Comparação de duas ferramentas de triagem nutricional para predição de desfechos clínicos negativos em pacientes com excesso de peso hospitalizados

Abstract

The application of nutritional screening tools allows detecting risks early. In general, the association of nutritional risk with some negative clinical outcomes is already well established among adult and elderly hospitalized patients, but this relationship is still not well understood when overweight individuals are specifically assessed. Thus, the aim of this study was to compare two nutritional screening tools for assessing nutritional risk and their relationship with the occurrence of negative clinical outcomes in hospitalized overweight patients. This is a retrospective study conducted in a medium-sized hospital, which included patients ≥ 20 years old, admitted to the medical and surgical clinics between July 2017 and December 2019. Data from patients with Body Mass Index (BMI) ≥ 25.0 kg / m² for adults and> 27.0 kg / m² for the elderly and who had medical records of the results of two nutritional screening tests: Nutritional Risk Screening-2002 (NRS-2002) and Nutritional Mini Assessment Reduced (MAN-R). Also, sociodemographic, clinical, and anthropometric characteristics of the patients and clinical outcomes (prolonged hospital stay - ≥3rd quartile, readmission during the study period, and death in hospitalization and during the study period) were obtained. The Kappa index assessed the agreement between NRS-2002 and MAN-R, and the differences between groups at risk and without nutritional risk were analyzed using the Chi-Square and Mann-Whitney tests. The performance of the tools in predicting outcomes was analyzed using characteristic operating curves of the receiver (ROC) (p <0.05) for the study population and the elderly subgroup. Data were collected from 643 of these patients, 56.7% (n=387) female, with a median age of 62.0 (47.0-76.0) years [54.9% (n=353) elderly] and median BMI of 29.4 (27.6-32.4) kg / m² The prevalence of nutritional risk among overweight patients was 17.7% (n=114) by NRS-2002 and 36.1% (n = 232) by MAN-R and agreement of k=0.390 (p<0.001). In the elderly, the prevalence was 26.9% (n=95) by NRS-2002 and 44.1% (n=156) by MAN-R and agreement of k = 0.383 (p <0.001). All negative clinical outcomes were significantly more frequent among patients at nutritional risk using both tools (p <0.05). The area on the ROC curve of MAN-R (0.638) was significantly larger (p=0.023) than that of NRS-2002 (0.585) for the prolonged hospital stay. Only MAN-R obtained a significant percentage of prediction (p <0.05) for readmission (57.2%) and death during hospitalization (65.7%). For mortality throughout the study, the area under the ROC curve was significant and similar for the two tools [MAN -R (0.605); NRS-2002 (0.607)] (p=0.057). MAN-R was able to detect a higher proportion of risk among hospitalized overweight individuals and predict all negative clinical outcomes in the total population compared to NRS-2002 according to the area on the ROC curve.