Correlation of end-tidal CO2 measurements to arterial PaCO2 in nonintubated patients

To determine the accuracy of end-tidal carbon dioxide levels as a measure of arterial carbon dioxide levels in nonintubated patients presenting to an emergency department for care.

A prospective, cross-sectional analysis.

University hospital ED.

Nonintubated adult patients presenting to the ED for care of a variety of problems.

Patients who had arterial blood gas samples taken as part of their ED evaluation were asked to breathe normally through an endotracheal tube adapter or a modified nasal cannula connected to a side port sampling capnometer while a sample for arterial blood gas was drawn from the radial artery.

End-tidal carbon dioxide levels (mm Hg) were recorded at the time of arterial blood gas sampling. The difference between end-tidal carbon dioxide and PaCO2 was tested with the paired t-test at a significance level of .05. The correlation of end-tidal carbon dioxide to PaCO2 was tested in all patients and in subgroups using simple linear regression.

Seventy-six patients were enrolled. In all patients, end-tidal carbon dioxide was 3.5 mm Hg lower than PaCO2 and correlated well with PaCO2 (r2 = .772). In patients with hypocapnia, there was no significant difference between end-tidal carbon dioxide and PaCO2 (P = .17), and the correlation of end-tidal carbon dioxide to PaCO2 was stronger (r2 = .838). In patients with a respiratory or metabolic acidosis, the difference between end-tidal carbon dioxide and PaCO2 was 6 mm Hg (P = .005), but end-tidal carbon dioxide correlated well to PaCO2 (r2 = .899).

Measurements of end-tidal carbon dioxide concentrations correlate well with PaCO2 values in nonintubated patients presenting with a variety of conditions to EDs. End-tidal carbon dioxide measurements may be sufficient measures of PaCO2 in selected patients and obviate the need for repeat arterial blood gas determination. Further study is warranted.