Why I Won't Stop Treating Patients at the End of Life

When I walked into the patient's room, I could tell that things weren't going well.

Ravi B. Parikh, MD, MPP

If there was anybody I would be nervous to treat with further therapy, it would be her.

This was during my first block of consults on the solid tumor oncology service. I'd been asked by the primary team to have an end-of-life conversation with an elderly woman with metastatic non–small cell lung cancer.

Despite her shortness of breath, weight loss, exhaustion, and cancer progression, it didn't take long for her to ask me what the next step was in her treatment.

It was the moment for which I'd received extensive training in palliative care throughout my fellowship. I began going through the list of questions and phrases I'd learned to convey my concerns about her receiving more treatment.

Feeling confident that I had explored her end-of-life wishes and communicated that more treatment may not be the best idea, I then asked her what her thoughts were.

"I'd like to give treatment one last shot," she said.

When I pointed out to her outpatient oncologist afterward that I was nervous about offering her more therapy, given her current symptoms and age, she said, "I agree with you, but if she is healthy enough to make it to her appointment, then I will consider treating her if she wants it."

A Shifting Mindset About End-of-Life Treatment

During my time as an internal medicine resident, I saw the sickest cancer patients on our solid tumor oncology service, many of whom were near the end of their cancer journey. I had to manage the acute complications of their disease or therapy—sometimes in the ICU with ventilators and chest compressions—while they were hospitalized. But of course, these were not my primary patients. I always spoke with the primary oncologist, and it often baffled me and the other residents when more systemic therapy was recommended—the very same therapy that often caused the patient to come in to the hospital.

Why would any doctor recommend therapy near the end of life? And why would any patient want it? As a resident, it was easy to get a jaded view of treatment near the end of life.

But today, as a faculty member in an outpatient setting, I see end-of-life care differently.

Patients vote with their feet: If they come to the appointment, then for the most part I can assume that they are at least interested in hearing about the benefits and risks (physical, psychosocial, and financial) of further cancer therapy. And, if after hearing about those risks they are hopeful about cancer treatment, then who am I to deny them that? Telling patients, "I worry that you might not live long enough to have a benefit from your therapy" is relatively ineffective at shifting their mindset. It also presumes a sort of omniscience and paternalism from the oncologist that usually doesn't go over well.

Why Does End-of-Life Treatment Happen?

A rash of metrics, including from the American Society of Clinical Oncology, the National Quality Forum, and the Centers for Medicare & Medicaid Services, have been designed to shift behavior near the end of life for patients with cancer.

I've begun to think that these metrics, while well intentioned, miss the mark. They presume that treatment near the end of life is preventable and always a bad thing. But there are many cases where end-of-life treatment is legitimate.

Prognosis. Cancer metrics related to end-of-life care presume that oncologists can reliably identify the end-of-life phase. However, studies have shown that even among patients with advanced cancer who are on hospice, oncologists accurately approximate a patient's life expectancy only about 30% of the time. Imagine how difficult it is to identify those with a greater than 6-month life expectancy in routine practice. Perhaps this is why, when asked about prognosis by their patients, many oncologists respond with wide ranges like "months to a year" or "days to weeks." It is a tall order—and one made even more difficult when patients desire more therapy irrespective of a doctor's prognosis.

Much of my research focuses on designing tools like machine learning to identify patients at risk for shorter-term mortality to make prognostication better. But some studies suggest that even if we had a perfect prognostication tool—one that identified every patient with a high probability of death—we would still not identify most patients who die in a given year. Why? Because death is inherently unpredictable, even among those with cancer. We should stop designing metrics that presume we know more about prognoses than we do.

Communication. Many physicians argue that most end-of-life treatment is a function of poor communication. If we just explored patients' wishes more, or were more honest about the side effects and likelihood of benefit, then maybe fewer patients would opt for end-of-life treatment.

I've often been brutally honest with patients—provided that they want it. If they wish to know their life expectancy, I'm usually conservative with my prognoses, and I tend to be someone who goes through most of the potential side effects of a therapy, no matter how rare. But as I consider withdrawing therapy, I spend a lot of time exploring patients' wishes and goals to see whether they view therapy as just the "next step" or whether they truly have realistic expectations of what more therapy will offer.

Here is the problem with presuming that poor communication is the cause: We have no idea whether good communication will meaningfully reduce end-of-life treatment. In one of the largest randomized trials of early palliative care consultations in patients with advanced cancer, those who received early palliative care had only about a 10-percentage-point reduction in end-of-life therapy. This is by no means trivial, but even with an intensive palliative care intervention, about one third of patients still received chemotherapy near the end of life.

Do we think that relying on primary oncologists—who often barely have time for a 15-minute appointment, let alone a 30-minute conversation—to have better communication will move the mark on end-of-life therapy?

Therapies. In my opinion, the reason that many patients with cancer receive treatment near the end of their lives is because we have made it easy for them to do so. Immunotherapies and targeted therapies have significantly better side-effect profiles than chemotherapy. And because many clinical guidelines now endorse better awareness and prophylaxis of pain, nausea, and infection, we have even gotten better at controlling chemotherapy-related side effects.

Perhaps this is why end-of-life treatment has doubled for cancers like bladder cancer in the immunotherapy era. When counseling patients near the end of life, it often becomes difficult to justify why immunotherapy wouldn't be worth a try. Some physicians have even argued that, in cancers where it is applicable, no patient should die without a trial of immunotherapy. I usually cite that the response rates for immunotherapy are only around 20%, and are even less for particularly sick patients who may be near the end of life; but many, if not most, patients still opt for it. I would probably advise my family member to do the same if I were in that scenario. What if it works?

What to Do?

Ultimately, I think that delineating an "end-of-life phase" in cancer care is a faulty exercise.

Rather than shaming doctors and patients with metrics and vignettes about treatment near the end of life, we should focus on better ways to support cancer patients irrespective of when they are going to die.

For example, we should make it easier for patients to receive chemotherapy at home—no matter when they are going to die. We should allow patients to receive concurrent hospice and curative-intent therapy. Immunotherapy, targeted therapy, advanced diagnostics, and other innovations in oncology should not bankrupt patients or healthcare systems. Changes like these would help ensure that patients get the care they want—and could reduce costs along the way.

We shouldn't view treatment at the end of life as a failure in and of itself. Instead, treating patients near the end of life can be seen as one aspect of a well-designed system that gives patients and families what they want as their cancer journey comes to an end.

Ravi B. Parikh, MD, MPP, is a medical oncologist and faculty member at the University of Pennsylvania and the Philadelphia VA Medical Center, an adjunct fellow at the Leonard Davis Institute of Health Economics, and senior clinical advisor at the Coalition to Transform Advanced Care (C-TAC). His research and writing focus on policy and innovation in cancer care, with specific interests in advanced illness and predictive analytics.

Follow Medscape on Facebook, Twitter, Instagram, and YouTube