Friday, August 27, 2010
You had me at 'improves HRQOL'...
(Editor's note: I discovered last night that Blogger doesn't allow really long comments, so I have taken the liberty of posting my comments about Lyle's post on the NEJM early palliative care trial and the subsequent comment discussion as a full post.)
Well it's been one of those weeks-after-vacation in which it takes you an entire week of madness just to catch up with what you didn't do the week you were gone. Along those lines, I finally read the full paper today, and had the following reflections:
- I think the core, talking-point message from the paper is that early palliative care specialist consultation for 'high risk' cancer patients helps our patients: it improves their symptoms and quality of life. These were consistent and moderate effects across the board for this intervention. When it comes to HRQOL improvement for incurable cancer - moderate improvements are a very very positive and gratifying outcome.
- Along these lines, the big 'selling point' of this to our friends in oncology who don't particularly object to us but aren't sure what we add is that we are needed/helpful, we improve their patients' lives which is our mutual goal.
- I think the real limitation of this study, as others have mentioned as do the authors, is one of generalizability - one center, one patient population, pre-screened (in essence) by their oncologists, and it's unclear what the pall care team did. That said, reading the appendices, what they outline sounds very much like routine palliative care assessment/interventions, nothing fancy: whole person symptoms, disease understanding, coping, communication preference, goals. I don't know how you pre-specify further what it is we do in real life.
- Survival: I disagree with those who suggest that the survival finding was some sort of fluke - there's nothing in the study design to suggest this is an error *within the constraints of this protocol*. The fact that it's a secondary outcome doesn't mean it's not a real phenonomenon: it's a basically unbiasable outcome - you're dead or not - and the effect here is huge - 3 months median. Not subtle at all for this patient population. Yes, there could be a flaw in the study design, or this is a statistical mishap, but I don't see anything in particular which would lead me to that conclusion. If this was a RCT of a hypertension drug in which a drop in mean SBP greater than 10mmHg at 3mo was the primary outcome, and a secondary outcome was all-cause mortality at 3 months, and it so happened that there 30% more drug patients died than placebo, you wouldn't dismiss it. In addition, if median survival in the palliative care group was 3 months shorter than non-PC I hope people wouldn't dismiss it outright, b/c they shouldn't. It'd be 'bad press' I guess, but if you believe that patients should have expert asssistance in maintaining 'quality' of life regardless 'quantity' then you should be comfortable with that quantity being shorter. Cf. the recent study looking at abx for pneumonia in advanced dementia that we blogged about (to be sure this was observational not a trial, I'm just using this to illustrate my point) in which those without antibiotics had no worse/and maybe better symptoms and died sooner: that should be ok with all of us. Anyway, I seriously doubt for cancer there'll ever be an association for palliative care specialist interventions with increased mortality for a lot of reasons (patients, generally, want to live and want treatment; anticancer treatments for those near the end are generally ineffective and not life prolonging anyway so even if palliative group patients didn't get them they'd still live as long, etc.). Anyway, for this study, the big question in my mind about survival has to do with the fact that 30% of patients were still alive at the time of censuring - enough to substantially alter the survival outcomes at, say, 5 years. That concern doesn't invalidate the findings within the scope of their protocol.
- All of this is to say that I think this study really does raise the question of whether certain palliative interventions in certain patient populations really do help people live longer. Why not, after all? Is there any reason to think expert symptom control, anticipatory guidance/planning, etc. are any more likely to alter survival than, say Tarceva for NSCLCA patients with ECOGs of 3? Nonetheless touting improved survival outside of the specifics of this study are - generalizing it - to be sure, premature.
- One of the probably starry-eyed ideas going through my head after reading this is that the real impact of this study should be to, as others have argued in the past (I can't remember who exactly at the moment, maybe Abernathy and Currow et al), do away with the nebulous 'routine care' or 'routine supportive care' as a control intervention for cancer therapy trials. It should be replaced with, instead, specialist palliative care. I know that's practically impossible - there's not enough people to do this etc. - but I think this study, if it does nothing else, should be the beginning of the end of that practice. Routine supportive care, whatever that means, is not a standard.
- The media will be the media, but I agree with those who think some in the HPM community mis-fired in playing up the survival benefit (the authors of the paper did not play it up). This is not because the benefit, within the confines of this study, was bogus, but because as others have said the focus on survival shifts focus from our core values and creates unrealistic expectations.
- I don't think all the heat generated however is really warranted, and part of me felt like 'Gee, so it's come to this, a skirmish about whether we should be touting or cringing from findings showing we prolong life.' Part of the lack of sobriety in reacting to the survival results, I wonder, is from a tacit recognition of how young we are as a field, how misunderstood many of us feel sometimes, and how sick we are of people outside our field assuming things about us that aren't true (we are the death doctors, we casually hasten death, we don't support our patients' will to live despite the odds, etc.) - such that there is a little triumphant 'See! We don't kill people!' aspect to this....
- That aside - the cult of mortality (as a meaningful outcome divorced from other considerations, particularly in patients with chronic, life-limiting, morbid conditions) is pernicious to medicine, research, society, and our patients. And, I think as a community we don't want to lose site of the larger war, just to celebrate this single battle. *Forgive the martial metaphor - I can't come up with anything more appropriate at the moment.
- What we do improves the lives of our patients and their families, it helps them get the medical care that they actually want and helps them meet their goals. If it improves mortality, great; if not, no problem because adding a few months of life is not what all our patients want given the consequences.
- One doesn't want to be too naive, however. "You had me at 'palliative care improves HRQOL'..." is what you hope oncologists, other referrers, and payers will say, but I know data about mortality and cost savings are taken seriously by many of our key constituents, so it should be discussed, and goddammit we should be proud of what we do, and this study gives us good reason to be, and I cannot imagine how long term this trial will be nothing other than a boon to our field and to our patients (by improving access to our services).
I hope David Weissman will forgive me for paraphrasing his response to the discussion about this trial: "If I'm a life-prolonging intervention, does that mean I'll have to discontinue myself when a patient is ready to die?" Funniest thing I've heard all week.