When a team of researchers at University of California, San Francisco, started collecting tools for predicting the likelihood of death, they thought their work would be used primarily by physicians. But the project ended up as an interactive tool that would be of interest to medical professionals, elderly patients—and the morbidly curious alike.
The site, ePrognosis.org, displays 16 different methods for determining a person’s chances of dying in the near future. The team designed the site so that doctors could have something better to go on than average life expectancy and intuition when deciding what treatments to recommend for elderly patients. The hope is that a better understanding of life expectancy will help patients and doctors decide on treatments—for instance, sparing a patient with advanced cancer from an invasive procedure for an ailment that likely will never have the chance to become a problem.
The tools aren’t new. Many were publicly available before, or kept behind medical-journal paywalls. But this is the first time so many have been assembled in one place, ranked according to their accuracy, and made so user friendly. A doctor—or anyone who clicks a button saying she’s a doctor—can plug in the relevant medical information and get a prognosis: 59 percent chance of dying within four years for an elderly diabetic male smoker with a history of congestive heart failure, for example.
Some worry that the apparent clarity of that number will mislead laypeople. “It’s not a crystal ball,” says Carol Levine, director of the families and health-care project at the United Hospital Fund. She says that prognosis is a serious matter and should be discussed one-on-one with a professional.
Ken Covinsky, also at UCSF, has a similar concern. “Your radiology report is shrouded in medical terminology, so you’d go over it with your doctor,” he says by way of comparison. “But the number on the prognosis calculator is easily understood.” He cautions that these indices are developed for populations, not individuals. The number is the average life expectancy for a population, which has a huge range within it. Individuals often outlive their life expectancy by quite a lot, he says. “Maybe they have an illness the index accounts for, but it’s a mild case.” Users might also undershoot their prognosis, if they have one of the many illnesses the index doesn’t account for. “I’d tell someone using it that they’re an individual, not a population.”
The researchers behind ePrognosis debated making the tool publicly available. In the end they decided that anything that got people talking about life expectancy was for the good. “Medicine is focused on diagnosing and treating, but not discussing prognosis,” says Eric Widera, one of the researchers behind the project, partly because that’s where the money is. Funding goes to research that yields patentable results, says Widera, not to studies that help doctors decide when expensive, risky procedures might not be necessary. At the clinical level, doctors have a financial incentive not to discuss end-of-life care with patients. Those conversations tend to be lengthy and hard to bill for.
They also tend to be hard to have. Karl Steinberg, a long-term geriatrician in Oceanside, Calif., knows this first-hand. He says that patients being admitted to the nursing home are often taken aback when he asks them about their end-of-life wishes. “I’ll be talking to a 95-year-old lady who’s just been hospitalized for pneumonia, and she’ll stop and say, ‘Wait, Doc, is something wrong?’ She’s been going to the same doctor for 25 years, and they’ve never had a talk about the end of life.”
Steinberg says his patients are relieved to finally be talking openly about it, and that they appreciate being given the chance to choose whether to be kept alive at all costs or to be let go when the time comes. Family members, however, are not always as appreciative. “Even after a long talk about the situation, someone will say, ‘No, no, my mother’s going to live to be a hundred years old.’ I’ve been accused of having an agenda. You can understand the denial.”
Steinberg says that ePrognosis may help start a difficult conversation. Like the nursing-home admission paperwork, or Physician Orders for Life-Sustaining Treatments, the tool could be an impersonal way to spark a discussion about the most personal of decisions. “Our culture is averse to talking about death, and it would help to have the data, to say, look, your mother isn’t a statistic, but these are the odds for someone with her condition.”
In 2010, a group of UCSF researchers, including several who later worked on ePrognosis, set out to determine whether elderly people want to know how long they have to live, or whether they’d rather the information be kept from them. Suppose your doctor knew you had fewer than five years to live, they asked their subjects. Would you want to know? Three quarters said they would. Their reasons ranged from financial planning to planning for the afterlife. “I would take money out of the bank and see who I could give it to,” said an 84-year-old man. “I need to prepare for eternity,” said a 78-year-old woman. People who didn’t want to know said it would make their last years too emotionally difficult. Others said they wouldn’t believe it, so why bother.
Michael D. Pollock is firmly in the camp of those who want to know, though he also takes his prognosis with a healthy dose of skepticism. After all, he says, when he was 8 years old, doctors said he’d be lucky to reach 50, but then penicillin was invented and his chronic bone infection was cured. Now he’s 94 and going strong. Nevertheless, he says he’d like to know how much time he has left “to get his house in order.” His eldest son, who’s his executor, and with whom he has dinner with every Sunday, has been asking him about his last wishes. “I thought I might as well find out how many years the experts think I have, so I can tell him not to worry so much,” he says. “I’m not ready to go.”
When he asked his doctor, she told him not to worry. But when he asked how she knew, she said no doctor could predict life expectancy, and that she was going by his firm handshake and good cheer.
EPrognosis isn’t much more helpful for him. It gives him a 50/50 chance of dying within the next four years, but he’s still so ailment-free that none of the surveys quite fit. The Lee Index tracks body mass index—Pollack has “a bit of a tummy”—and it also asks whether you have trouble walking several blocks, which Pollack does, due to a knee transplant.
That new knee is another reason Pollack takes any prognosis with a grain of salt—and why he’s glad people are working on more accurate ones. When he was 88 he couldn’t walk, and his doctor said no surgeon would bother operating on him because he was too old. “If I’d listened, I would have spent the last seven years in a wheelchair.” The average is just an average, he says, and he has the good fortune to be an outlier. Plus, he says, the indices overlook the most important factor. “The nanogenerians I know are all cheerful people. They keep busy, and they’re happy. They need a question or two about that.”
