Validation of a Second-Generation Near-Infrared Spectroscopy Monitor in Children With Congenital Heart Disease.

Abstract

Cerebral oximetry using near-infrared spectroscopy is a noninvasive optical technology to detect cerebral hypoxia-ischemia and develop interventions to prevent and ameliorate hypoxic brain injury. Cerebral oximeters are calibrated and validated by comparison of the near-infrared spectroscopy-measured cerebral O2 saturation (SctO2) to a "field" or reference O2 saturation (REF CX) calculated as a weighted average from arterial and jugular bulb oxygen saturations. In this study, we calibrated and validated the second-generation, 5 wavelength, FORE-SIGHT Elite with the medium sensor (source-detector separation 12 and 40 mm) for measurement of SctO2 in children with congenital heart disease. After institutional review board approval and written informed consent, 63 children older than 1 month and ≥2.5 kg scheduled for cardiac catheterization were enrolled. Self-adhesive FORE-SIGHT Elite medium sensors were placed on the right and left sides of the forehead. Blood samples for calculation of REF CX were drawn simultaneously from the aorta or femoral artery and the jugular bulb before (T1) and shortly after (T2) baseline hemodynamic measurements. FORE-SIGHT Elite SctO2 measurements were compared to the REF CX (REF CX = [0.3 SaO2] + [0.7 SjbO2]) using Deming regression, least squares linear regression, and Bland-Altman analysis. Sixty-one subjects (4.5 [standard deviation 4.4] years of age; 17 [standard deviation 13] kg, male 56%) completed the study protocol. Arterial oxygen saturation ranged from 64.7% to 99.1% (median 96.0%), jugular bulb venous oxygen saturation from 34.1% to 88.1% (median 68.2%), the REF CX from 43.8% to 91.4% (median 76.9%), and the SctO2 from 47.8% to 90.8% (median 76.3%). There was a high degree of correlation in SctO2 between the right and left sensors at a given time point (within subject between sensor correlation r = 0.91 and 95% confidence interval [CI], 0.85-0.94) or between T1 and T2 for the right and left sensors (replicates, within subject between time point correlation r = 0.95 and 95% CI, 0.92-0.96). By Deming regression, the estimated slope was 0.966 (95% CI, 0.786-1.147; P = .706 for testing against null hypothesis of slope = 1) with a y intercept of 2.776 (95% CI, -11.102 to 16.654; P = .689). The concordance correlation coefficient was 0.873 (95% CI, 0.798-0.922). Bland-Altman analysis for agreement between SctO2 and REF CX that accounted for repeated measures (both in times and sensors) found a bias of -0.30% (95% limits of agreement: -10.56% to 9.95%). This study calibrated and validated the FORE-SIGHT Elite tissue oximeter to accurately measure SctO2 in pediatric patients with the medium sensor.