PURPOSE

Arterial blood gas measurement is frequently performed in critically ill patients to diagnose and monitor acute respiratory failure. At a given metabolic rate, carbon dioxide partial pressure (PaCO2) is entirely determined by CO2 elimination through ventilation. Transcutaneous partial pressure of carbon dioxide (PtcCO2) monitoring permits a noninvasive and continuous estimation of arterial CO2 tension (PaCO2). The accuracy of PtcCO2, however, has not been well studied. To assess the accuracy of different CO2 monitoring methods, we compared PtcCO2 and end-tidal CO2 concentration (EtCO2) to PaCO2 measurements in nonintubated intensive care unit (ICU) patients with acute respiratory failure.

METHODS

During a 2-month period, we conducted a prospective observational cohort study in 25 consecutive nonintubated and spontaneously breathing patients admitted to our ICU. Arterial blood gases were measured at study inclusion, 30, 60, and 120 minutes later. At each sampling time, EtCO2 was continuously monitored using a Philips Smart Capnoline Plus, and PtcCO2 was measured using was measured using SenTec device. The aim of the study was to assess agreement between PtcCO2 and PaCO2 and between EtCO2 and PaCO2 in nonintubated ICU patients with acute respiratory failure. Bland-Altman techniques and Pearson correlation coefficients were used. The differences over time (at 30, 60, and 120 minutes) between PaCO2 and EtCO2 and between PtcCO2 and PaCO2 were evaluated using 1-way analysis of variance.

RESULTS

Transcutaneous partial pressure of carbon dioxide and PaCO2 were well correlated (R = 0.97), whereas the correlation between EtCO2 and PaCO2 was poor (R = 0.62) probably due to the presence of an alveolar dead space in a few patients, most notably in the group with chronic obstructive pulmonary disease. The difference over time remained stable for both PaCO2 vs EtCO2 (analysis of variance; P = .88) and PaCO2 vs PtcCO2 (P = .93).

CONCLUSION

We found large differences between EtCO2 and Paco2 in spontaneously breathing nonintubated ICU patients admitted for acute respiratory failure. Our study argues against the use of EtCO2 monitoring in such patients but raises the possibility that PtcCO2 measurement may provide reasonable estimates of PaCO2.