Skip to Main Content

Dr. Nancy Keating has seen what too much health care looks like. The Harvard Medical School professor recalls a patient with metastatic lung cancer who, despite not wanting chemotherapy, was convinced by her doctor to try a more targeted treatment. The patient spent the end of her life in the hospital – the exact place she had hoped to avoid.

Keating, also a physician at Brigham and Women’s Hospital, studies how to deliver high-quality care to patients with cancer. Her latest work examines the factors that contribute to large hospital-by-hospital differences in end-of-life spending for cancer patients. The new study reveals that the variation in the intensity of treatment stems more from the availability of services and physicians’ discomfort navigating end-of-life choices than from patients’ wishes.

From surveys, conducted between 2003 and 2005, Keating found that physicians in higher-spending areas reported less comfort addressing end-of-life issues. They felt less equipped to treat end-of-life symptoms, to discuss “do not resuscitate” status, and to present care options such as hospice to their patients.


“We really can discount the idea that high-spending areas are responding to patient demand,” said Keating. Rather, physicians’ beliefs are a major driver of higher spending in the last month of life.

Keating and her co-authors — researchers at Harvard, Kaiser Permanente Northwest, the Dana-Farber Cancer Institute, and Indiana University School of Medicine — acknowledge that their survey data may be outdated. Physicians’ attitudes toward palliative and hospice care have likely shifted in the past decade. But other researchers argue that the findings still matter, particularly because newly developed therapies have opened up more avenues for intensive and expensive end-of-life care.


Dr. Amber Barnato, a palliative care specialist and researcher at the Dartmouth Institute for Health Policy and Clinical Practice, views immunotherapy as the counterweight to palliative care in cancer treatment. “Although it’s possible that attitudes could have improved over time toward a focus on a more palliative approach at the end of life, it could also be that there’s renewed hope and magical thinking that surrounds this new treatment,” she said.

Barnato’s earlier research laid the groundwork for Keating’s study, published Monday in Health Affairs. Barnato showed that health care spending in the last month of life — a proxy for the intensity of care — varies substantially across regions served primarily by individual hospitals, though the people living in these areas are equally sick and have similar wishes for end-of-life treatment.

Barnato and her team asked Medicare beneficiaries to respond to hypothetical situations: If you were dying of a terminal disease, would you want to spend your final days at home or in the hospital? Would you worry about receiving too little or too much medical care? Would you choose to take comfort drugs that could shorten your life or life-extending medications that could cause pain?

The survey revealed a similar range of preferences across hospital regions, and raised a new question: If patients don’t drive the variation in spending, then who does?

Keating’s study points to physicians.

Keating and her fellow researchers first replicated Barnato’s findings in a novel way. Instead of using data on hypothetical situations, they drew from surveys conducted between 2003 and 2005 of patients with either advanced lung or colorectal cancer facing the end of their lives. When faced with real mortality, patient preferences for care did not differ across hospital regions. Spending, however, still did.

Linking the survey data to Medicare claims, the researchers found a nearly twofold difference in expenditures during the last month of life in the lowest-spending areas compared to the highest ones.

Higher-spending regions had different characteristics than lower-spending ones. They were more populous and had a greater percentage of nonwhite residents, fewer primary care providers, and fewer hospital beds and hospice facilities per 10,000 people. Patients in higher-spending areas also had lower levels of social supports.

Another significant difference: Their doctors felt less prepared to care for dying patients.

Barnato told STAT that the new study probably underestimates how much physician attitudes contribute to care decisions. Her more recent work, conducting case studies in different hospitals, shows that there are major differences in the norms for decision-making between hospitals that provide high-intensity end-of-life care and those that provide low-intensity care. Many aspects of physician bias play into their care choices, she said, including: how physicians define dying, whether they view respiratory machines as a temporary tool used in recovery or as life-sustaining machines, and the degree to which they defer decision-making responsibility to patients and their family. All of these attitudes and beliefs change the way doctors provide care for their patients.

These attitudes might also be influenced by how physicians would choose their own care. To her surprise, Keating found that doctors in higher-spending areas were less likely to agree that they would enroll themselves in hospice if diagnosed with a terminal disease.

“I’ve always had the idea that most doctors think really highly of hospice, and would certainly want to enroll in hospice themselves if they were terminally ill,” Keating said.

Maybe, she said, those physicians have seen patients undergo bad experiences with hospice care. Or maybe, it is a purely educational issue – doctors in those areas lack information on the benefits of hospice. Either way, “we need to understand why some doctors have more negative views about hospice and what can we do to improve those.”

Health care leaders have for years been calling for giving patients a greater voice in their care, so it may be disheartening to learn that physicians’ beliefs can outweigh patient values at the end of life. Dr. Gabriel Brooks, a medical oncologist at Dartmouth-Hitchcock, said in cancer care, there will always be some surrogate decision-making by physicians. For many patients, he said, a cancer diagnosis is their first brush with a serious, life-changing illness. Patients don’t immediately know what their preferences are, so they lean on their doctors to help them make decisions about care.

Still, doctors can do a better job of incorporating their patients’ desires into those decisions, said Brooks, who has previously collaborated with some of the authors of the new study.

There are so many different players who participate in care decisions, Brooks added. “You have the patient, you have the family, you have the doctor — and we can all be guilty at different times of pushing beyond a point when our treatments are delivering what we want them to deliver.”

When new treatment options are presented to patients and their families, “it’s easy to say ‘yes’ and much harder to say ‘no,’” said Keating — even when saying “no” is more consistent with their wishes.

Keating said she is hopeful about the future of palliative and hospice care. She said that, in the decade since the data were collected, there has been an increase in hospice use. And she thinks that the next generation of physicians will feel more comfortable with saying “no” to aggressive treatment when it’s the right call for patients. Recent literature shows that training physicians to discuss end-of-life issues can go a long way in helping patients receive the kind of care they want.

“I’m optimistic that we can help with a more holistic view that isn’t only about what more can I do to prolong someone’s life but to really think about how can I help people with serious illness have a better experience at the end of life,” said Keating.

  • Even DNR isn’t foolproof. I clearly had a heart attack, but everyone went busy trying to intubate me. Why???? Because it’s not covered by DNR!
    There is a simple solution, I think: Medical insurance should not pay for treatment beyond that called for by patient, at least via a Living Will.

  • I think the important finding here is the disparity in the end-of-life treatment depending on the demographics of the hospital. Even the premise for the article, do physicians or patients have more influence on end-of-care treatment is dependent on these demographic characteristics. This fits well with an article from the Chicago Tribune on how geography becomes destiny as it pertains to healthcare.

Comments are closed.