Friday, July 14, 2017

Palliative Care & CHF: PAL-HF trial

The main results of PAL-HF - a randomized, controlled trial of specialty palliative care team involvement in advanced heart failure patients -  have just been published in the Journal of the American College of Cardiology (DOI: 10.1016/j.jacc.2017.05.030. registration here). 

This is an important, well-done study, with encouraging results - specialty PC improved the quality of life of patients with HF. I'll discuss the results in more details in this post.

The study was done by a multi-disciplinary team of palliative & cardiology investigators at Duke. This week's publication looks at QOL results which were the main, pre-specified outcomes. Of note, in the registration for the study, they do pre-specify healthcare resource utilization outcomes as one of their secondary outcomes, but this paper doesn't  present those data - presumably those will take them longer to collect and will be forthcoming. 

PAL-HF enrolled 150 hospitalized patients with HF at high risk of mortality or rehospitalization, and randomized the patients to receiving specialty palliative care vs usual care. They identified patients using the ESCAPE risk model, which I hadn't heard of - key reference is here. The inclusion criteria involved the patients having an ESCAPE model predicted 6 month of mortality - notably in this study the actual 6 month mortality was 30%.

Intervention-arm patients received a palliative care visit, I think by a palliative NP - the methods section is a little cagey about who exactly saw the patient apart from a palliative NP, if anyone. The NP did a comprehensive palliative evaluation (physical, psychoemotional, spiritual evaluation), had a goals of care discussion, did advance care planning, and presumably made recs about what to do. The patients were followed for 6 months, which is the length of the data collected for this study. They methods say PC remained involved in the patients' care, although the exact nature of that involvement is opaque to me - eg it's not clear the patients actually saw PC in clinic or anything, and the role of PC may have been advisory to the cardiology team. The full methods of the trial were published in a different paper but even reading that it's not entirely clear to me who from the palliative team saw the patient in addition to the NP, nor the real nature of the 6 month planned follow-up. 
The primary outcomes were QOL on the Kansas City Cardiomyopathy Questionnaire and the FACIT-Pal QOL scales st 6 months. I need to say: 6 months, whoa. This is rare to see a study of this kind look at outcomes over such a long period, and it's one of the reasons I think PAL-HF will set a new standard in these sort of complex palliative-intervention trials. In the methods they note that a 5 point improvement in the KCCC and a 10 point improvement in the FACIT-Pal are believed to be clinically relevant outcomes. Their power estimate showed they needed 200 patients to show such a difference - they ended up enrolling only a 150 (but still showed a difference). It's not totally clear to me why they capped enrollment at 150 – they mention survival was better than anticipated. Nonetheless they met their primary outcome with less than the estimated number of patients (which is an argument for the robustness of their findings). Key here is that unlike a lot of  lower quality research they designed their outcomes to explicitly be patient-relevant, and powered their study as such. 

Patients had a mean age of about 70 years, ~50% were women, and ~40% Black. 

The outcomes at 6 months were good: improvement of 9-10 on the KCCC and 11 on the FACIT-Pal in the intervention group compared to the usual care group. These exceeded the thresholds for what is considered clinically relevant. Some secondary outcomes were improved too: depression and anxiety symptoms, and spiritual well-being (as measured using FACIT). 

As above, I am assuming the resource utilization outcomes are forthcoming. They do mention that survival and rehospitalization rates were similar between the groups (30% for each outcome at 6 months). 

My summary thoughts:

This is a well-designed and executed study - the sort of thing I read and say Gosh we need a lot more of these. Eg, fewer retrospective chart reviews and case control studies, more well-thought out, well-designed, prospective studies testing if and how palliative care improves our patients' lives. Besides wishing the methods write-up was a little clearer about the exact nature of the palliative intervention, I am few complaints about the paper. 

The limits of the study are intrinsic to its methods - and I want to talk about one of the major challenges we face in our research base, which is the challenge of generalization. Unlike a drug trial, team-based palliative care interventions are inherently complex, and presumably very sensitive to very local factors. Eg, what the good folks at the Duke palliative care program do may be somewhat different than what my teams do, and any other team in the country. It's well documented that while palliative care has become ubiquitous in larger American hospitals, that does not mean every program is populated by well-trained, competent, interprofessional teams. In fact, we know that many are not. What this means is that we know what the Duke team did really did improve these patients' quality of life. Generalizing, exporting, what they did to other programs is difficult. Ideally the next steps in research like this would be to do a similar study that is national, involving many regions and types of hospital populations (not just academic referral centers). This is not to criticize the PAL-HF trial, it's freaking great, but more to acknowledge that we can't just all go around claiming Palliative Care Improves QOL for HF Patients, Full Stop. As if palliative care is one simple thing, simple intervention, the same everywhere, etc. This is in contradistinction to drug trials. Generalization is a big issue with drug trials too, but it's mostly an issue of extrapolating results to different patient populations (eg community patients to academic center patients, etc). But it's not really a matter of thinking that there's something importantly different about, say, enoxaparin, administered in Loma Linda vs Durham NC. There may very well be important differences in palliative care between Loma Linda and Durham, however. 

I hope those sorts of multi-center trials become more common.

At the end of the day, this is a landmark study. I am really, really glad it was published in a heart journal. We have over a decade of decent research showing the improvements palliative care provides in cancer patients' QOL, but far less in other patient populations, including HF. PAL-HF is a big step towards making the belief that PC benefits patients with HF less up to debate.  

Pallimed | Blogger Template adapted from Mash2 by Bloggermint