Estudo de validação de ferramenta preditora de prognóstico neurológico favorável para pacientes pediátricos em cuidados intensivos

Abstract

INTRODUCTION: The mortality of patients in need of intensive care has been constantly decreasing in recent years. In proportion to this drop, we observed an increase in cases of patients who survive severe insults, but with some associated morbidity. There is no consensus on what are the risk factors for this outcome, however it is known that there may be an association with the initial disease, complications or some therapies instituted during hospitalization. The fact is that morbidity, especially neurological, is a limiting condition for the patient, as it affects the quality and perspective of life of these children, as well as their family structure. OBJECTIVE: To translate, cross-culturally adapt and validate a predictive tool for favorable neurological prognosis. Identify risk factors associated with neurological prognosis. METHODS: the first phase of the study included translation, cross-cultural adaptation and content validation of the instrument. In the second phase, the use of the tool was validated with a historical cohort based on the review of medical records of patients aged between 1 month and 12 years old who were admitted to two Pediatric Intensive Care Units (PICUs) in southern Brazil, between November 2016 and February 2018. The tool to be validated can estimate the probability of a favorable neurological outcome. To define this outcome, the Pediatric Cerebral Performance Category (PCPC) is used, a scale that classifies the patient according to their neurological status. The two possible definitions for the outcome are: a) primary: no change, with a 1-level decline in PCPC or improvement, between admission and discharge from the PICU and b) secondary: without change or with improvement in the PCPC between admission and discharge from the PICU, the option for one or the other definition changes the result. The tool consists of twenty variables, segmented in definition of the outcome, general data, value of the PIM-2 mortality score, diagnoses and interventions. In the Diagnostics section, it contains acute respiratory failure, shock / hypotension, seizures, cardiac arrest, renal insufficency, stroke, trauma and acquired heart disease. Items such as the use of nitric oxide, indication of oxygenation by extracorporeal membrane (ECMO), type of mechanical ventilation used, time in days of mechanical ventilation and hospitalization in the unit, comprise the Interventions section. Each patient was classified according to the PCPC and all the variables mentioned above were collected in two moments: admission and discharge from the PICU. Based on the estimated and observed probabilities of the outcome, performance, discrimination, calibration and goodness adjustment test of the tool were evaluated. RESULTS: The tool was translated without disagreement, with few adaptations for the Brazilian context. The translation obtained a total Content Validation Coefficient for the 0.92 scale, thus confirming the validation of the translation. In the second phase, 653 patients were included, totaling 1306 records in the database, since each patient had one record referring to the moment of admission and the other referring to discharge. In the sample, males predominate, with a median of 23 months (IQ 7-69). Most hospitalizations were due to clinical pathologies (60%), mainly respiratory, such as acute viral bronchiolitis and pneumonia, followed by seizures and sepsis. The rate of patients who declined by at least one level of the PCPC was 9%. As for risk factors, it was found that stroke (p <0.01), seizures (<0.01), mortality score PIM-2 at admission (p = 0.04), shock / hypotension (p < 0.01) and days on mechanical ventilation (p <0.01) are associated with the neurological outcome. There was good performance, discrimination, calibration and adjustment of the tool in the sample. CONCLUSION: The predictive tool for favorable neurological outcome was validated with good results in the sample, demonstrating the instrument has external validity.