Studies are only as goods as their methods and the integrity of their basic premise. Here, I think the basic premise is suspect. Inherent in such a study is the assumption that there are sufficient similarities between all cancers and all chemo protocols to make valid conclusions possible. IMHO, this is incorrect for numerous reasons.

Only certain cancers are treated curatively with chemo only (i.e. lymphoma, leukemia, etc). Many others (including OC) are treated primarily with surgery/radiation, with chemo used as adjunctive therapy to facilitate radiation and eliminate "stray" cancer cells. For these cancers, a much higher percentage of chemo only patients are being treated palliatively, so you would expect decreased survival.

I agree with you that it is infuriating when studies with marginal premises are then used to support improper inferences.