Accuracy of Transcutaneous Carbon Dioxide Levels in Comparison to Arterial Carbon Dioxide Levels in Critically Ill Children

Abstract

BACKGROUND:

Widespread use of transcutaneous P_CO2 (P_tcCO2 ) monitoring is currently limited by concerns many practitioners have regarding accuracy. We compared the accuracy of P_tcCO2 with that of P_aCO2 measurements in critically ill children, and we investigated whether clinical conditions associated with low cardiac output or increased subcutaneous tissue affect this accuracy.

METHODS:

We performed a single-center prospective study of critically ill children placed on transcutaneous monitoring.

RESULTS:

There were 184 children enrolled with paired P_aCO2 and P_tcCO2 values. Subjects had a median age of 31.8 mo (interquartile range 3.5–123.3 mo). Most children were mechanically ventilated (n = 161, 87.5%), and many had cardiac disease (n = 76, 41.3%). The median P_aCO2 was 44 mm Hg (interquartile range 39–51 mm Hg). The mean bias between P_aCO2 and P_tcCO2 was 0.6 mm Hg with 95% limits of agreement from −13.6 to 14.7 mm Hg. The P_tcCO2 and P_aCO2 were within ±5 mm Hg in 126 (68.5%) measurements. In multivariable modeling, cyanotic heart disease (odds ratio 3.5, 95% CI 1.2–10, P = .02) and monitor number 2 (odds ratio 3.8 95% CI 1.3–10.5, P = .01) remained associated with P_tcCO2 ≥ 5 mm Hg higher than P_aCO2 . Serum lactate, fluid balance, renal failure, obesity, vasoactive-inotrope score, and acyanotic heart disease were not associated with high or low P_tcCO2 values. In 130 children with a second paired P_tcCO2 and P_aCO2 measurement, predicting the second measured P_aCO2 by subtracting the initial observed difference between the P_tcCO2 and P_aCO2 from the subsequent measured P_tcCO2 decreased the mean bias between observed and predicted P_aCO2 to 0.2 mm Hg and the 95% limits of agreement to −9.4 to 9.7 mm Hg.

CONCLUSIONS:

P_tcCO2 provides an acceptable estimate of P_aCO2 in many critically ill children, including those with clinical conditions that may be associated with low cardiac output or increased subcutaneous tissue, although it does not perform as well in children with cyanotic heart disease. P_tcCO2 may be a useful adjunct monitoring method, but it cannot reliably replace P_aCO2 measurement.