268 - Reliability and Accuracy of Bioelectrical Impedance Analysis for Measuring Body Composition in the Neonatal Intensive Care Unit

Poster Number: 268
Publication Number: 268.131

Katherine Bell, Brigham and Women's Hospital, Boston, MA, United States; Kaitlin Drouin, Harvard Medical School, Boston, MA, United States; Laurie P. Foster, Brigham & Women's Hospital, Boston, MA, United States; Jordan O'Brien, Brigham and Women’s Hospital, Boston, MA, United States; Andrea Klein, Brigham & Women's Hospital, Boston, MA, United States; Deirdre M. Ellard, Brigham and Women's Hospital, Boston, MA, United States; Hunter L. Pepin, Brigham and Women's Hospital, Boston, MA, United States; Mandy B. Belfort, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, United States

Background: Body composition is emerging as a useful measure of nutritional status for preterm infants, but its monitoring in the neonatal intensive care unit (NICU) is challenging because the current gold standard method—air displacement plethysmography (ADP)—is expensive, non-portable, and not suitable for infants on respiratory support. Bioelectrical impedance analysis (BIA) holds promise as a rapid, low-cost, portable method. Valid prediction equations for estimating fat and fat-free mass from BIA data exist for adults and school-age children, but little is known about the validity of BIA to estimate body composition for preterm infants.

Objective: Among very preterm infants in the NICU: 1) Determine the reliability of BIA measurements, and 2) Evaluate the accuracy of 5 published equations for estimating body composition from BIA data, as compared with ADP.

Design/Methods: In a cohort of infants born < 31 weeks’ gestation, we measured body composition by BIA and ADP concurrently at 2 timepoints (36 and 40 weeks postmenstrual age [PMA]). Infants unable to undergo ADP had BIA measured alone. At each timepoint, we performed 4 repeated BIA tests using the Quadscan 4000. We excluded BIA data not meeting quality standards (appropriate fall in impedance, as described by manufacturer). We determined intra-rater reliability of BIA raw data (resistance at 50 kHz [R50]) using intra-class correlation coefficient (ICC). Using the median R50, we estimated fat and fat-free mass using 5 published equations. We then used Pearson correlations and Bland-Altman analysis to compare fat and fat-free mass estimated by BIA to ADP results.

Results: 27 infants (70% male, mean gestational age 28.6 weeks) had a total of 184 BIA measurements (Table). Of these, 15 also underwent ADP (24 concurrent assessments). 84% of BIA measurements passed data quality inspection. Intra-rater reliability was excellent: ICC 0.98 (95% confidence interval [CI]: 0.96, 0.99). One published equation (Lingwood et al) was superior, showing high correlation of BIA with ADP (r=0.8 for both fat and fat-free mass; Figure 1). Relative to ADP, BIA overestimated fat-free mass (average bias 370g, 95% CI: 180g, 560g) and underestimated fat mass (Figure 2).Conclusion(s): BIA was feasible and reliable among very preterm infants in the NICU. Estimates of fat and fat-free mass at 36 and 40 weeks PMA were highly correlated with the gold standard ADP. Biased estimates of fat and fat-free mass relative to ADP suggest the need for preterm infant-specific BIA prediction equations. Overall, BIA is a promising approach to assess body composition is small, sick newborns.
Table
Figure 1