POS-180 EXTERNAL VALIDATION OF A CLINICAL SCORING SYSTEM TO PREDICT MORTALITY IN LEPTOSPIROSIS PATIENTS: A SINGLE-CENTRE RETROSPECTIVE STUDY

Abstract

Leptospirosis is a widespread and prevalent zoonotic disease with an incidence 10 times higher in tropical countries, such as the Philippines. It spreads through contact of the Leptospires via ingestion of contaminated water or soil. It may progress from a mild fever to severe multi-organ damage leading to mortality in 5-20%. Although it is a potentially fatal disease, it remains underreported with no reliable global incidence figures. Based on the Philippines’ Department of Health disease surveillance, there were 3,011 cases reported nationwide in 2019 with the majority (38%) located in the National Capital Region (NCR). This study aims to externally validate a clinical prognostication tool for risk of progression to mortality in symptomatic leptospirosis patients, thereby helping in the triaging of patients. The initial 2018 derivation cohort was composed of 383 patients seen from January to October 2018 at the National Kidney and Transplant Institute (NKTI). It then combined potential predictive factors based on history, initial physical examination and baseline laboratory results at the emergency room level. Logistic regression was performed to identify the independent predictors that resulted in mortality. By univariate comparison,13 risk factors (all with p <0.05) were then selected for the final scoring system model and assigned a score with a minimum of 0 to maximum of 18 points.This retrospective validation study of the clinical scoring system included 169 subjects admitted at NKTI from January to December 2019. All patients with complete medical records, aged over 18 years with a clinical diagnosis of leptospirosis based on NKTI guidelines were included. Receiver operating characteristic (ROC) analysis was used to assess the predictive ability of the scoring system for mortality progression. Sensitivity, specificity, predictive values, and likelihood ratios and its 95% confidence intervals were then computed via STATA 15.0. Model calibration in the external validation cohort was evaluated using the Hosmer–Lemeshow goodness-of-fit test. In this validation study, majority of the patients were male, in both the survivors (n=122) and non survivors (n=21). The mean age of included patients is 37 years old (range 19-79). The following factors based on p-value were shown to be significant predictors of mortality: hemorrhage (54.5%), tachypnea (63.6%), thrombocytopenia (77.2%), and acidosis (54.5%). The cutoff score of 4 points, based on Youden’s J-index, showed an 86.36% sensitivity and 62.59% specificity in predicting mortality. The highest mortality rate is seen for risk scores >9, where there is a 25.8% chance of mortality. The scoring system's ability to discriminate between subjects in different risk categories was assessed by C-statistic, and is equivalent to the area under the receiver operating characteristic curve (AUC). Based on the ROC curve analysis, the validation cohort achieved a good discriminatory AUC of 0.77. The clinical scoring system has moderate accuracy, is well calibrated, and has a sensitivity of 86% and specificity of 63% in predicting progression to mortality. The risk scoring system can be useful for immediate triaging for leptospirosis patients at the emergency room level.