The Tirzepatide Trial Library

Every trial, graded on the same rubric

An evidence grade here measures how much weight a result can carry, not whether the result was favourable. A large event-driven outcome trial outranks a small open-label comparison even when the smaller trial reports a more dramatic number. Every input is visible, so you can disagree with the weighting and recompute it.

Two things this grade deliberately does not do

It does not reward a favourable result, and it does not penalise industry funding beyond a single five-point independence criterion. Almost every trial in this field is manufacturer-funded; treating that as disqualifying would leave nothing to read. What matters is that sponsorship is disclosed on every page, along with the sponsor's role in design, analysis and drafting.

It also cannot grade what was never published. Selective reporting is invisible to a rubric applied to published reports, which is why prospective registration carries points and why every trial here links to its own registry record.

Estimands, in one paragraph

Modern trials report the same result under more than one estimand — a precise statement of the question being answered. The treatment-regardless-of-adherence estimand (often called the treatment policy estimand) asks what happened to everyone randomised, including people who stopped the drug early or started another one. The efficacy estimand asks what would have happened if everyone had stayed on treatment as assigned. The efficacy estimand almost always produces a larger number, because it removes the people for whom the drug did not work well enough to keep taking. Marketing tends to quote the efficacy figure; a patient deciding whether to start should generally weigh the treatment-regardless figure, because it includes the possibility of being someone who stops.