Clinical evaluation of noninvasive monitoring of oxygen saturation in critically ill patients.

OBJECTIVE

To examine the effect of abnormal cardiac index on the accuracy of measurement of oxygen saturation by pulse oximetry.

METHODS

Forty-six patients (mean age, 49 years) in a 9-bed medical ICU were studied. Measurements of oxygen saturation obtained with pulse oximeters and with a functional cooximeter were collected at baseline and 4, 8, 16, 24, 32, 40, and 48 hours later. Hemodynamic and cardiopulmonary parameters were recorded.

RESULTS

The Bland-Altman technique yielded upper and lower limits of agreement of 2.53% and -7.11%. Most (95.7%) of the differences between the measurements of oxygen saturation obtained with the 2 methods were within these limits, although some of these differences may be clinically unacceptable. The bias was -2.29%, and the precision was 2.41%. The clinical conditions associated with inaccurate tracking of saturation by pulse oximetry across the range of actual arterial oxygen saturation values were abnormal cardiac index, partial pressure of carbon dioxide, heart rate, and pulmonary capillary wedge pressure.

CONCLUSIONS

In patients with abnormal cardiac index, the pulse oximeter measurements exceeded the actual oxygen saturation by up to 7%. Pending prospective studies, clinicians should be aware that when certain cardiopulmonary parameters are abnormal, the margin of error in measurements of oxygen saturation obtained with a pulse oximeter may be greater than when those parameters are normal.