In neurosurgery, estimation of PaCO2 from PETCO2 has been questioned. The aim of this study was to reevaluate the accuracy of PETCO2 in estimating PaCO2 during neurosurgical procedures lasting >3 h and to measure the effect of surgical positioning on arterial to end-tidal CO2 gradient (P[a-ET]CO2) over time. One hundred four neurosurgical patients classified into four groups (supine [SP], lateral [LT], prone [PR], sitting [ST]) were included in a prospective study. PaCO2, PETCO2, and P(a-ET)CO2 were measured after induction of anesthesia (T0), after positioning (T1), each following hour (T2, T3, T4), and at the end of the procedure after return to the SP position (T5). Data are expressed as the mean +/- SD, and statistical analysis used linear regression, the Bland-Altman method, and analysis of variance. The mean durations of positioning and surgery were 4.1+/-1 h and 3.7+/-1.3 h, respectively. We performed 624 simultaneous measurements of PaCO2 (33+/-5 mm Hg) and PETCO2 (27+/-4 mm Hg), leading to a mean P(a-ET)CO2 of 6+/-4 mm Hg. P(a-ET)CO2 of the LT group (7+/-3 mm Hg) was larger (compared with the SP, PR, and ST groups) because of a lower PETCO2 (26+/-4 mm Hg). Negative P(a-ET)CO2 (PETCO2 > PaCO2) occurred 22 times, only in the SP (n = 9) and ST groups (n = 13). Changes in opposite directions of PETCO2 and PaCO2 between two successive measurements were found in 26% of the cases. Correlation coefficients in the four groups (PaCO2 versus PETCO2) were not in good agreement (0.46 to 0.62; P < 0.001). The mean bias was between 5 and 7 mm Hg. The superior (13-15 mm Hg) and inferior (-5 to 0 mm Hg) limits of agreement were too large to expect PETCO2 to replace PaCO2. In conclusion, during neurosurgical procedures of >3 h, capnography should be performed with regular analysis of arterial blood gases for optimal ventilator adjustment.
Implications: This study, which aimed to reevaluate the ability of PETCO2 to estimate PaCO2 during neurosurgical procedures according to surgical position, indicates that PETCO2 cannot replace PaCO2 for the following reasons: scattering of individual values; occurrence of negative arterial to end-tidal CO2 gradient (P[a-ET]CO2; PaCO2 and PETCO2 variations in opposite directions; large changes in P(a-ET)CO2 between two samples; and instability of P(a-ET)CO2 over time.