NAION is a sudden, usually painless loss of vision in one eye, caused by interrupted blood supply to the optic nerve, and it is usually permanent. That is a serious enough outcome that the numbers below deserve reading in full rather than as a headline — starting with which population each one describes, which is the distinction most figures about these drugs lose on the way to being quoted.
Where a weight-loss reader sits
The pooled figure across all eight studies is a relative effect of 1.93, 95% CI 1.22 to 3.08. [1] Split by population, the association held among people with diabetes — 1.84, 95% CI 1.21 to 2.80 — and did not reach significance among overweight or obese people: 1.68, 95% CI 0.72 to 3.91, P = .23.
Almost everybody reading a price page for these drugs is in that third row. So the honest summary for this audience is that the association is not established in people like them — which is the opposite of what a headline drawn from the pooled figure would say.
The studies disagree with each other
Heterogeneity was 91% overall and above 83% in both subgroups. That is close to the ceiling: it means the eight studies are not estimating one common number, they are producing genuinely different answers.
A random-effects pooled figure in that situation is an average of results that conflict, and it may be a number no individual study found — the same problem as an average nobody in the trials experienced, and a reason to treat 1.93 as a summary of disagreement rather than as a measurement.
The number that is missing
How many people this would actually affect. NAION is rare in the general population, so even a doubled relative risk can amount to a small number of extra cases.
This abstract publishes no baseline rate and no absolute difference, so the arithmetic cannot be done here, and this desk will not invent a denominator to do it with. That absence is the single most important limitation on the page, for the same reason a relative figure alone settles nothing.
What this supports
Every one of the eight studies is observational, and the authors describe the finding as a possible association and ask for randomized trials.
If you have diabetes, this is the subgroup where the signal held, and it is worth raising with whoever prescribes for you — particularly alongside any existing eye disease. If you are taking it for weight without diabetes, the honest position is an open question rather than a finding, which is a different thing from a settled real-world result.