External validation of the clinical prediction score for predicting 28-day mortality of older sepsis patients in emergency department

Background: Sepsis in older patients is associated with a high mortality rate, presenting a considerable clinical challenge. Existing mortality prediction scoring systems have demonstrated limited accuracy in this population. To address this limitation, the Ramathibodi Older Sepsis Score (ROSS) was developed. However, external validation is necessary to assess its efficacy in predicting 28-day mortality. Objective: This study aimed to validate the Ramathibodi Older Sepsis Score (ROSS) for predicting 28-day mortality in older sepsis patients in the emergency department (ED). Methods: Data for the development cohort were retrospectively collected from August to December 2018, while data for the validation cohort were collected from January to June 2022. Seven prespecified prognostic factors for 28-day mortality were used to calculate a predictive score. Results: A total of 500 older sepsis patients were included in the validation cohort, and 599 patients were included in the development cohort. The predictive ability of the ROSS model in the validation cohort (Area under receiver operating characteristic curve; AuROC: 0.69, 95% CI: 0.61–0.77) decreased compared to the development cohort (AuROC: 0.87, 95% CI: 0.82–0.92); P<0.01. This performance was compared with other scoring models: SIRS (AuROC: 0.50, 95% CI: 0.42–0.58; P < 0.01), qSOFA (AuROC: 0.70, 95% CI: 0.64–0.77; P = 0.75), NEWS (AuROC: 0.68, 95% CI: 0.60–0.76; P = 0.81), and REWS (AuROC: 0.66, 95% CI: 0.57–0.75; P = 0.51) Conclusion: The external validation of the ROSS demonstrated moderate performance in predicting 28-day mortality in older sepsis patients, with AuROC values similar to qSOFA and NEWS.