We Know What They Do, Not What They're Worth

The American College of Physicians commissioned two evidence reviews to build a guideline on these drugs. The one on whether they work pooled 69 studies and 112,511 participants and reached clear conclusions.^[1] The one on whether they are worth the money found nine studies, none of its 42 comparisons at high certainty, and concluded that conclusions cannot be drawn.^[2]

What the efficacy review concluded

This is a living systematic review, meaning it is designed to be updated continuously as trials report rather than fixed at publication. It covered thirteen drugs including the newest — orforglipron, retatrutide, cagrilintide–semaglutide — through October 2025.

• Nearly all treatments beat placebo or lifestyle alone for weight loss, and nearly all produced more discontinuation due to adverse events.
• Semaglutide probably reduced mortality and major cardiovascular events.
• Semaglutide and tirzepatide produced the greatest weight loss in both pairwise and network comparisons.
• Evidence on mortality, cardiovascular events and serious adverse events was limited, and direct head-to-head comparisons were scarce.

That corroborates a separate synthesis of 262 trials published in the BMJ, which reached the same two conclusions independently — that semaglutide carries the mortality signal and that the two leaders on weight are not the same as the leader on outcomes. Two guideline-grade reviews, different teams, same answer, in what 262 trials say together.

What the cost review found instead

The economic review applied one inclusion criterion worth pausing on: it accepted only non-industry-sponsored US cost-effectiveness evaluations. That is a deliberate methodological choice, and it tells you something about the literature it excluded.

None of the forty-two comparisons reached high certainty. Not one.

The reviewers’ conclusion is that the evidence is hampered by poor-quality studies, limiting the ability to draw conclusions. For a drug class whose defining practical problem is that people cannot afford to keep taking it, that is a remarkable gap.

The finding inside the six usable studies

Six studies reached moderate certainty, and what they say does not follow the efficacy ranking at all.

That is what cost-effectiveness analysis exists to surface, and it is exactly the comparison absent from almost every conversation about these drugs. It does not mean anyone should take phentermine–topiramate instead. It means the question “is this worth it” has a different answer from “does this work best,” and only six moderate-certainty studies have attempted the first.

Why the gap exists, and why it matters

Every one of the nine studies was model-based — projecting lifetime costs and outcomes from trial inputs rather than measuring what people actually spent and gained. Models require assumptions about how long people stay on treatment, what happens when they stop, and how much of the benefit persists. Every one of those assumptions is contested, and this register has covered why.

• Most people stop within a couple of years, and modeling lifetime treatment assumes something that mostly does not happen.
• Weight returns when they do, which determines whether earlier benefit is banked or lost — what happens when you stop.
• The cardiovascular benefit does not track weight loss, so a model that credits benefit to weight is modeling the wrong thing — as SELECT found.

Our own coverage of one manufacturer-adjacent cost-effectiveness model is in the cost-effectiveness case, and the assumption that undoes it is the same one.

None of which is a reason to distrust these drugs. It is a reason to notice that the question a patient most often asks — can I justify this — is the one the evidence base has answered least.