Detection of carbon dioxide thresholds using low-flow sidestream capnography in ventilated preterm infants

Abstract

Monitoring CO2 levels in preterm infants receiving mechanical ventilation is designed to avoid the harmful consequences of hypocapnia or hypercapnia. Capnography is of questionable accuracy for monitoring PCO2 in preterm infants. To determine the accuracy of sidestream capnography in ventilated preterm infants by comparing end-tidal carbon dioxide (EtCO2) values to mixed venous carbon dioxide pressure (PvCO2) and to transcutaneous carbon dioxide pressure (TcPCO2). Simultaneous recordings of EtCO2, TcPCO2 and PvCO2 in 37 ventilated preterm infants. The PvCO2-EtCO2 gradient was calculated. The Bland-Altman technique and the intra-class correlation coefficient (ICC) were used to assess agreement between methods. The area under the curve (AUC) was calculated. Ninety-nine EtCO2/PvCO2 pairs were studied from 37 preterm infants with a mean gestational age of 27.7 +/- 1.9 weeks and a mean birth weight of 1,003 +/- 331 g. The mean PvCO2-EtCO2 gradient was 11.2 +/- 8.0 mmHg, and the ICC was 0.28. The mean PvCO2-TcPCO2 gradient was 0 +/- 7.8 mmHg, and the ICC was 0.78. AUCs for EtCO2 and TcPCO2 were similar in detecting high or low PvCO2. Despite an insufficient correlation between EtCO2 and PvCO2, capnography was able to detect low and high CO2 warning levels with a similar efficacy to that of TcPCO2, and may therefore be of clinical interest.