Validity of BIA prediction equations in determining the fat mass of 11-13 year old Sri Lankan girls

Abstract

INTRODUCTION AND OBJECTIVES: Bioelectrical impedance analysis (BIA) is a simple body composition assessment method, based on use of prediction equations. Validation of equations for the specific populations is important for accurate assessment. This study aimed to determine the validity of available BIA equations in assessing the fat mass (FM) in Sri Lankan girls aged 11-13 years. METHODS: Forty-six 11-13 year-old healthy school girls were purposively selected. Weight, height and impedance using BIA were measured. Total body water was determined and FM was derived through the criterion Deuterium dilution technique. Twelve BIA prediction equations applicable to the age and sex were identified from literature. Predicted FM calculated according to each equation was compared with reference FM (assessed through isotope dilution), and validity was assessed using correlation coefficients, paired samples T-test and Bland-Altman plots. RESULTS: FM predicted by all twelve equations was significantly correlated (r>0.93, p<0.05) with reference FM. Mean (±SD) bias of predicted FM ranged from -5.32 (±1.79) kg to 5.8 (±2.1 l) kg. Only four equations predicted mean FM values that were not significantly different from the mean reference FM values, the mean bias (±SD) ranging from -0.21 (±2.23) kg to 0.06 (±l.72) kg. Of these four prediction equations, only one had a symmetric, uniform distribution of error within the ±l .96 SD limits in the Bland-Altman analysis. CONCLUSION: Most available BIA prediction equations are unsatisfactory for use in the local context. Cross validation of existing prediction equations before use or development of BIA prediction equations to suit the local populations is recommended.