Secondary Logo

Journal Logo

Technology, Computing, and Simulation: Research Reports

Improved Accuracy of Methemoglobin Detection by Pulse CO-Oximetry During Hypoxia

Feiner, John R. MD; Bickler, Philip E. MD, PhD

Author Information
doi: 10.1213/ANE.0b013e3181f46da8

Pulse oximetry noninvasively estimates arterial oxygen saturation (SpO2) by shining red and near-infrared light through blood-perfused tissue and analyzing the modulation of light through the cardiac cycle. In clinical practice, elevated dysfunctional hemoglobin may be present in addition to, rather than instead of, hypoxemia. Thus it becomes useful to understand the interactions that exist in the measurement of the various hemoglobin species. Recently, an alternative pulse oximeter–like device for the continuous and noninvasive measurement of the percentage of carboxyhemoglobin (COHb) (%COHb) and methemoglobin (MetHb) (%MetHb) content in the arterial blood has become available (Masimo Rainbow SET® Radical 7 pulse CO-oximeter, Masimo Corp., Irvine, California). Seven or more LED wavelengths in the sensor are used to estimate arterial oxygen saturation (SaO2), %COHb, and %MetHb, reporting values of SpO2, SpCO®, and SpMet™, respectively.1

We recently reported that Masimo Rainbow SET® Radical 7 pulse CO-oximeter (Masimo Corp., Irvine, California) reports falsely high MetHb during hypoxemia, apparently confusing deoxyhemoglobin with MetHb.2 Additional testing in our laboratory has allowed Masimo to refine its system and correct errors. Among the changes made to the revised equipment include revisions to software and separation of probes for COHb and MetHb.

The goal of our study was to determine the ability of Masimo's revised pulse CO-oximeter to estimate %MetHb accurately during hypoxia.


This study was approved by the University of California at San Francisco Committee on Human Research, and informed consent was obtained from all subjects. Twelve healthy adult subjects were included, with 5 men and 7 women, spanning a range of skin pigmentation. Rainbow ReSposable R2-25 Sensor System oximeter probes (adult R2-25a disposable optical sensors (adhesive portion) with adult R2-25r reusable optical sensor, Sensor Lot: E10A398) were placed on the middle fingers of each hand of each subject and connected to 2 Radical-7 oximeters (SET software version A radial arterial cannula was placed in either the left or right wrist of each subject. Blood gas analysis to determine SaO2 and %MetHb was performed on a multiwavelength optical blood analyzer (ABL800 FLEX, Radiometer Medical A/S, Copenhagen, Denmark).

Each subject had 2 blood samples drawn while breathing room air at the beginning of each experiment. Hypoxemia was then induced to different targeted SaO2 levels (between 70% and 100%, on the basis of end-tidal gas analysis) by having subjects breathe mixtures of nitrogen, air, and carbon dioxide according to a protocol that we have described in detail previously.3 Each SaO2 plateau level was maintained for at least 30 s and until pulse oximeter readings stabilized, at which point, 2 arterial blood samples were obtained approximately 30 seconds apart. Elevated MetHb was then induced by slow IV administration of approximately 300 mg of sodium nitrite to produce a target %MetHb level of 7%–10%. During sodium nitrite infusion, blood samples were obtained every 5 minutes to confirm elevated MetHb. Hypoxemia was then induced to similar plateaus after reaching the target level of MetHb. Serial outputs from the oximeters were recorded at 1 Hz on a computer using custom software developed with LabVIEW 2009 (National Instruments, Austin, Texas).

SpMet performance was analyzed by calculating mean bias (SpMet − %MetHb), precision (SD of the bias), and root-mean-square error (Arms) over different ranges of SaO2. The effects of SaO2 and %MetHb were examined by both univariate analysis, or with both variables, either as an analysis of variance (ANOVA) (decadal SaO2 range) or as linear regression. SpMet performance was also analyzed by observing the incidence of excessive reading bias at the various levels of SaO2. Positive and negative predictive values for detecting methemoglobinemia were calculated from the observed data. SpO2 reading bias, precision, and Arms were also determined. The effects of SaO2 and %MetHb on SpO2 bias were examined in both univariate and multivariate analysis, either as an ANOVA (decadal SaO2 range) or as a linear regression. Tukey–Kramer HSD was used for any multiple comparison testing. For all statistical tests, P < 0.05 was considered significant. Data were analyzed with JMP 7.0.2 (SAS Institute, Cary, North Carolina).


Three-hundred seven total blood samples were taken from 12 subjects, with 23 to 31 samples per subject. Data covered a span of SaO2 from 74.2% to 100% and of %MetHb from 0.4% to 14.4%. The 2 oximeters (device 1 and device 2) did not differ, so data were pooled.

SpMet Reading Accuracy

Mean SpMet bias (SpMet − %MetHb) over the full range of SaO2 was 0.16%, with precision of 0.83% and Arms of 0.84%.

Effect of Hypoxia on SpMet Bias

Table 1 summarizes the SpMet reading characteristics over different ranges of SaO2. SpMet bias differed significantly over the decadal ranges of SaO2 (P = 0.02), but no range was different from the others by multiple comparison testing. Figure 1 shows the SpMet bias as a function of SaO2. With all the data combined (panel A), SpMet bias did not change with SaO2 (P = 0.06), and by examining only the data with elevated MetHb levels (%MetHB >4%, panel B), we found again that SaO2 did not affect the SpMet bias (P=0.1). Analyzing the SpMet bias with respect to both SaO2 and %MetHb with multivariate analysis showed a small but statistically significant effect of SaO2 (P = 0.02).

Table 1:
SpMet Reading Summary
Figure 1:
Pulse CO-oximeter SpMet reading bias (SpMet − percentage of methemoglobin in arterial blood [%MetHb]) as a function of measured arterial oxygen saturation (SaO2). Circles and squares indicate, respectively, readings from the 2 Rainbow devices placed simultaneously. A, All data. B, Data only for elevated methemoglobin levels (>4%). SpMet bias was significantly higher at elevated methemoglobin levels (P < 0.0001), although the slope of the bias with respect to SaO2 was not significant (P = 0.07 in panel A and 0.1 in panel B).

Effect of %MetHb on SpMet Bias

Figure 2 shows the SpMet reading bias as a function of %MetHb. Bias significantly increases as %MetHb increases (P < 0.0001). SpMet bias was significantly lower for SaO2 >95 (P = 0.003). For just data on room air (SaO2 >95%), SpMet bias was not significantly changed by increasing MetHb (P = 0.07). With hypoxia, the SpMet bias was higher at higher %MetHb (P = 0.009).

Figure 2:
Pulse CO-oximeter SpMet reading bias (SpMet – percentage of methemoglobin in arterial blood [%MetHb]) as a function of measured arterial %MetHb. Circles and squares indicate, respectively, readings from the 2 Rainbow devices placed simultaneously. A, All data. B, Data only for arterial oxygen saturation (SaO2) >95% (room air). Bias significantly increases as %MetHb increases (P = 0.0001). SpMet bias was significantly lower at SaO2 >95 (P = 0.003), although the slope of the bias with respect to %MetHb was not significant (P = 0.07).

The highest absolute SpMet bias was 3.1%, which occurred at a %MetHb level of 9.7%. There were no “reading errors,” defined as a displayed SpMet value with an absolute bias in excess of 5% in comparison with the measured %MetHb (Table 1). Performance of pulse oximetry in detecting methemoglobinemia in terms of sensitivity, specificity, accuracy and predictive value is shown in Table 2. This is also illustrated in Figure 3, which shows SpMet plotted against %MetHb, and identifies values for false and true positives and negatives for detecting %MetHb ≥10%.

Table 2:
SpMet Predictive Performance (0.4% < %MetHb ≤ 14.4%)
Figure 3:
Pulse CO-oximeter SpMet readings as a function of measured arterial percentage of methemoglobin in arterial blood (%MetHb). Values measured in room air (arterial oxygen saturation [SaO2] >95%) are shown as Xs, and data obtained during hypoxemia are shown with open squares. The horizontal and vertical lines separate the data into quadrants, representing true or false positives or negatives for detecting a methemoglobin level of 10% or more.

SpO2 Reading Accuracy

In the absence of methemoglobinemia (%MetHb <2%), SpO2 reading bias (SpO2 minus measured SaO2) showed no change over the range of SaO2 values (70%–100%), with a mean bias of 0.34% and an root-mean-square error (Arms) of 1.39%.

As %MetHb increased, SpO2 bias became more negative (P < 0.0001). In the presence of MetHb, SpO2 bias was more negative at higher SaO2 values (P < 0.0001). Figure 4 shows the SpO2 bias as a function of measured SaO2. Panel A shows all the data, and panel B shows data only at elevated MetHb (%MetHb >4%). This same interaction between MetHb and hypoxia is shown in Figure 5, where SpO2 bias is plotted as a function of %MetHb. Panel B shows only data at SaO2 >95% (room air), where increasing %MetHb makes the SpO2 bias more negative. Table 3 summarizes SpO2 bias, which is only noticeably negative in the higher SaO2 ranges.

Figure 4:
Pulse CO-oximeter SpO2 reading bias (SpO2 – arterial oxygen saturation [SaO2]) as a function of measured SaO2. Circles and squares indicate, respectively, readings from the 2 Rainbow devices placed simultaneously. A, All data. B, Data only for elevated methemoglobin levels (>4%). Bias increased as SaO2 decreased (P < 0.0001). For percentage of methemoglobin in arterial blood (%MetHb) >4%, the slope of this bias was increased (P < 0.0001), and bias was significantly more negative at higher SaO2 (P < 0.0001).
Figure 5:
Pulse CO-oximeter SpO2 reading bias (SpO2 – arterial oxygen saturation [SaO2]) as a function of measured arterial percentage of methemoglobin in arterial blood (%MetHb). Circles and squares indicate, respectively, readings from the 2 Rainbow devices placed simultaneously. A, All data. B, Data only for SaO2 >95% (room air). Bias became significantly more negative as %MetHb increased (P < 0.0001). This slope was steeper at SaO2 >95% (P < 0.0001).
Table 3:
SpO2 Reading Summary (4% < %MetHb < 15%)


Our data indicate a high level of accuracy of the new Masimo pulse CO-oximetry measurement for MetHb. Mean bias was <1%, even in the presence of hypoxia. Although there was a statistically significant effect of MetHb combined with hypoxia on the bias, the size of this effect was small. This is in marked contrast to our study of the previous version of the Rainbow oximeter, which demonstrated an extreme interaction of combined hypoxemia and methemoglobinemia.

Sensitivity, specificity, predictive value, and accuracy were all substantially improved over the earlier version of the Rainbow oximeter. The lower value of positive predictive value (66%) for detecting %MetHb ≥10% at SaO2 >95% was not clinically significant. All the false positive readings, for which the Rainbow read SpMet ≥10%, represented true %MetHb levels of >9% (yet still below 10% and therefore false positives). The large number of values very near 10% occurred because of our chosen MetHb target level of just over 10%. The value of 10% was also chosen to compare to our previous study.

MetHb still shows a statistically significant effect on the measurement of SpO2 by the Rainbow system. This effect is most noticeable in room air, in which producing methemoglobinemia creates a consistently negative SpO2 bias. This is most obviously apparent in Figure 5B, and in Table 3 where the bias at SaO2 >95% is −2.91% with an Arms of 3.28%, below the threshold for accuracy in Food and Drug Administration testing. Increasing hypoxia significantly alters this bias, in that the SpO2 bias becomes less negative during hypoxia. However, the Rainbow oximeter clearly detects and measures developing hypoxemia even in the presence of methemoglobinemia. The negative bias in room air that occurs in the presence of MetHb occurs with standard pulse oximeters as well. It could be argued that this is an advantage, because an abnormal SpO2 would alert the clinician to a clinical problem that requires investigation. This lower SpO2 accuracy is certainly not a safety concern.

The probe used with the Rainbow oximeter does not currently measure COHb. Our previous study demonstrated that elevated MetHb levels degraded the accuracy of SpCO measurement. Probes measuring COHb will require further study. Because of the similar light absorption of oxyhemoglobin and COHb in the 600- to 800-nm wavelength range,4 accurate measurement of COHb during hypoxia may be more technically challenging than that for MetHb.

Multiwavelength pulse oximetry can now measure SpO2, SpMet, SpCO, and hemoglobin. The measurement of any 1 hemoglobin species can be affected by the presence of abnormal levels of any of the other hemoglobin species. These complex interactions present significant challenges to both the development of these devices and their testing.

In conclusion, the current Masimo Rainbow pulse CO-oximeter demonstrated a high level of accuracy for measuring MetHb up to %MetHb of 14.4%, which did not degrade during hypoxia. Methemoglobinemia also does not impair the ability of the Rainbow pulse CO-oximeter to detect hypoxemia, although as with most pulse oximeters, elevated MetHb levels create lower SpO2 readings during normoxia.


1. Masimo. Radical-7 Signal Extraction Pulse CO-Oximeter with Rainbow Technology: Operator's Manual. Irvine, CA: Masimo Corporation, 2008
2. Feiner JR, Bickler PE, Mannheimer PD. Accuracy of methemoglobin detection by pulse CO-oximetry during hypoxia. Anesth Analg 2010;111:143–8
3. Feiner JR, Severinghaus JW, Bickler PE. Dark skin decreases the accuracy of pulse oximeters at low oxygen saturation: the effects of oximeter probe type and gender. Anesth Analg 2007; 105:S18–23
4. Zijlstra WG, Buursma A, Meeuwsen-van der Roest WP. Absorption spectra of human fetal and adult oxyhemoglobin, de-oxyhemoglobin, carboxyhemoglobin, and methemoglobin. Clin Chem 1991;37:1633–8
© 2010 International Anesthesia Research Society