Bias between capnometry and venous carbon dioxide during initial assessment of pediatric emergency department patients: A video-based study

ObjectiveThe bias of capnometry (ETCO2) and venous carbon dioxide (vpCO2) among pediatric emergency department (PED) patients triaged to critical care areas is unknown. We aimed to explore correlations and bias between ETCO2 and vpCO2,and identify predictors of bias.MethodsThis was an observational, video-based, retrospective study comparing ETCO2 and vpCO2. Pediatric patients with simultaneous ETCO2 and vpCO2 data were included. Our primary aim utilized linear regressions to determine correlations and Bland-Altman analysis to assess bias. Our secondary aim utilized multiple regression to identify clinical covariates contributing to bias. Covariates included age, respiratory rate, heart rate, mean arterial blood pressure, capnometry interface, PED diagnosis, and PED disposition.ResultsA total of 200 PED patients with ETCO2 and vpCO2 data were included. The median (interquartile range [IQR]) ETCO2, vpCO2, and Delta CO2 in mmHg were 38 (32, 46), 49 (41, 61), and 11 (4, 20), respectively. ETCO2 (r = 0.76) and Delta CO2 (r = 0.71) were highly correlated with vpCO2. The mean bias between ETCO2 and vpCO2 was -14.1 mmHg (95% confidence interval [CI], -41.9 -13.7). The bias between ETCO2 and vpCO2 increased at higher values of each measure. ETCO2 sampling interface was the only independent predictor of vpCO2 in our multivariate analysis. Patients requiring bag-valve mask (BVM) ventilation had the highest median bias between ETCO2 and vpCO2 (29 mmHg, IQR 15, 37).ConclusionETCO2 and vpCO2 were highly correlated. However, bias increased at higher levels of both ETCO2 and vpCO2. Among PED patients, ETCO2's ability to approximate vpCO2 diminishes with worsening hypercarbic respiratory failure.