Abstract Background The reliability of end-tidal carbon dioxide (EtCO₂) as a surrogate for arterial arterial carbon dioxide partial pressure (PaCO₂) during one-lung ventilation (OLV) in infants is uncertain. This study evaluated EtCO₂-PaCO₂ agreement in infants undergoing OLV and identified factors associated with clinically significant discrepancy. Methods This secondary analysis of a prospective, randomized controlled trial included 172 infants (aged ≤ 2 years) undergoing thoracoscopic surgery. Paired PaCO₂ and EtCO₂ measurements were obtained 20 min after initiating OLV. Clinical agreement was defined a priori as an absolute difference of ≤ 5 mmHg. Statistical evaluations included Bland–Altman analysis and multivariable logistic regression to identify predictors of disagreement, and correlation analyses. Results During OLV, the mean EtCO₂–PaCO₂ bias was 2.8 mmHg with wide 95% limits of agreement (-4.7 to 10.3 mmHg), exceeding the clinically acceptable threshold. Oxygenation impairment (PaO₂/FiO₂ < 200 mmHg) significantly reduced the proportion of clinically acceptable measurements (55.9% vs. 81.7%, P = 0.0001). In multivariable analysis, EtCO₂–PaCO₂ disagreement occurred exclusively in infants ventilated under the non–protective protocol (VT 8 mL/kg with ZEEP). After excluding ventilatory variables that were structurally confounded with randomized allocation, lower lung compliance was independently associated with disagreement (adjusted OR 1.66, 95% CI 1.05–2.64, P = 0.030). Conclusions EtCO₂ provides only limited reliability as a stand–alone substitute for arterial blood gas analysis in infants during OLV, particularly under non–protective ventilation. The lung–protective ventilation strategy as a composite intervention is associated with substantially improved EtCO₂–PaCO₂ agreement during OLV in infants. Among individual patient-level factors, lower lung compliance independently predicts EtCO₂–PaCO₂ disagreement and may serve as a bedside indicator for clinicians to consider supplementary arterial blood gas analysis.
Zhu et al. (Fri,) studied this question.