Use of Oxyhemoglobin Saturation, Rather Than Oxygen Tension, as a Marker of Oxygenation in Cyanotic Patients

Abstract

This study analyzes clinically indicated arterial blood gas values from patients with cyanotic congenital heart disease to determine whether oxyhemoglobin saturation or arterial oxygen tension provides a better measure of oxygenation.

In the modern era, 2 methods are available to assess oxygenation: arterial oxygen tension (Pao₂) and oxyhemoglobin saturation (Sao₂). Pao₂ represents oxygen dissolved within the plasma and is measured amperometrically, whereas Sao₂ represents hemoglobin-bound oxygen and is measured by absorbance spectroscopy (ie, co-oximetry). Assessment of the degree of hypoxemia in critically ill patients may influence medication or ventilator management, the use of extracorporeal life-support, or the need for catheterization or reoperation.^1,2 Historically, the determination of hypoxemia was made using the Van Slyke apparatus manometric method³ or Pao₂, as co-oximetry was not broadly available.⁴ However, Pao₂ and Sao₂ may yield disparate information, particularly in severely cyanotic patients.

Methods

This study was approved by the Institutional Review Board at Boston Children's Hospital under a waiver of informed consent. We extracted 70 743 time-stamped clinically indicated arterial blood gas (ABG; ABL800 FLEX Co-Ox, Radiometer America) values from 2163 admissions in patients with univentricular, cyanotic congenital heart disease who were treated at Boston Children’s Hospital between January 1, 2006, and December 31, 2016.

First, we studied the relationship between Pao₂ and Sao₂ using a generalized estimating equations model to account for the correlation among repeated measurements, hemoglobin, temperature, pH, and Paco₂ (arterial partial pressure of carbon dioxide).

Second, we examined changes in Pao₂ and Sao₂ in sequential ABGs to determine the relationship of changes, stratified by degree of hypoxemia (lowest Pao₂).

Third, we evaluated the precision of Pao₂ and Sao₂ measurements on 190 separate ABGs rerun within 5 minutes on different blood gas analyzers. All analyses were completed in SPSS (IBM).

Results

First, the relationship between Pao₂ and Sao₂ was nonlinear, as expected for the oxyhemoglobin dissociation curve (Figure 1A). Within the cyanotic range, a narrow range of Pao₂ values corresponded with a wide range of Sao₂ values; the more hypoxemic the patient, the wider the standard deviation in Sao₂ (Figure 1B).

Figure 1. Clinically Derived Oxyhemoglobin Dissociation Curve and Variance of Oxyhemoglobin Saturation (Sao₂) Within Narrow Arterial Oxygen Tension (Pao₂) Increments.

A, Pao₂ and Sao₂ become increasingly variant as Pao₂ decreases. B, Within the actionable range of hypoxemia, Sao₂ correlates progressively less closely with Pao₂ as Pao₂ decreases. Note the increasingly broad ranges and SDs of Sao₂ within progressively decreasing increments of 5 mm Hg in Pao₂. ABGs indicates arterial blood gases.

Second, among sequential ABGs, Sao₂ exhibited a larger change per unit change in Pao₂, moreso in severe cyanosis (Figure 2A). When controlling for temperature, hemoglobin, pH, and Paco₂, every 1 mm Hg increase in Pao₂ was associated with a 1.30 (95% CI, 1.27-1.33) percentage point increase in Sao₂. In serial ABGs, Pao₂ and Sao₂ changed in opposite directions in more than 25% of ABGs, at times even when changes in Pao₂ exceeded 4 mm Hg between sequential ABGs (Figure 2B).

Figure 2. Relationship Between Changes in Oxyhemoglobin Saturation (Sao₂) and Changes in Arterial Oxygen Tension (Pao₂) in Serial Arterial Blood Gases (ABGs).

A, The relationship between changes in Sao₂ and Pao₂ on sequential ABGs are shown, stratified by the lowest Pao₂ of the pair. Slopes of linear regression lines (m) increase with increasing hypoxemia (P < .001). The lines indicate linear regression ±95% CI. B, The frequency of opposite changes in Pao₂ and Sao₂ was stratified by degree of hypoxemia.

Third, the precision of both Pao₂ (interclass coefficient [ICC], 0.987; 95% CI, 0.980-0.991) and Sao₂ (ICC, 0.982; 95% CI, 0.962-0.990) was excellent. The median (interquartile range) change in Pao₂ between repeats was 0.5 (0.3-0.7) mm Hg and in Sao₂ was 0.7 (0.6-1.2) percentage point.

Discussion

Within the clinical range that we studied, the correlation between Pao₂ and SaO₂ worsened with hypoxemia. Narrow ranges of Pao₂ (eg, 30-35 mm Hg) were associated with wide ranges of Sao₂ (eg, 28%-85%), with widely disparate prognostic and treatment implications. Because the magnitude of change in Sao₂ in the cyanotic range was greater than that in Pao₂, and because the precision of Pao₂ and Sao₂ measurements was similar, the two may trend in opposite directions; a patient believed to be worsening on the basis of Pao₂ may frequently have improved based on Sao₂, and vice versa.

The arterial Po₂ range of many cyanotic patients is normally that of the tissues rather than arterial blood. In this range, the hemoglobin system is designed to be responsive to changes in surrounding temperature, pH, Pco₂, and [2,3-diphosphoglycerate], optimizing oxygen release to tissues.⁵ However, in the arterial system, these factors create significant variance between Pao₂ and Sao₂, as does the concentration of hemoglobin F.⁶ Because (1) arterial oxygen content is predominantly determined by Sao₂, (2) the magnitude of change in Sao₂ is higher than that in Pao₂ (particularly in profound hypoxemia), (3) Pao₂ and Sao₂ sometimes trend in opposite directions on the same AGBs, and (4) the precision of Sao₂ by co-oximetry is excellent, we believe that measured Sao₂ provides a superior assessment of oxygenation in cyanotic patients.