Risk-prediction tool calculates chance of dying within six months of hospitalization
A calculated risk: Researchers develop way for doctors to tell which heart attack patients to reassure, and which to warn
Simple pocket card calculates risk of dying within six months of hospitalization
ANN ARBOR, Mich. -- When patients go home from the hospital after a heart attack or sudden chest pain episode, they often face an uncertain future and a lot of worry. Even though they've survived one heart-related crisis, another potentially fatal one could be just around the corner. Or they could be fortunate and live for many years.
Now, new research may give these heart patients, and their doctors, a better sense of who's really at risk -- and who can be reassured that they'll probably be fine.
In the June 9 issue of the Journal of the American Medical Association, an international group of researchers provides a simple way for doctors to calculate the chances that a particular patient will die within six months of going home from the hospital after a heart attack or unstable angina episode.
The calculating tool, which can fit on a pocket card or be programmed into an ordinary handheld data device, is based on data from 22,645 patients treated at 94 hospitals in 14 countries. Its developers hope that doctors everywhere will adopt it as a way of guiding treatment decisions and counseling recommendations for patients.
The paper's first author, University of Michigan Cardiovascular Center Clinical Director Kim A. Eagle, M.D., explains that the risk-predicting tool could help doctors decide early on how aggressively to treat a particular patient, to reduce his or her risk of dying soon after being discharged from the hospital. And, he says, it could ease the minds of many patients, while helping others face the reality of their situation.
The tool creates a score for each patient based on nine variables. The higher their score, the higher their chance of dying within six months of leaving the hospital.
Older age, a history of previous heart attack or heart failure, or a lack of angioplasty or stenting during hospitalization boost patients' scores the most. But so do results from exams and blood tests conducted when they first arrive at the hospital: Patients with faster pulse rates, lower systolic blood pressures, certain electrocardiogram readings, and high levels of blood creatinine and cardiac enzymes score higher.
"Every patient is an individual, and we can never predict everything that will happen to him or her, but this tool has been proven very accurate because it combines a person's own characteristics and compares it with data from thousands of others who have had the same experience," says Eagle. "It's also more current, and more broadly applicable, than other tools developed in the past."
The new tool is based on data from GRACE, the Global Registry of Acute Coronary Events, which pools information on people who have had heart attacks and unstable angina episodes, and allows researchers to analyze their in-hospital symptoms and care, medical history, demographics and survival rates. Taken together, the various forms of heart attack (myocardial infarction) and sudden severe chest pain (unstable angina) are known as acute coronary syndromes, or ACS.
The GRACE prediction model, as the new tool is called, is available online for free use by any clinician, at www.outcomes-umassmed.org/grace.
Eagle and his co-authors developed the GRACE model based on data from 15,007 patients who were discharged alive from the participating hospitals between April 1999 and March 2002, and followed for at least six months after leaving the hospital. They used sophisticated statistical methods to determine which factors were common to those in this "development group" who lived through or died during that period, and how often those various factors occurred in each group.
They then validated the tool, or measured its ability to predict six-month mortality risk, by using it on 7,638 patients treated between April 2002 and December 2003. They found that the tool offered an excellent gauge of which patients were most at risk. The nine variables stood out as consistently different between those who died soon after leaving the hospital and those who didn't. This allowed them to assign a "hazard ratio" or relative point value for each characteristic, and create the risk-prediction tool to allocate points to patients.
The researchers showed through statistical methods that the tool was highly reliable for predicting whether a patient will die or not, including in subgroups of patients (those with ST-segment elevated myocardial infarction, and non-ST-segment elevated myocardial infarction and unstable angina).
The GRACE data pool also included records for 5,000 patients whose fate after leaving the hospital was not known. These patients' in-hospital characteristics were similar to those of other patients.
In general, Eagle hopes the tool can help doctors evaluate patients while they are still in the hospital and determine how much of a post-hospitalization risk they face. This, in turn, can guide treatment. For example, since patients who had angiopasty or stenting to open a clogged artery did better than those who didn't, a doctor may want to consider ordering this kind of revascularization procedure for patients who haven't had it -- if the patient is a good candidate.
And, doctors may want to pursue more aggressive drug treatment, post-hospital monitoring and rehabilitation programs for patients who score high on the model. Alternately, they may be able to reassure a patient who scores low that he or she has a low risk of dying in the next few months -- and help that patient understand how diet, exercise and medication can help keep that risk low.
"The idea is to inform practitioners, and help steer their decision-making, not to make the decisions for them," says Eagle. "Regional variations in medical practice are important and will always be present, but we hope that the broad GRACE population and the use of a key endpoint, mortality, will make this prediction tool applicable in most cases. This model, in particular, helps us move forward in applying the knowledge we gain by studying large populations to benefit the individual patient. It is just part of our effort to improve cardiac care in Michigan, the United States and the world."
Source: Eurekalert & othersLast reviewed: By John M. Grohol, Psy.D. on 21 Feb 2009
Published on PsychCentral.com. All rights reserved.
People ask for criticism, but they only want praise.
-- W. Somerset Maugham