Abstract
Body mass index (BMI) kg/m2 is a key screening tool for under-nutrition in adults, but difficult to obtain in immobile or unwell patients, particuarly in low resource settings, due to inability to accurately measure both weight and height. Mid-upper arm circumference (MUAC) is used to assess under-nutrition in children under 5 years but no standardised cut-off values exist for adults. In a cohort of adult Filipino patients admitted to a tuberculosis ward we assessed (i) cut-offs for MUAC to predict moderate under-nutrition (BMI <17kg/m2), (ii) the performance of limb lengths to predict height and; (iii) associations of body fat percentage from skinfolds and hand grip-strength with BMI. In 303 patients with MUAC and BMI at admission, aged 18-80 years (mean = 45.5, SD:14.8), BMI ranged from 11.2-30.6 kg/m2 and 141 (46.5%) had BMI <17.0 kg/m2. Using receiver operator curves, MUAC cut-offs were identified as <20.5cm for males (sensitivity: 89%, specificity: 84%) and <18.5cm for females (sensitivity: 91%, specificity: 89%), for BMI<17.0 kg/m2. Using published equations, knee height had the lowest mean difference between predicted and measured heights compared to ulnar or demi-span: (-0.98 cm, 95% CI: -1.51/-0.44). Both grip-strength and body fat percentage were positively associated with BMI, in separate linear regression models with exposure-age-sex interactions (adjusted-R-squared values: 0.15, 0.66, respectively). MUAC can predict moderate acute under-nutrition with high positive predictive value. Further research is required to determine the performance of alternative measures to BMI to predict mortality or adverse outcomes in acutely unwell patients.