Mortality prediction for palliative care

Predicting death seems to be a morbid undertaking; yet there is real value in this information. Previously, I have described predicting death at different time frames, from the long-term to the short-term to impending death (i.e. those likely to die within the next few days). For the long-run, people have used statistical and machine learning techniques to calculate your biological age, which can be used to estimate roughly the number of years you have left to live. This prediction scheme relied on various bits of information including biomarker levels (blood sugar), fitness (VO2 max), BMI, etc. to evaluate your overall health.

For the more near term, researchers have developed a rule-based algorithm that used vital signs (e.g. heart and breathing rates) and biomarker data to pick out the most sick patients. The algorithm was able to identify 90% of the patients that were likely to die the next day without special medical intervention.

Why is this information useful? In the long-run, it can encourage people to adopt life-style changes to slow down their biological aging (i.e. become healthier). In the short-run, identifying the most sick can help direct hospital resources to the patients with greatest need.

There is another use-case for predicting mortality and that is for identifying patients that may be suitable for palliative care. What is Palliative Care (WSJ)?

"Palliative care focuses on easing symptoms such as pain and shortness of breath that are often overlooked amid aggressive efforts to save seriously ill patients. Unlike hospice, patients receiving palliative care aren’t required to forgo potentially curative treatments like chemotherapy. But many do, especially if the regimens are unpleasant and unlikely to buy much time."

Thus, palliative care fills the gap between hospice care and hospital care (Figure 1). In a hospice, you forgo curative treatments (e.g. chemotherapy against cancer) as you prepare to die. In a hospital, they keep you alive at all costs no matter how bleak the long-term prognosis. Many with serious, chronic,, worsening maladies would benefit from medical care focused on comfort while not forgoing life-preserving treatments (WebMD):

"But palliative care is actually a new medical specialty that has recently emerged -- and no, it's not the same as hospice. It doesn't serve only the dying. Instead, it focuses more broadly on improving life and providing comfort to people of all ages with serious, chronic, and life-threatening illnesses. These diseases may include cancer, congestive heart failure, kidney failure, chronic obstructive pulmonary disease, AIDS, and Alzheimer's, among others."

One societal benefit of palliative care is that it is significantly cheaper than hospital care:

"Health-care experts often lament that one-quarter of all Medicare spending—$150 billion annually—goes to treating patients in their last year of life. But identifying those patients in advance and cutting back on futile care has been difficult. Can an algorithm help?"

A new startup company called Aspire Health has sprung up to fill this breach. The key technology is an algorithm that tries to predict which patients would benefit most from palliative care (i.e. those who are likely to die in the not-so-distant future):

A startup called Aspire Health says that it can predict which patients are likely to die in the next year and reduce their medical bills substantially by offering them palliative care at home, keeping them comfortable while avoiding costly ER visits and hospital stays.
“We can tell which patients will die in one week, six weeks or one year,” says former Senate Majority Leader William Frist, a transplant surgeon who co-founded Aspire. “We can say to health plans, ‘How much are these patients costing you? We can care for them for less, and have higher patient satisfaction rates too.’ ”

How does this prediction algorithm work?

"To identify target patients, Aspire’s algorithm sorts through medical claims, looking for diagnoses such as congestive heart failure or late-stage cancer or for a pattern of frequent hospitalizations. Its clinicians also consult with patients’ primary care physicians to see if palliative care would be appropriate."

The basic source of data are medical claims. There are certain conditions that are associated with poor prognoses which include cancer, as well as heart, kidney, and lung failure. The algorithm searches for these patients along with signs that their status is worsening (e.g. more frequent hospitalizations). Patients with these illnesses are not likely to live for much longer, and frequent trips to the hospital are costly while not improving their long-term survival. Thus, pushing the patients to palliative care could save lots of money:

"Hospitals that offer in-house palliative-care programs find that they save an average of $7,000 per patient, according to the National Palliative Care Research Center. The handful of hospitals that provide palliative care in patients’ homes can save even more—as much as $2,000 a month in one study—by preventing return trips to the hospital."

Why would a patient agree to palliative care? The point is that palliative care in the home would greatly reduce frequent hospital trips that are largely unnecessary and do not significantly prolong life. A greater emphasis on comfort and consistent care would improve the patient experience while not denying curative treatments and the ability to go to the hospital if necessary.

The psychology of enrolling in palliative care is tricky. Effectively, the patient is admitting that she has only a limited amount of time left to live. Indeed the goal of the Aspire Health algorithm is to pick out patients with roughly one year left to live. After living with a chronic illness that is worsening, most patients recognize that the end may not be far off. So the physicians of Aspire Health basically ask the patient about their own long term prognosis to clear the air. Then they assure the patient that they will not be denied any treatment or going to the hospital, and that they are not giving up:

"About 15% of those approached decline because they have home-care already or don’t think they need it. Many of those who do enroll are aware of their prognosis. “At their initial assessment, the first thing we say is, ‘Tell us about your illness,’ ” says David Thimons, Aspire’s lead physician in the Pittsburgh area. “Sometimes they say, ‘I’ve got about four months to live—don’t tell my kids.’ ” A key part of the program is discussing patients’ individual “goals of care,” including what treatments they want and don’t want. “We emphasize that not wanting to go to the hospital anymore doesn’t mean you’re giving up,” says Tiffany Lunsford, a nurse practitioner and clinical director with Aspire. “And if they do want to want to go to the hospital, we go.”

Palliative care is a concept that we will hear more about in the future in order to fill the gap between hospitalization and hospice care. In theory, it can save money without affecting the quality of care, and indeed the goal is to provide more comfortable and consistent care to the patient avoiding repeated hospitalizations. Ultimately palliative care will be adopted by a wide array of medical care providers, but for now startups such as Aspire Health will lead the way to gain greater acceptance for this approach.

Figure 1. A continuum of medical care for chronic illness. For those less ill, one can hope for cures or substantial life extension through aggressive treatment. For those who are terminally ill, there is hospice care (followed by bereavement) in which curative treatments are halted. In between is palliative care which allows for curative treatments but in a less aggressive fashion, and tries to minimize the number of trips to the hospital while providing comfort for the patient.