Emergency Department SpO2/FiO2 Ratios Correlate with Mechanical Ventilation and Intensive Care Unit Requirements in COVID-19 Patients

Author Affiliation
Gary Zhang, MD Maine Medical Center, Department of Surgery, Portland, Maine
Michael J. Burla, DO Maine Medical Center, Department of Surgery, Portland, Maine;
Southern Maine Healthcare, Department of Emergency Medicine, Biddeford, Maine;
Tufts School of Medicine, Boston, Massachusetts
Benjamin B. Caesar, MD Maine Medical Center, Department of Surgery, Portland, Maine
Carolyne R. Falank, PhD Maine Medical Center, Department of Surgery, Portland, Maine
Peter Kyros, MD Maine Medical Center, Department of Surgery, Portland, Maine
Victoria C. Zucco, BS Maine Medical Center, Department of Surgery, Portland, Maine
Aneta Strumilowska, BS University of New England College of Osteopathic Medicine, Biddeford, Maine
Daniel C. Cullinane, MD Maine Medical Center, Department of Surgery, Portland, Maine
Forest R. Sheppard, MD Maine Medical Center, Department of Surgery, Portland, Maine

Introduction
Materials and methods
Results
Discussion
Limitations
Conclusion

ABSTRACT

Background

Patients with coronavirus 2019 (COVID-19) are at high risk for respiratory dysfunction. The pulse oximetry/fraction of inspired oxygen (SpO2/FiO2) ratio is a non-invasive assessment of respiratory dysfunction substituted for the PaO2:FiO2 ratio in Sequential Organ Failure Assessment scoring. We hypothesized that emergency department (ED) SpO2/FiO2 ratios correlate with requirement for mechanical ventilation in COVID-19 patients. Our objective was to identify COVID-19 patients at greatest risk of requiring mechanical ventilation, using SpO2/FiO2 ratios.

Methods

We performed a retrospective review of patients admitted with COVID-19 at two hospitals. Highest and lowest SpO2/FiO2 ratios (percent saturation/fraction of inspired O2) were calculated on admission. We performed chi-square, univariate, and multiple regression analysis to evaluate the relationship of admission SpO2/FiO2 ratios with requirement for mechanical ventilation and intensive care unit (ICU) care.

Results

A total of 539 patients (46% female; 84% White), with a mean age 67.6 ± 18.6 years, met inclusion criteria. Patients who required mechanical ventilation during their hospital stay were statistically younger in age (P = 0.001), had a higher body mass index (P < .001), and there was a higher percentage of patients who were obese (P = 0.03) and morbidly obese (P < .001). Shortness of breath, cough, and fever were the most common presenting symptoms with a median temperature of 99°F. Average white blood count was higher in patients who required ventilation (P = <0.001). A highest obtained ED SpO2/FiO2 ratio of ≤300 was associated with a requirement for mechanical ventilation. A lowest obtained ED SpO2/FiO2 ratio of ≤300 was associated with a requirement for intensive care unit care. There was no statistically significant correlation between ED SpO2/FiO2 ratios >300 and mechanical ventilation or intensive care unit (ICU) requirement.

Conclusion

The ED SpO2/FiO2 ratios correlated with mechanical ventilation and ICU requirements during hospitalization for COVID-19. These results support ED SpO2/FiO2 as a possible triage tool and predictor of hospital resource requirements for patients admitted with COVID-19. Further investigation is warranted.

INTRODUCTION

The coronavirus 2019 (COVID-19) pandemic profoundly impacted hospital systems worldwide. Identifying patients presenting with COVID-19 in the emergency department (ED) at greatest risk for requiring mechanical ventilation or intensive care unit (ICU) care is of paramount importance since this would facilitate more efficient allocation of limited medical resources. Severe COVID-19 infection can be life-threatening and is associated with significant hypoxemia and the development of acute respiratory distress syndrome (ARDS).1,2 Identifying early predictors of respiratory failure and ICU need is vital both for patient care and logistics in the setting of a global pandemic with limited ICU resources.

The pulse oximetry/fraction of inspired oxygen (SpO2/FiO2 ratio) has previously been used as a predictor of high-flow nasal cannula failure, need for intubation, and mechanical ventilation.3 The SpO2 value has been demonstrated to be a reliable surrogate for partial pressure of oxygen in the arterial blood (PaO2),4,5 and the SpO2/FiO2 ratio does not require any blood tests. The SpO2/FiO2 ratio is a non-invasive assessment of respiratory dysfunction that can be quickly obtained at the bedside. Measured at the time of presentation, the SpO2/FiO2 ratio has been demonstrated to be an independent indication of ARDS development.6 The ability to quickly determine required level of care for vulnerable patients is essential to prevent poor outcomes, particularly in resource-limited environments. The COVID-19 pandemic led to ED crowding and a decrease in ventilator and ICU availability.7 A validated prognostic indicator tool akin to the systematic inflammatory response syndrome or Sequential Organ Failure Assessment criteria for sepsis8 is vital for ED use to identify COVID-19 patients at highest risk of ventilator and ICU need. The SpO2/FIO2 ratio predictive value has previously been validated in ARDS,6 and early measurement may serve as an indicator and triage tool in COVID-19 with regard to respiratory failure/ventilation risk and ICU need.

Our objective in this study was to evaluate ED SpO2/FIO2 ratios in COVID-19 patients and correlate them with subsequent respiratory failure, necessitating the need for ICU level of care and/or mechanical ventilation during hospitalization. Use of this ratio may help hospital systems more efficiently use resources and effectively prepare for a patient’s need for ICU care or mechanical ventilation.

MATERIALS AND METHODS

Study Design and Participant Selection

This was a retrospective study that evaluated admission encounters from both Maine Medical Center (MMC) and Southern Maine Health Care (SMHC). These institutions work closely together, with MMC being the region’s tertiary care center with over 70,000 annual ED visits and a total of 45 multipurpose ICU beds. The SMHC is a community hospital within close proximity to MMC, averaging ≈55,000 total ED visits and nine ICU beds. COVID-19 patients who were ≥18 years old and required admission to either hospital met inclusion criteria. Encounters were collected between March–December 28, 2020; thus, no patients had been vaccinated against COVID-19. Patients were excluded if they did not require admission. This study was performed under approval of the institutions’ review boards.

Data Variables

We performed retrospective chart review to identify patient demographics, diagnoses, level of hospital care, and hospital outcomes data from electronic health records. The FiO2 values were calculated using nasal cannula flow rate.9 We recorded the patient’s lowest and highest SpO2 and FiO2 values in the ED and calculated SpO2/FiO2 ratios.

Outcomes

The primary outcome was the need for mechanical ventilation. Secondary outcomes included ICU level of care, ventilator days, in-hospital complications, escalation of care following initial triage, ICU length of stay (LOS), hospital LOS, and in-hospital mortality.

Analysis

We analyzed data using RStudio 2020 (RStudio Inc, Boston, MA). Descriptive statistics were presented as frequency and percentage. Normally distributed continuous data were reported as mean with SDs, and ordinal non-normally distributed continuous data were described with medians with interquartile ranges. We used multivariable logistic regression to assess the association between either low or high SpO2/FiO2 ratios within the ED, anticoagulation use, asthma, coronary artery disease (CAD), congestive health failure (CHF), chronic obstructive pulmonary disease (COPD), diabetes, hyperlipidemia, hypertension, and gastroesophageal reflux disease (GERD), or the need for mechanical ventilation, adjusted for age and body mass index (BMI). Bivariable analysis of categorical variables was done using the χ2 test, and nonparametric variables by the Kruskal-Wallis test. Regression models controlled for both age and BMI.

RESULTS

A total of 539 patients, with a mean age 67.6 ± 18.6 years, met inclusion criteria. Patients were stratified into two cohorts based on the need for mechanical ventilation (Table 1). As shown in the table, patients who required mechanical ventilation during their hospital stay were statistically younger in age (P = 0.001), had a higher BMI (P < .001), and there was a higher percentage of patients who were obese (P = 0.03) and morbidly obese (P < .001). Shortness of breath, cough, and fever were the most common presenting symptoms, with a median temperature of 99°F. The average white blood count was higher in patients who required ventilation (P = <0.001) (Table 1). Patients requiring mechanical ventilation had higher diagnoses of ARDS (P < .001), pneumonia (P < .001), shock (P < .001), respiratory and renal failure (P < .001), and worse hospital outcomes with an in-hospital mortality of 32% vs 8% (P < .001) and a median hospital LOS of 17.5 vs 6 days (P < .001).

Table 1.Baseline characteristics of patients with coronavirus 2019.

Demographic data Mean ± SD, median, range or n (%)
Not mechanically ventilated n = 451 Mechanically ventilated n = 88 P-value
Age (median, IQR) 72, 26 66, 19.75 0.001
BMI (median, IQR) 28.9, 9.4 32.3, 10.9 <.001
Gender
 Female 217 (48%) 31 (35%) 0.03
 Male 234 (52%) 57 (65%) 0.03
Race
 Asian 11 (2%) 5 (6%) 0.03
 Black 31 (7%) 5 (6%) 0.73
 Native Hawaiian or other Pacific Islander 1 (0.2%) 0 (0%) 0
 Unknown/not reported 2 (0.4%) 2 (2%) 0.003
 More than one race 3 (0.6%) 0 (0%) 0.47
 White 397 (88%) 73 (83%) 0.20
 Other 6 (1%) 3 (3%) 0.13
Ethnicity
 Hispanic or Latino 9 (2%) 3 (3%) 0.56
 Not Hispanic or Latino 440 (98%) 84 (95%) 0.10
 Unknown/not reported 2 (0.4%) 1 (1%) 0.46
Origin
 Home 282 (63%) 54 (61%) 0.72
 Nursing home 61 (14%) 9 (10%) 0.31
 Skilled nursing home 31 (7%) 0 (0%) 0.01
 Rehab 1 (0.2%) 2 (2%) 0.03
 Other* 76 (17%) 23 (26%) 0.05
Comorbid conditions
 Alcohol use 23 (5%) 8 (9%) 0.14
 Anticoagulation therapy 52 (12%) 13 (15%) 0.44
 Asthma 66 (15%) 14 (16%) 0.81
 Cerebrovascular accident 41 (9%) 4 (5%) 0.22
 COPD 71 (16%) 16 (18%) 0.64
 Chronic heart failure 67 (15%) 13 (15%) 1
 Chronic kidney disease 73 (16%) 12 (14%) 0.64
 Cancer 57 (13%) 9 (10%) 0.44
 Coronary heart disease/heart failure 105 (23%) 19 (22%) 0.84
 Current smoker 30 (7%) 2 (2%) 0.08
 Dementia 75 (17%) 5 (6%) 0.01
 Diabetes mellitus 156 (35%) 38 (43%) 0.15
 GERD 132 (29%) 26 (30%) 0.85
 Myocardial infraction 39 (9%) 5 (6%) 0.36
 Hypertension 282 (63%) 57 (65%) 0.72
 Hyperlipidemia 222 (49%) 49 (56%) 0.23
 Morbidly obese 14 (3%) 11 (13%) <.001
 Obese 81 (18%) 25 (28%) 0.03
Presenting symptoms
 Fever 176 (39%) 34 (39%) 1
 Myalgia 72 (16%) 15 (17%) 0.82
 Arthralgias 21 (5%) 2 (2%) 0.22
 Headache 50 (11%) 4 (5%) 0.09
 GI symptoms 140 (31%) 17 (19%) 0.02
 Cough 229 (51%) 54 (61%) 0.09
 Shortness of breath 253 (56%) 57 (65%) 0.12
 Other 233 (52%) 43 (49%) 0.61
Average temperature in the ED ± SD (Fahrenheit) 97.1 ± 12.4, 99.1, 7.3–104.5 98.1 ± 10, 99, 37–103 0.48
WBC count in the ED (median, IQR) 6.2, 4.7 8, 7.4 <.001
Diagnoses
 ARDS 24 (5%) 57 (65%) <.001
 Pneumonia 183 (41%) 60 (68%) <.001
 Neurological diagnoses 128 (28%) 40 (45%) 0.002
 Renal diagnoses 129 (29%) 55 (63%) <.001
 Liver diagnoses 44 (10%) 23 (26%) <.001
 Heart diagnoses 168 (37%) 56 (64%) <.001
 Pulmonary diagnoses 280 (62%) 69 (78%) 0.004
 Shock 10 (2%) 46 (52%) <.001
 Respiratory failure 153 (34%) 75 (85%) <.001
 Renal failure 29 (6%) 22 (25%) <.001
ICU
 Patients who required ICU care at any point 75 (17%) 84 (95%) <.001
 Required more than one ICU admissions 2 (0.4%) 6 (7%) <.001
 ICU LOS (median, IQR) 2, 3 13, 16 <.001
Intubated
 Patients who were intubated 0 (0%) 84 (95%) <.001
 Days intubated n/a 2, 4
 Non-procedure based intubation 0 (0%) 51 (58%) <.001
Mechanical ventilators
 Ventilator days (median, IQR) n/a 9, 13
 Required reintubation n/a 7 (8%)
 Escalation of care from initial triage 60 (13%) 56 (64%) <.001
 Hospital LOS (median, IQR) 6, 6 17.5, 19 <.001
Discharge disposition
 Home or self-care 193 (43%) 8 (9%) <.001
 Home with services 97 (22%) 17 (19%) 0.53
 Hospice/palliative care unit 11 (2%) 1 (1%) 0.52
 Mental health/psychiatric hospital 8 (2%) 0 (0%) 0.18
 Nursing home 17 (4%) 1 (1%) 0.16
 Other 62 (14%) 34 (39%) <.001
 Rehab 15 (3%) 24 (27%) <.001
 Skilled nursing facility 48 (11%) 3 (3%) 0.02
 In-hospital mortality 36 (8%) 28 (32%) <.001

*
Other includes homeless, transfers in, group home, Primary care physician follow up, mental health facility.
BMI, body mass index; IQR, interquartile range; COPD, chronic obstructive pulmonary disease; GERD, gastroesophageal reflux disease; WBC, white blood count; ED, emergency department; ARDS, acute respiratory disease syndrome; GI, gastrointestinal; ICU, intensive care unit; LOS, length of stay.

The SpO2/FiO2 ratios in the ED and their associations with mechanical ventilation or need for ICU care are presented in Table 2. A highest obtained ED SpO2/FiO2 ratio of 300 or below was statistically associated with a requirement for mechanical ventilation during hospitalization. A lowest obtained ED SpO2/FiO2 ratio of 300 or below was statistically associated with a requirement for ICU care during hospitalization. There was no statistically significant relationship between ED SpO2/FiO2 ratios above >300 and mechanical ventilation or ICU level of care.

Table 2.SpO2/FiO2 ratios and their association with intensive care unit or mechanical ventilation needs.

Variable SpO2/FiO2 ratios* No mechanical ventilation N (%) Required mechanical ventilation N (%) 95% CI OR P-value
Lowest ED SpO2/FiO2
 0–100 18 (4) 13 (18) 2.4–10.9 5.1 <.001
 101–200 5 (1) 3 (4) 0.75–12.6 3.1 0.05
 201–300 37 (9) 14 (20) 1.2–4.7 2.4 0.005
 301–400 113 (27) 19 (27) 0.6–1.8 1.0 0.86
 401–500 251 (59) 22 (31) 0.2–0.5 0.31 <.001
Highest ED SpO2/FiO2
 0–100 7 (2) 7 (10) 2.2–19.2 6.5 <.001
 101–200 4 (1) 3 (4) 1.0–21.2 4.6 0.05
 201–300 14 (3) 11 (15) 1.7–8.1 3.7 0.002
 301–400 106 (25) 20 (28) 0.72–2.2 1.3 0.47
 401–500 293 (69) 30 (42) 0.2–0.6 0.3 <.001
Variable SpO2/FiO2 ratios* No ICU admission N (%) ICU admissionN (%) 95% CI OR P-value
Lowest ED SpO2/FiO2
 0–100 8 (2) 23 (17) 3.8–20 8.8 <.001
 101–200 1 (0.3) 7 (5) 2.3–158 19.2 <.001
 201–300 27 (8) 24 (18) 1.4–4.5 2.5 0.001
 301–400 94 (26) 36 (26) 0.6–1.5 0.93 0.66
 401–500 228 (64) 47 (34) 0.2–0.5 0.32 <.001
Highest ED SpO2/FiO2
 0–100 3 (1) 11 (8) 2.8–10 10.3 <.001
 101–200 0 (0) 7 (5) 2.3–19 19.2 <.001
 201–300 11 (3) 14 (10) 1.3–2.8 2.80 0.01
 301–400 88 (25) 38 (28) 0.78–1.2 1.21 0.66
 401–500 256 (72) 67 (49) 0.26–0.39 0.39 <.001

*
For patients who had ED SpO2/FiO2 values.
CI, confidence interval; OR, odds ratio; ED, emergency department; ICU, intensive care unit.

Chronic obstructive pulmonary disease was a confounding factor for COVID-19 patients who required mechanical ventilation (adjusted R2 value = 0.1132; P < .001). No statistically significant associations were identified between the following co-morbidities: anticoagulation use; asthma (adjusted R2 = 0.096, P = 0.75); CAD (adjusted R2 = 0.102; P = 0.07); CHF (adjusted R2 = 0.096; P = 0.95); diabetes (adjusted R2 = 0.10; P = 0.07); hyperlipidemia (adjusted R2 = 0.11; P = 0.08); hypertension (adjusted R2 = 0.096; P = 0.58); and GERD (adjusted R2 = 010; P = 0.28) for the requirement of mechanical ventilation.

DISCUSSION

This study demonstrated that the highest obtained ED SpO2/FiO2 ratio of 300 or below correlated with the need for mechanical ventilation during hospitalization. Additionally, a lowest obtained ED SpO2/FiO2 ratio of 300 or below was associated with a requirement for ICU-level care. Although COPD was a confounding factor for patients requiring mechanical ventilation, other co-morbidities were not independently associated with higher rates of mechanical ventilation and the ED SpO2/FiO2. This suggests that the SpO2/FiO2 ratio can be used as a prognostic indicator to stratify severity of illness in patients with COVID-19 during their initial evaluation in the ED. Since the SpO2/FiO2 ratio is non-invasive and can be quickly obtained and trended during a patient’s evaluation, this ratio could be an important factor in patient triage and disposition.

Multiple prognostic indicators have been proposed in the previous literature to help stratify ARDS severity and predict outcomes.1013 The PaO2:FiO2 (P:F) ratio is a widely used measure of ARDS severity; however, multiple studies have shown that the P:F ratio is not an independent predictor of mortality.1013 Another prognostic tool, the oxygenation index, (OI [FIO2/PaO2 × mean airway pressure × 100]) has been demonstrated to be an independent risk factor for mortality in adults with ARDS,11,12 but it requires mechanical ventilation and arterial blood gas analysis for calculation. Oxygen saturation index (OSI [FIO2 × mean airway pressure × 100)/SaO2]) is a measure that correlates to OI and is an independent predictor of clinical outcomes.12 Although OSI calculation does not require blood analysis, it still requires mechanical ventilation. Another prognostic tool, the Lung Injury Prediction Score (LIPS), has applicability in the ED.13 However, the LIPS tool requires a detailed past medical history (e.g, alcohol use disorder) and the patient’s pH, requiring a blood gas. Although all these tools provide some prognostic value, each has limitations, resulting in barriers to deployment for triaging patients in the ED.

In contrast, the SpO2/FiO2 ratio requires no blood tests and is quickly and easily obtained at the bedside. Measured at the time of presentation, it has been shown to be an independent indication of ARDS development.6 This study suggests that the SpO2/FIO2 ratio may offer an estimate of disease severity in patients with COVID-19 before progression to overt respiratory failure, serving as a triage tool to identify those at greatest risk for needing mechanical ventilation and critical care. The SpO2/FiO2 ratio can be used as a tool or part of a protocol to assess whether a patient meets transfer criteria within a hospital system. Many regional health systems operate under a “hub and spoke” model where a large central institution supports a network of smaller hospitals. Rapid identification of patients at risk for decompensation and with need for higher level care would facilitate access to limited critical care resources while also decreasing the incidence of over-triage to the hub hospital.

LIMITATIONS

The study is retrospective with inherent limitations in controlling confounding variables. The cohort was limited to one hospital system, and thus cannot account for practice variations in other healthcare systems. The hospitals evaluated in this study may have had different criteria for ICU admission. Additionally, FiO2 values were based largely on nasal cannula flow rates; limiting to high flow nasal cannula would permit more accurate FiO2 but would also limit applicability. At the time of data collection, no patients were vaccinated, thus limiting the applicability of findings to populations with some form of COVID-19 vaccination.

CONCLUSION

In summary, ED SpO2/FiO2 ratios correlate with mechanical ventilation and ICU requirements during hospitalization for COVID-19 infection. These results support ED SpO2/FiO2 as a triage tool and predictor of hospital resource requirements for patients admitted with COVID-19. Further study is required with a prospective analysis assessing accuracy of the SpO2/FiO2 ratio in predicting mechanical ventilation and need for ICU-level care.

Footnotes

Section Editor: Christopher Tainter, MD

Full text available through open access at http://escholarship.org/uc/uciem_westjem

Address for Correspondence: Michael J. Burla, DO, Southern Maine Health Care, Department of Emergency Medicine, Southern Maine Health Care, 1 Medical Center Dr, Biddeford, ME 04005. Email: Michael.burla@Mainehealth.org
05 / 2024; 25:325 – 331

Submission history: Revision received March 31, 2023; Submitted January 8, 2024; Accepted January 10, 2024

Conflicts of Interest: By the WestJEM article submission agreement, all authors are required to disclose all affiliations, funding sources and financial or management relationships that could be perceived as potential sources of bias. This work was supported in part by the Northern New England Clinical and Translational Research grant U54GM115516. There are no conflicts of interest or other sources of funding to declare.

REFERENCES

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.