# Prognosis and Risk Stratification of Patients with Advanced Heart Failure Followed-Up on an Outpatient Clinic

## TL;DR

This study examines the outcomes of advanced heart failure patients in an outpatient clinic and finds that existing risk scores are not very accurate for predicting short-term mortality.

## Contribution

The study introduces logistic regression-based models that improve the accuracy of predicting 12- and 30-month mortality in advanced heart failure patients.

## Key findings

- Existing risk scores (MAGGIC, SHFM, BCN-Bio) had limited prognostic value for 12-month mortality.
- Logistic regression models incorporating clinical factors showed higher accuracy for predicting 12- and 30-month mortality.
- Mortality was associated with factors like blood pressure, sodium levels, and presence of coronary artery disease.

## Abstract

Background/Objectives: Advanced heart failure (AdvHF) characterizes patients with impaired functional capacity, severe systolic or diastolic cardiac function, unplanned visits or hospitalizations, raised natriuretic peptides, and increased mortality. Methods: Ninety-five consecutive AdvHF patients followed in a tertiary academic center in Northwestern Greece (2nd Department of Cardiology, University Hospital of Ioannina) were enrolled over a 30-month period. Three distinctive patterns of management were recognized and assessed: intermittent levosimendan administration to 33 patients, intermittent intravenous furosemide administration to 17 patients, and 45 patients were followed up exclusively on an outpatient basis with frequent visits. MAGGIC, SHFM, and BCN-Bio scores were assessed in all patients and mortality was also assessed. Results: Mean age was 73 (±10) years, and 38% were females, 41% had diabetes mellitus, 41% had chronic obstructive pulmonary disease, 59% had coronary artery disease (CAD), 73% had a history of atrial fibrillation, and 82.1% had a cardiac device implanted. The median duration of follow-up was 24 months (IQ range 14, 30). The 12-month and 30-month mortality rates were 19% and 49%, respectively. Higher rates of 1-year mortality were observed in the levosimendan group (30%). The median 12-month mortality of the three scores was comparable to the actual mortality, but their prognostic value was not satisfactory (AUC < 0.540 and p > 0.05 for all), while they performed better for 30-month mortality (AUC < 0.756 and p > 0.05 for all). In the current study, mortality at 12 months was associated with decreasing diastolic blood pressure (DBP) and sodium levels; the presence of CAD (p < 0.05 for all) and mortality at 30 months was associated with decreasing systolic blood pressure, as well as DBP and left ventricle ejection fraction, but also with the presence of CAD and the use of renin–angiotensin–aldosterone system blockers. Logistic regression-based models incorporating these factors have a greater diagnostic accuracy (AUC = 0.824 and 0.817 for 12 and 30 months, respectively; p < 0.001 for both). Conclusions: AdvHF patients represent a complex population requiring close follow-up and novel strategies to improve survival. Larger studies are needed to refine and update predictive scores in this population.