Two variants of the Nutritional Risk in the Critically Ill Score as predictors of mortality in Intensive Care Unit patients at a Brazilian University Hospital/Duas variantes do Escore de Risco Nutricional em Pacientes Criticos como preditoras de mortalidade em pacientes admitidos em uma Unidade de Terapia Intensiva de um Hospital Universitario Brasileiro

To evaluate the agreement between the modified version of the Nutritional Risk in the Critically Ill Score (without Interleukin-6) and a variant composed of C-Reactive Protein as well as its capacity to predict mortality. A prospective cohort study was carried out with 315 patients in an Intensive Care Unit of a university hospital from October 2017 to April 2018. The agreement between the instruments was evaluated using the Kappa test. The predictive capacity for estimating mortality was assessed with the Receiver Operating Characteristic curve. The critical patients involved in the study had a mean age of 60.8[+ or -]16.3 years and 53.5% were female. Most patients had C-Reactive Protein levels [greater than or equal to]10 mg/dL (n=263, 83.5%) and their admission in the Intensive Care Unit was medical (n=219, 69.5%). The prevalence of mortality was observed in 41.0% of the evaluated patients. The proportions at high nutritional risk according to Nutritional Risk in the Critically Ill without Interleukin-6 and with C-Reactive Protein were 57.5% and 55.6%, respectively. The tools showed strong and significant agreement (Kappa=0.935; p=0.020) and satisfactory performances in predicting mortality (area under the curve 0.695 [0.636-0.754] and 0.699 [0.640-0.758]). Both versions of the Nutritional Risk in the Critically Ill tool show a satisfactory agreement and performance as predictors of mortality in critically ill patients. Further analysis of this variant and the association between nutrition adequacy and mortality is needed. Avaliar a concordância entre a versão modificada do Escore de Risco Nutricional em Pacientes Críticos (sem Interleucina-6) e uma variante composta de Proteína C-Reativa, bem como a capacidade de ambas as versões para predizer mortalidade em pacientes críticos. Os pacientes apresentaram idade média de 60,8[+ or -]16,3 anos, dos quais 53,5% eram mulheres. A maioria dos pacientes apresentou níveis de Proteína C-Reativa [greater than or equal to]10 mg/dL (n=263; 83,5%). O tipo de admissão na Unidade de Terepia Intensiva foi clínica (n=219; 69,5%), sendo que a prevalência da mortalidade foi observada em 41,0% dos pacientes avaliados. O alto risco nutricional, avaliado pelo Nutritional Risk in the Critically Ill sem Interleucina-6 e pela variante com Proteína C-Reativa, foi demonstrado em 57,5% e 55,6% dos pacientes críticos, respectivamente. Os instrumentos demostraram concordância forte e significativa (Kappa=0,935; p=0,020) e desempenho