For many older Americans with advanced and incurable cancer, where and how they die — at home with their family or sedated in an ICU with a tube down their throat — may not be based on their final preferences and wishes, but on customs, care patterns, and even the financial incentives and number of beds in the hospital they and their loved ones entrusted with their care.
That’s the conclusion of a new addition to the Dartmouth Atlas, a compendium of 20 years of research on how health care usage and practices vary tremendously from one place to another.
“The bottom line is the care patients receive has less to do with what they want and more with the hospital they happen to seek care from,” said Dr. David Goodman, lead author of the Dartmouth Atlas Project and director of Dartmouth’s Center for Health Policy Research. The study found “no consistent pattern of care or evidence that treatment patterns follow patient preferences, even among the nation's leading academic medical centers.”
Overall 1 in 3 of these patients died in the hospital, sometimes in the ICU and sometimes on life support, but there was significant variation from one region or even one hospital to another. Six percent of the patients received chemotherapy in the last two weeks of life, but in some regions and academic medical centers the rate went above 10 percent. Half got hospice but often for just a few days, too little for them and their families to fully benefit from the medical and psychosocial assistance and comfort hospice can offer.
In addition to that wide variation, the overall message was that these very sick patients — elderly patients with advanced cancers nearing the end of their lives — are getting lots of aggressive care. Given the risk of infections and complications when frail people undergo invasive inpatient procedures, this aggressive care often does not help them and may even harm them, noted Rosemary Gibson, a health care consultant who has worked extensively on end-of-life and health care quality.
“We still don’t know when to stop,” said Gibson, who read the Dartmouth report but is not part of that research team.
Goodman said the Dartmouth data was adjusted to take socioeconomic aspects of a hospital population and location into account (although some health researchers have, in the past, faulted the Dartmouth methodology). The researchers also focused on patients who have a fairly predictable poor prognosis — patients with such diseases as advanced lung or pancreatic cancer. They looked at records of more than 235,000 Medicare patients who died from 2003 to 2007.
Some of the wide variation occurred in places that have cropped up in previous reports documenting treatment disparities. Nearly 7 out of 10 of these patients were hospitalized in their last month of life in McAllen, Texas — the community depicted as a profligate health care spender in Atul Gawande’s New Yorker article that influenced the national health reform.
But in La Crosse, Wis., which has achieved national renown as a community where the medical culture emphasizes advanced planning and thoughtful conversations with patients and families, fewer than half were admitted in that final month.
That may not be a surprise; similar patterns have been identified before. But what can be quite startling is the variation between hospitals so near each other. At Lenox Hill Hospital on New York’s Upper East Side, half of the older advanced cancer patients died in the hospital. But within walking distance on the Upper East Side, at Memorial Sloan-Kettering, the rate was 1 in 3.
And Sloan-Kettering is one of those magnets that draw people from afar in hope of a miracle. The same is true of Johns Hopkins — another center known for leading-edge cancer care that still has a fairly low rate of hospital use for the dying. And Hopkins is in Baltimore, socioeconomically a far cry from the Upper East Side.
People don’t seek care at Hopkins or Sloan-Kettering to die. They go in hopes of a cure. But when a cure isn’t possible, Goodman noted, “part of the very best care in cancer is also care for comfort.” And that includes open conversations about the likely course of the disease, a patient’s options, choices and wishes.
“We understand those conversations are difficult for physicians,” said Jon Radulovic, a vice president at the National Hospice and Palliative Care Organization. But patients and families may already know in their heart what the doctors are reluctant to say; they know when their chemo is failing, when options are running out. “[Patients] might not be as surprised as many physicians think they may be.” And they may be grateful for a chance to talk about their wishes, fears and concerns.
The reasons for the wide variation in hospitalization and hospice patterns aren’t always clear. It can be the local business model and bed supply of a hospital with a pricy cancer center. It can derive from the way Medicare pays hospitals; they get a lot more for chemo and CPR than they do for palliative care and conversations.
And it can be local practice patterns. One group of doctors learns to practice in a certain way, and they pass that on to their colleagues, institutionalizing it over the years. There’s not necessarily any solid evidence to prove that one group of physicians’ approach is better than another — one reason the federal government has increased its investment in comparative effectiveness research both through the 2008 economic stimulus law and health reform.
In addition, some communities have more hospice and palliative care available than others. And to make it even more confounding, a couple of hospitals with nationally recognized palliative care teams still had a lot of patients in the hospital shortly before or at the time of death.
According to Dr. Sean Morrison, director of the National Palliative Care Research Center based at Mt Sinai Hospital in New York, subtle differences in community norms also play a role. Hasidic Jews in Brooklyn, Morrison noted, make different end-of-life decisions than Latinos in the Bronx, a few miles away physically, but a chasm culturally. Midwesterners, one hospice administrator once told me, tend to respond to a bad prognosis with a plan for how to live out their life. Manhattanites getting the same upsetting news, she told me, want to schedule a meeting to strategize against death.
“It’s great that we know about this variation,” Morrison said. “But what are the true drivers?” Morrison’s major worry isn’t precisely what is the “right” amount of care for the dying. He wants to make sure that they are getting the care that they want, that appropriately trained health care providers talked to them — and listened to them.
“Is the care consistent with patient preferences?” he asked. “Are physicians really understanding what the values and goals are? Until we really understand that, we aren’t going to change the observed variation.”
One aspect of palliative care — which unlike hospice does not require people to give up chemotherapy, radiation or similar treatments — is to have those conversations and identify the goals of care. Palliative care services at hospitals and cancer centers have expanded greatly since 2003, the start year for the study. But the Dartmouth data doesn’t tease out which of the hospitalized patients may have had those services or whether the patients who died in the hospital were also being treated and helped by a palliative team.
“What patients really want is for their physicians to be honest with them, to share the full range of treatment choices. Patients want to live long, but they also need to live well,” said Goodman.
“Our preferences certainly can change. And they do change,” Goodman added. “But opening up that conversation early allows for that conversation as well to evolve over time. The tragedy is when those conversations only start near the end. The longer you wait, like many things, the harder it is to do.”