The two study settings answer different questions
A randomized trial assigns treatment under a defined protocol. An observational study examines care as recorded in practice. Both can be valuable, but the circumstances differ. Real-world data can reveal treatment interruptions and access patterns, while randomization helps reduce certain biases in a comparison. A lower or higher average in one setting does not automatically prove the other study is wrong.
Persistence changes who contributes to the result
A clinical-practice study examined weight and glucose outcomes according to discontinuation status among people starting semaglutide or tirzepatide. That distinction matters because a result among continuing users is not the same as a result among everyone who started. Before using an average, identify whether the analysis requires people to remain on treatment or includes those who stopped.[1]
A final prescription record may not mean treatment stopped
A 2026 study of weight trajectories after the last recorded prescription examined 4,182 patients. In a 300-person subset whose notes were reviewed, discontinuation was documented for 40% near that point. This illustrates a key records problem: the last prescription visible to a data system is not automatically the last dose a person used. Care outside the system or incomplete documentation can change interpretation.[2]
Selection can make an average look better or worse
If an analysis includes only people with a follow-up weight, it may differ from one that accounts for those without a later measurement. If it requires continuous use, it may exclude people facing cost, tolerability or access problems. These choices can be reasonable for a particular question, but they need to be visible. Ask who was excluded before treating the reported average as the expected experience of every new patient.
Do not turn an observational comparison into randomization
The SHAPE study reported real-world weight changes for semaglutide and tirzepatide users. Treatment choice in clinical practice is not assigned randomly, so groups may differ in ways that affect outcomes. Statistical adjustment can help, but it does not automatically remove every difference. A numerical advantage in an observational dataset should be described with that design in mind, rather than as definitive proof of comparative efficacy.[3]
Check the medicine and exposure definition
Ask how researchers determined that someone started treatment, continued it or stopped. A prescription order, a pharmacy fill and confirmed administration are different kinds of evidence. Also check the indication and formulation. A study of approved injectable products cannot establish results for an untested preparation just because a marketing page uses the same ingredient name. Related reading: product-status distinctions.
Use the evidence to ask practical care questions
Real-world studies can prompt useful questions about affordability, refill continuity and follow-up. They do not provide an individual guarantee of weight change. Ask the clinician how progress will be assessed and what happens if access is interrupted. A provider’s payment and cancellation terms may influence continuity, but a commercial promise alone cannot determine the medical outcome.
A comparison checklist for the next headline
Write down the population, comparator, exposure definition, follow-up period, missing-data approach and whether the result includes discontinuation. Then identify the actual question being answered. This makes it easier to compare a trial with a practice-based study without forcing them into one ranking. Evidence becomes more useful when differences are explained rather than hidden behind a single percentage. Related reading: how to read GLP-1 research.
Continue your research
Sources & scope
Sources checked October 13, 2026. This guide explains published information and consumer research questions; it is not a clinical evaluation.
- Clinical-practice weight change by discontinuation status
- Weight trajectories after the last documented prescription
- SHAPE real-world weight-loss study
Prices and terms can change. Confirm your specific offer directly with the provider.