Transcutaneous monitors during one-lung ventilation: are they reliable?

During one-lung ventilation, levels of oxygen and carbon dioxide in the blood are commonly assessed by intermittent blood gas sampling. Transcutaneous PO2 (tcPO2) and transcutaneous PCO2 (tcPCO2) have been reported to accurately reflect arterial PO2 (PaO2) and arterial PCO2 (PaCO2) in hemodynamically stable patients. Transcutaneous monitors appear to be ideal for detecting trends toward hypoxia and hypercarbia, conditions that may not be evident when using intermittent blood gas sampling, while pulse oximetry, since it reflects saturation, may not detect hypoxia until it has already occurred. Thirty: one patients undergoing one-lung ventilation were monitored using both transcutaneous electrodes applied to the upper arm (group 1) or chest (group 2) and arterial blood gas sampling. Arterial blood gases were sampled while tcPO2 and tcPCO2 values were being recorded. Regression, correlation, and covariance analyses were performed. Correlation coefficients of PaO2 to tcPO2 varied from .05 to .99 for each patient. The slopes of individual regression lines varied from 0.03 to 1.16. Correlation coefficients of PaCO2 to tcPCO2 varied from .01 to .99, while the slopes of individual regression lines ranged from 0.02 to 5.89. Covariance analyses revealed considerable variation in PaO2 to tcPO2 and PaCO2 to tcPCO2 in individual patients even under stable hemodynamic conditions. Analysis of covariance also demonstrated that in group 2 the slopes comparing arterial and transcutaneous values were significantly different for PaO2 <100 mmHg and PaO2 >200 mmHg. In group 1, for PaO2 <100 mmHg, there was no difference in slopes but y-intercepts were significantly different (P < .05). However, transcutaneous indices were significantly different in both groups for PaO2 <100 mmHg and PaO2 >200 mmHg. It is concluded that transcutaneous monitoring is useful to indicate trends in arterial values in some patients, but blood gas analysis is still necessary to verify the reliability of such monitoring.