults with Overweight or Obesity (STEP 1)New England Journal of Medicine / PubMed · Randomized controlled trialAccessed 2026-11-04

The next useful question.

Research guide

How to read a result.

Keep the outcome, the denominator, the time period and the uncertainty together.

7 min read
Substance profile

Semaglutide.

A plain-language guide to semaglutide: one peptide medicine, different products, different authorised uses, and evidence that should not be mixed together.

10 min read