Anthropometric markers for assessing nutritional status
The above study had demonstrated that both BMI and MAC were relatively good markers for assessing nutritional status. Then the ROC curve analysis was used to analyze whether the two markers could be better predictors for nutritional status using MQSGA as the reference standard (Figure (Figure2A,2A, A,2B).2B). The area under the curve (AUC) was 0.664 (95% CI = 0.531 - 0.798) for BMI (Figure (Figure2A),2A), and 0.726 (95% CI = 0.609 - 0.843) for MAC (Figure (Figure2B).2B). High level of BMI and MAC were shown to be the appropriate predictors for better nutritional status (P < 0.05). BMI with a threshold value of 21.19 kg/m2 provided 85.7% sensitivity and 52.8% specificity for the prediction of malnutrition. MAC with a threshold value of 25.05 cm provided 85.7% sensitivity and 41.5% specificity for the prediction of malnutrition.