Gabriela Guerron-Gomez, Eduardo Tuta-Quintero, Alirio Bastidas, Luis F Giraldo-Cadavid, Maria Pérez-Escobar, Luisa F Martínez, Maria Castillo-Páez, Isabella Criado-Quintero, Manuela Trujillo-Herrera, Angie Sandoval-Blanco, Gabriela Osorio-Betancourt, Laura Medellín-Ortiz, Laura Chaves-Pauwels, Paola Martínez-Sáenz, Luis F Reyes
The new score demonstrated weak-to-moderate discriminatory capacity. The variables included in the new score reflect multiorgan involvement, disease severity, and comorbidity burden, all of which are associated with long-term mortality.
BACKGROUND: Community-acquired pneumonia (CAP) causes significant long-term morbidity and mortality. Existing clinical scores focus on short-term outcomes, highlighting the need to validate tools that accurately predict 12-month mortality in hospitalized patients.
MATERIALS AND METHODS: A retrospective cohort of adults hospitalized with CAP from 2012 to 2020 was analyzed. Clinical, laboratory, radiological, and hospitalization-related data were collected. A Cox proportional hazards model was developed to predict post-acute mortality between 30 days and 12 months after hospital admission among patients with CAP who survived the first 30 days, with the cohort split 50:50. Model performance was evaluated using Area Under the Receiver Operating Characteristic Curve (AUROC) and standard diagnostic accuracy metrics.
RESULTS: A total of 13,851 patients with CAP were included. In the derivation cohort, independent predictors were altered mental status (HR 2.08; 95% CI 1.49-2.90; p < 0.05), elevated BUN (>30 mg/dL; HR 1.98; 95% CI 1.48-2.63; p < 0.05), temperature extremes (<35 °C or >39.9 °C; HR 1.98; 95% CI 1.43-2.74; p < 0.05), corticosteroid use (HR 1.85; 95% CI 1.40-2.44; p < 0.05), neoplasia (HR 1.58; 95% CI 1.08-2.32; p < 0.05), and hospital stays longer than 8 days (HR 1.39; 95% CI 1.06-1.82; p < 0.05). The new score achieved the highest AUROC (0.69; 95% CI: 0.66-0.73), followed by PSI (0.65; 95% CI: 0.62-0.69), CURB-65 (0.63; 95% CI: 0.59-0.67), and CAPSI (0.62; 95% CI: 0.58-0.67). The optimal cutoff point for the new score was 3, as determined by the Youden index (0.339). The model sensitivity was 83.0%, specificity: 50.9%, PPV: 8.9%, and NPV: 99.0%. The LR+ was 1.69 (95% CI: 1.38-2.06), and the LR- was 0.33 (95% CI: 0.27-0.41).
CONCLUSIONS: The new score demonstrated weak-to-moderate discriminatory capacity. The variables included in the new score reflect multiorgan involvement, disease severity, and comorbidity burden, all of which are associated with long-term mortality.