Energy Malnutrition Risk Factors Assessment

Data were represented as median (interquartile range (IQR)), range, or number. Values for differences between the energy malnutrition and non-energy malnutrition groups were analyzed by use of Wilcoxon rank-sum tests. In addition, logistic regression analysis was used to analyze the independent factors related to energy malnutrition. For the selection of candidates for logistic regression analysis, we used the single factor regression analysis (Spearman’s tests). Factors associated with energy malnutrition were profiled using decision tree analysis. A decision-tree algorithm is a data-mining technique that reveals a series of classification rules by identifying priorities and therefore allows clinicians to choose an option that maximizes the benefit for the patient [30 (link)]. Decision trees are a popular modeling technique in economics and clinical practice and have proved their usefulness in human medicine [31 (link),32 (link),33 (link)]. Finally, the Receiver Operating Characteristic (ROC) curve using the Youden index was used to determine the best cutoff value of independent factors to discriminate RQ < 0.85. All statistical analyses were conducted using Statistical Analysis Software (JMP Pro version 15.0; SAS Institute, Cary, NC, USA). The statistical significance level was set at p < 0.05.