139 Body composition measurement methods in preterm infants 7 SFT measurements Table 4 describes the predictive value of SFT. Schmelzle et al. showed that the sum of SFT measured at four sites had a good predictive value for fat mass (g) in a study population that included both preterm as well as term infants.26 However, only 10 out of 104 infants in this study population were born preterm. Koo and colleagues also assessed the predictive value of SFT in a mixed population of term and preterm infants (respectively n = 68 and n = 52). They reported that SFT, when added to weight and length, explained an additional 13% of the variance in fat free mass (g).17 Thus, SFT had a poor predictive value for fat free mass (g). Schmelzle et al. as well as Koo et al. did not assess the predictive value of SFT for fat (free) mass percentage. In contrast, Daly-Wolfe et al. did and reported a poor predictive value of SFT for fat mass percentage (see table 4).24 Bioelectrical impedance measurements Table 5 describes the predictive value and validity of bioelectrical impedance measurements. The impedance index (height2 in cm2/impedance in Ω) measured with BIA adds little to the variance in fat-free mass already explained by weight.27 (see table 5) Indeed Raghavan et al. reported that the least bias was obtained when weight alone was used to estimate TBW.28 Table 5 shows that models used to estimate body composition based on the impedance index alone showed poor agreement. In contrast, the predictive equation by Dung et al. based on weight and the impedance index showed good agreement.27 (see table 5) Table 6 describes the predictive value of ultrasound measurements. Ultrasound measurements of muscle and fat mass showed high reliability (intraclass correlation coefficient 0.874 - 0.975; technical error of measurements 0.251 - 0.628 mm), but had a poor predictive value for fat mass percentage measured by ADP. 29 (see table 6) Table 7 describes the predictive value of MRI. Dyke and colleagues assessed the accuracy of body composition measured with rapid whole body MRI. Repeated scans showed good agreement of fat mass percentage (95% limits of agreement 1.3%). 30 However, body compositionmeasurement was not compared to a reference method. (see table 7) Table 8 describes the validity of ADP. Compared to isotope dilution, ADP showed good agreement when measuring fat free mass density. 31 Forsum et al. and Roggero et al. demonstrated that there was a small bias in the measurement of fat mass percentage.31,32 Nevertheless, the limits of agreement were relatively wide and thus there was poor agreement between fat mass percentage measured with ADP compared to fat mass percentage measured by isotope dilution. (see table 8)
RkJQdWJsaXNoZXIy MTk4NDMw