Background. Continuous noninvasive monitoring of arterial carbon dioxi
de (CO2) in neonatal intensive care unit (NICU) patients would help cl
inicians avoid complications of hypocarbia and hypercarbia. End-tidal
CO2 monitoring has not been used in this population to date, but recen
t technical advances and the introduction of surfactant therapy, which
improves ventilation-perfusion matching, might improve the clinical u
tility of end-tidal monitoring. Objective. To determine the accuracy a
nd precision of end-tidal CO2 monitoring in NICU patients. Design. Non
randomized recording of simultaneous end-tidal and arterial CO2 pairs.
Setting. Two university NICUs. Patients. Forty-five newborn infants r
eceiving mechanical ventilation who had indwelling arterial access, an
d a predefined subsample of infants who were <1000 g birth weight, <8
days of age, and who received surfactant therapy (extremely low birth
weight [ELBW] <8). Outcome Measures. The correlation coefficient, degr
ee of bias, and 95% confidence interval were determined for both the o
verall population and the ELBW <8 subgroup. Those factors which signif
icantly influenced the bias were identified. The ability of the end-ti
dal monitor to alert the clinician to instances of hypocarbia or hyper
carbia was determined. Results. There were 411 end-tidal/arterial pair
s analyzed from 45 patients. The correlation coefficient was 0.833 and
the bias was -6.9 mm Hg (95% confidence interval, +/- 11.5 mm Hg). Th
e results did not differ markedly in the ELBW <8 infants. Measures of
the degree of lung disease, the ventilation index and the oxygenation
index, had small influences on the degree of bias. This type of capnom
etry identified 91% of the instances when the arterial CO2 pressure wa
s between 34 and 54 mm Hg using an end-tidal range of 29 to 45 mm Hg.
End-tidal values outside this range had a 63% accuracy in predicting h
ypocarbia or hypercarbia. Conclusion. End-tidal CO2 monitoring in NICU
patients is as accurate as capillary or transcutaneous monitoring but
less precise than the latter. It may be useful for trending or for sc
reening patients for abnormal arterial CO2 values.