Clustering of Deceased Patients with COVID-19 in Iran Based on Clinical Features in Hospitalization

Abstract

Background: Effective clinical care and identifying susceptibility to COVID-19-related mortality requires rapid recognition of risk factors and their relations with disease outcomes. This study aimed to cluster and identify various subgroups of COVID-19 patients and examine the relationship between these subgroups and the causes of death. Methods: This retrospective study assessed the risk factors contributing to COVID-19 patients’ death (n = 128) by evaluating deceased patients' demographic, clinical, and laboratory features and clustering various subgroups of individuals to investigate any correlation. Results: The mean age of deceived patients was 69.7 years, and the majority of them were male (65.6%). The levels of blood urea nitrogen, creatinine, alkaline phosphatase, erythrocyte sedimentation rate, lactate dehydrogenase, and C-reactive protein were high at admission and increased during the hospitalization. Shortness of breath (68%) and cough (62.5%) were the most common symptoms, and hypertension (50.8%) was the most common comorbidity among deceased patients. The clustering quality based on the underlying disease and symptoms was not acceptable. However, clustering based on vital signs showed significant differences in body temperature, pulse rate, respiratory rate and oxygen saturation (P<0.001). Furthermore, disseminated intravascular coagulation (DIC) was significantly higher in patients with weaker vital signs than those with better vital signs (15.36% vs. 0.0% P = 0.002). Conclusion: Older age, male sex, hypertension, and high inflammatory markers might be the risk factors for COVID-19-related mortality. Furthermore, considering that patients with poor vital signs were susceptible to develop DIC, prevention of these consequences might be helpful in COVID-19 management.

Keywords

• Cluster Analysis; COVID-19; Mortality; Risk Factors