End-tidal and arterial carbon dioxide measurements correlate across all levels of physiologic dead space

Abstract

Background: End-tidal carbon dioxide (P(ETCO(2))) is a surrogate, noninvasive measurement of arterial carbon dioxide (P(aCO(2))), but the clinical applicability of P(ETCO(2)) in the intensive care unit remains unclear. Available research on the relationship between P(ETCO(2)) and P(aCO(2)) has not taken a detailed assessment of physiologic dead space into consideration. We hypothesized that P(ETCO(2)) would reliably predict P(aCO(2)) across all levels of physiologic dead space, provided that the expected P(ETCO(2))-P(aCO(2)) difference is considered.

Methods: Fifty-six mechanically ventilated pediatric patients (0-17 y old, mean weight 19.5 +/- 24.5 kg) were monitored with volumetric capnography. For every arterial blood gas measurement during routine care, we measured P(ETCO(2)) and calculated the ratio of dead space to tidal volume (V(D)/V(T)). We assessed the P(ETCO(2))-P(aCO(2)) relationship with Pearson's correlation coefficient, in 4 V(D)/V(T) ranges.

Results: V(D)/V(T) was <or= 0.40 for 125 measurements (25%), 0.41-0.55 for 160 measurements (32%), 0.56-0.70 for 154 measurements (31%), and >0.7 for 54 measurements (11%). The correlation coefficients between P(ETCO(2)) and P(aCO(2)) were 0.95 (mean difference 0.3 +/- 2.1 mm Hg) for V(D)/V(T) <or= 0.40, 0.88 (mean difference 5.9 +/- 4.3 mm Hg) for V(D)/V(T) 0.41-0.55, 0.86 (mean difference 13.6 +/- 5.2 mm Hg) for V(D)/V(T) 0.56-0.70, and 0.78 (mean difference 17.8 +/- 6.7 mm Hg) for V(D)/V(T) >0.7.

Conclusions: There were strong correlations between P(ETCO(2)) and P(aCO(2)) in all the V(D)/V(T) ranges. The P(ETCO(2))-P(aCO(2)) difference increased predictably with increasing V(D)/V(T).

Figures

Figure 1

End-tidal carbon dioxide versus arterial carbon dioxide at different ranges of physiologic dead space. A) At a physiologic dead space to tidal volume ratio (Vd/Vt) ≤ 0.4 the correlation is very strong (ρ = 0.95). B) At Vd/Vt of 0.40-0.55 (ρ = 0.88), C) 0.55-0.7 (ρ = 0.86), and D) > 0.7 (ρ = 0.78), the correlation coefficients decrease slightly but remain moderately strong.

Figure 1

End-tidal carbon dioxide versus arterial carbon dioxide at different ranges of physiologic dead space. A) At a physiologic dead space to tidal volume ratio (Vd/Vt) ≤ 0.4 the correlation is very strong (ρ = 0.95). B) At Vd/Vt of 0.40-0.55 (ρ = 0.88), C) 0.55-0.7 (ρ = 0.86), and D) > 0.7 (ρ = 0.78), the correlation coefficients decrease slightly but remain moderately strong.

Figure 2

Bland-Altman plots of ETCO2 versus PaCO2. These plots do not take into account the expected change in the ETCO2-PaCO2 gradient seen with increasing physiologic dead space. So it is impossible to tell whether the variation seen in these plots is due to variation in physiologic dead space within each subset or to unreliability of ETCO2 as a surrogate measure.

Figure 2

Bland-Altman plots of ETCO2 versus PaCO2. These plots do not take into account the expected change in the ETCO2-PaCO2 gradient seen with increasing physiologic dead space. So it is impossible to tell whether the variation seen in these plots is due to variation in physiologic dead space within each subset or to unreliability of ETCO2 as a surrogate measure.