Non-invasive SpO2/FiO2 ratio (SFR) as surrogate for PaO2/FiO2 ratio (PFR): A scoping review

The study protocol was drafted concerning the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMAScR) [8]. We did a comprehensive search for articles by accessing databases such as Scopus, PubMed, Web of Science, and CINAHL database. Keywords used to find articles included SFR, PFR, oxygenation status, oxygen, and non-invasive tool. Articles from 2000 to 2024 were searched and Rayyan.ai software was used to delete any duplicates.

Population: Adult, pediatric, and neonates

Intervention: Comparison of SFR with PFR in determining the utility of non-invasive tools. Use of SFR in combination with other parameters to determine the success or failure of any critical care treatment and in diagnosing any pulmonary conditions. Title and abstract screening, followed by full-text screening, was conducted by two reviewers using a standardized template aligned with the inclusion and exclusion criteria.

A total of 575 articles were initially screened, focusing on the title, abstract, and availability of full text for this scoping review. The search databases used were Web of Science (433), Scopus (112), PubMed (29), and CINAHL (1). Data extracted from the articles included the year of publication, study location, population size, diagnosis, use of the SFR, and outcomes. After applying the inclusion and exclusion criteria, 32 articles were included in the final review. Articles excluded during title selection primarily focused on medication effects on critical care diseases, therapeutic interventions for critical care disorders, or other parameters impacting oxygenation that did not specifically examine the SFR or PFR. After screening titles, abstracts were reviewed, leading to further exclusions of studies that did not align with our inclusion criteria. Only one article, initially deemed relevant based on title and abstract, was excluded due to the unavailability of the full text. The article screening process is outlined in detail in Figure 1.

Fig. 1.

PRISMA flow diagram

Most studies included in the review were published between 2020 and 2024. Geographically, most studies were conducted in the United States of America, India, South Korea, Spain, Thailand, France, Italy, Japan, Pakistan, Brazil, China, and Iran. The majority of the studies were retrospective in design [9], with a few prospective observational studies and two cross-sectional studies. No randomized controlled trials were included. To statistically evaluate the reliability of the SFR, univariate and multivariable logistic regression models were used, along with receiver operating characteristic (ROC) analysis and area under the curve (AUC) to assess the performance of this non-invasive tool. Regression analysis was also employed to establish the relationship between the SFR and PFR. The studies included are detailed in Table 1.

A total of 81,637 patient records were collected, including arterial blood gas values, primarily the PFR, vital signs (mainly pulse oximetry values), mechanical ventilator settings such as F_iO₂, and patient diagnoses. The study population included adults, pediatrics, and neonates admitted to critical care units for various respiratory illnesses. The primary focus was on the intensive care setting, where most respiratory-related diseases are observed. The most common pathologies/diagnoses among these patients were acute hypoxemic respiratory failure, acute lung injury, and acute respiratory distress syndrome (ARDS)[10,11,12,13,14,15,16,17,18,19,20,21,22] followed by COVID-19 (SARS-CoV-2) [23,24,25,26,27,28,29,30], cardiothoracic surgeries [31,32], sepsis [33], pulmonary embolism [34], asthma [35], and interstitial lung disease [36]. Some studies included data from all patients admitted to critical care during a specified period, covering all age groups [37,38,39,40,41].

Almost all the studies included in this review have utilized SFR in some capacity to assess its utility in daily clinical practice. Many studies have evaluated SFR as a surrogate for PFR, while others have examined its effectiveness as a diagnostic tool, particularly for diagnosing acute lung injury and ARDS in both adult and neonatal populations. Some studies have investigated SFR as a predictor for success or failure in high-flow nasal cannula (HFNC) therapy. A few studies have explored the use of SFR in combination with other tools, such as blood glucose [25], ROX index [8,24, 26,29,41] and oxygenation saturation index (OSI) [17,19,21,24,37] to predict weaning outcomes. One study focused on using SFR to predict mortality in critically ill patients [39]. Additionally, some studies evaluated the SFR role in determining hemodynamic deterioration in pulmonary embolism [34]. Few studies assessed whether SFR could serve as a stand-alone tool for diagnosing ARDS [10,14,21,28,29], the success of High flow nasal cannula (HFNC) therapy [24,26] and for screening hypoxia [28,31,37]. In essence, all these studies aimed to replace PFR with SFR to assess the efficacy of the latter as a non-invasive alternative and found that SFR can be replaced and used to determine the various outcomes in critical care set-up. The use of SFR was not only limited to critical care settings but was also employed in emergency departments [23,28] to quickly assess the need for oxygen therapy and identify the worsening of lung conditions leading to ARDS or the potential need for mechanical ventilation. In the COVID-19 population, SFR was primarily used to monitor the response to therapy, evaluate whether the patient’s condition was improving or deteriorating, and predict the potential need for higher levels of ventilation.

The statistical test used to assess the correlation between SFR and PFR was Spearman’s correlation in most of the studies. The area under the curve (AUC) for sensitivity and specificity was employed to evaluate the performance of both SFR and PFR. With a 95% confidence interval (CI), all the studies included in this review demonstrated statistically significant sensitivity and specificity (p value <0.05) for SFR, highlighting the utility of this non-invasive tool.

Various cut-off values were identified for different levels of PFR. In addition, cut-off values were assessed for other purposes, such as predicting ICU admission and determining the success or failure of HFNC therapy. The majority of studies found that an SFR of <300 indicated worsening of respiratory condition or hypoxemia and a need for mechanical ventilation. However, the various cutoff values identified were; for a PFR of 300, the identified cut-off values for SFR were 241(AUC 0.74; 95% CI) [33], 315 (Positive likelihood ratio 2.06, 95% CI of 1.64 – 2.76) [10], 263 [for ALI – 93% sensitivity; 95% CI (264 – 272)] [11], 201 201 [for ARDS – 84% sensitivity, 95% CI 206 – 211)] [11], 296 [95% CI (292–314)] [37], 321 (AUC 0.87; sensitivity −68.9%, specificity – 95%) [21], and 311 (Sensitivity 90%, Specificity 80%) [32]. Another study found that a cutoff of SFR <433 (AUC 0.95; p < 0.001) accurately identified ARDS (as defined by PFR of <300) [27]. These values varied depending on the disease condition and the time of data collection. A threshold of 350 indicates PFR < 300 [AUC: 0.91 (CI 95% 0.885–0.950)], while a threshold of 470 indicates PFR < 400 [AUC:0.9 (95% CI 0.872–0.930)] among COVID-19 population [28]. One study found that an SFR value of 236 [95% CI (230–245)] [37], 252 (AUC 0.83; sensitivity −68.9%, specificity – 95%) [21] and 235 (positive likelihood ratio 5.64; 95% CI of 4.69 – 7.08) [10] corresponds to a PFR of 200 or less, while an SFR value of 146 [95% CI (140–148)] corresponds to a PFR of 100 or less [37]. To identify moderate ARDS (identified by PFR <200), SFR of <336 was found to be moderately accurate (AUC 0.71, 95% CI 0.944–0.969, p < 0.001) [27]. One study found an SFR of 154 (AUC 0.92, 95% CI) [33] corresponded to a PFR of 100. No single cut-off value was established for a specific PFR.

Several studies in the review identified cut-off values for different clinical purposes. One study found that an SFR of 230 predicted failure of high-flow nasal cannula (HFNC) therapy, with a sensitivity of 78.0% and specificity of 68.7% (95% CI 5.06–35.84) [16]. The SFR was also used to predict ICU admission in pediatric patients, with a cut-off value of <300 for predicting PICU admission in wheezy children, and a best predictive cut-off of <178.79 [AUC (95% CI) 0.841 (0.767 – 0.914)] for determining the need for intensive care [35]. In the context of COVID-19 in adults, one study suggested a SFR cut-off of 300, which indicated a threshold for predicting the imminent need for mechanical ventilation²⁵. Additionally, a study using the SFR to predict mortality found that mortality significantly increased when the SFR dropped below 200, identifying this as a critical cut-off for assessing hypoxemia [39]. A cutoff value of <98.5 (AUC 0.68, p <0.05) predicted non-invasive positive pressure ventilation in one of the studies [12]. Another finding suggests that a cutoff value of <260 (AUC 0.81; p < 0.05) predicts hemodynamic deterioration, demonstrating a sensitivity of 74% and a specificity of 88% [34].

Our review highlights SFR as a valuable alternative tool for assessing oxygenation status, supported by the findings of multiple studies. SFR offers several advantages; first, as a non-invasive method, it avoids complications associated with arterial punctures, such as infection, hemorrhage, and increased costs. Second, it allows for quick acquisition of values, enabling timely evaluation of a patient’s response to therapy and overall oxygenation status.

SFR has demonstrated utility in various clinical scenarios, aiding physicians in decisions such as patient admission to triage, predicting the success or failure of mechanical ventilation strategies, and serving as a diagnostic and screening tool in ARDS settings. While different studies proposed varying cut-off values, a collective cut-off of SFR <300 was commonly used to predict worsening conditions and the potential need for higher levels of mechanical ventilatory support. The application of the SFR across these scenarios underscores its benefits as a practical and effective tool in clinical practice.

Several factors can influence the accuracy and reliability of the SFR in assessing oxygenation status. The PFR provides a more precise measure of oxygenation as it is derived from direct blood gas analysis. However, various clinical factors, such as patient agitation, endotracheal suctioning, and positioning, can impact P_aO₂ measurements [33]. Similarly, pulse oximetry readings used to calculate the SFR can be affected by skin pigmentation, nail polish, shock, anemia, and conditions like methemoglobinemia [33, 39]. While these factors may not cause significant variations in cut-off values for predicting deterioration, they should be considered when interpreting results. Additionally, in diseases that severely impair respiratory function, a high prevalence of oxygenation dysfunction can enhance the positive predictive value in statistical analyses. Risk factors contributing to respiratory conditions may further influence the accuracy of the SFR. Therefore, it is crucial to account for these variables before making clinical decisions based solely on SFR measurements [38].

In patients with an S_pO₂ of >97%, the values lie on the flat portion of the oxygen dissociation curve. While an S_pO₂ >97% is generally indicative of good oxygenation, there are instances where the P_aO₂ value may be lower than expected. Since pulse oximetry has an upper saturation limit of 100%, this can influence SFR readings, potentially masking underlying hypoxemia. However, this does not imply that the SFR is an unreliable tool; rather, it underscores the importance of considering other clinical factors and disease conditions when interpreting results to ensure accurate assessment and clinical decision-making [37, 38].

Certain studies have utilized heterogeneous samples, leading to the generalization that the SFR is a reliable tool for assessing oxygenation status, particularly in the pediatric population, regardless of mechanical ventilation status. However, its reliability may be compromised in specific conditions, such as carbon monoxide poisoning, where pulse oximetry readings can be misleading due to the presence of carboxyhemoglobin [37].

Further research is needed to establish a standardized SFR cut-off value comparable to the PFR for practical implementation in assessing oxygenation status. Prospective observational studies and clinical trials focusing on its use in early ARDS detection could help achieve this goal.

A few limitations were identified during this review. A significant number of included studies were retrospective, which inherently limits control over confounding variables. While a few prospective observational studies were reviewed and included to strengthen our findings, their number was relatively small. Additionally, not all studies exclusively focus on assessing the correlation between SFR and PFR. Many incorporated SFR in combination with other indices, such as ROX index, or utilized it as a substitute for PFR, leading to variability in findings. Moreover, while substantial evidence supports the use of SFR in adult populations, there remains a limited number of studies addressing its application in pediatric and neonatal populations. This lack of uniformity and limited data for younger age groups highlight the need for further research to establish more consistent and comprehensive findings.