Identifying Limitations of Near-infrared Spectroscopy in Clinical Situations. Ter Minassian et al. (page 985) 

Although studies have shown that near-infrared spectroscopy responds rapidly to cerebral oxygen desaturation during cerebral hypoperfusion or systemic hypoxia in normal human subjects, its value in more complex situations has not been proven. Accordingly, Ter Minassian \E et al.  examined the relationships between near-infrared spectroscopy cerebral oxygen saturation (Sc^O2) and jugular venous oxygen saturation (Svj^O2) during manipulation of arterial carbon dioxide partial pressure and blood pressure in adults with severe head trauma.

Nine patients with closed head injuries and with multifocal contusions or diffuse brain swelling confirmed by computed tomography scan were studied within the first 10 days after injury. All patients were placed in the supine position with the head and thorax tilted upward at 30°. After sedation with midazolam and fentanyl, patients were mechanically ventilated to achieve 100% arterial oxygen saturation and moderate hypocapnia. Patients with a mean arterial pressure < 70 mmHg were given a norepinephrine infusion to maintain arterial pressure above this threshold.

Near-infrared spectroscopy was used to record Sc^O2. The system used two wavelengths, 730 and 810 nm, and the sensor contained a light-emitting diode and two light detectors located 30 and 40 mm from the light-emitting diode, to help distinguish between the cerebral and extracerebral signals and thus monitor Sc^O2in the underlying area of the brain. Blood flow velocity in the middle cerebral artery was recorded unilaterally by pulsed Doppler ultrasound. Intracranial pressure, mean arterial pressure, end-tidal CO², middle cerebral artery blood flow velocity, and Sc^O2were recorded at 200 Hz and averaged every 4 s. All signals were stored for off-line analysis.

All patients underwent two or three CO²challenges, with a minimum interval of 24 h between challenges. After the end of the first CO²challenge, systemic arterial pressure was pharmacologically altered using norepinephrine infusion (in four patients, starting or increasing the rate of infusion; in five, discontinuing or reducing the infusion rate). Data of Sc^O2and Svj^O2were plotted according to Bland and Altman and showed a uniform scatter. However, a Bland and Altman plot of changes in Sc^O2versus  changes in Svj^O2had a negative slope. Regression analysis showed that changes in Sc^O2were positively correlated with changes in Svj^O2during the CO²challenge, but negative during arterial pressure challenge. As assessed by near-infrared spectroscopy, Sc^O2does not adequately reflect changes in Svj^O2in patients with severe head injury, such as those studied here. These findings may be attributable, in part, to changes in arteriovenous partitioning of intracranial blood, infrared-spectroscopy contamination of the signal by extracerebral blood, algorithm errors, and dissimilar tissue sampling.