ABSTRACT
OBJECTIVES: For population-level screening of malnutrition among adults-especially in developing-country settings-the body mass index (BMI) can be impractical because of logistical requirements for weight and height measurement. We analyzed anthropometric data collected from a large-scale nutritional survey on women of rural Bihar to determine the mid-upper arm circumference (MUAC) cutoffs corresponding to standard BMI cutoffs and the predictive accuracies of the determined cutoffs. STUDY DESIGN: It was a cross-sectional study using multistage cluster sampling. METHODS: The current analysis used anthropometric data from a study on dietary practices of rural women (adolescents, lactating mothers, and women in the interpregnancy period). The MUAC (cm) cutoffs corresponding to four standard BMI (kg/m2) values were determined using receiver operating characteristic (ROC) curve analysis. RESULT: We detected a significant positive correlation between BMI and MUAC (r = 0.81, P < 0.0001). In ROC curve analysis, the MUAC cutoffs corresponding to BMI cutoffs of 18.5, 23, 25, and 30 kg/m2 were estimated to be 23.2, 26.0, 27.3, and 30.5 kg/m2, respectively. The predictive accuracy of the determined cutoffs was good, as indicated by the area under the ROC curve for the four different cutoffs-which ranged between 88% and 97%. Other than the cutoff for 'obese' (BMI, 30 kg/m2), the Kappa coefficients for the rest of the MUAC cutoffs showed 'substantial' agreement (>0.6) with their BMI counterparts. CONCLUSION: The results suggest that the cutoffs based on MUAC-a less resource-intensive measure than BMI-can be used for community-based screening of malnutrition among women of Bihar.
