August 03, 2016
2 min read

In-hospital mortality risk model for acute MI validated in new study

You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact

The ACTION Registry-Get With The Guidelines risk model for in-hospital mortality for patients with acute MI has been updated to include cardiac arrest and validated as a robust tool for risk adjustment and benchmarking of mortality outcomes.

“Adjusting for cardiac arrest among heart patients is critically important and enables a fairer assessment for hospitals that care for these patients,” Robert L. McNamara, MD, MHS, associate professor of medicine at Yale School of Medicine, said in a press release. The model “should enhance research into best practices to further reduce mortality in [MI] patients.”

Updated risk model

The updated model is based on data from 2012 and 2013 and replaces an earlier version, which was based on 2007 and 2008 data, according to the release. The revised risk model is more robust and generalizable to a specific population of patients with MI. The new model includes age, heart rate, systolic BP, presentation after cardiac arrest, presentation in cardiogenic shock, presentation in HF, type of MI, creatinine clearance and troponin ratio. McNamara and colleagues analyzed data from 243,440 consecutive patients at 655 hospitals from January 2012 to December 2013 to develop and validate this new risk model for prediction of in-hospital mortality after acute MI.

Overall, the in-hospital mortality rate was 4.6%. Risk scores varied considerably depending on patients’ demographics and clinical features. Observed mortality rates increased with rising risk scores. Patients with a risk score of less than 30 had an observed mortality rate of 0.4%, whereas those with a risk score of greater than 59 had an observed mortality rate of 49.5%.

The model showed “high discrimination in both the derivation and validation populations,” according to the researchers. The C statistic was 0.88 for both groups. The researchers also reported “excellent calibration” in the validation cohort (slope, 0.996 [P = .74]; intercept, –0.025 [P = .42]). The model also performed well in subgroups based on age; sex; race; presence of cardiac arrest, cardiogenic shock, diabetes, renal dysfunction and STEMI; and transfer status.

“This new model should enable improved assessment of hospital quality and enhance research into best practices to further reduce mortality with acute MI,” McNamara and colleagues wrote in the Journal of the American College of Cardiology.

One risk score ‘does not fit all’

In an accompanying editorial comment, Peter W.F. Wilson, MD, of the Atlanta Veterans Administration Medical Center and Emory Clinical Cardiovascular Research Institute in Atlanta, and Ralph B. D’Agostino Sr., PhD, from the statistics and consulting unit of the department of mathematics and statistics, Boston University, pointed out that there are now several risk models to choose from when making treatment decisions. Wilson and D’Agostino recommended that “a comprehensive cross validation and comparison across at least some of the algorithms — TIMI, GRACE, HEART, DAPT and ACTION — would help at this point. Interventions and decision points have evolved over the past 15 years, and evaluation of relatively contemporary data would be especially helpful.

“For example, the HEART score is likely to be used in situations in which the negative predictive capabilities are most important. The ACTION score is likely to be most useful in severely ill patients and to provide guidance for newer interventions. If detailed information concerning stents is available, then the DAPT score should prove helpful,” they wrote.

However, “it is likely that [one] score does not fit all,” Wilson and D’Agostino wrote. “Each algorithm provides a useful summary of risk to help guide decision-making for patients with ischemic symptoms, depending on the severity of the signs and symptoms of presentation and the duration of the follow-up interval.” – by Tracey Romero

Disclosure: McNamara reports serving on a clinical trials endpoint adjudication committee for Pfizer. Wilson and D’Agostino report no relevant financial disclosures. Please see the full study for a list of all other researchers’ relevant financial disclosures.