Chest
Volume 93, Issue 3, March 1988, Pages 515-517
Journal home page for Chest

Bias and Precision of Pulse Oximeters and Arterial Oximeters

https://doi.org/10.1378/chest.93.3.515Get rights and content

We evaluated measurements of oxygen saturation from four noninvasive pulse oximeters, and two conventional arterial oximeters. Simultaneous measurements were obtained on each instrument on three different occasions in five healthy subjects breathing gas with an FIO2 of 1.00, 0.50, 0.21, 0.17, 0.15, 0.13 and 0.11. Mean bias relative to the sum of oxy-, carboxy-, and methemoglobins ranged from –0.4 to –2.6 percent for the pulse oximeters and +1.8 to –4.2 percent for the conventional oximeters. Two pulse oximeters performed well at all saturations down to 65 percent, while the others showed either increasing or decreasing bias below 80 percent saturation. Precision was approximately 2 percent for all instruments except one conventional oximeter with a precision of 0.7 percent. In the clinically relevant range, the performance of the noninvasive pulse oximeters was similar to conventional oximeters using arterial samples.

Section snippets

MATERIAL AND METHODS

The Human Subjects Review Board of Children's Hospital of Oakland gave approval for this study. Five 31- to 34-year-old nonsmoking men free from lung, heart, or blood disease gave informed consent. All completed the test and none complained of excessive discomfort during the procedure.

After local anesthesia, a 20-gauge radial arterial catheter was placed in the nondominant wrist for collection of arterial blood samples. Four different pulse oximeters (Biox/Omeda 3700, Boulder, CO, designated as

RESULTS

We obtained 165 arterial samples and simultaneous pulse oximetry measurements. The mean and standard deviation of pH measurements were 7.40 ± 0.03 and of the PaCO2 were 41.3 ± 5.50. The mean carboxyhemoglobin was 2.0 percent and the mean methemoglobin was 0.3 percent. Thus, an oximeter with a bias of –2.3 relative to the sum of oxy-, carboxy- and methemoglobin would closely approximate the oxyhemoglobin saturation reading of the IL282 device. In no case was the carboxyhemoglobin greater than

DISCUSSION

Before accepting a new device in clinical practice, clinicians should be aware of the bias, precision, reproducibility, and limitations. This study confirms that noninvasive pulse oximeters perform adequately for clinical decision-making over the range of saturations likely to be encountered. Indeed, their performance compared favorably with that of conventional oximeters which use arterial blood samples. We found that caution should be exercised in interpreting pulse oximeter readings of less

ACKNOWLEDGMENT

The authors thank Drs. Marshall Klaus, Herman W. Lipow, Gabriella E. Molnar, Lily C. Kao and David Durand for their suggestions, and Joy Childers and Ceres C. Larkrith for preparation of the manuscript.

REFERENCES (11)

There are more references available in the full text version of this article.

Cited by (0)

Supported by grants from the Novametrix Corporation, the Fund for Pediatric Care and Research, and the Northern California Sudden Infant Death Syndrome Foundation.

Presented in part at the Annual Meeting, American Thoracic Society, New Orleans, May, 1987.

Manuscript received May 22; revision accepted August 11.

View full text