Agreement between arterial and transcutaneous PCO2 in patients undergoing non-invasive ventilation

Aim: Transcutaneous carbon dioxide (PtCO2) monitoring offers a potentially non-invasive and continuous means to determine the arterial carbon dioxide tension (PaCO2). ED studies of agreement between PtCO2 and PaCO2 have had conflicting findings and have not been targeted to subgroups with severe ventilatory disturbance such as those requiring non-invasive ventilation [NIV]. Our aim is to determine agreement between PtCO2 and PaCO2 for patients undergoing NIV for respiratory failure. Methods: This prospective observational study included a convenience sample of patients undergoing NIV for respiratory failure who required arterial blood gas analysis as part of their care. Data collected included patient demographics, indication for NIV, diagnosis, vital signs, and pH, PaCO2 and PtCO2. The outcome of interest was agreement between PaCO2 and PtCO2. Analysis was made using descriptive statistics, Bland-Altman techniques, Mann Whitney U test and Fisher/Chi square tests. Results: 46 comparisons were analysed. Median age was 69 [IQR 65-79], 67% male; median PaCO2 60 mmHg [IQR 46-70] and median pH 7.35 [IQR 7.30-7.38]. Average difference between PaCO2 and PtCO2 was 6.1 mmHg with 95% limits of agreement -10.1-22.3 mmHg. Thirty seven comparisons [80%] were within 10 mmHg [95% CI 66-90%]. Difference >10 mmHg was associated with increasing PaCO2 [p = 0.001; median difference 19.6 mmHg, 95% Cl 9.2-30.4 mmHg]. All cases with difference >10 mmHg had PaCO2 > 60 mmHg. Conclusion: In patients undergoing NIV, agreement between PaCO2 and PtCO2 was sub-optimal, with unacceptably wide 95% limits of agreement. PtCO2 cannot be recommended as a substitute for PaCO2 testing in this group. (C) 2010 Elsevier Ltd. All rights reserved.