A team led by researchers from the San Francisco VA Medical Center (SFVAMC) and the University of California, San Francisco (UCSF) has completed the first systematic review of prognostic indices used to calculate a patient’s life expectancy, and created a website that puts these indices in one central location.
The review concludes that the most accurate and usable indices might have value when used in conjunction with other clinical information.
The review appears in the January 11 issue of the Journal of the American Medical Association. The prognostic indices are collected at www.eprognosis.org.
Many medical interventions have guidelines recommending that doctors take a patient’s life expectancy into account, said senior investigator Alexander K. Smith, MD, MPH, a palliative medicine doctor at the UCSF-affiliated SFVAMC.
Given this goal, he said, “It would be ideal if there were one index that would allow you to plug in your patient’s information – age, diseases, functional impairments – and get an accurate long-term prognosis,” Smith said. “Unfortunately, there is not. In the absence of that, we have this systematic review and corresponding online compendium, which we hope physicians will find a useful adjunct, along with patient preferences and their own professional judgments, in making clinical decisions that involve life expectancy.”
The authors note that the 16 indices need further independent testing for accuracy in different settings, and that further studies are needed to show whether use of the indices improves clinical outcomes. In the meantime, they have made the indices more accessible to clinicians and patients who are interested in the information they provide.
“We often don’t talk about prognosis with our patients, and, as clinicians, we are, frankly, not trained to think about it,” said lead author Lindsey Yourman, MD, a medical student at UCSF at the time of the study. “This can lead to unnecessary suffering when we order invasive interventions for patients who may not live long enough to benefit from them.”
At the same time, she noted, prognostic indexes are not intended to limit care for elders. “In some instances,” she said, “they may lead to more interventions. For example, some older patients may not be offered cancer screening due to their age, but a prognostic index may suggest they are healthy and likely to benefit from cancer screening because of long life expectancy.”
Website Prognosticates for Typical Geriatric Patients
There are already websites with prognostic calculators, said Smith, but they exist for patients with specific diseases. “This is the only site we know of that helps physicians prognosticate for older adults with multiple conditions,” he said. “So it’s applicable to the typical geriatric patient.”
Yourman said that while the indices are of varying value and applicability, most have one component in common: “Functional status – the ability of the patient to carry out certain tasks and activities of daily living – was found to be statistically significant in predicting life expectancy in almost all of the indices we reviewed,” she said.
In compiling and analyzing the indices – winnowed down from an initial field of more than 21,000 related research titles – the researchers created a set of criteria, included on the website, that physicians can use to evaluate the quality of prognostic tools.
The first criterion is accuracy, which is the degree to which predicted outcomes match observed outcomes, said Smith, who is also a Greenwall Faculty Scholar in Bioethics and an assistant professor of medicine in the Division of Geriatrics at UCSF.
The second criterion is what the researchers term transportability – the extent to which an index was tested independently in patients with essentially different characteristics from the patient population in which it was first tested. “The more transportable the index, the more generalizable it is to other types of patients,” said Smith.
The third criterion, said Smith, is applicability: “Are the patients in which this index was studied similar to my own patient in key characteristics, such as age, functional impairments and co-morbidities?”
The research team also evaluated the bias of each index. “Let’s say that in developing a five-year mortality index, 10 percent of your subjects left the study before data collection was complete, and you weren’t able to assess their mortality over five years,” Smith explained. “It’s quite possible that they differed from the other 90 percent, and that may bias your results.”
Yourman said she hopes that other researchers will take advantage of the website to further test the indices in diverse patient populations. “We want to do our best to make sure that we help older patients make informed decisions about whether they will benefit from a treatment or not,” she said.
Co-authors of the study are Sei J. Lee, MD, MAS, of SFVAMC and UCSF; Mara A. Schonberg, MD, MPH, of Beth Israel Deaconess Medical Center, Boston, Mass.; and Eric W. Widera, of SFVAMC and UCSF.
The study was supported by funds from the National Institute on Aging, the Greenwall Foundation and the Clinical and Translational Science Institute at UCSF.
SFVAMC has the largest medical research program in the national VA system, with more than 200 research scientists, all of whom are faculty members at UCSF.
UCSF is a leading university dedicated to promoting health worldwide through advanced biomedical research, graduate-level education in the life sciences and health professions, and excellence in patient care.
