We got the milnacipran result backwards — here’s the fix
A small confession, because it matters more than the fact itself.
We had written that milnacipran — an SNRI used in fibromyalgia — “did not reduce ventricular lactate,” and used that to argue the drug’s benefit must be non-metabolic. Then we actually read the abstract.
The trial found the opposite. Milnacipran lowered ventricular lactate compared with both baseline and placebo (F1,18 = 8.18, P = .01), and patients who improved in pain were more likely to also drop in lactate (Natelson et al. (2015)).
Why this happened. Early in a long project, a summary was inferred from a citation’s title rather than the abstract. That is exactly the failure mode a rigorous pipeline is meant to catch — and in this case it took a deliberate re-read to catch it.
Why we’re telling you. Not for credit. Because the correction changes the argument: the drug’s effect may have a metabolic component after all — a different claim from the one we first made. (Being careful about direction: the lactate drop could also follow improvement rather than drive it.)
The lesson we apply. Every “did not” in our text gets checked against the source abstract, not the title. A claim that a trial failed is a claim, and it deserves the same scrutiny as one that says a trial worked.
This is what a patient-commissioned document can do that a press release will not: publish its own corrections, and keep the trail visible.