ObjectiveTo examine the agreement and association of a noninvasive method of measuring CO2(using end-tidal PCO2) with PaCO2in mechanically ventilated adults with severe head trauma.DesignA prospective, quasi-experimental, repeated-measures study was used to compare end-tidal PCO2and PaCO2at two time points: before and after a standardized endotracheal suctioning procedure.InterventionsControlled intervention of endotracheal suctioning.SettingThe study was conducted at two intensive care units designated as Level 1 trauma centers.PatientsA consecutive sample of 35 severe head-injured patients with a Glasgow Coma Scale score of less than equals 8.Measurements and Main ResultsEnd-tidal PCO2and PaCO2values were simultaneously obtained and compared. End-tidal PCO2was measured, using a sidestream sensor placed in line of the ventilator circuit's deadspace. Arterial gases were drawn from an indwelling arterial catheter.No relationship was found between arterial and end-tidal measures (range r sup 2 equals .09 to r2equals .11). Using the Bland-Altman technique, agreement decreased as the amount of positive end-expiratory pressure increased. When a subset of patients (mechanically ventilated, with positive end-expiratory pressures of less than 5 cm H2O, paralyzed, and sedated) were examined (n equals 12), the correlation between the CO2measures improved (r2equals .77).ConclusionsThis study indicated that end-tidal PCO2monitoring correlates well with PaCO2in patients without respiratory complications or without spontaneous breathing, resulting in rebreathing of gases. However, its clinical validity is questionable in patients who have the greatest need for end-tidal PCO2monitoring (i.e., patients who have respiratory distress or who are breathing spontaneously and overriding the ventilator).(Crit Care Med 1996; 24:785-790)