Objectives To analyze the association between glycemia levels upon pediatric intensive care unit admission and mortality in patients hospitalized.MethodsA retrospective cohort of pediatric intensive care unit patients admitted to the Instituto Nacional de Salud del Niño between 2012 and 2013. A Poisson regression model with robust variance was used to quantify the association. Diagnostic test performance evaluation was used to describe the sensitivity, specificity, positive predictive value, negative predictive value and likelihood ratios for each range of glycemia.ResultsIn total, 552 patients were included (median age 23 months, age range 5 months to 79.8 months). The mean glycemia level upon admission was 121.3mg/dL (6.73mmol/L). Ninety-two (16.6%) patients died during hospitalization. In multivariable analyses, significant associations were found between glycemia < 65mg/dL (3.61mmol/L) (RR: 2.01, 95%CI 1.14 - 3.53), glycemia > 200mg/dL (> 11.1mmol/L) (RR: 2.91, 95%CI 1.71 - 4.55), malnutrition (RR: 1.53, 95%CI 1.04 - 2.25), mechanical ventilation (RR: 3.71, 95%CI 1.17 - 11.76) and mortality at discharge. There was low sensitivity (between 17.39% and 39.13%) and high specificity (between 49.13% and 91.74%) for different glucose cut-off levels.Conclusion There was an increased risk of death at discharge in patients who developed hypoglycemia and hyperglycemia upon admission to the pediatric intensive care unit. Certain glucose ranges (> 200mg/dL (> 11.1mmol/L) and < 65mg/dL (3.61mmol/L)) have high specificity as predictors of death at discharge.