"1. ICU, kidney transplant, chemo, the works
I'm guessing the prognosis here is actually a pretty hasty and expensive demise. Is she even stable enough for a kidney transplant? Might the drugs given to stave off rejection accelerate her cancer? Would a new kidney survive chemo? I wouldn't seriously consider this option, but I'm just wondering on how many levels it is a bad idea."
This isn't even a consideration, as you outline. Transplantation takes a while to set up; in a sudden liver failure we can rush it and save a rare person who's otherwise well, but generally it takes at least weeks to months. There's a need for psychiatric and infectious disease reviews of what could go wrong on the meds, there's multiple tests, preoperative "clearance," and the person has to be able to tolerate a major surgery. This individual's kidneys are only failing because she's dying, and surgery would kill her, the organ wouldn't likely work if she didn't die, and if everything went great infection would ensue rapidly and she'd die of her cancer shortly anyway. This comes up only because in the USA we talk about doing everything and so families want everything as a way of showing how much they care. It makes having discussion about realistic options that more difficult because people start off with not unrealistic expectations, but wholly impossible ones.
Oh the chemo. Forgot the chemo. Chemo is poison. It's hard on you (some more than others; there are some oral options that work a little, and there are special circumstances with drugs that are growth inhibitors with relatively few side effects). People are not candidates unless they are generally well, and then, you have to wonder what the result will be. For example, in the bad cancers, like pancreatic or metastatic lung, you're looking at prolongation of life by a few weeks or a month or two. In some situations people with certain types of lymphoma or even metastatic breast, colon, or pancreatic cancer, therapies can keep people alive and functional for months to even many years. Chemo can cure acute leukemia (in kids, it usually does, whereas a few decades ago it was 100% fatal) and new drugs can keep chronic leukemia at bay for who knows how long. It's a bit like the progress against HIV/AIDS. These are wonderful therapies and we are beginning to see the era of living with cancer rather than dying of it. I am in awe of what the oncologists can sometimes do. This patient has sequentially failed sequentially less effective therapies; there is nothing she can take for this disease, even if she were entirely well apart from the cancer when it started. Again, a complete disconnect with what the family wants--and why they have no idea what's possible after several failed chemo cycles and years of illness I have no idea.
"2. ICU, dialysis
With an almost but not quite non-functional liver, she could last weeks, months? Expensive, not a lot of fun."
We're notoriously bad at prognosis, and I think it would be fair to say she might be closer to "days to a week or two" than "hours" because she responded to pressors before, but the body is rapidly failing, and nothing can change that. This is the American death: huge costs (financial, as well as born by the patient) for a few more hours or days of life in the ICU grievously ill (often at advanced age). Outside of the hospital, most of us ignore our health, and millions can't get basic, cost effective, preventive care. It did take internship to teach me that there are things worse than death, and that prolonging life is a laudable goal--but prolonging dying is often a disservice to the patient.
"3. offer ICU, but no dialysis
Well, without dialysis, what's the point?"
Maybe the other care helps the kidneys recover? I agree, not something I would recommend, but one could reason, "maybe if we fix some other intervening problem she could go back to stability for a while." This would be the key potential indication for the first ICU stay she had.
"4. offer dialysis, but no ICU
With dialysis and no ICU, how long is she likely to last?"
Again, largely out of our hands. We support a failing organ but don't fix the underlying problems, and we're still looking at a few days, if we're lucky then she could live a week or two at most.
"5. offer maximal conservative care on the floor
I don't know what this means."
Excellent question... this would mean, let's offer noninvasive care--fluids, nutrition, antibiotics, things like that. But let's NOT put a dialysis catheter in her neck (this is no IV line; it's more like a sharpie or other marker in thickness, and she's not well enough to sedate--scary plus pain). I don't think this is unreasonable. A lot of people would go for this--it's a way to do "everything" that's not too burdensome. The thing is though, it's mostly a show. It's like putting out a 3 alarm fire with a garden house. It may serve as a demonstration of care, but rather than recommend something I know won't work, I usually advise we show our care in other ways. For example, most people don't say they want to die getting maximal conservative care in the hospital. Most people say they'd like to die at peace, surrounded by family, without a lot of medical fuss--so why not give them that (or whatever it is, including being at a hospital, but with appropriate resource use)? Further, some of this care may be harmful. Fluids seem humane but often increase secretions and can lead to gagging. Food seems humane but can cause distention and cramps and pain when a failing body is fed food it no longer needs or wants (there's a reason we lose our hunger). Care that causes harm for the purpose of showing care to me speaks of 1) a lack of preparation 2) a lack of familiarity with the risks and benefits 3) difficulty facing the issue directly.
A family from another culture crystallized the issue nicely for me one time when they requested fluids, nutrition, and antibiotics for a dying matriarch. I explained that these things would hurt and not help and why, and they said quite simply that she would be dead soon and their duty was to help and that means doing something and if that something was harmful, no matter, because they cared more about feeling that THEY tried everything rather than what her outcome was. This was another informative story because I will NOT deliver care I know is riskier than helpful, and yet, I managed to meet their requests. In this situation let's just say a soft block worked far better than a hard block.
addendum: astute readers might be wondering what the difference is between maximal conservative care and fluids and antibiotics might be. These are all artificial distinctions I whipped up, as even with chemo and dialysis the dose can be varied. But two things might be that 1) sometimes medicines necessary for pain can lower blood pressure and respiratory effort and that can hasten death (which is never the goal, but in dying people it does happen, and is viewed as acceptable by every ethicist and religion I know of), so one could withhold pain medication to prolong life and 2) some "ICU" drugs to support blood pressure (pressors) can be given in low dose on the floor, in some situations, so if someone wanted to try some dopamine to "do something" without addressing the underlying problems, that would be an option.
I hope this is illustrating the futility in spelling out all your wishes on paper. It's fine to try, but there is no way to capture all potential situations and options. A statement of general goals and specific examples is a good start, as is a next of kin who cares (but not too much).