It's just that it's an estimate that hinges on a few judgement calls (e.g. whether or not to cost services that AMF receive pro bono) — we're in agreement with Givewell's estimate. For comparison, the WHO think that LLINs work out to be around ~$30/DALY and AMF run a pretty tight ship.
Given the tricky nature of cost-effectiveness estimates I don't think the numbers are hard enough that it'd be wise to talk about the difference implying anything as exact as an intervention being '20% more efficient'. It's more to get a reasonable ballpark for comparisons between other interventions.
(Also, seems there was a typo in the OP, should be DALY not QALY).
Given the tricky nature of cost-effectiveness estimates I don't think the numbers are hard enough that it'd be wise to talk about the difference implying anything as exact as an intervention being '20% more efficient'. It's more to get a reasonable ballpark for comparisons between other interventions.
(Also, seems there was a typo in the OP, should be DALY not QALY).