Presented at the Neonatal Society 2010 Autumn Meeting.
Abdelhamid AE, Ratnavel N, Sinha A
Neonatal Transport Team, London
Background: Non-invasive carbon dioxide (CO2) monitoring in mechanically-ventilated preterm neonates in the NICU-based and during inter-hospital transport management is essential to avoid risks associated with levels out of range, while minimising repeated arterial sampling with its known complications. End-tidal CO2 (EtCO2) monitoring is of value in identifying some ventilation problems, but the reports on its use for ventilatory management in neonates are controversial. The current use of microstream devices has allowed neonatal capnography to be studied with a greater degree of exactitude.
Objective: To assess the accuracy of measurements of EtCO2 during neonatal transport of mechanicallyventilated preterm infants compared to the partial pressure of arterial CO2 (PaCO2) measurements.
Methods: Data analysis of 221 paired EtCO2 and PaCO2 recordings taken during stabilisation/road transport of 125 mechanically-ventilated preterm infants was run. EtCO2readings were obtained by a microstream non-dispersive infrared spectroscopy to continuously measure the amount of CO2 during every breath. The EtCO2 reading displayed on the set screen was timely recorded during arterial sampling for the blood gas reading paired with/compared to. All the arterial samples, were analysed by the blood gas machine in the referring or the receiving hospital, and any ventilatory change made at that time was based on the result. The paired CO2values were compared and the differences were investigated. The Bland-Altman method was used to assess bias and repeatability. The trend of change of the successive readings (i.e. the difference between each consecutive EtCO2 readings and that between each corresponding PaCO2 readings) obtained from the same patient at different stages of management were analysed by simple linear regression to find the correlation (if any) between the 2 trends to see if the change in the PaCO2 (decrease or increase) between 2 consecutive readings was associated with a change in the EtCO2 readings in terms of magnitude and direction.
Results: EtCO2 significantly correlated with PaCO2 (figure 1). However, the correlation was extremely poor (r=0.39, p<0.0001, 95% limits of agreement: 0.1996 – 0.4428) (figure 2). EtCO2underestimated PaCO2 at a significant level (mean [SD] 2.595 [1.418] kPa), and did not trend reliably over time within individual subjects (r=0.15, p=0.12). The EtCO2 bias was independent of the PaCO2 level range and lung disease severity in terms of gas exchange or shunting. After correction of EtCO2 with mean bias, 58% of EtCO2 values fell within 1 kPa of PaCO2 values.
Conclusion: EtCO2 has an under-recording bias when compared to the PaCO2. The trend values of EtCO2 do not obviate the need for blood gas measurements following ventilator changes. Alternative methods of non-invasive CO2 monitoring for absolute value or trend analysis should also be considered for neonatal transport.
1. Donald MJ et al (2006) Emerg Med J. 23: 728-730
2. DeIorio NM (2005) Emerg Med J. (22): 490-493
3. Moses JM et al(2009) BMC Pediatrics. (9): 20-25
4. Sullivan KJ (2005) Pediatric Emergency Care. 21 (5): 327