64 / 208
Chapter 3
62
A
B
C
Figure 1.
Bland-Altman plots for agreement a) between VCO
2
values derived from indirect calorimetry
(IC) and Servo-I® measurements with a mean relative difference of 15.6% (panel A), b) between REE values
derived from IC and Servo-I® measurements with a mean relative difference of 19% (Panel B), and relative
differences between VCO
2
values derived from IC (Deltatrac®) and Servo-I® measurements by weight (kg)
(panel C). The mean relative difference is represented by the solid horizontal line; dotted lines indicate
95% limits of agreement (panels A and B) or 10% difference between methods (panel C)
In the 20 children weighing 15 kg or more, the agreement of the REE values on the Bland-
Altman plot was considerably better. The mean relative difference between methods was 1.3%
(p=0.668), with 95% limits of agreement of -24.6% and 27.2% (Fig. 2a).
This mean relative difference in children ≥15 kg was significantly smaller than that of the
Schofield-weight (-13.4%, p=0.03), Schofield-weight/height (-13.2%, p=0.03) andWHO (-15.0%,
p=0.07) equations. As shown by the Bland-Altman plots (Fig. 2), limits of agreement were
narrowest for Servo-I® based REE when compared with the REE based on predictive equations.
In children ≥15 kg, the proportion of comparable measurements (difference ≤10% with IC-
derived REE) was highest, whereas the proportion of measurements with a difference >30%
was lowest for the Servo-I® based REE, when compared with proportions of the equations-
based REE (Fig. 3).