Separate trials do not automatically make a league table

Two percentages can look comparable long before the trials behind them are comparable. A weight-change result at one time point in a selected population is often placed beside a result from a different programme, then used to rank the medicines. The arithmetic may be correct while the implied contest is not. Differences in eligibility, baseline characteristics, accompanying care, follow-up and analysis can all shape the numbers in that table.

A direct randomised comparison offers a different kind of evidence because the alternatives are studied within one trial framework. SURMOUNT-5 compared tirzepatide and semaglutide over 72 weeks in adults with obesity without type 2 diabetes. It was an open-label trial, an important design detail to retain alongside its results. Its existence improves the evidence for that specific comparison; it does not create a universal ranking across all conditions, outcomes or individuals.

For this November briefing, the aim is to make comparisons more useful by making their basis explicit. A reader should know whether a statement comes from a head-to-head randomised trial, an observational cohort or a casual juxtaposition of separate publications. Those approaches answer different questions and carry different limitations. Before asking which compound wins, ask which comparison the available evidence can actually support and whether that is the question you intended to ask. An honest comparison states its basis before presenting its strongest number, so readers can judge the match.