“Fault” is way too strong a word. I’d say “imprecise” is more what I’m thinking. What do I mean?
(I assume you’re a health care professional, so will use medical terminology)
When I see the term "hepatotoxicity with respect to a drug, I think of it as meaning either acute liver damage or a more insidious process leading to cirrhosis or fibrosis. In the former case, as might occur with almost any drug, the manifestations may be purely biochemical (eg. raised transaminases) or may be overt (jaundice, encephalopathy). Edema, if it even occurred in the acute setting, would be an irrelevant and nonspecific finding. Most importantly, I cannot conceive of edema being the sole manifestation of acute drug-induced hepatotoxicity.
In the latter, insidious-onset type of drug-induced liver damage, such as might occur with methotrexate, the manifestations are quite likely to be part of a constellation of signs (eg. variceal bleeding, encephalopathy, jaundice, ascites, etc.). Certainly there can be edema, but, once more, it is difficult for me to imagine a clinical scenario where the only manifestation of drug-induced cirrhosis is edema. So, to single out edema as a side effect of methotrexate, for example, is a real stretch and seems almost misleading.
Regarding anemia leading to edema, that would be even more of a stretch, or at least would require a number of intervening developments which would be expected to have their own, prominent, manifestations. It is true that a drug could cause anemia and that could lead to high output failure, and that heart failure leads to edema, ergo anemia can cause edema. But, again, a more straightforward way of listing the drug’s side effect in this regard would be to simply say “can cause anemia”. More importantly, if the anemia becomes severe enough to cause high output failure, I’d bet you’d know about it way before there was edema.
I should also add that anemia causing heart failure in an otherwise health person would be very rare.
I’ll close with an example. Let’s say a potential side effect of a drug is hyperglycemia (eg. prednisone, dilantin). I would certainly agree that listing polyuria and polydipsia is fair and helpful. However, listing retinopathy, or neuropathy, or any other diabetic complication, while theoretically true is misleading and imprecise more than anything else.
I 'm not sure that I’ve made myself as clear as I would have liked but hope you can see what I’m getting at.