Facebook
Twitter
LinkedIn
Email
 |
Checking for direct PDF access through Ovid | |
Excerpt
We are grateful for the comments on our recent article by Huang et al (1). The issue of spontaneous variability for the measurements of arterial blood gases has been addressed in the past by Thorson et al. (2), Hess and Agarwal (3), and Sasse et al. (4) and for the measurement of cardiac output (CO) by Sasse et al (5). All these previous articles calculated their intraclass correlation by averaging the coefficients of variation (CV) from individual subjects. To express the central location and the CV dispersion of the CV distribution in our study and to compare our results with the results of Sasse et al. (5), we, instead of taking the random effects into account, also derived our spontaneous variability of CO by averaging the individual CV values. Although the naïve estimator underestimated the true within-subject CV (w CV), the corrected w CVs by random-effects model for analysis of variance (5.0% for setting A: high positive end-expiratory pressure [PEEP], inspiratory/expiratory [I:E] ratio 1:2; 5.3% for setting B: low PEEP, I:E 2:1; and 5.5% for setting C: low PEEP, I:E 1:2) were still much less than the values from Sasse et al. (7.7% for overall, 6.4% for covariable stable, and 9.9% for covariable unstable) (5). Therefore, we thought that the conclusion in our study was valid.
Because the intrathoracic positive pressures were not similar during the three settings, the resulting impedance to venous return was different also. The absolute values of CO measurements will likely vary between each other. However, the aim of our study, rather than to look into the influences that different ventilatory settings have on CO measurement, was to investigate the spontaneous fluctuations of CO during various ventilatory settings. What we were interested in is the intersetting CO CV difference; therefore, to discuss the limit-of-agreement of CO measurements during different ventilatory settings is not necessary.
However, we do greatly appreciate the comments from Dr. Nguyen, and we will consider the within-subject coefficient of variation with random effects model in the next studies regarding the issue of variability.
