Likely the best controlled and accessible report on OA dribble is the thesis by Nicolas Aliano
http://digitalcommons.unl.edu/cgi/viewcontent.cgi?article=1000&context=entomologydiss
and the article he published on one aspect of his thesis work.
http://digitalcommons.unl.edu/cgi/viewcontent.cgi?article=1186&context=entomologyfacpub
Some points: when brood is present Aliano found a slight, but ultimately insignificant effect of OA dribble. (His protocol was weekly applications for three weeks). He did not detect cumulative toxicity, but he also didn't observe control of VM.
He was able to reproduce very effective Varroa control using dribble and/or spray on **broodless** fall colonies, and experimental "packages" of shook bees.
A key finding of Aliano is illustrated by this figure from his publication (modified by JWC).
OA:Sugar solution is toxic to mites, **and** toxic to bees. Control requires an effective toxic dose on the mites, while under-treating the bees.
Industry sources describe using a 3.5% solution dosing at a rate of 50 ml per 10 frame box, this is metered to 5 ml per frame gap to distribute the solution evenly. The Aliano figure shows that 5 ml (of 2.8%) has a highly effective impact on mites, but only lightly affects the bees. He calculated impacts at a "per 1000 bees" value. His experimental subjects were shook packages of 1,600 bees each.
In order to extrapolate in hive loadings, a realistic estimate of bees per hive is needed. The drawings on the Cushman memorial page are the best resource I know of:
http://www.dave-cushman.net/bee/beesest.html My own estimates based on photo's of my frames and sub-sample counting is a frame with "top cover" -- or bees crawling over the top bar -- has 1600-2000 bees per frame (2 sides counted). I don't have good estimates for cluster size counting. One can estimate hive population by counting top cover frames x 2000 (dispersed bees ignored). One can further estimate loading needed in a **broodless** dribble or spray by top cover/2 * 5 ml. This yields a max of 25 ml per deep, which is half the volume typically applied. Aliano used a painting airbrush with a pigment cup, and "airbrushed" his packages evenly. Perhaps his lower loadings illustrate at least half the hive dribble is immediately wasted. The industry practice of 50 ml per deep is likely within (but near) the practical limit reported by Aliano to avoid direct bee mortality.
As an aside, the mode of OA is poorly understood. OA is a simple 2 carbon chain --
when placed in solution (sugar, moisture, and injested with a H20 source) it immediately sheds H+ protons to become a conjugate base--
The base is termed Oxalate -- and its reaction potential is such that this transformation is rapid and complete. It is a massive H+ donor and as a highly reactive anion-- forms compounds with 2 valance metals such as Ca++
If the key mode of action is a rapid acidification of the Varroa hemolymph through absorbtion directly or injestion from the bee brood -- then consider that the those who promote ascorbic acid additions to feed solutions (and/or) highly acid honey development might be using analogues the OA effect. The impact of H+ in 5 liters of "acidified syrup" and 50 ml of OA might be similar, as OA is a 1000 times stronger acid (H+ donor) than ascorbic or acetic acid.
My irony meter pegs when I consider that "acidified syrup" advocates (scornful of treatment) and OA devotees (frustrated with TF zombies) are simply using the same acid impact treatment, with different donors -- OA is a simple natural acid, and ascorbic -- a more complex carbon ring.
Within a moist hive, OA will form Calcium Oxalate crystals (and similar insoluble salts). These crystals are very sharp, and might present a mechanical danger to the mites.