This is a single-center, retrospective study conducted in a large academic medical center, Fondazione Policlinico Universitario A. Gemelli IRCCS (Rome, Italy).

Study population

We identified all patients 65 years of age and older consecutively admitted to the Emergency Department (ED) due to AHF and hospitalized in internal medicine wards, over a 4-year period between January 1, 2016, and December 31, 2019. We excluded data from 2020 onwards to avoid potential confounding factors related to the COVID-19 pandemic.

The criteria for identifying cases included an admission diagnosis of AHF, either de novo or acutely worsening HF, adjudicated by the emergency physician and based on a set of standardized parameters including clinical symptoms, physical examination, laboratory parameters, biomarkers, and radiological findings. In addition, cases needed to have AHF coded as the primary diagnosis in the discharge record. Diagnoses at hospital discharge were based on ICD-10 codes [International Classification of Disease, 10th revision].

Among these patients, we evaluated only those who underwent a 12-derivation electrocardiogram (EKG) and a cardiac ultrasound during the initial hospital stay. In the final sample, we included all patients classified as HFpEF—ejection fraction (EF) ≥ 50%—according to the European Society of Cardiology (ESC) guidelines [12] [Figure S1]. The diagnosis of AF was adjudicated according to the clinical history of AF and the EKG rhythm at enrollment. The final sample was divided into two groups: HFpEF patients “with AF” and “without AF”. In the patients “with AF” group we included patients with permanent, persistent, and paroxysmal AF.

Patients presenting to the ED with AHF due to acute coronary syndromes and requiring catheter-based interventions, those with advanced atrioventricular blocks or cardiac tamponade, patients with pacemakers or implantable cardioverter defibrillator, and those who were otherwise admitted to an intensive care unit (ICU), were excluded from the study.

Study variables

Data were obtained from electronic medical records. Each patient’s record was used to collect demographics and clinical characteristics, data regarding ED presentation, as well as any information related to hospital stay, including diagnostic tests and procedures, treatments, and outcome. To gain further information, relevant medical documentation was reviewed to reach a complete account of all comorbidities.

The data considered in the study included:

Demographic data: age and sex.

Clinical presentation at ED admission, including vital signs [blood pressure, heart rate, oxygen saturation], body mass index (BMI), clinical symptoms (dyspnea, chest pain, syncope, fatigue), and physical signs such as the presence of peripheral edemas and oliguria. The New York Heart Association (NYHA) classification was used to categorize all patients according to the severity of HF symptoms [13].

Echocardiographic variables, including LVEF, evaluated with conventional 2D-trans-thoracic echocardiogram according to international standard criteria [14], left ventricular end-systolic diameter (LVSd), left ventricular end-diastolic diameter (LVDd), tricuspid annular plane systolic excursion (TAPSE), pulmonary artery systolic pressure (PASP), E/e’ ratio and left atrial dimension index (LAVI).

Laboratory parameters and biomarkers, including blood N-terminal pro-B-type natriuretic peptide (NT-proBNP), high-sensitivity cardiac troponin I (hs-cTnI), hemoglobin (Hb), white blood cell count (WBC), platelet count (PLT), creatinine, glucose levels, procalcitonin and C-reactive protein (CRP).

Comorbidities, including hypertension, ischemic heart disease (IHD), peripheral artery disease (PAD), cerebrovascular disease (history of previous stroke), dementia, chronic obstructive pulmonary disease (COPD), diabetes, chronic kidney disease (CKD) and malignancy. The number of comorbidities and their severity were assessed by the Charlson Comorbidity Index (CCI) [15].

Medications prescribed at hospital discharge, including loop diuretics, beta-blockers, mineralocorticoid antagonists (MRA), angiotensin-converting enzyme inhibitors (ACEi), angiotensin receptor blockers (ARB), antiarrhythmic and anticoagulants (including both oral and parenteral anticoagulants).

Outcome measures

The primary endpoint of the study was the all-cause, in-hospital mortality. In addition, we analyzed separately the occurrence of cardiovascular (CV)- and non-CV-related deaths.

Based on electronic health records and on the hospital-based death certificates, the causes of death were distinguished between CV- and non-CV death. CV death events were defined as deaths occurring due to terminal HF and cardiogenic shock, acute myocardial infarction, arrhythmias, acute pulmonary embolism, cardiac tamponade, and acute cerebrovascular disease. Non-CV-related events were defined as deaths occurring due to respiratory failure, severe sepsis/septic shock, renal failure, and to bleeding with hemorrhagic shock.

The secondary endpoint was the length of hospital stay (LOS), calculated as the time from ED admission to hospital discharge or death.

Statistical analysis

Categorical variables were presented as numbers and percentages. Continuous normally distributed variables were presented as mean ± standard deviation, non-normally distributed data were presented as median (inter-quartile range), and binary or ordinal variables were presented as absolute frequency (%). Parametric variables were compared by the Mann–Whitney U test, whereas categorical variables were compared by the Chi-square test (with Fisher test if indicated). Significant variables at univariate analysis were entered into a multivariate logistic regression model to identify independent predictors for the outcomes. To avoid overfitting and overestimation of the parameters, the variables with high collinearity were excluded from the multivariate models. If possible, categorical variables were preferred to continuous. The single items composing cumulative variables (i.e., Charlson index) were excluded from the model to avoid redundancy. The results of the logistic regression analysis are reported as odds ratio (OR) (95% confidence interval). Survival analysis was performed according to the Kaplan–Meier approach.

All data were analyzed by SPSS v26® (IBM, NY, USA). A two-sided p value of 0.05 or less was considered statistically significant.

Statement of ethics

The investigation conforms to the principles outlined in the Declaration of Helsinki and was approved by the local ethical committee (IRB #0051814/19).