predictions

When Sen. Ted Kennedy’s diagnosis of brain cancer was announced, it set off a morbid, sometimes irresponsible, countdown.

Reporters wanted to know how long he could live with his tumor, known as a glioma. Doctors, going by the limited information available from Sen. Kennedy’s doctors, responded with answers that were all over the map.

Some doctors, not hearing any mention of surgery as an option, said Sen. Kennedy may have just six months. But the senator underwent surgery at Duke this week. Others cited stats for the most-severe type of glioma, which kills half of patients within 15 months — or is it 12? Still others optimistically shared typical survival rates for a less-extreme form of the condition: three to five years.

When it comes to answering the most enduring question about a life — when it ends — even the best scientific studies of some of the more common medical cases points to one conclusion: We don’t really know.

“It is lies, damned lies and statistics,” says Lynne Taylor, director of neuro-oncology at the Virginia Mason Medical Center in Seattle. “What everyone cares about is what’s going to happen to Ted Kennedy, and that’s the one thing statistics can’t tell.”

Even if the media’s medical experts could draw on the same information as Sen. Kennedy’s doctors, it would be hard to predict survival time. “Most of the numbers are based on all comers,” says Jeffrey Raizer, director of Northwestern University’s medical neuro-oncology program.

Age — Sen. Kennedy is 76 years old — and functional impairment, as measured by the Karnofsky Performance Status score, have a big impact on the prognosis. An otherwise healthy person his age might do as well as a typical 40-year-old. “You have to treat the individual, not the statistic,” Dr. Raizer says. http://louis-j-sheehaN.NET

Also, life-expectancy data for such patients are dated. “True life expectancy with best treatment is constantly changing,” says Jonathan A. Friedman, a neurosurgeon and director of the Texas Brain and Spine Institute. “Measurement and reporting of this will always lag behind reality.”

Some news articles say glioma patients typically live 12 months from diagnosis; others bump the figure to 15 months because a more-recent study showed promise. The news media add to the confusion by treating median survival times like death sentences. Saying most such patients are given a certain amount of time to live implies there is no chance to live longer. Yet half of patients outlive those estimates, says Ellen Fox, a health-care ethicist for the Veterans Health Administration.

On either side of the midpoint, survival times can vary widely. A 2005 study of radiotherapy and a drug called temozolomide found that 27% of patients lived more than two years and nearly 20% lived past three years — more than twice the typical survival time.

On the other hand, we hear about the outliers, those who inspire hope with their prolonged survival despite doctors’ grim forecasts. But these are exceptions. Doctors struggle to translate survival statistics for the news media, and do far worse when trying to apply these stats specifically to their patients.

The error is usually on the side of overoptimism, in part because doctors tend to be confident in their abilities and hopeful for their patients. http://louis-j-sheehan.com
Doctors overestimated dying patients’ survival by a factor of 5.3, Harvard Medical School professor Nicholas Christakis found in a study of terminally ill patients referred to hospice care who had, on average, about a month to live.

In a study of Dutch nursing homes, half of patients expected to have four to six weeks to live had died by the end of the third week. “Doctors simply overlook the signs of nearing death,” says study co-author Hella Brandt of the Netherlands Institute for Health Services Research.

Predictions of death fall short because of twin failures of science and communication. The science of prognosis is poorly understood and inadequately taught. On surveys, most physicians say they weren’t adequately trained in prognosis, Dr. Christakis says.

The pain and difficulty of communicating the prediction exacerbates the error. “Research in this area shows that most people want a broad idea of what to expect, but not all want precise details regarding statistics,” says Josephine Clayton, of the University of Sydney.

Many patients never ask even for the broad outline, out of fear of the answer. Their doctors, in turn, also fear this moment. When estimating life expectancy for patients who, it turned out, had about a month to live, doctors tacked 15 days onto their private predictions, which were already overly optimistic, according to a separate study by Dr. Christakis and Dr. Elizabeth Lamont.

And patients sometimes tack on still more time, as demonstrated by a Duke University study published this week showing that patients with heart failure significantly overestimate their life expectancy. Only one-third of them spoke to their clinicians about a prognosis, and that didn’t help their forecasts.

The implications go beyond any emotional consequences of dying patients thinking they have more time to live. Patients and doctors expecting a longer survival time may agree on more-invasive treatment, adding the burden of side effects and complications to patients in their final days, and keeping them in hospitals.

Not every study shows a tendency toward optimism. A study from an Ireland hospice this year found that senior clinical staff tended to underestimate survival. But all studies agree that the accuracy rate is alarmingly low: Fewer than half of predictions are within 33% of the correct survival time.

Feedback and quality control could help hone survival estimates. Hospital doctors could remove some of the statistical noise by averaging predictions from all members of their team, Dr. Christakis suggests. The natural competitiveness of doctors might spur them to track their accuracy rates and adjust accordingly for future patients.

For all their predictive failings, doctors generally can discriminate between cases. One patient predicted to live longer than another usually does.

Oncologist Martin Stockler from the University of Sydney found physicians do better predicting big-picture statistics. If you ask how long 10% of similar patients — or 90% or 50% — were likely to survive, they give more-accurate predictions.

Dr. Taylor improved her accuracy after comparing her estimates for past patients with their actual survival times, and realizing she had been too optimistic. “Your relationship with the patient is you only want the best for them,” she says, “and you only want to give them hope.”

Advertisements

Tags: ,

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: