Risk stratification of dyspnoeic patients: prognostic implications for all-cause mortality and heart failure readmission

Abstract Background Acute dyspnoea is a leading cause of hospitalisation, often due to heart failure (HF). Within a year, over half of patients admitted due to acute dyspnoea die or require hospital readmission. Yet, prognoses vary widely. Some are terminally ill and require end-of-life care, while others have conditions that respond well to prompt intervention and, particularly for HF patients, guideline-directed therapy (GDMT). We lack objective tools to categorise these diverse patient needs. Purpose We developed a risk stratification model using available clinical, biochemical, and radiological markers to classify dyspnoeic patients into three distinct prognostic groups: Life-threatening, urgent and non-urgent. Methods This prospective observational study included 233 consecutive emergency patients aged 50 years with acute dyspnea. All underwent immediate chest radiography, lung ultrasound, chest CT, echocardiography, and biomarker assessment. We excluded patients requiring intensive care, those with acute coronary syndrome, and those unable to consent. Using backward selection in a multivariable Cox model, we identified key predictors of 365-day mortality from study examinations. We defined the life-threatening group as those with a predicted 365-day mortality risk 50%. After excluding the life-threatening group, we built a second Cox model to assess the composite risk of death or HF readmission (median follow-up: 2.3 years, range: 0–3.4). Using model coefficients, we calculated a risk score, and patients with a score above the median were classified as urgent, while those at or below were non-urgent. Results Mean age was 74.2 years; 79 (34%) patients died within 365 days. The 365-day mortality risk score (AUC 0.75) included troponin T, echocardiographic dilated or compressed inferior vena cava, known chronic obstructive pulmonary disease (COPD), age, and systolic blood pressure (Pict. 1). The life-threatening group (N=72, 50% predicted risk) had a hazard ratio of 3.68 for 365-day mortality compared to the remaining patients. The second analysis included the remaining 161 patients, of whom 61 (38%) experienced the composite endpoint of death or HF readmission. In this model (AUC 0.74), CT pleural effusion was the strongest predictor, followed by COPD, troponin T, eGFR, and age (Pict. 1). A risk score above the median (-0.02; range: -1.60 to 1.66) defined the urgent group (N=81) with hazard ratio of 3.65 vs. the non-urgent (N=80) for the composite endpoint. When applied to the full dataset, this model accurately identified the three distinct prognostic groups (Pict. 2). Conclusion This novel risk stratification tool incorporates readily available biomarkers to identify three risk groups with potentially distinct needs after initial stabilisation: a non-urgent group suitable for routine workup and GDMT, an urgent group requiring in-hospital diagnosis and GDMT, and a life-threatening group, for whom palliative care should also be considered.