Diagnostic accuracy of capnography during high-frequency ventilation in neonatal intensive care units

Background and Objective High-frequency ventilation (HFV) is a powerful tool for CO₂ elimination, and thus requires careful monitoring of CO₂. Our aim was to assess the diagnostic accuracy (correlation, agreement, and trending) of continuous distal capnography (dCap) with PaCO₂ in infants ventilated with HFV. Design This was a prospective, observational, multicenter study. dCap was compared with simultaneous PaCO₂ ("gold standard") drawn from indwelling arterial line for patient care in term and preterm infants ventilated with HFV. dCap was obtained via the side-port of a double-lumen endotracheal-tube by a Microstream capnograph with specially designed software for HFV. Results Twenty-four infants participated in the study (median [range] gestational age [GA]: 26.8 [23.6-38.6] weeks). Analysis included 332 measurements. dCap was in correlation (r = 0.70, P < 0.001) but with less than adequate agreement (mean difference ± SD of the differences: -11.7 ± 10.3 mmHg) with PaCO₂. Comparable findings were found in the subgroup of infants <1,000 g (n = 240 measurements). Correlations were maintained in severe lung disease. Changes in dCap and in PaCO₂ for consecutive measurements within each patient were correlated (r = 0.63, P < 0.001). Area under the receiver operating curves (ROC) for dCap to detect high (>60 mmHg) or low (<30 mmHg) PaCO₂ was 0.83 (CI: 0.76-0.90) and 0.88 (CI: 0.79-0.97), respectively; P < 0.001. Conclusions Our prospective study suggests that continuous dCap in infants ventilated with HFV may be helpful for trends and alarm for unsafe levels of PaCO₂. dCap is only a complimentary tool and cannot replace PaCO₂ sampling because the agreement between these measurements was less than adequate.