Paediatric Risk of Mortality III Score to Predict Outcome in Patients Admitted to PICU with Dengue Fever

ABSTRACT

Background: The objective of the current study was to evaluate the ability of PRISM III score calculated within 24 hours of PICU admission to predict outcome in patients with dengue fever. Materials & Methods: The prospective cohort study included children admitted to PICU with diagnosis of ‘Dengue with warning sign’ and ‘Severe Dengue’. Outcome included PICU mortality, length of PICU stay (LOS), need for mechanical ventilation and renal replacement therapy (RRT). PRISM III score was calculated and compared with outcome groups. Calibration of the score was measured using Hosmer‑Lemeshow modification of chi square test and discrimination using Area under the curve of Receiver Operating Characteristic curves. Results: This study included 151 patients with 54.3% Dengue with warning signs and 45.7% severe Dengue. Median PRISM III-24 score of patients who died (p-0.001), required RRT (p-0.006), mechanical ventilation (p-0.032) and those with prolonged LOS (p-0.003) were significantly higher. Hosmer‑Lemeshow modification of chi square test to assess calibration showed good fit of PRISM III-24 model to predict mortality (χ2-2.022; p-0.846), need for RRT (χ2-3.564; p-0.614), prolonged LOS (χ2-4.360; p-0.499) and need for mechanical ventilation (χ2-7.497; p-0.186).  ROC curve for the PRISM III-24 model to predict the discriminating power yield an AUC of 0.923 (95% CI 0.829-1.000) for mortality, 0.953 (95% CI 0.896-1.000) for need for RRT, 0.682 (95% CI 0.494-0.870) for need for mechanical ventilation and 0.663 (95% CI 0.563-0.764) for prolonged LOS. Conclusion: PRISM III is an effective tool to predict mortality and need for RRT in patient with dengue fever