Abstract
OBJECTIVE: This study aimed to derive and validate a prognostic scoring system to identify patients with hematological malignancies (HMs) and sepsis who have a high mortality rate. METHODS: Cohorts for derivation and validation were created from all data. Using univariate and multivariate analysis, the independent variables connected to 28-day mortality in the derivation cohort were found. A receiver operating characteristic (ROC) curve was used to compare the predictive power and determine their cutoff points. These risk variables were given a score weighted by risk prediction function, and a new scoring system was also developed. The area under the ROC curve (AUROC) and sensitivity and specificity for mortality of the risk category of the new scoring system were compared with Sequential Organ Failure Assessment (SOFA) score. RESULTS: 90 (45.22%) of the 199 patients passed away within 28 days. Ninety-nine patients made up the derivation cohort, with 47 (47.47%) fatalities. Ages in the non-survival group were higher (61.47 ± 14.53 vs 55.13 ± 15.66) than in the survival group. As independent predictors of death, multivariable analysis identified SOFA score (OR 1.442, 95% CI 1.035, 2.009), age (OR 1.242, 95% CI 1.026, 1.503), and prothrombin time (PT) (OR 1.213, 95% CI 1.030, 1.430). The AUROC with 95% CI of the new scoring system and its sensitivity and specificity to mortality were virtually all superior to SOFA score in both derivation and validation cohorts: AUROC (0.757 vs 0.716), Sensitivity (75 vs 67.3%), and Specificity (68.1% vs 63.8%) are the Derivation cohort; Validation cohort: Sensitivity (91.2% vs 84.2%), AUROC (0.792 vs 0.733), and Specificity (58.1% vs 58.1%). The model was correctly calibrated, according to the Hosmer-Lemeshow test. CONCLUSION: The new scoring system was more accurate in predicting 28-day mortality among patients with HMs and sepsis than the SOFA score.