Infrared end‐tidal CO 2 measurement does not accurately predict arterial CO 2 values or end‐tidal to arterial P CO 2 , gradients in rabbits with lung injury

End‐tidal P co 2 (P   ET   co   2) measurements from two commercially available neonatal infrared capnometers with different sampling systems and a mass spectrometer were compared with arterial P   co   2(P   a   co   2) to determine whether the former could predict the latter in mechanically ventilated rabbits with b d without lung injury. The effects of tidal volume, ventilator frequency and type of lung injury on the gradient between P   ET   CO   2and P   a   co   2(ΔP   (a‐ET)   CO   2) were evaluated. Twenty rabbits were studied: 10 without lung injury, 5 with saline lavage and 5 with lung injury by meconium instillation. Paired measurements of P   ET   CO   2by two infrared capnometers and a mass spectrometer were compared to P   a   co   2. In the rabbits without lung injury, the values from the infrared capnometers and mass spectrometer correlated strongly with P   a   co   2(r ± 0.91) despite differences in the slopes of the linear regression between P   ET   CO   2and ±P   a   co   2and in ΔP   (a‐ET)   CO   2. (P < 0.05). Values from the mainstream IR‐capnometer more closely approdmated the line of identity than the regression between the sidestream IR‐capnometer values or the mass spectrometer and P   a   co   2but tended to overestimate P   a   co   2. The ΔP   (a‐ET)   CO   2was similar at all tidal volumes and ventilator frequencies, regardless of capnometer type. In the rabbits with induced lung injury, while there was a positive correlation between the slopes of the regression between P   ET   CO   2and P   a   co   2for both capnometers (r ≥ 0.70), none of the regression slopes approximated the line of identify. The ΔP   (a‐ET)   CO   2was greater in rabbits with injured than noninjured lungs (P < 0.05). The ΔP   (a‐ET)   co   2was similar among capnometers regardless of tidal volume, ventilator frequency, or type of lung injury. The 95% confidence interval of plots P   a   co   2against P   ET   co   2was large for rabbits with injured and noninjured lungs. The ΔP   (a‐ET)   CO   2changed in an erratic manner with changing P   a   co   2. We conclude that P   ET   co   2measurements by infrared capnometer or mass spectrometer does not accurately predict P   a   co   2or ΔP   (a‐ET)   CO   2in the presence of lung injury, and should be used with caution in the management of critically ill infants. Pediatr Pulmonol. 1994; 17:189–196. © 1994 Wiley‐Liss, Inc.