Saturday, June 30, 2018
Let's Stop Claiming That Palliative Care Improves Survival
by Drew Rosielle
Let's just stop doing this.
There has never been any actual evidence that palliative care (PC) interventions improve survival in patients, but since the landmark Temel NEJM 2010 RCT of early outpatient palliative care for lung cancer patients showed a clinically and statistically significant improvement in longevity in the PC arm, I have heard and all read all sorts of statements by palliative people and all sorts of others (hospital executives, policy makers, oncologists) in all sorts of venues (local talks, national talks, webinars, newspapers, etc) along the lines of 'palliative care helps patients with cancer live longer.' I've even heard the results discussed as evidence that hospice helps cancer patients live longer.
We should have never done this, and if you're still doing it, please stop.
To begin with, 'palliative care' isn't a single thing. It's not like studying enoxaparin, or nivolumab, or olanzapine, where you can to a reasonable extent assume that if it helps patients in Lille, France or Boston, MA, it will likely help your similar patients in your practice wherever. Palliative care is just not like that - it is complex, and local conditions are very important, and it is impossible to make broad generalizations about PC in general from a single study at a single institution. Some PC research involves full interdisciplinary teams doing their thing; some are one or two disciplines; some involve telemedicine, care coordination, etc. You just can't generalize these sorts of interventions to 'PC' in general because it can mean so many different things.
Plus, the Temel study was only a single disease, and their nicely done follow-up study which broadened the patient population presumably failed to show any survival difference (because they haven't published that result that I'm aware of).
There is an exceedingly thin quantity of additional PC research showing improved survival. Most importantly I think is Bakitas' ENABLE III study which, notably, did not compare PC to usual care but early PC to PC 3 months later in cancer patients and showed prolonged survival. Curiously, it didn't show any difference between groups in any other outcome (not in QOL, symptoms, intensity of EOL care, chemo in the last 14 days of life, hospice enrollment)! There is this secondary analysis of the study arguing that the survival benefit was mediated by depression, which sure maybe, except that the actual stable itself didn't show any change in mood between groups. So, it's messy, right, and at the end of the day one is left thinking that the survival improvement is curious, and you don't really know what to make of it (and not left to strongly endorse broad, unqualified claims that PC makes people live longer).
There is also this somewhat famous study purporting to show that hospice extends survival in CHF and cancer patients. Besides being industry funded (NHPCO) and using data that are nearly 20 years old now, it uses a statistically complex, opaque, retrospective design using Medicare data, that is really difficult to understand for us non-biostaticians. It's a tough question to study after all -- without randomization which would be ethically and practically challenging, how do you compare survival in patients without a clear time zero. I.e., if time zero is when someone enrolls in hospice, what is the comparable time zero for someone who never enrolls in hospice? How do you capture them at the time when they are 'equally sick' to someone who happens to enroll in hospice. You can't, thus the statistically complex study design. I'm not criticizing the study, but I am very much arguing that it's not the sort of research we translate into broad claims that 'people live longer on hospice.'
The vast majority of studies of PC intervention either don't report survival, or if they do have a neutral affect on it. Although it's notable that while everyone knows about the Temel 2010 study, hardly anyone mentions the really well done randomized, controlled trial of home-based palliative care which showed all these great outcomes (patient satisfaction, health care utilization), but also, oops, showed shorter survival in the palliative care intervention group. (You can enjoy this cranky letter to the editor about this written by Andy Billings and Craig Blinderman, as well as the authors' response, here. I really miss Andy's constant, erudite crankiness.)
Given how heterogeneous PC is, what would constitute adequate evidence that 'PC' actually prolongs survival? I think it would be one of two things, neither of which exists.
One would be a large number of single institutional studies of PC interventions which show prolonged survival (and, concomitantly, an absence of numerous single institution studies showing the opposite). How many? I don't know, but a lot more than we have now. Think about how many studies we have of PC interventions, from all over the world, in different patient populations, inpatient and outpatient, including both trials and observational research, which show improvements in some patient-centered outcome like quality of life. A lot, so many in fact that it's notable when one doesn't show such an outcome. We don't have anything close to this for survival outcomes.
In fact, someone went ahead and did a very nice meta-analysis of all these (mostly single institution) palliative trials, which shows just how great we are at improving all sorts of awesome outcomes, just not survival.
The second would be a large, multi-institutional, multi-regional study of some sort of PC intervention showing improved survival. That doesn't exist. We do have it for resource utilization - such as this multi-regional study by Sean Morrison published in 2008. Additionally, resource utilization outcomes are like QOL outcomes - they are a dime a dozen (we have dozens of studies of PC interventions, of many different shapes and sizes, from all over the world, showing differences in resource utilization).
At the end of the day I of course hope PC interventions prolong survival, but I'd be pretty surprised if that panned out. Most of our patients want to live longer with a reasonable quality of life, and I think we should be satisfied with the 'helping them have better QOL part.' We should react with curiosity about any new single-institutional study which shows improved survival, just as we should with similar studies that show decreased survival. Not anything to celebrate or bemoan.
Thing is, we have a lot to be proud of. A lot that we do well, and that we have all sorts of evidence supporting us in. Without qualification, we can go around saying Palliative care improves the quality of life of patients with serious illness. Seems good enough to me.
Drew Rosielle, MD is a palliative care physician at the University of Minnesota Health in Minnesota. He founded Pallimed in 2005. You can occasionally find him on Twitter at @drosielle. For more Pallimed posts by Drew click here.