Commentary

Article

Dr Ravi Parikh Presents Groundbreaking BE-a-PAL Data on Default Palliative Care Referrals

The BE-a-PAL study investigated potential of an algorithm-based default palliative care referral among patients who have stage III or IV lung or noncolorectal gastrointestinal cancer.

Today, at the 2024 American Society of Clinical Oncology annual meeting, Ravi Parikh, MD, MPP, presented the final results of the BE-a-PAL study. This study investigated the potential of an algorithm-based default palliative care referral among patients who have stage III or IV lung or noncolorectal gastrointestinal cancer,1 building on his prior work of using electronic nudges to increase referrals to palliative care.

These newest findings show physicians allowed referrals in 89% of the cases in the intervention arm, which encompassed weekly default electronic health record notifications to oncologists as prompts for specialty palliative care referrals in high-risk patients.2

Parikh is assistant professor of medicine, medical ethics and health policy, and director, Human-Algorithm Collaboration Lab at the Perelman School of Medicine. Here he speaks to the data he presented in the abstract, “BE-a-PAL: a cluster-randomized trial of algorithm-based default palliative care referral among patients with advanced cancer.”

Transcript

What is the main takeaway from your default approach to palliative care referral?

So currently, in standard practice, we use more of an opt-in approach to palliative care referral, meaning that if I, a clinician, feel like a palliative care referral is warranted for a patient, then I, usually, or my nurse practitioner, will place that referral. That's what gets patients into the palliative care field. And we know that when that happens, that has big benefits for our patients. It generally gets them better, more specialized expertise in their symptom management, and it oftentimes eases the transition and their goals of care when we have support from specialty palliative care–trained practitioners.

The challenge is that we don't do it often enough. Less than half of people have a palliative care counsel with cancer prior to death, despite the fact that it's an evidence-based practice recommended for anyone with cancer regardless of stage sometimes. And so in an effort to increase uptake of an evidence-based practice, default referrals that don't require a physician to opt in or actually place a referral are a promising strategy. So, with that in mind, I would say that there are 3 big takeaways from this study.

First, default palliative care works; it increases rates of completed palliative care visits by nearly 4-fold in a busy community oncology setting consisting of 15 practices, where palliative care is resource-constrained. It's not like we staffed up palliative care for this trial. So, that is a promising sign of real-world effectiveness. The second thing that I think is a big takeaway is that default palliative care doesn't just increase rates of completed palliative care visits, which is more of a process outcome. It actually had meaningful impacts on, most notably, receipt of chemotherapy near the end of life, which is a bad low-value-care practice. That decreased by over 2-fold in our intervention computer. And it also had some promising signals that didn't reach statistical significance in increasing hospice utilization.

And then the third thing that I would say—and this is more from a practicality standpoint—is that in both quantitative and qualitative data that we've gotten from this trial, you might think that a default referral would risk overwhelming community oncology palliative care providers, and that did not happen. Largely, it was because we only use default referrals for an upper segment of the risk distribution, a sort of high-risk population of patients with cancer that was identified by an automated electronic health record predictive algorithm—that's near what we've tested in the past similarly for other things like advanced care planning. But in this setting, where you might envision a whole flush of new referrals could risk overwhelming the system, it really didn't because we use more of a risk-targeted approach for that.

So those, I think, are the 3 big takeaways that are most relevant for practicing oncologists and palliative care providers.

When physicians opted out, can you discuss the reasons they gave for doing so?

So across all of our trials with defaults, whether those be default, sort of prompts for advanced care planning, default referrals for certain care management interventions, or default referrals for palliative care in this setting, we've always found that there is a relatively large group of accepting physicians and then a relatively small group of physicians or nurse practitioners who just opt out of everything because they want to preserve their own autonomy in deciding whether a patient gets here or not. And that's okay. We know that these care delivery interventions aren't accepted by everybody. But the fact that they are accepted by most is a strong signal that 1, I think our algorithm generally got it right most of the time because physicians were largely accepting, and then 2, it suggests that oftentimes physicians just need a little bit of behavioral nudge, a push, toward doing the right thing, rather than relying on them to refer into what should be an evidence-based practice.

You brought up what are the reasons. We actually tracked it because we used a concept called accountable justification to track reasons why clinicians did not accept a certain palliative care referral. What we found was sort of interesting. We found that, at least when it came to opting out of the intervention, the most common reasons why physicians opted out wasn’t because they didn’t feel that patient's condition was severe enough to warrant a palliative care referral or that they felt that there were certain patient-specific barriers that the algorithm couldn't see—for example, known reluctance toward discussing advanced care planning and end-of-life care. The physicians wanted to avoid putting that in the hands of palliative care specialists by just opting out of the referral entirely.

Now, it should be noted that even if physicians opted in, patients could still refuse to seek palliative care because we offered patients the option, and, in fact, only in about 63% of the time did patients actually agree to a palliative care referral, even if a physician did not opt-out. And the reasons for that were a little more variable, but they included being overburdened with visits and the health care system and feeling that their own symptoms were relatively well controlled.

Did having a coordinator set up the referral increase the likelihood that physicians agreed to this process?

We've run a qualitative study that assesses what worked and what didn't work well. We interviewed close to 20, 25 oncologists and palliative care clinicians that participated in the trial. The number 1 thing that they felt was most effective was having the coordinator use standardized language to introduce palliative care. If you think about how physicians, oncologists, nurse practitioners introduce palliative care, I bet if you survey 20 of them, you'd get 20 different answers. Some of them palliative care as anchored to end-of-life care, whereas others use it as an adjunctive supportive care service, or that's how they describe it.

And so without using standardized language, then you can imagine that you'd have very wildly different rates of acceptance of palliative care, depending on how patients perceive the intervention. If they feel palliative care is only appropriate near the end of life, then maybe they'd be a little more reluctant to accept it than if they saw it as more of a supportive care service that could be used concurrently with systemic therapy—which is very much how we message language and which has been language advocated through both ASCO and the Center to Advance Palliative Care. Having that standardized risk language from a coordinator introducing it is a big motivator, I would think, maybe just as important as the algorithm or the default itself.

The other thing I'll say that's come up in some criticism of the work is, well, does every clinic have the staff to be able to do this and to have a dedicated staff member that's responsible for introducing palliative care. I'd have 2 responses there. One is that this wasn't the coordinator's only job; they had a bunch of other things that they were doing during the day than introducing palliative care. And so managing it into the workload is more of a matter of what clinic level priorities are than whether there's a dedicated staff member who's available. But then the second thing is that there's no reason why this has to be a research coordinator that does the introduction of palliative care. It could be a scheduling nurse, it could be a triage nurse, it could be any number of staff that are part of a routine clinic, as long as they're using standardized messaging to introduce the palliative care intervention. I think that is a key facilitator toward getting broader uptake rather than relying on submissions. We’re busy, time-strapped, and think of palliative care in different ways to be introducing the actual referral.

References

1. Predictive analytics and behavioral nudges to improve palliative care in advanced cancer. ClinicalTrials.gov. Updated February 1, 2024. Accessed June 2, 2024. https://clinicaltrials.gov/study/NCT05590962

2. Parikh RB, Ferrell WJ, Li Y. BE-a-PAL: a cluster-randomized trial of algorithm-based default palliative care referral among patients with advanced cancer. Presented at: ASCO 2024; May 31-June 4, 2024; Chicago, IL. Abstract 12002. https://meetings.asco.org/2024-asco-annual-meeting/15800?presentation=232443#232443

Related Videos
Rachel Dalthorp, MD
Andrew Cournoyer
Quint and Petris
Andrew Cournoyer
Rebecca Flynn, Peter Lio, Timothy Caulfield, and Nichole Halliburton at SPD 2024
timothy caulfield, JD
Nichole Halliburton, APRN, CNP
Related Content
CH LogoCenter for Biosimilars Logo