Performance of subjective global nutritional assessment in predicting clinical outcomes: data from the Brazilian survey of pediatric oncology nutrition

Abstract

Background: Methods for assessing nutritional status in children and adolescents with cancer is a difficult in clinical practice. The study aimed to evaluate the performance of Subjective Global Nutritional Assessment (SGNA) in predicting clinical outcomes in children and adolescents with cancer in Brazil. Methods: This was a prospective cohort multicenter study. It was included 723 children and adolescents with cancer aged 2–18 years between March 2018 and August 2019. Nutritional assessment was performed according to World Health Organization recommendations and using SGNA within 48h of hospitalization. Unplanned readmission, length of hospital stay, and post-discharge death were analyzed. Cohen’s kappa coefficient was used to ascertain the agreement between body mass index for age (BMI/A) and SGNA. The sensitivity, specificity, positive and negative predictive values, and accuracy of SGNA were estimated. Odds ratios (ORs) with 95% confidence intervals (CIs) were evaluated using multiple logistic regression. Results: The mean patient age was 9.4 ± 4.9 years. SGNA showed that 29.7% (n = 215) and 6.5% (n = 47) patients had moderate and severe malnutrition,respectively. Considering the concurrent validity criterion, SGNA had an OR (95% CI) of 6.8 (3.1–14.9) for predicting low and very low weight for age at admission, with a sensitivity and specificity of 72.4% (59%–82.1%) and 72% (64.2%–78.9%), respectively. SGNA could predict death in children with severe/moderate malnutrition, with an accuracy of 63.8% (63%–65.1%). Logistic multivariate analysis showed that the adjusted effect of death; hematological tumor; living in the northeast, southeast, and midwest regions of Brazil; and older age was associated with malnutrition according to SGNA. Conclusion: Based on concurrent validity between SGNA and anthropometry, SGNA performed well and had a good ability to predict death in Brazilian children with cancer.