Effects of Skin Pigmentation on Pulse Oximeter Accuracy at Low Saturation
Nov 07, 2018
PULSE oximetry theoretically can compute arterial hemoglobin oxygen saturation from the ratio of the pulsatile to the total transmitted red light divided by the same ratio for infrared light transilluminating a finger, ear, or other tissue. The derived saturation should be independent of skin pigmentation, and many other variables, such as hemoglobin concentration, nail polish, dirt, and jaundice. Several large controlled studies comparing black and white patients (380 subjects)1,2 reported no significant pigment-related errors in pulse oximeters at normal saturation. However, Severinghaus and Kelleher3 reviewed data from several investigators who had reported anecdotal errors (+3 to +5%) in black patients.4–7 Model simulations of errors due to various pigments were reviewed by Ralston et al. 8 Cote et al. 9 reported that nail polish and ink on skin surface can cause errors, a finding confirmed anecdotally by others from fingerprinting ink,10 henna,11 and meconium.12 Intravenously injected dyes cause transient errors.13 Lee et al. 14 found overestimation of saturation, especially at low saturation in pigmented patients (Indian, Malay vs. Chinese). The Technology Subcommittee of the Working Group on Critical Care, Ontario Ministry of Health,15 reported unacceptable errors in pulse oximetry at low saturation in pigmented subjects. Zeballos and Weisman16 compared the accuracy of the Hewlett-Packard (Sunnyvale, CA) ear oximeter and the Biox II pulse oximeter (Ohmeda, Andover, MA) in 33 young black men exercising at three different simulated altitudes. At an altitude of 4,000 m, where arterial oxygen saturation (Sao2) ranged from 75 to 84%, the Hewlett-Packard underestimated Sao2by 4.8 ± 1.6%, whereas the Biox overestimated Sao2by 9.8 ± 1.8% (n = 22). It was stated that these errors, previously reported in whites, were both exaggerated in blacks.
During our many years of testing pulse oximeter accuracy at oxygen saturations as low as 50%, we have occasionally noted unusually high positive bias, particularly at very low saturation levels, in some but not in other deeply pigmented subjects. This investigation was therefore specifically designed to determine whether errors at low Sao2correlate with skin color.
All pulse oximeters marketed in the United States are required by the US Food and Drug Administration to have been tested and to be certified as accurate to less than ±3% root mean square error at Sao2values between 70 and 100%. The great majority of calibration and confirmation tests have been conducted in volunteer subjects with light skin pigmentation.
The Food and Drug Administration has recently suggested that studies of pulse oximeter accuracy submitted for Food and Drug Administration device approval include subjects with a range of skin pigmentations, although no quantitative requirement has been distributed. We are aware of no data that support this action.
If there is a significant and reproducible positive bias at low saturation in dark-skinned subjects, inclusion of dark-skinned subjects will increase test group mean root mean square errors, perhaps enough to cause rejection by the Food and Drug Administration. If a reproducible bias is found at low saturation in dark-skinned subjects in all pulse oximeters, warning labels should be provided to users, possibly with suggested correction factors.
Related Industry Knowledge