Proefschrift Kerklaan

Chapter 3

Table 1b. Ventilator settings and metabolic measurement data for all children and for subgroups based on compatibility of the indirect calorimetry (IC) and ventilator method (defined as a difference ≤10% or >10% between IC and ventilator-derived VCO 2 values) All children P-value

Difference ≤ 10% N=18

Difference >10% N=23

N=41

Ventilation mode Support Control

%

0.94 c

34.1 65.9

33.3 66.7

34.8 65.2

FiO

Median (IQR) 0.26

0.28 (0.21-0.36)

0.25 (0.21-0.30)

0.57 b

2

(0.21-0.31)

Tidal volume (ml) Inspiratory Expiratory

Median (IQR)

89 (34-157) 91 (34-180)

142 (96-185) 143 (95-198)

36 (20-90) 35 (19-97)

0.002 b 0.001 b

Tube leak (%) (inspiratory-expiratory tidal volume)

Mean ± SD -1.9 ± 8.6 -4.8 ± 10.2 0.3 ± 6.4 0.07 a

Respiratory rate (/min)

Median (IQR) 38 (25-47)

26 (23-35)

43 (38-50) 128 (101-142)

0.02 b 0.27 b

Heart rate (/min)

Median (IQR) 116

113 (99-124)

(103-139)

LOS at moment of measurement (days)

Median (IQR) 2 (1-4)

2 (1-4)

2 (1-4)

0.75 b

Length of measurement (min)

Mean ± SD

30 ± 9.5

34 ± 8

33 ± 11

0.71 a

VCO

2 Deltatrac (ml/min)

Median (IQR) 75.2

108.6 (94-132.9) 111.6 (94.8-135.9)

45.9 (28.3-78.6) 33.4 (12.6-75.3)

<0.001 b

(41.6-116.1)

VCO

2 Servo-I (ml/min)

Median (IQR) 74.0

<0.001 b

(27.5-118.0)

P-values are calculated for differences between the two subgroups using the following tests: a Independent samples t-test, b Mann-Whitney test, c Chi-square test. IQR= Interquartile Range, FiO 2 = Fraction inspired Oxygen, LOS= Length of Stay, SD= Standard Deviation

Correlationbetween Servo-I®-derivedVCO 2 values and IC-derived valueswas excellent [ρ=0.965 (95% CI: 0.935 to 0.981)]. However, despite being a necessary condition for agreement, high correlation does not automatically imply good agreement. Figure 1a shows the Bland-Altman plot for agreement between the two methods for measuring VCO 2 values. Measurements were not comparable with a mean relative difference of 15.6% (p=0.002, one-sample t-test), due to underestimation of VCO 2 values by the Servo-I®. The 95% limits of agreement were wide (-42.4 % and 73.6%). Linear regression of Servo-I®-derived VCO 2 values on IC-derived values showed a regression line with a slope of 1.15 (95% CI: 1.07 to 1.23) and an intercept at -15.66 (95% CI: -23.32 to -8.01). The slope of this regression line was significantly different from 1 (p<0.001), and the intercept was significantly different from 0 (p<0.001), reflecting both proportional and fixed bias.

60