Oxygen Reserve Index: A Novel Noninvasive Measure of Oxygen Reserve—A Pilot Study

THE relationship between the arterial oxygen saturation (Sao₂) and the partial pressure of oxygen (Pao₂) is characterized by the oxyhemoglobin dissociation curve that incorporates three distinct ranges: hypoxia, normoxia, and hyperoxia. The relationship follows a well-described sigmoidal curve until Pao₂ reaches about 80 mmHg or greater and the corresponding Sao₂ remains nearly constant at 100%. But at higher oxygen partial pressures, Sao₂ cannot increase further. Thus, pulse oximetry (Spo₂) cannot assess oxygen partial pressure once hemoglobin is fully saturated. That the relationship between oxygen partial pressure and hemoglobin saturation plateaus when Pao₂ exceeds about 80 mmHg is generally of little consequence because anesthesiologists are most concerned about low partial pressures of oxygen and rather less so about the large range of partial pressures that fully saturate hemoglobin.

However, there are situations in which it is helpful to know the partial pressure of oxygen in full saturated blood. For example, very high partial pressures of oxygen—even for minutes—promote atelectasis¹ ; and over days, high partial pressures cause pulmonary injury in critical care patients²  and retinal injury in premature infants.³  But there is also a common situation in which having an estimate of oxygen partial pressure of fully saturated blood would be clinically useful: during induction of anesthesia. Most patients are given 80 to 100% oxygen for several minutes before induction of anesthesia. The primary reason is that filling the functional residual capacity of the lungs provides several minutes worth of oxygen “reserve” in case of airway compromise.

Once desaturation starts during an airway crisis, it typically progresses rapidly to potentially lethal levels.⁴  Desaturation during anesthetic induction is especially rapid in premature babies, infants, and young children,^5–7  as well as in obese children⁸  and those suffering from upper respiratory infection.⁹  Obtaining Pao₂ values from arterial blood samples are not helpful in such situations because analysis takes far too long, and securing the airway rather than obtaining arterial blood is the appropriate priority. During a crisis, having a reliable estimate of time remaining before hypoxemia becomes critical would help guide management. For example, it would be perfectly reasonable to again attempt intubation knowing that several minutes remained before desaturation, whereas some other ventilation strategy might be more appropriate if only 30 s remained.

Recently, Masimo Corporation (USA) developed a novel continuous and noninvasive measurement called the Oxygen Reserve Index (ORI). The ORI is a nondimensional index that ranges from 1 (much reserve) to 0 (no reserve) and is measured by optically detecting changes in Svo₂ after Sao₂ saturates to the maximum (see  appendix). At this time, the device is not cleared in the United States and is limited to investigational use.

In this pilot study, we tested the hypothesis that the ORI provides a clinically important warning of impending arterial desaturation in pediatric patients during induction of anesthesia. Our primary outcome was the time that elapsed between activation of the Oxygen Reserve alarm until saturation reached 98% without ventilation, that is, the warning time the index provided of impending desaturation.

After obtaining approval from the institutional review board (University of Texas Southwestern, Dallas, Texas), we obtained written informed consent from the parents of 33 pediatric patients and assent from 8 patients older than 10 yr. This prospective cohort study was conducted at Children’s Medical Center in Dallas, Texas with patients enrolled during a 4-month period extending from January to April, 2014.

Inclusion criteria were pediatric surgical patients with American Society of Anesthesiologists physical status I and II, scheduled for general anesthesia with orotracheal intubation. We excluded patients with substantial cardiorespiratory compromise and those with known or anticipated difficult intubation.

Anesthesia was induced with 8% sevoflurane in 100% O₂ supplemented by intravenous propofol (2 to 3 mg/kg) and fentanyl (1 to 2 μg/kg). Muscle relaxants were not used. After patients became apneic, the trachea was intubated, and the endotracheal tube position was confirmed by the end-tidal carbon dioxide response to a single tidal volume breath. The anesthesia circuit was then disconnected to eliminate apneic oxygenation, and saturation was allowed to decrease to 90%. Subsequently, the anesthesia circuit was reconnected, and patients were ventilated with 100% oxygen until Spo₂ returned to 99 to 100%. Thereafter, anesthesia continued per routine.

The ORI and the Spo₂ were measured simultaneously at 1-s interval with a pulse co-oximeter sensor (R1 25L) placed on the patient’s toe and connected to a Radical-7 pulse oximeter (Masimo, USA). The monitor averaging time was set at 8 s, per the manufacturer’s default.

The Radical-7 pulse monitor, used for this study, was equipped to collect optical raw data needed to compute ORI but was not equipped to display it. Throughout the study, investigators were, thus, aware of oxygen saturation but not the ORI values that were subsequently provided to the investigators by the manufacturer after offline analysis of the collected raw data. The time at which the ORI alarm would have started was also calculated offline using the manufacturer’s proprietary algorithm. Alarm activation was based on the fractional rate of change in ORI rather than on a specific oxygen reserve value. The alarm stopped when the ORI reached its minimal value of 0.

The ORI and Spo₂ values were recorded from a Masimo Pulse Co-Oximeter Sensor at the beginning of apnea, beginning and end of intubation, beginning and end of the Oxygen Reserve alarm, and 2 min after reoxygenation. The beginning of induction (mask application) was designated elapsed time 0. The ORI was also recorded when Spo₂ decreased to 98 and 90% or recovered from 90 and 98%. We also recorded the total apnea time defined as the time from the beginning of apnea until Spo₂ reached 90% and ventilation was reinstated. The early warning time was defined from the beginning of the rapid decrease of the oxygen reserve values (indicated by the start of the ORI alarm), until saturation reached 98%.

The warning time is presented a as median warning time (interquartile range [IQR]) and 95% CI for mean, whereas the rest of the results are presented as mean ± SDs. SAS 9.3 software (SAS Institute Inc., USA) was used to conduct the analyses.

Thirty-three patients were enrolled. Eight resumed spontaneous ventilation during the study period and therefore did not reach the target Spo₂ of 92%. Therefore, these eight were excluded from analysis, leaving 25 patients whose results we report. Demographic and morphometric characteristics of the participating patients are presented in table 1. All patients had a Spo₂ of 100% at the beginning of the apneic period.

The induction time (from the beginning of induction with mask application until patients became apneic) was 3.7 ± 1.9 min. Thereafter, the ORI slowly and progressively decreased over a mean of 5.9 ± 3.1 min from 0.73 ± 0.2 at the beginning of apnea to 0.4 ± 0.1 when the ORI began to decrease rapidly. Spo₂ remained 100% throughout this initial period (fig. 1).

Concurrently with alarm activation, the ORI began to decrease rapidly, and in median of 31.5 s (IQR, 19 to 34.3 s), saturation decreased to 98%. Thus, the alarm would have provided a median of 31.5 s (IQR, 19 to 34.3 s; 95% CIs for mean, 23.4 to 60.2 s) warning of impending desaturation had the system been working in real time. The mean (SD), median (IQR), and 95% CI for mean data at all study times are presented in table 2. Among 22 of the 25 patients included in our analysis, the warning time was between 0 and 50 s. The remaining three patients’ warning times were at 98, 112, and 221 s (fig. 2).

After reinstitution of ventilation, Spo₂ values continued to decline to a mean saturation of 88 ± 3% before recovering to a Spo₂ of 98% 34 ± 65 s later. The oxygen saturation nadir was 84 ± 2% (table 3).

The ORI was at its minimum value (0) when oxygen saturation was less than 98% and increased rapidly to 0.5 ± 0.3 as saturation increased to 99% and above during reoxygenation.

Mean blood pressures and heart rates at baseline, end of intubation, and at reoxygenation are presented in table 2, along with end-tidal Pco₂ at the end of intubation and at reoxygenation.

Pulse oximetry provides continuous, noninvasive assessment of arterial oxygen saturation and is a sensitive detector of hypoxemia¹⁰  and major hypoxic events.¹¹  However, oxygen supplementation delays detection of hypoventilation by pulse oximetry in both adults^12,13  and children¹⁴  because oxyhemoglobin saturation remains 100% over a wide range of oxygen partial pressures exceeding about 80 mmHg.¹⁵ 

For the same reason, it is difficult to predict when desaturation will start in apneic preoxygenated patients. For example, our patients maintained 100% oxygen saturation for an average of more than 6 min of apnea, which is consistent with previous clinical^16,17  reports and computational models.⁴  However, there was a substantial variability among patients, which means that in any individual, it would have been difficult or impossible to predict just when the rapid desaturation phase would begin.

Our major finding is that monitoring the ORI before and during intubation detected impending desaturation in median of 31.5s (IQR, 19 to 34.3 s) before noticeable changes in Spo₂ occurred. This represents a clinically important warning time that might give clinicians time for corrective actions. We are unaware of any other method that reliably provides warning that desaturation is imminent.

The ORI is a novel pulse oximeter–based nondimensional index that ranges from 1 to 0 as Pao₂ decreases from about 200 to 80 mmHg. The ORI is based on the Masimo Rainbow SET technology in which the pulsatile signal is extracted, enabling detection of changes in Pao₂ relative to the changes in Svo₂ at the measurement site. The ORI algorithm combines the Fick principle with the absorption properties of both arterial and venous hemoglobin wavelength. Specifically, as Pao₂ increases beyond 100 mmHg, Svo₂ continues to increase, even though Sao₂ has effectively saturated at 100%. This modest increase in Svo₂ above its normal value of approximately 75% will eventually stop as Pao₂ reaches significantly higher values (i.e., more than 200 mmHg). The consequent wavelength change makes it possible to detect changes in Pao₂ up to about 200 mmHg if not even higher.

Potential confounding factors include oxygen consumption, cardiac output, blood pH, Pco₂, temperature, the amount of perfusion (venous pulsation), and the presence of abnormal hemoglobins. For a more detailed explanation of the relationship between pulse oximetry and ORI, see  appendix.

The ORI was validated in an institutional review board–approved study conducted by Masimo per ISO-80601 guidelines (not peer reviewed or published)¹⁸  in 11 adult volunteers who underwent a variety of interventions to change their Pao₂ and Spo₂ levels. A total of 1,885 paired sets of ORI and Pao₂ values were collected. An index of 0.3 provided more than or equal to 80% specificity for a Pao₂ less than 150 mmHg. Although we did not evaluate Pao₂, our results suggest a consistent and clinically useful relationship between the ORI and blood partial pressure of oxygen.

This being the initial report about the ORI, there remain many questions that will have to be addressed in future studies. For example, we did not evaluate the oxygen index in the setting of abnormally high oxygen consumption states such as fever or hypermetabolic stress. We expect that the relationship between Pao₂ and ORI will remain similar, but that the “warning time” will be reduced because Pao₂ will decrease more rapidly during apnea. More importantly, we did not evaluate the correlation between ORI and Pao₂, which remains of obvious clinical interest.

Furthermore, all our patients were preoxygenated before apnea was instituted, and ORI values were recorded. Thus, it is probable that at lower Fio₂ (room air), the warning time will be significantly shorter. Future studies will also have to quantify the accuracy or reliability of ORI and the extent to which hemodynamic and other common clinical perturbations might influence the index.

The ORI measurements were obtained from a pulse co-oximeter sensor placed on the patient’s toe. Thus, sensor placement in other sites might provide different advanced warning times. Our study group was too small to permit evaluating the relationship between the ORI, age, and weight. But although both likely influence the duration of apnea that can be maintained with full saturation, neither seems likely to directly influence accuracy of the ORI, any more than pulse oximetry is much influenced by morphometric and demographic characteristics.

We evaluated the ORI in one clinical scenario, but believe that warning of impending desaturation is likely to also prove useful during airway maneuvers (suctioning, bronchoscopy, etc.) in critically ill patients. The ORI may also be helpful for assessing the effectiveness of preoxygenation before rapid sequence induction and in avoiding hyperoxia in neonates.

In conclusion, most intubations—fortunately—are rapid and smooth. But when they are not, clinicians need to make life-sustaining strategic decisions. In this pilot study, we found that during prolonged apnea in healthy anesthetized children, the ORI detected impending desaturation in median of 31.5 s (IQR, 19 to 34.3 s) before noticeable changes in Spo₂ occurred. Knowing even roughly how much time remains before the rapid desaturation phase begins seems likely to guide proper decisions.

This study was supported by Masimo Corp, Irvine, California.

Dr. Szmuk is a member of the Masimo Scientific Advisory Board and has received grant support from Masimo (Irvine, California) for previous and current research projects. The other authors declare no competing interests.

The Fick principle¹⁹  relates oxygen consumption (Vo₂) with cardiac output (CO) and oxygen content of arterial blood (Cao₂) and deoxygenated venous blood (Cvo₂).

The oxygen content equation for arterial (Cao₂) and venous (Cvo₂) blood is given by

where tHb is total hemoglobin.

Substituting the oxygen content equation for the arterial (Cao₂)²⁰  and venous (Cvo₂) blood, we are left with the following equation (where Svo₂ is the oxygen saturation in the venous blood and Pvo₂ is the partial pressure of oxygen in the venous blood):

This equation can be modified via oxygen saturation equations to the following format:

The oxygen dissociation curve provides a relationship between Sao₂ and Pao₂ as given by the equation below.

Substituting equation 2 in equation 1, we get:

Hence, for a constant oxygen consumption and cardiac output, Svo₂ is directly proportional to Pao₂, as ƒ (defined in eq. 2) is an increasing function. This results in the following relationship:

Pulse oximeters work by measuring the absorption of pulsatile blood at the measuring site (finger). Pulsatile changes are observed at the arteries, capillaries, and in the venules, although to a lesser degree in the venules. A pulse oximeter absorption measurement at wavelength λ, denoted by A(λ), is thus affected by both arterial and venous blood absorption changes.

where α << 1 and α is dependent on perfusion at the measurement site.

In the absence of dyshemoglobins:

where A_a^O₂^Hb is the absorption of oxyhemoglobin in arterial blood, A_v^O₂^Hb is the absorption of oxyhemoglobin in venous blood, A_a^HHb is the absorption of deoxyhemoglobin (reduced) in arterial blood, and A_v^HHb is the absorption of deoxyhemoglobin (reduced) in venous blood.

Substituting equations 5 and 6 into equation 4:

Combining equation 3 and equation 7, we observe that A(λ) changes as a function of Pao₂.

The units of measurement for the different components in the equations are as follows:

• Sao₂, Svo₂ = percent saturation expressed as decimal fraction (e.g., 0.98).

• tHb is measured in g/dl.

• Pao₂ and Pvo₂ are expressed in mmHg.

• Cardiac output is expressed in ml/min.

• Vo₂ is expressed in ml/min.