Prediction of Mortality From Symptom Cluster Patterns Among Patients Hospitalized With Decompensated Heart Failure: A Multi-Institutional Database Study

BACKGROUND: Whether symptom clusters predict in-hospital death and subsequent rehospitalization or death is unknown in patients with heart failure (HF) in Taiwan. Our aims were to (1) identify symptom clusters in patients with HF hospitalized for an acute exacerbation of HF, (2) determine whether these symptom clusters predicted in-hospital death, and (3) determine whether these symptom clusters predicted HF rehospitalization and all-cause death during a 5-year follow-up period. METHODS: We used the electronic medical record database from Chang Gung Memorial Hospitals in Taiwan. We included 8161 patients admitted to hospitals for HF exacerbations between 2008 and 2013 and followed cardiac outcomes of each patient for 5 years. Cluster analysis was used to identify symptom clusters. Logistic regression was used to determine in-hospital death, and Cox regression was used to predict HF rehospitalization and all-cause mortality. RESULTS: We identified and named 4 distinct symptom clusters (cough, congestion, perfusion, and multiple symptoms) characterized by a high incidence of cough, edema, dyspnea, and many symptoms, respectively. The crude rates of in-hospital death in the cough, congestion, perfusion, and multiple symptoms clusters were 1%, 5.6%, 7.2%, and 33.6%, respectively. Membership in the multiple symptoms cluster significantly predicted higher risk of in-hospital death, rehospitalization for HF, all-cause death, and the composite outcomes of rehospitalization and all-cause death compared with those in the other 3 symptom clusters. CONCLUSIONS: Symptom clusters predict in-hospital death and subsequent cardiac events. Monitoring and managing these clusters promptly could potentially decrease mortality rates and prevent unnecessary hospitalizations in patients with HF.