SILVER-AMI Bests GRACE for Predicting 6-Month Mortality After Acute MI in Elderly

A novel risk calculator that incorporates functional impairments associated with aging shows better ability to predict the 6-month risk of mortality after acute MI in older patients than does the standard GRACE score, new data suggest.

“There are not a lot of risk models that are available specifically at the time of discharge from  hospitalization for heart attack,” senior author Sarwat I. Chaudhry, MD (Yale University School of Medicine, New Haven, CT), told TCTMD. “We basically wanted to take everything that had been out there in the previous traditional risk models and add the functional impairments to see if we could do a better job of predicting mortality in these older patients.”

A limitation of the GRACE score is that it was derived from an ACS registry of patients with a mean age of 65 years, calling into question its predictive value for considerably older patients, she added.

By taking into account age-related factors such as hearing and mobility issues, in addition to traditional demographic and clinical variables, Chaudhry and colleagues led by John A. Dodson, MD, MPH (New York University School of Medicine, NY), found the new score had better predictive accuracy and quantitative ability for assessing individual risk for people who were primarily in their 70s and 80s.

The study was published online ahead of print this week in Annals of Internal Medicine.

SILVER-AMI

The new risk score is derived from the SILVER-AMI study of patients ages 75 years and older. Using data on the 3,006 patients who were hospitalized with acute MI and discharged alive, a list of 72 potential predictive variables for 6-month mortality was compiled. Of these, 15 were selected for inclusion in the new score, which is available online. While it has many of the same features of GRACE, the four notable exceptions are the addition of hearing impairment, mobility impairment, weight loss, and patient-reported health status, all of which the researchers say can be obtained at the patient’s bedside in less than 10 minutes.

In an accompanying editorial, Dhruv S. Kazi, MD (Beth Israel Deaconess Medical Center and Harvard Medical School Boston, MA), and Kirsten Bibbins-Domingo, MD, PhD (University of California, San Francisco), note that the four measures can either be automatically captured from electronic health records in some cases, or obtained from brief surveys or tests.

“Routine collection of these measures might be facilitated by their inclusion in an annual wellness visit covered for Medicare recipients,” Kazi and Bibbins-Domingo add. “Measures of frailty and the prediction rules that include them should be integrated into the electronic health record in a form that can influence clinical decision-making.

To TCTMD, Chaudhry said that while it would be “a great idea to start considering ways that we might automate this information,” there is one major concern. “There might be a change from a wellness visit to the time that a person has a heart attack,” she observed. “Our risk score used information that had been collected at the time of hospitalization. Having said that, the more information that we have about patients’ function, the better. For many patients, just having more information is valuable in and of itself.”

Another important issue raised by the editorialists, however, is uncertainty over whether the predictors of 6-month mortality are themselves the targets for intervention, or whether they are markers for other features that need to be addressed.

“For instance, if a patient with limited mobility is at increased risk for death after an MI, does providing transportation services become more effective than referring them to physical therapy?” they write.

“It is true that while it's very important to be able to have goal-directed therapy and use information from the risk score to actually do something to decrease mortality, you could also imagine a scenario where an older person might have a very high risk for mortality and that might inform their goals of care conversation,” Chaudhry said.

In other situations, things such as hearing impairment, which may increase mortality risk through difficulty in understanding and carrying out complex medication and follow-up instructions, could prompt interventions to ensure that patients are hearing and understanding all of the information correctly and/or that a caregiver is present when the information is given.

Until external validation of the risk score is conducted in larger populations, these questions will likely remain unanswered, Chaudhry noted.

“Our study was merely developing and validating the risk score; sort of putting it out there,” she told TCTMD. “Whether putting that into the hands of clinicians actually makes a difference, that's an important future study that's outside the bounds of what we did.”