Purpose: To evaluate the agreement between end-tidal carbon dioxide (PETCO2) and arterial CO2 (PaCO2) in patients with traumatic brain injury and to document the course of the (PaCO2-PETCO2) gradient over time.
Methods: Twenty one traumatic brain injury patients (Coma Glasgow Scale < or = 8) were included in this prospective observational study over a period of six months. Simultaneous determinations of PaCO2 and PETCO2 (by infrared capnometry) were recorded. Agreement between PaCO2 and PETCO2 was determined by the statistical method described by Bland and Altman. Changes in PETCO2 over time were compared with changes in PaCO2. Factors likely to explain a gradient superior to +/- 4 mmHg were explored.
Results: One hundred and eleven data pairs were obtained. The bias was 5.5 mmHg with a precision of 5.1 mmHg and limits of agreement ranged from -4.5 mmHg to 15.5 mmHg. The latter exceeded the predefined limits of agreement established to determine interchangeability between methods (+/- 4 mmHg). PETCO2 and PoCO2 changed in opposite directions in 20% of 90 consecutive measurements. Only the duration of ventilation was found to be significantly associated with a gradient superior to +/- 4 mmHg.
Conclusions: In this selected population of patients with severe traumatic brain injury, measurements of PETCO2 and PaCO2 are not interchangeable. Further the PoCO2-PETCO2 gradient is not stable over time and cannot predict variations of PaCO2. The use of PETCO2 instead of PaCO2 could be deleterious in patients in whom strict control of PaCO2 values is required.