1 Introduction

Conventional pulse oximetry (SpO2) provides valuable information about the oxygen saturation in blood which has been proven to have a good correlation with arterial saturation of oxygen (SaO2) [1, 2]. However, despite it being a most significant advance in patient monitoring, SpO2 is unable to differentiate between states of normoxia and hyperoxia in healthy patients receiving supplemental oxygen [3] where, for example, a reading of more than 97% can be associated with a wide range of PaO2 (80mmHg ~ 600mmHg). Although pulse oximetry has become accepted universally as part of the standard of care in clinical practice as it can detect states of hypoxemia, it remains limited in not being able to detect hyperoxia, which in extreme situations could cause harm through upregulating oxidative stress pathways or paradoxically impairing oxygen delivery to organs [4, 5]. Conversely, moderate hyperoxia could be desirable in certain situations like pre-oxygenation before induction of anesthesia [6] or during off pump coronary artery bypass grafting (CABG) surgery [7] to counteract the impaired oxygen supply and delivery caused by those situations.

The Oxygen Reserve Index (ORI) is a novel noninvasive and continuous dimensionless parameter intended to provide insight into the oxygenation status of a patient within the moderate hyperoxic range (PaO2 > 100 and ≤ 200 mm Hg) [8,9,10]. Previous studies in the pediatric population have shown that ORI detects impending desaturations before noticeable changes in SpO2 occur [11] and can be used as an indicator of moderate hyperoxia intraoperatively [12]. In this study, we investigated if the relationship between ORI and PaO2 is preserved in the setting of open-heart surgery. An intact relationship would allow for tighter control of oxygen levels to achieve moderate hyperoxia while potentially allowing early detection and preemptive interventions to mitigate imminent hypoxemia.

2 Methods

2.1 Trial Design

This study was a prospective observational, proof-of-concept, single center study conducted at the Babtain Cardiac Center, Dammam, Saudi Arabia, over a 24-month period between September 1st 2019 and August 31st 2021. The study was approved by the local Institutional Ethics Committee (IRB-2018-17). Written informed consent for enrollment and to use the data was obtained from each patient or from legal surrogate.

2.2 Patients

Eligible patients were adults scheduled to undergo non-emergent open heart surgery including valve, aortic root, and CABG surgeries either on or off cardiopulmonary bypass. Exclusion criteria were severe restrictive or obstructive respiratory disease, emergency cardiac surgery, and patients coming on inotropic support, vasodilators or established mechanical ventilation.

2.3 Interventions

Enrolled patients were continued on all their preoperative medications, except ACE Inhibitors, until the morning of the surgery. Additionally, all patients received anxiolysis with oral diazepam (5-10mg) the night before the surgery and 1-2mg lorazepam on the morning of surgery. All patients were connected to a 5 lead ECG and noninvasive blood pressure monitors. A pulse co-oximeter sensor (R1 25L Masimo Rainbow Disposable RD Lite; Masimo Inc, Irvine, CA, USA) was placed on the right index finger and connected to the Radical 7 monitor (part of the Massimo Root Platform) for SpO2 and ORI measurements. An opaque shield to prevent exposure to ambient light was applied over the sensor.

A 20 gauge arterial cannula was inserted under local anesthesia in the one radial artery using an aseptic technique. All paired arterial blood gas (ABG) samples were drawn from the radial artery catheter and were collected into standard 1-ml heparinized syringes. To ensure the accuracy of blood gas analysis, the samples were deaerated, carefully mixed and analyzed promptly after drawing them. A baseline ABG sample along with ORI and SpO2 readings were collected and recorded on room air before induction of anesthesia.

Subsequently, general anesthesia was induced in a standardized technique using midazolam, fentanyl, and etomidate with rocuronium used for paralysis. Ventilation was facilitated using a Drager Primus anesthetic machine and maintenance of anesthesia was done using sevoflurane. Invasive and noninvasive hemodynamic monitoring was maintained all throughout the procedure and hemodynamic stability was ensured with maintenance of Mean Arterial Pressure (MAP) of 60 mmHg or more by judicious use of vasopressors.

After induction of general anesthesia and endotracheal intubation, FiO2 was reduced to 0.22 and the first blood gas sample was drawn along with a recording of the ORI and SpO2 at that moment. Then, FiO2 was increased in steps of 0.08 and blood gas samples were drawn after maintenance of each interval step for 5 minutes at least. Sample analysis was paired with recordings of ORI and SpO2 at each step.

2.4 Data Collection and Analysis

Electronic data was retrieved and stored from the standard anesthesia (Drager Primus) and Radical-7 monitors and imported into Microsoft Excel 2010 (Microsoft, Redmond, WA) for analysis. A complete set of data was collected as per protocol. A visual inspection of the data plots was performed to detect any artifact-induced outliers or missing data.

2.5 Statistical Analysis

Data was tabulated and analyzed using SPSS program (Statistical Package for Social Sciences) version 26 (IBM Inc, Chicago, IL) in the following domains:

  • Descriptive analysis (mean ± SD) for numerical data

  • Frequency and percentage of non-numerical data

  • Pearson correlation coefficient (r) was used to correlate two numerical parameters

  • Linear regression curves were done for all correlations

  • p value was considered significant if it was < 0.05 and highly significant if it was < 0.01, this applied to all tests.

  • Scatter plot was created to visualize the relationship between ORI and PaO2 using Microsoft Excel 2016

3 Results

During the period between September 1st 2019 and August 31st 2021, 50 patients who met the inclusion criteria were recruited after giving consent to be included in the study. A total of 500 paired readings (PaO2 with corresponding SpO2 and ORI) were collected. Characteristics of the patients and respective procedures are presented in the Table 1.

Table 1 Patient Characteristics
Full size table

3.1 ORI Sensitivity Range

Where PaO2 was in the range of 100–200 mmHg, there was a highly statistically significant correlation with ORI (r = 0.8193, p < 0.001, Fig. 1). Additionally, a significant correlation was observed between ORI and FiO2 (r = 0.693, p < 0.001, Fig. 2) as well as ORI and SpO2 in the range of 95% and 100% (r = 0.529, p < 0.05, Fig. 3). Additionally, we can observe that where SpO2’s sensitivity and specificity tails off drastically once it reaches 100%, ORI continues to be more sensitive and specific at higher PaO2 values and thresholds (Fig. 4).

Fig. 1
figure 1

- Relationship between PaO2 and ORI

Full size image
Fig. 2
figure 2

Relationship between FiO2 and ORI

Full size image
Fig. 3
figure 3

- Relationship between ORI and SpO2

Full size image
Fig. 4
figure 4

– AUC comparing SpO2 and ORI specificity and sensitivity at PaO2 thresholds of 120mmHg and 130mmHg

Full size image

3.2 Association of PaO2 with ORI Outside the Sensitive Range

An additional finding of the study was that an ORI of 0 consistently correlated with a PaO2 of less than 90mmHg and an SpO2 ranged between 87% and 96% (n = 44 data points)

4 Discussion

The main finding of this prospective, observational study is a significant relationship between the ORI and SpO2, FiO2 and PaO2 in high risk, adult cardiac surgical patients. Previous data showed a linear correlation between ORI and PaO2 values between 62 mmHg and 240 mmHg [10] and 100 mmHg and 200 mmHg [12]. The findings from these studies in addition to ours suggest that ORI reasonably correlates to PaO2 in the mild and moderate hyperoxic range.

When subjected to extremely elevated oxygen tensions, cardiac surgery patients on CPB exhibit undesirable sequalae including a reduction in lung vital capacity and an elevation in proinflammatory markers leading to oxidative myocardial damage [6] [13]. The use of ORI in this patient population can allow more careful titration of oxygen levels to avoid potentially dangerous hyperoxic states.

In the wider surgical patient population, hyperoxia may cause vasoconstriction and increase microcirculatory heterogeneity, thereby compromising perfusion and, hence, actual oxygen delivery [5], which in a wide range of conditions, such as ischemic heart disease, stroke and after cardiac arrest, is associated with increased morbidity and mortality [7]. This information also reflects the potential of ORI monitoring in the critical care setting. Physicians can utilize it in patients receiving supplemental oxygen through modalities including invasive and noninvasive ventilation, preventing undesirable elevated PaO2 levels that have been associated with increased mortality [15].

About 4% of surgical patients have been shown to develop postoperative respiratory failure resulting in more than 25 times the in-hospital mortality from respiratory failure seen in other patients (10.3% vs 0.4%) [1]. These patients as well as many other groups of patients at risk of hypoxaemia, could benefit from ORI in addition to pulse oximetry as it can provide an early warning of impending oxygen desaturation, allowing for early intervention and possibly improving outcome [2]. When used as a monitoring tool in the process of induction and intubation, an ORI value of 1, can allow the physician to expect PaO2 levels to be at a level equal to or higher than 200mmHg, this information can be used as an indicator that denitrogenation is complete and the process of intubation can proceed. During this apneic period, ORI will provide an early warning of impending hypoxia ahead of SpO2 changes, allowing the clinician to make a better-informed decision on whether to persist with the attempt of intubation, or abandoning the procedure in order to oxygenate the patient [16]. This utility of ORI has been demonstrated previously in different patient populations, including pediatric [12] and obese patients [14]. The same data in Fig. 1 alludes to a potential utility of ORI in procedures requiring apneic oxygenation (such as bronchoscopy), or deflation of the lungs (such as sternotomy or Internal Mammary Artery harvesting in cardiac surgery), allowing the physician to act earlier before impeding hypoxia occurs.

In conclusion, this study shows that the consistent relationship between ORI and PaO2 in the normoxic and moderately hyperoxic range can allow more precise titration of oxygen therapy to prevent unintentional hyperoxia, and facilitate oxygen therapy in the mild and moderate hyperoxic range even in patients undergoing major open heart surgery. Additionally, there is a potential for ORI to be used as an early warning system for desaturation, but that would require further studies.