Almost every trial on this site compares a drug against placebo. That answers whether the drug does something, and leaves unanswered the question a patient and a doctor actually face: is this better than what we would otherwise have done. SURPASS-EARLY asked the second question [1]. It is the design this site asked for in what happens in ordinary care.
The comparator arm received intensified conventional care as practiced locally, guided by local treatment guidelines, and permitted to include other GLP-1 receptor agonists. Only tirzepatide itself was excluded from that arm. So this is a drug against real-world practice, including real-world practice that reaches for a drug in the same family — the comparison this site could only approximate indirectly in two meta-analyses built on the same six trials.
Over two years, HbA1c fell 1.99 percentage points on tirzepatide against 1.32 on conventional care, an estimated treatment difference of 0.68 points. Weight differed by 8.0 kg and waist circumference by 6.2 cm, both favoring tirzepatide. The share reaching an HbA1c below 5.7% was 60.2% against 24.0%.
The trial was open-label, which the authors list as their limitation. Everyone knew which arm they were in. In a trial whose outcomes are a lab value and a scale reading, that matters less than it would for a symptom score, but it still colors adherence and intensification decisions on both sides. Gastrointestinal events were the most common adverse events in both groups, which is expected when the comparator arm also contains GLP-1 drugs.
What makes this worth reading is the comparator, not the effect size. A 0.68-point HbA1c advantage over real practice is a smaller claim than the arm numbers this site sees in placebo-controlled trials, and a much more useful one. It also sits alongside the finding in where metformin lands when ranked: the older drug is not nothing, and the honest question is always the margin, not the headline.