ED Tool Predicts Prognosis for Those With Acute Heart Failure

Diana Phillips

October 02, 2017

A risk score based on 13 independent risk factors can accurately predict the prognosis of patients admitted to the emergency department (ED) with acute heart failure (AHF), a study has shown.

Developed on the basis of data from the Epidemiology of Acute Heart Failure in Emergency Departments (EAHFE) Registry, the MEESSI-AHF (Multiple Estimation of Risk Based on the Emergency Department Spanish Score in Patients With AHF) score predicted 30-day mortality with excellent discrimination and calibration and, as such, may help ED physicians stratify patients with AHF according to risk. Oscar Miro, PhD, from the Hospital Clınic, University of Barcelona, Spain, and colleagues report their findings in an article published online October 3 in the Annals of Internal Medicine. The predictions are most accurate for patients at very high risk for 30-day mortality and those at low risk, the authors write.

The EAHFE registry collects detailed information on consecutive patients diagnosed with AHF treated in the 34 Spanish EDs that range in patient volume and setting from community facilities to academic centers. The registry includes all patients with a confirmed HF diagnosis except those with acute ST-segment elevation myocardial infarction, the authors report.

To develop the model, the authors analyzed data from a derivation cohort of 4867 consecutive patients, mean age 79.7 years, who were admitted to an ED with AHF during May 2009 and from November to December 2011.

On the basis of baseline demographic characteristics, medical history, and status at admission, they identified 88 possible prognostic variables. Of the derivation cohort, 10.3% (500) died within 30 days of ED admission. Using logistic regression and forward stepwise variable selection, the authors identified the following 13 highly significant independent predictors of death, which they included in MEESSI-AHF risk score.

Ordered by their predictive strength, the 13 variables include:

  1. Barthel index score at admission

  2. systolic blood pressure

  3. age

  4. N-terminal pro B-type natriuretic peptide level

  5. potassium level

  6. positive troponin level

  7. New York Heart Association class IV disease at admission

  8. respiratory rate

  9. low-output symptoms

  10. oxygen saturation

  11. episode associated with acute coronary syndrome

  12. hypertrophy on electrocardiogram

  13. creatinine level

The multivariable risk score for a specific patient, including those who do not have values for Barthel index score, troponin level, or N-terminal pro B-type natriuretic peptide level, can be calculated using an online tool. "On the Web site, values for the relevant items can be entered, and the percentage of patients with these values who are predicted to die within 30 days will be calculated," the authors write.

An analysis of cumulative 30-day mortality for patients classified in the bottom 4 quintiles and the top 2 deciles of the risk score's distribution showed good discrimination (c-statistic, 0.836; 95% confidence interval, 0.818 - 0.853), the authors report. "There was a steep gradient in 30-day mortality across risk groups, with 45% mortality for the top decile and about 0.7% for the bottom quintile."

To validate the risk model, the authors tested it in a cohort of 3229 patients recruited during January and February 2014. "The model fit and extent of risk discrimination were similar in the derivation and validation cohorts," they write. "For example, the c-statistic in the validation cohort was 0.828 (CI, 0.802 to 0.853), and the Hosmer–Lemeshow test for the validation cohort yielded a P value of 0.122."

The MEESSI-AHF score also demonstrated better overall discrimination than the Emergency Heart Failure Mortality Risk Grade tool, which was developed to predict 7-day mortality in these patients (c-statistic, 0.830 vs 0.750; P < .001), the authors note.

With the MEESSI-AHF risk calculator, "emergency physicians will now be able to determine whether a patient is at high (or low) risk for death within 30 days, which in turn might allow for better patient management. Our score may be particularly useful in the 10% of patients at very high risk (around 45%) for death at 30 days and in the 40% of patients at low risk (<2%)," the authors write. "Identification of both groups has important management implications. For a patient with very high risk, attention should be focused on ensuring that the patient and their relatives are aware of the severity of the situation." The high-risk patient should also receive aggressive treatment if appropriate and early admission to an intensive care unit.

Management for low-risk patients should focus on evidence-based treatment that will lead to early discharge from the ED to home, the authors state.

"We believe that physicians can consider using this tool to inform clinical decisions as we conduct further studies to determine whether the tool enhances physician decisions and improves patient outcomes," the authors write, although they advise caution with respect to extrapolating the findings of the current study to other countries.

In an accompanying editorial, Peter S. Rahko, MD, from the University of Wisconsin School of Medicine and Public Health, Madison, describes the predictive study as "rigorous and well-conceived," although he notes that it is the fourth major study aiming to define ED prognosis. Although all four studies claim excellent discrimination between high and low risk, only three variables are common to all of them: positive troponin level, oxygen saturation, and renal function.

"If any of these models are to gain acceptance, they will need to be prospectively tested in diverse populations," Dr Rahko stresses. Further, he notes, "[i]f 40% of ED patients with HF are truly at very low risk, we must find commonalities among them. This information may guide development of an alternate infrastructure to successfully treat these patients out of the hospital."

The study authors report financial relationships with Novartis, Pfizer, Merck Sharp & Dohme, AstraZeneca, Abbott, Servier, Janssen, and Sanofi. Dr Rahko has disclosed no relevant financial relationships.

Ann Intern Med. Published online October 3, 2017.

For more news, join us on Facebook and Twitter


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.