P0017 / #879: TOOL VALIDATION FOR NEUROLOGICAL PROGNOSIS IN PEDIATRIC ICU

Abstract

Aims & Objectives: To validate a prediction tool for favorable neurologic outcome and identify the possibility of its use as a new PICU quality indicator. Determine risk factors that impair neurological prognosis in order to reduce new morbidities of survivors. Methods: Retrospective cross-sectional study by reviewing medical records of patients aged 1 month to 12 years incomplete. A 22-variable database was constructed and patients were classified according to baseline neurological level by PCPC (Pediatric Cerebral Performance Category) at admission and discharge. Used tool discrimination and calibration method, based on expected and obtained favorable prognoses. The favorable prognosis was defined in two classifications: a) primary: no change, 1-level decline in PCPC or improvement, between admission and discharge from PICU, and b) secondary: no change or improvement in PCPC between admission and discharge from PICU. Results: A total of 714 patients from two southern PICUs were included in a 18-month period, with 11.4% of deaths and 16.2% of patients with PCPC decline. The tool presented good calibration for both definitions, with area under the ROC curve of 0.91. Days on mechanical ventilation, shock, stroke, cardiopulmonary resuscitation, and seizures are independent factors associated with unfavorable neurologic outcome. Conclusions: The prediction tool for favorable neurologic outcome demonstrated good calibration and discrimination with both definitions, able to predict the neurological prognosis of PICU patients.