Clinical and CT image features for survival prediction in severe pneumonia during the SARS-CoV-2 Omicron wave

Background Identifying prognostic factors for severe COVID-19 pneumonia during the Omicron wave remains crucial for early risk stratification and improving patient outcomes. This study aimed to identify and analyze key clinical and CT imaging features associated with survival in patients with severe pneumonia caused by the SARS-CoV-2 Omicron variant. Methods This retrospective study included patients presenting to the emergency department of Shandong Provincial Hospital (December 2022–January 2023) with confirmed SARS-CoV-2 Omicron infection and severe pneumonia. Clinical/laboratory data and CT imaging features were systematically collected and evaluated. Patients were randomly divided into training (70%) and validation (30%) cohorts. Univariate and multivariate analyses were rigorously applied to identify significant baseline clinical and CT imaging features associated with survival. A predictive nomogram was constructed based on the selected feature combination. Results Among 1,739 COVID-19 patients, 151 (8.68%) had severe pneumonia (median age 75, 70.1% male). Multivariate logistic regression analysis identified a critical combination of features independently associated with survival: CT findings of pleural effusion ( p = 0.008) and cardiac enlargement ( p = 0.008), along with clinical/laboratory factors including reduced baseline pulse oxygen saturation ( p = 0.034), elevated SAA ( p = 0.020), elevated GLU ( p = 0.022), and reduced Ca concentration ( p = 0.029). The nomogram integrating these combined features demonstrated good predictive performance for in-hospital mortality (AUC: training cohort 0.914, validation cohort 0.802). Conclusion This study identifies a distinct combination of clinical and CT imaging features (pleural effusion, cardiac enlargement, low SpO2, elevated SAA, elevated GLU, low Ca) as key independent prognostic factors for survival in severe Omicron pneumonia. The predictive tool based on this feature combination shows significant clinical utility. These preliminary findings provide critical insights for early risk assessment and targeted management, facilitating improved patient prognosis.