125 DXA, ADP & SFT in preterm infants 6 Figure 3. Bland-Altman plot of fat mass percentagemeasured by DXA compared with fat mass percentage estimated based on the formula by Schmelzle and Fusch [18]. DXA dual-energy X-ray, ΣSFT sum of skinfold measurements Average of fat mass percentage measured with DXA and estimated with the formula is depicted on the x-axis and the difference between the fat mass percentage measured with DXA and estimated by the formula is depicted on the y-axis. Mean difference: 4.9 ± 5.4%, lower limit of agreement − 5.6%, upper limit of agreement 15.5%, maximum allowed difference 2% Predictive model for fat mass measured with ADP Within our cohort, fat mass, measured with ADP, could be estimated with gestational age, waist circumference, head circumference, weight SDS, head circumference SDS and the ∑SFT: Fat mass (g) = -3013.0 – 9.4*gestational age + 39.1*waist circumference + 65.9*head circumference + 67.6*weight SDS – 59.3*head circumference sds + 15.1*∑SFT (Gestational age in weeks, waist and head circumference in mm and ∑SFT in mm). These factors explained 72% of the variance (R2 = 0.716, S.E. of the estimate 138.4 g, p <0.001). Fourty-two percent of the fat mass percentage measured with ADP could be explained by the ΣSFT and waist circumference(R2 =0.426, S.E. of the estimate 3.1 % <0.001) In multivariate analysis other potential confounders, were found to not be significant.
RkJQdWJsaXNoZXIy MTk4NDMw