Validating a linear regression equation using mid-upper arm circumference to predict body mass index

Abstract

Estimating body mass index (BMI) in hospitalised patients for nutritional assessment is challenging when measurement of weight and height is not feasible. The study aimed to validate a previously published regression equation to predict BMI using mid-upper arm circumference (MUAC). We also evaluated the proposed global MUAC cut-off of ≤24cm to detect undernutrition. We measured standing height, weight, and MUAC prospectively in a sample of stable patients. Agreement between calculated and predicted BMI was evaluated using Bland-Altman analysis. We studied 201 patients; 102 (51%) were male. Median (IQR age was 42 (29-50) years. 95% limits of agreement between predicted and calculated BMI were+0.6767 to+1.712 and the bias was+1.076. MUAC ≤24cm was 97% sensitive and 83% specific to detect undernutrition. BMI derived from MUAC had poor calibration for estimating actual BMI. However, low MUAC has good discriminative accuracy to detect undernutrition.