Dynamic blood oxygen indices in mechanically ventilated COVID-19 patients with acute hypoxic respiratory failure: A cohort study.

BACKGROUND: Acute hypoxic respiratory failure (AHRF) is a hallmark of severe COVID-19 pneumonia and often requires supplementary oxygen therapy. Critically ill COVID-19 patients may require invasive mechanical ventilation, which carries significant morbidity and mortality. Understanding of the relationship between dynamic changes in blood oxygen indices and clinical variables is lacking. We evaluated the changes in blood oxygen indices-PaO2, PaO2/FiO2 ratio, oxygen content (CaO2) and oxygen extraction ratio (O2ER) in COVID-19 patients through the first 30-days of intensive care unit admission and explored relationships with clinical outcomes. METHODS AND
FINDINGS: We performed a retrospective observational cohort study of all adult COVID-19 patients in a single institution requiring invasive mechanical ventilation between March 2020 and March 2021. We collected baseline characteristics, clinical outcomes and blood oxygen indices. 36,383 blood gas data points were analysed from 184 patients over 30-days. Median participant age was 59.5 (IQR 51.0, 67.0), BMI 30.0 (IQR 25.2, 35.5) and the majority were men (62.5%) of white ethnicity (70.1%). Median duration of mechanical ventilation was 15-days (IQR 8, 25). Hospital survival at 30-days was 72.3%. Non-survivors exhibited significantly lower PaO2 throughout intensive care unit admission: day one to day 30 averaged mean difference -0.52 kPa (95% CI: -0.59 to -0.46, p<0.01). Non-survivors exhibited a significantly lower PaO2/FiO2 ratio with an increased separation over time: day one to day 30 averaged mean difference -5.64 (95% CI: -5.85 to -5.43, p<0.01). While all patients had sub-physiological CaO2, non-survivors exhibited significantly higher values. Non-survivors also exhibited significantly lower oxygen extraction ratio with an averaged mean difference of -0.08 (95% CI: -0.09 to -0.07, p<0.01) across day one to day 30.
CONCLUSIONS: As a novel cause of acute hypoxic respiratory failure, COVID-19 offers a unique opportunity to study a homogenous cohort of patients with hypoxaemia. In mechanically ventilated adult COVID-19 patients, blood oxygen indices are abnormal with substantial divergence in PaO2/FiO2 ratio and oxygen extraction ratio between survivors and non-survivors. Despite having higher CaO2 values, non-survivors appear to extract less oxygen implying impaired oxygen utilisation. Further exploratory studies are warranted to evaluate and improve oxygen extraction which may help to improve outcomes in severe hypoxaemic mechanically ventilated COVID-19 patients.

Introduction

The novel SARS-CoV-2 viral infection (coronavirus disease (COVID-19)) is currently imposing an unprecedented challenge for the medical community worldwide. A global pandemic was declared by the World Health Organisation (WHO) in January 2020 and continues to cause significant burden from multiple waves of varying lineages accounting for around 4.5 million case fatality to date [1]. The majority of patients develop mild illness without any significant respiratory sequelae [2]. Hypoxic respiratory failure is the hallmark of severe COVID-19 pneumonia and often requires supportive oxygen therapy via various delivery methods [3]. Development of Acute Respiratory Distress syndrome (ARDS) and persistent hypoxaemia necessitating admission to an intensive care unit (ICU) for invasive mechanical ventilation carries substantial mortality in the region of 50% [4].

As a novel cause of acute hypoxic respiratory failure, COVID-19 offers a unique opportunity to study a relatively homogeneous cohort of patients with similar underlying pathology. Compared to other critically unwell patients, this group requires high concentrations of inspired oxygen for prolonged periods and tends not to display the typical features of respiratory distress despite profound hypoxia [5]. Moreover, the presence of acute hypoxic respiratory failure and the degree of hypoxaemia, defined by the ratio of arterial partial pressure of oxygen (PaO2) to the fractional inspired oxygen (PaO2/FiO2 ratio), are independently associated with increased mortality [6]. Consequently, effective oxygen therapy via mechanical ventilation remains the mainstay of critical care management of patients with severe hypoxic respiratory failure. However, it is unclear if increments in the fractional inspired oxygen improve blood oxygen indices such as total arterial oxygen content (CaO2) and oxygen utilisation or impact the overall clinical outcomes of mechanically ventilated COVID-19 patients with severe hypoxic respiratory failure.

One method of assessing tissue-level oxygen utilisation is to examine the balance between oxygen delivery (DO2) and oxygen uptake (VO2), by calculating the oxygen extraction ratio (O2ER). In health, O2ER at rest is approximately 25% and therefore is usually ‘supply independent’. It may increase in well-trained athletes and may exceed 75% under conditions of exceptional metabolic stress. Recent work examining venous oxygen saturation (SvO2) suggests oxygen extraction may be compromised in patients with COVID-19 and that such compromise may be associated with reduced survival, although this work examined blood oxygen indices immediately after admission to ICU and not throughout the admission course [7]. As both hypoxaemia and hyperoxemia can be associated with adverse outcomes in critically ill patients [8], it is imperative to assess tissue level oxygen availability and extraction. The aim of this study was to describe trends in blood oxygen indices (PaO2/FiO2 ratio, CaO2 and O2ER) in patients with COVID-19 throughout the first 30-days of intensive care admission and explore the relationship between these indices and clinical outcomes.

Methods

Ethical approval was provided as part of the REACT COVID-19 observational study (a longitudinal cohort study to facilitate better understanding and management of SARS-CoV-2 infection from hospital admission to discharge across all levels of care): REC reference 17/NW/0632, SRB reference number; SRB0025 [9]. Due to the retrospective and observational nature of the study and there were no identifiable patient’s source data, the need for individual informed patient consent was waived. The data analysed were already routinely collected and electronically stored as part of clinical care. All data were anonymised and handled according to the local institutional and national policies. The study used STROBE guidelines for reporting observational studies [10].

We performed a retrospective observational cohort study in a single centre University Teaching Hospital in the UK. We included all patients admitted to the General Intensive Care Unit, between 1st March 2020 and 31st March 2021 inclusive. Eligible participants were aged 18 years or over, tested positive for COVID-19 by reverse transcriptase-polymerase chain reaction (RT-PCR) nasal and throat specimens, required mechanical ventilation, and had one or more arterial blood gas (ABG) samples performed. As a pragmatic retrospective study without intervention, we evaluated the merit of various oxygen indices under a real-life, generalised intensive care setting which may be applicable to routine practice. We therefore did not exclude any patients based on the presence of comorbidities that may have contributed to their death or those enrolled in other clinical trials.

Suitable patients were identified using admission records by a combination of manual and semi-automated data extraction. We collected baseline patient characteristics (age, gender, comorbidities), Clinical Frailty Scale (CFS) [11] and Charlson Comorbidity Index (CCI) [12] and ICU severity indices, including Acute Physiology and Chronic Health Evaluation–II (APACHE II) [13] and sequential organ failure assessment score (SOFA) [14]. The Intensive Care and National Audit Centre (ICNARC) UK summary data were used for comparison [4]. Additional data were extracted from our institution’s electronic patient record (EPR) (MetaVision, iMDSoft, Tel Aviv, Israel). At our institution, blood gas and laboratory results, ventilation parameters and vital signs are recorded automatically or by the bedside nurse and stored within the EPR. These were extracted for the entire duration of ICU stay for all included patients. Data underwent exploratory analysis and data cleaning to remove erroneous values and ensure data quality for the final analyses. The median days of invasive mechanical ventilation were 15 (interquartile range 8, 25). Therefore, to capture the entire intubated duration and initial recovery, results are reported for day one to seven and day one to 30. The primary outcome was hospital mortality at 30 days.

FiO2 was extracted directly from the ventilator to avoid labelling errors. PaO2/FiO2 ratios were calculated for each arterial blood gas sample. Total arterial oxygen content (CaO2) is the sum of the oxygen bound to haemoglobin and oxygen dissolved in plasma. It is calculated by [15]; where 1.34 is Hüfner’s constant, Hb is the amount of haemoglobin in grams per decilitre (gdl1), SaO2 the arterial haemoglobin saturation in fraction, 0.023 the solubility coefficient of oxygen at body temperature (i.e., the number of millilitres of oxygen dissolved per 100ml of plasma per kilopascal (ml O2 100ml-1 plasma kPa-1), and PaO2 the partial pressure of oxygen in arterial blood in kilopascals (kPa). Both point of care (ABG) and lab haemoglobin values were available, with laboratory values used across all calculations.

Venous blood gas (VBG) samples taken from central venous catheters within 30 minutes of an ABG on the same FiO2 were studied for central venous saturations (ScvO2), to ensure a strict temporal relationship been arterial and central venous samples, and used as a surrogate marker for pulmonary artery mixed blood saturation (SvO2). All other venous gas samples were excluded. The oxygen content of mixed venous blood (CvO2) was then calculated:

Oxygen extraction ratio (O2ER) was then calculated by the following equation:

These formulas were selected in order to avoid reliance on cardiac output monitoring. All calculations for O2ER were also compared to those produced when using (SaO2 –SvO2) / SaO2, producing similar results.

Statistical analysis

Statistical analysis and data processing were performed using R (R Core Team, Vienna, Austria) and GraphPad Prism version 9.0.0 for Windows, (GraphPad Software, San Diego, California USA, www.graphpad.com). Demographics variables were presented as medians and interquartile ranges. The statistical raw data for blood oxygen indices is presented as means, as they were normally distributed for day one to 30 of admission. For demographic comparisons between survivors and non-survivors, we used Mann-Whitney U test for continuous variables and Fisher’s Exact test for categorical data. For blood oxygen indices with normal distribution, we used Welch-two sample t-test to compare survivors and non-survivors. Pearson’s correlation coefficient was used to assess the relationship between individual blood oxygen indices and haemoglobin. The accuracy of individual blood oxygen indices in predicting mortality was assessed using area under receiver operating curves (AUROC). Median values are presented with the interquartile range (IQR), mean values are presented with confidence intervals (95% CI) and categorical data are presented with percentage (%). Statistical significance was assumed when p value of <0.05. We used the Benjamini-Hochberg (BH) adjustment to reduce the false discovery (type I error) rate when performing multiple statistical tests [16,17].

Results

During this study period, there were 1835 SARS-CoV-2 positive hospital admissions, of which 340 required admissions to the critical care unit. 184 patients required invasive mechanical ventilation, providing a total of just over 36,383 serial arterial or venous blood gas data points over the course of the first 30 days, all of which were included in the analysis (Fig 1). For these patients received invasive mechanical ventilation, the 30-day hospital survival rate was 72.3%. Baseline demographic, laboratory and ICU interventions and outcomes of all patients and comparison between survivors and non-survivors are presented in Table 1.

Fig 1

Flow diagram of included participants.

Table 1

Patient characteristics and outcomes of all patients meeting inclusion criteria (n = 184).

Variables	All patientsN = 184	SurvivorsN = 133	Non-survivorsN = 51	p-value
Age	59.5 (51.0, 67.0)	57.0 (49.0, 64.0)	65.0 (59.5,72.0)	<0.01
Sex, n (%)MaleFemale	115 (62.5%)69 (37.5%)	79 (59.4%)54 (40.6%)	36 (70.6%)15 (29.4%)	0.18
BMI (kg/m 2 )	30.0 (25.8, 35.5)	30.1 (25.2, 35.8)	29.4 (26.6, 33.8)	0.75
Ethnicity, n (%)WhiteAsianBlackMixedUnknown	129 (70.1%)31 (16.8%)14 (7.6%)6 (3.3%)4 (2.2%)	95 (71.4%)21 (15.8%)10 (7.5%)5 (3.8%)2 (1.5%)	34 (66.7%)10 (19.6%)4 (7.8%)1 (2.0%)2 (3.9%)	0.590.521.01.00.31
Clinical Frailty Score	2.0 (2.0, 3.0)	2.0 (2.0, 3.0)	3.0 (2.0, 4.0)	0.06
Charlson Comorbidity Index	2.0 (1.0, 3.0)	2.0 (1.0, 3.0)	3.0 (3.0, 4.5)	<0.01
Comorbidities, n (%)AsthmaChronic obstructive pulmonary diseaseDiabetes mellitusHypertensionIschaemic heart diseaseChronic kidney diseaseImmunosuppression	19 (10.3%)11 (6.0%)56 (30.4%)78 (42.4%)16 (18.7%)11 (3.3%)22 (12.0%)	12 (9.0%)6 (4.5%)40 (30.1%)57 (42.9%)7 (5.3%)5 (3.8%)16 (12.0%)	7 (13.7%)5 (9.8%)16 (31.4%)21 (41.1%)9 (17.6%)6 (11.8%)8 (15.7%)	0.420.180.860.870.020.070.63
Admission arterial blood gaspHPaO2 (kPa)PaCO2 (kPa)PaO2/FiO2HCO3- (mmol/L)Base excess (nmol/L)Lactate (mmol/L)	7.44 (7.38, 7.48)9.4 (8.5, 11.1)5.0 (4.5, 5.9)15.0 (12.1, 19.1)25.7 (23.2, 28.1)1.4 (-1.0, 3.6)1.2 (0.9, 1.6)	7.44 (7.40, 7.48)9.3 (8.4, 11.5)5.0 (4.5, 5.7)15.0 (12.3, 18.9)25.9 (23.9, 28.2)1.6 (-0.1, 3.8)1.1 (0.8, 1.5)	7.43 (7.35, 7.48)9.6 (8.7, 10.5)5.0 (4.5, 6.5)15.4 (12.1, 19.7)24.9 (20.7, 27.2)1.2 (-3.2, 2.9)1.4 (1.0, 1.8)	0.220.980.580.510.010.030.01
Admission lab variablesBilirubin (μmol/L)Creatinine (μmol/L)Creatinine kinase (IU/L)CRP (mg/L)D-Dimer (μg/L)Ferritin (μg/L)INRLDH (IU/L)Lymphocytes (x109/L)Neutrophil/lymphocyte ratioProcalcitonin (ng/L)Troponin (ng/L)	11 (8, 14)73 (55, 98)128 (57, 386)125 (67, 192)619 (340, 1283)687 (381, 1168)1.1 (1.0, 1.2)968 (760, 1276)0.7 (0.5, 1.0)10.5 (6.7, 18)0.3 (0.1, 1.0)15 (9, 53)	11 (7, 15)68 (51, 96)132 (64, 393)133 (77, 192)614 (324, 1148)656 (373, 1093)1.1 (1.0, 1.2)910 (755, 1231)0.7 (0.5, 1.0)10.3 (6.1, 17.5)0.3 (0.1, 0.9)15 (8, 37)	10 (8, 13)84 (66, 106)99 (54, 329)97 (51, 184)667 (358, 2017)871 (543, 1339)1.1 (1.0, 1.2)1098 (829, 1431)0.7 (0.5, 0.9)10.6 (6.8, 19.6)0.3 (0.1, 1.0)20 (10, 71)	0.930.010.750.290.310.140.250.050.700.580.670.11
ICU severity scores on admissionAPACHE IISOFA	18.0 (12.0, 23.0)4.0 (3.0, 7.0)	16.0 (11.0, 23.0)4.0 (3.0, 6.0)	19.0 (14.3, 24.8)5.5 (4.0, 8.0)	0.010.09
ICU interventionsPre-intubation NIV/CPAP, n (%)Prone positioning, n (%)Renal replacement therapy, n (%)	113 (61.4%)147 (79.9%)50 (27.2%)	93 (69.9%)109 (82.0%)30 (22.6%)	32 (62.7%)38 (74.5%)20 (39.2%)	0.380.300.03
Duration of mechanical ventilation (days)	15 (8, 25)	16 (11, 29)	8 (5, 15)	<0.01
Duration of ICU length of stay (days)	20 (11, 36)	24 (17, 42)	11 (6, 17)	<0.01
Duration of hospital length of stay (days)	28 (19, 53)	41 (27, 64)	16 (10, 20)	<0.01

All scores and laboratory variables were performed at the time of ICU admission. APACHE II: Acute physiology and chronic health evaluation; BMI: Body mass index; CRP: C-Reactive protein; ICU: Intensive care unit; INR: International normalised ratio; LDH: Lactate dehydrogenase; SOFA: Sequential organ failure assessment.

The median age was 59.5 years (IQR 51.0, 67.0), and survivors were significantly younger, 57.0 (IQR 49.0, 64.0) compared to non-survivors, 65.0 (IQR 59.5, 72.0) with male predominance (62.5%). Gender was not associated with increased mortality. 129 patients were of white ethnic origin (70.1%), and 55 were from ethnic minority groups (29.9%). The median body mass index (BMI) was 30.0 (IQR 25.8, 35.5), with no significant difference between survivors, 30.1 (IQR 25.2, 35.8) and non-survivors 29.4 (IQR 26.6, 33.8). The median admission Acute Physiology and Chronic Health Evaluation II (APACHE II) score was 18 (IQR 12, 23) giving a predictive mortality of 25%. Survivors had a significantly lower score, 16.0 (IQR 11.0, 23.0) compared to non-survivors, 19.0 (IQR 14.3, 24.8) both of which also fall into the 25% expected mortality prediction. The median Sequential Organ Failure Assessment (SOFA) score on admission was 4.0 (IQR 3.0, 7.0) giving a predictive mortality of 36.1%. There was no significant difference in the SOFA score between survivors, 4.0 (IQR 3.0, 6.0) and non-survivors, 5.5 (IQR 4.0, 8.0). The median Clinical Frailty Score (CFS) was 2.0 (IQR 2.0, 3.0) meaning ‘well’, with no significant difference between survivors, 2.0 (IQR 2.0, 3.0) and non-survivors, 2.0 (IQR 2.0, 4.0). Median Charlson Comorbidity Index (CCI) was 2.0 (IQR 1.0, 3.0), with survivors scoring significantly lower, 2.0 (IQR 1.0, 3.0) classed as ‘moderate’, compared to non-survivors, 3.0 (IQR 3.0, 4.5) classed as ‘severe’. Presence of ischaemic heart disease was associated with non-survival (Table 1). Differences between survivors and non-survivors for other comorbidities (asthma, chronic obstruction pulmonary disease, diabetes, hypertension, chronic kidney disease and pre-existing immunosuppression) were not significant.

Median duration of mechanical ventilation was 15 days (IQR 8, 25). Survivors were mechanically ventilated for significantly longer, 16 days (IQR 11, 29) compared to non-survivors, 8 days (IQR 5, 15). Median duration of ICU length of stay was 20 days (IQR 11, 36). Survivors stayed on ICU for significantly longer, 24 days (IQR 17, 42) compared to non-survivors, 11 days (IQR 6, 17). Median duration of hospital stay was 28 days (IQR 19, 53), with survivors staying significantly longer, 41 days (IQR 27, 64), compared to non-survivors, 16 days (IQR 10, 20). 113 patients (61.4%) received non-invasive ventilation prior to intubation, with no significant difference between survivors, 93 (69.9%), and non-survivors, 32 (62.7%). 147 patients (79.9%) received prone positioning as part of their care, and there was no significant difference between survivors, 109 (82.0%), and non-survivors, 38 (74.5%). 50 patients (27.2%) required renal replacement therapy, with a significant difference between survivors, 30 (22.6%) and non-survivors, 20 (39.2%).

Median admission lactate was 1.2 (IQR 0.9, 1.6) and was significantly lower in survivors, 1.1 (IQR 0.8, 1.5) than non-survivors, 1.4 (IQR 1.0, 1.8). Median admission HCO3- was 25.7 (IQR 23.2, 28.2) and was significantly higher in survivors, 26.2 (IQR 23.9, 28.2) than non-survivors, 24.9 (IQR 21.9, 27.2). Median admission base excess was 1.4 (IQR -1.0, 3.6) and was significantly higher in survivors, 1.6 (IQR -0.1, 3.8), than non-survivors, 1.2 (IQR -3.2, 2.9). There was no other significant difference in admission arterial blood gas values. Admission creatinine was 73 (IQR 55, 98) and was significantly lower in survivors, 68 (IQR 51, 96) than non-survivors, 84 (IQR 66, 106). There were no other significant differences in baseline admission laboratory blood results between survivors and non-survivors. Detailed patient’s demographics and outcomes are presented in Table 1.

Arterial oxygen indices from 34,592 sampling points across days one to seven and days one to 30 are detailed in Table 2.

Table 2

Comparison of mean averaged blood oxygen indices at day one-seven, and day one-30; survivors (n = 133) vs. non-survivors (n = 51).

	Survivors	Non-survivors	Mean Difference	95% CI	p-value*
PaO2 (kPa)Day 1–7Day 1–30	9.809.73	9.499.21	-0.31-0.52	-0.41, -0.20-0.59, -0.46	<0.01<0.01
PaO2 (kPa) / FiO2 ratioDay 1–7Day 1–30	19.7421.19	17.5115.56	-2.23-5.64	-2.55, -1.91-5.85, -5.43	<0.01<0.01
CaO2 (ml/dL)Day 1–7Day 1–30	14.3312.78	14.6313.62	0.310.83	0.19, 0.420.75, 0.91	<0.01<0.01
O2ERDay 1–7Day 1–30	0.340.38	0.270.29	-0.07-0.08	-0.09, -0.04-0.09, -0.07	<0.01<0.01

* Using the Benjamini-Hochberg adjustment.

Non-survivors exhibited significantly lower PaO2 throughout the admission (Fig 2A). From day one to day seven of ICU admission there was an averaged mean difference of -0.31 kPa (95% CI: -0.41 to -0.20) and from day one to day 30 an averaged mean difference of -0.52 kPa (95% CI: -0.59 to -0.46). Moreover, non-survivors exhibited a significantly lower PaO2/FiO2 ratio, with improved separation over time (Fig 2B). Across day one to day seven of ICU admission there is an averaged mean difference of -2.23 (95% CI: -2.55 to -1.91) and across day one to day 30 an averaged mean difference of -5.64 (95% CI: -5.85 to -5.43). While both survivors and non-survivors exhibited sub-physiological CaO2 (trending down throughout admission, survivors exhibited significantly lower values (Fig 2C). Across day one to day seven of ICU admission there is an averaged mean difference in CaO2 of 0.31 (95% CI: 0.19 to 0.42) and for day one to day 30 an averaged mean difference of 0.83 (95% CI 0.75: to 0.91). For oxygen extraction analysis (Table 2), 1,791 data points were available with contemporaneous arterial and venous blood sampling. Non-survivors exhibited significantly lower oxygen extraction (Fig 2D). From day one to day seven of ICU admission there was an averaged mean difference in O2ER of -0.07 (95% CI: -0.09 to -0.04) and from day one to day 30 an averaged mean difference of -0.08 (95% CI: -0.09 to -0.07). As expected, there was a tight, linear correlation between CaO2 and haemoglobin concentrations (Fig 3).

Fig 2

Blood oxygen indices over time between survivors and non-survivors.

(A) PaO2 (kPa), (B). PaO2 / FiO2 ratio (kPa), (C). Total oxygen content (ml/dL), (D). Oxygen extraction ratio (O2ER).

Fig 3

The correlation between oxygen content and haemoglobin.

We studied the use of these blood oxygen indices in predicting hospital mortality of all mechanically ventilated patients. The area under the receiver operating characteristic curve (AUC) for FiO2 and PaO2/FiO2 were similar at 0.78 (95% CI: 0.77 to 0.79, p<0.01) and 0.78 (95%CI: 0.77 to 0.78, p<0.01) with a cut-off value of FiO2 > 0.54 and PaO2/FiO2 ≤ 16.6 kPa of respectively (Fig 4A). While PaO2 was less predictive with an AUC of 0.60 (95% CI: 0.60 to 0.61, p< 0.01, cut-off 9.3 kPa) (Fig 4A). O2ER was a better predictor of hospital mortality than CaO2 with an AUC of 0.70 (95%CI: 0.67 to 0.72, p<0.01, cut off ≤0.29) (Fig 4B and 4C). Although statistically significant, total oxygen content was less predictive of hospital mortality than other blood oxygen indices (Fig 4).

Fig 4

Blood oxygen indices operating characteristics analysis.

(A). AUC for FiO2, PaO2 and PaO2/FiO2, (B). Total oxygen content (CaO2), (C). Oxygen extraction ratio (O2ER).

Discussion

In this observational study, we have demonstrated that there were significant abnormalities in blood oxygen indices in mechanically ventilated adult COVID-19 patients. Of note, despite having higher total oxygen content, non-survivors exhibited lower oxygen extraction ratios. These findings support the notion that mechanically ventilated adult patients with COVID-19 may have impaired oxygen utilisation and that this is a marker of severity of disease.

The physiology of oxygen transport is well-described [15]. High-quality evidence to support the optimal measure of oxygenation in critically unwell patients is limited and most research has tended to consider SaO2, PaO2 or PaO2/FiO2 ratios in isolation and are conducted in heterogenous ICU cohorts with various underlying pathologies. Oxygenation targets for patients admitted to intensive care are conflicting [18], and while it appears in general that over-oxygenation might be harmful [19], diverse groups of patients with differing pathologies are unlikely to all benefit from a single approach. Some COVID-19 patients are at risk of profound hypoxemic respiratory failure and development of acute respiratory distress syndrome (ARDS), with a number of mechanisms proposed including intrapulmonary shunting, impaired lung perfusion regulation, intravascular microthrombi and impaired diffusion capacity at a tissue level [5]. Despite degree of hypoxaemia being predictive of mortality, oxygen targets for these patients are not yet well-established [20], though correction of oxygenation may improve survival [21] and some are calling for higher as well as lower targets [22].

Arterial partial pressure of oxygen (PaO2), though commonly used, gives no indication of the required inspired oxygen. Additional information from PaO2/FiO2 ratios can provide further indication as to the degree of hypoxic respiratory failure. These ratios are helpful in stratifying the severity of ARDS as part of the Berlin Definition [23], but are dependent on the operator input of correct FiO2 and temperature [24,25]. In patients with severe hypoxic respiratory failure, the PaO2/FiO2 ratio is often the primary blood oxygen index used to guide decisions regarding initiation of mechanical ventilation, escalation of ventilatory support or when to institute rescue measures such as prone positioning and extracorporeal membrane oxygenation (ECMO). In our patient cohort, the PaO2/FiO2 ratio continued to deteriorate over the 7-day period in non-survivors and was a better predictor of ICU survival in all mechanically ventilated COVID-19 patients than other blood oxygen indices. Despite a statistical difference, the association between PaO2/FiO2 ratio and the CaO2 was weak, suggesting that an increment in fractional inspired oxygen may not correspondingly influence oxygen content in all patients. Moreover, although oxygen content was low overall, there was relative preservation in non-survivors with no difference between the groups.

The commonly measured blood oxygen indices (SaO2, PaO2) may quantify the degree of COVID-19-related respiratory failure, but may not inform on oxygen delivery to the tissues. Recent studies have failed to demonstrate a survival benefit from optimisation of oxygen delivery (DO2) by the manipulation of supplemental oxygen, blood volume expansion, and cardiovascular supportive measures in sepsis [26-30], however as previously observed these studies represent more heterogenous underlying pathologies than COVID-19. In health, oxygen uptake (VO2) is well-maintained even with a decreasing DO2 due to a variety of compensatory mechanisms including increased O2ER and redistribution of blood flow to tissues with the highest oxygen demand. It has been suggested that VO2 decreases below a so-called “critical DO2 (DO2crit)”, where O2ER is maximal. Tissue hypoxia may occur if DO2 continues to decrease below a notional DO2crit, or if VO2 increases or fails, resulting in anaerobic respiration, lactate production and ultimately, ischaemia. This situation may be exacerbated by fever, rigors and sepsis [31], all features of severe COVID-19 infection.

Examination of SvO2 may indicate the balance between DO2 and VO2, where a value of >70% suggests adequate respiration. In severe sepsis, where tissue level oxygen metabolism is impaired due to microcirculation disorders and inflammatory mediator damage, a lower O2ER is strongly associated with increased mortality [32]. Our findings suggest O2ER is statistically lower in non-survivors than survivors. This finding contradicts a recent study which suggested an increased O2ER in non-survivors of severe COVID-19 infection [7]. However, this work is not directly comparable to our study as the authors only used admission blood sample data for calculations, to provide a ‘snapshot’ of admission parameters. It is unclear what proportion required respiratory support in this study, in contrast to our study in which all patients were mechanically ventilated. Moreover, the COVID-19 survivors had lower O2ER than healthy controls and the method of O2ER estimation was also different from our study by including estimates of cardiac output. Similar to our study, a small case series also reported reduced oxygen utilisation in patients with COVID-19 which may be associated with adverse outcomes [33].

The reasons for our finding of association of lower O2ER with mortality is uncertain, though are likely to be multi-factorial. Up-regulation of O2ER when DO2 is reduced may fail in severe pathology such as tissue hypoxia or acidosis, though curiously O2ER is not increased in healthy, acclimatised individuals at altitude (i.e., hypobaric hypoxia) [34], possibly due to hypoxia itself reducing the ability to extract oxygen. It has been documented that COVID-19 is implicated with a multi organ microangiopathic process with endotheliopathy, vascular thrombosis, overt inflammatory cytokine response and abnormalities of von Willebrand factor-platelet axis [35,36]. The interaction between virus and receptor is thought to downregulate ACE2 activation, thus increasing levels of angiotensin II with consequences of intense vasoconstriction, inflammation and oxidative stress enhancing thrombogenicity. The cumulative effect of both endotheliopathy and vasoconstriction may contribute to a scenario with tissue level reduced oxygen utilisation. This may explain our findings that a lack of ability to upregulate oxygen extraction led to poorer outcomes. Another most important consideration for our findings could be mitochondrial dysfunction [37]. While 90% of total oxygen consumption is accounted for by mitochondrial oxidative phosphorylation, mitochondrial function extends beyond adenosine triphosphate (ATP) production, playing critical roles in cellular messaging, apoptosis, autophagy, and calcium homeostasis [38]. There is increasing evidence that upon cell entry the SARS-CoV-2 virus hijacks host’s mitochondria which may contribute to mitochondrial dysfunction and cellular death [39,40]. The associations of increased mortality with advanced age, metabolic syndrome and immune deficiency may reflect existing mitochondrial dysfunction among these groups exposing their vulnerability [41].

To our knowledge this is the first study to establish the dynamic trend of PaO2, PaO2/FiO2, O2ER and CaO2 over the course of an ICU admission in patients with severe COVID-19 pneumonia.

Limitations include a retrospective design, with no prior power calculation, and potentially our significant results are the result of a type II error. Our sample size is reasonable, but only 51 patients died. This could lead to non-survivors being under-represented, though we tried to address this by only including intubated patients. Our data is clearly susceptible to sampling bias, as the sickest patients tend to get more frequent blood gas sampling than more stable patients. However, our results show that those who died were not sampled excessively when compared with non-survivors. The blood oxygen indices we describe are estimates for overall gas analysis, not for individual patients. Additional analysis designed to reduce this bias yielded similar results suggesting a degree of robustness. In addition, we analysed patients mechanically ventilated at any point, including those who were transferred to our centre under mutual aid, transferred out for ECMO, and those for who incidentally positive for COVID-19, but initially admitted for other pathology. Moreover, extraction ratios were calculated with ScvO2 from central venous lines, than SvO2 samples from pulmonary artery catheters. While ScvO2 correlates well with SvO2, it essentially reflects the oxygenation of the upper body and head, not including myocardial perfusion. This may be important for patients with COVID-19 due to cardiomyopathy and relatively high-output cardiac states.

Conclusions

The COVID-19 pandemic offers a unique opportunity to study a homogenous cohort of hypoxaemic critically unwell patients, with similar underlying pathology. In a cohort of mechanically ventilated adult ICU patients with hypoxaemic respiratory failure due to COVID-19, oxygen extraction is significantly lower in non-survivors compared to survivors during the first 30 days of ICU admission, despite having higher CaO2 values. This suggests COVID-19 may cause impaired oxygen utilisation. Urgent further evaluation of the relationship between impaired oxygen extraction and survival in COVID-19 is justified.

Transfer Alert

This paper was transferred from another journal. As a result, its full editorial history (including decision letters, peer reviews and author responses) may not be present.

7 Mar 2022

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Antonino Salvatore Rubino, M.D., Ph.D.

Academic Editor

PLOS ONE

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf.

2. Thank you for stating the following financial disclosure:

“The authors received no speicifc fubnding for this work.”

At this time, please address the following queries:

a) Please clarify the sources of funding (financial or material support) for your study. List the grants or organizations that supported your study, including funding received from your institution.

b) State what role the funders took in the study. If the funders had no role in your study, please state: “The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

c) If any authors received a salary from any of your funders, please state which authors and which funders.

d) If you did not receive any funding for this study, please state: “The authors received no specific funding for this work.

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

3. Please note that in order to use the direct billing option the corresponding author must be affiliated with the chosen institute. Please either amend your manuscript to change the affiliation or corresponding author, or email us at plosone@plos.org with a request to remove this option.

4. One of the noted authors is a group or consortium Tom Wilkinson, Anna Freeman, Hannah Burke, Michael Celinski, Saul N Faust, Gareth J Thomas, Christopher Kipps. In addition to naming the author group, please list the individual authors and affiliations within this group in the acknowledgments section of your manuscript. Please also indicate clearly a lead author for this group along with a contact email address.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: No

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Your work is applaudable. Please find below some comments and suggestions:

1. "Due to the nature of this study, the need for individual informed patient consent was waived." Can the authors provide further explanation as to why the waiver is applicable to this particular study?

2. "We did not exclude any patients based on the presence of co-morbidities that may have contributed to their death or those enrolled in other clinical trials." Can the authors provide further explanation as to why such patients were not within the frames of an exclusion criteria?

3. Can the authors provide their limitation in a separately headed segment? Or present it separately from the first statement of the paragraph that highlights limitation?

4. Can the authors provide statements in their conclusion that highlight the importance of their findings in terms of shaping clinical practice or global/public health practices?

Reviewer #2: This retrospective cohort study evaluated relationship between three oxygen indices and clinical outcomes in mechanically ventilated COVID-19 patients. Although concept of the study is reasonable and the manuscript is well written, methods have serious problem.

Major points:

1) As authors mentioned in a limitation part (P22, L364), the results of this study are affected by serious sampling bias, which has bearings on the fundamental point of the study design. Including all blood samples into the analysis cannot be reasonable, since more samples from sicker patients must lead to patently false conclusion. I totally agree with authors opinion that oxygen extraction rate is associated with mortality of COVID-19 patients, but I cannot agree this study is technically sound and appropriate, as long as they use quite biased samples. Since this study included 181 patients, number of blood gas samples must be multiples of 181.

Minor points:

2) P21, L353-357: Discussion about mitochondrial dysfunction looks too strong and too assertive, even though I agree with the opinion that mitochondrial function should affect oxygen utilization and clinical outcome of this patient population.

3) P22, L381-382: The same as above.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

20 Apr 2022

Thank you for the valuable comments. Please see our response to the reviewer’s comments.

Reviewer #1:

Your work is applaudable. Please find below some comments and suggestions:

1. "Due to the nature of this study, the need for individual informed patient consent was waived." Can the authors provide further explanation as to why the waiver is applicable to this particular study?

Response: Thank you for your comment. We have now modified this sentence to clarify this further as “due to the retrospective and observational nature of the study and there were no identifiable patient’s source data, the need for individual informed patient consent was waived. The data analysed were already routinely collected and electronically stored as part of clinical care”

2. "We did not exclude any patients based on the presence of co-morbidities that may have contributed to their death or those enrolled in other clinical trials." Can the authors provide further explanation as to why such patients were not within the frames of an exclusion criteria?

Response: Thank you. We have now modified this sentence to clarify this statement further. “As a pragmatic retrospective study without intervention, we evaluated the merit of various oxygen indices under a real-life, generalised intensive care setting which may be applicable to routine practice. We therefore did not exclude any patients based on the presence of comorbidities that may have contributed to their death or those enrolled in other clinical trials”.

3. Can the authors provide their limitation in a separately headed segment? Or present it separately from the first statement of the paragraph that highlights limitation?

Response: We have now presented the limitations as a separate paragraph.

4. Can the authors provide statements in their conclusion that highlight the importance of their findings in terms of shaping clinical practice or global/public health practices?

Response: Thank you. While this is an important clinical finding that oxygen utilisation may be compromised in severe fatal COVID-19 patients, this warrants further clinical and scientific explorations to shape any future clinical practice (i.e., measure to improve oxygen utilisation, is not straight forward). This a pragmatic observational study suggests that improving oxygen content alone may not be adequate. Any additional comments on change in clinical practice would be exploratory.

Reviewer #2:

This retrospective cohort study evaluated relationship between three oxygen indices and clinical outcomes in mechanically ventilated COVID-19 patients. Although concept of the study is reasonable and the manuscript is well written, methods have serious problem.

Major points:

1) As authors mentioned in a limitation part (P22, L364), the results of this study are affected by serious sampling bias, which has bearings on the fundamental point of the study design. Including all blood samples into the analysis cannot be reasonable, since more samples from sicker patients must lead to patently false conclusion. I totally agree with authors opinion that oxygen extraction rate is associated with mortality of COVID-19 patients, but I cannot agree this study is technically sound and appropriate, as long as they use quite biased samples. Since this study included 181 patients, number of blood gas samples must be multiples of 181.

Response: Thank you for your comment. This was a pragmatic observational study and consequently, we were unable to prescribe the timing and quantity of oxygen measurements that is suitable for an individual patient in ICU. We have clarified this in the limitations section. While we agree that might be variations in sampling episodes between patients and would have introduced precision bias, this is highly unlikely as the sampling episodes were similar between survivors and non-survivors (Table). Moreover, all the patients presented here were very sick and required mechanical ventilation, representing a unique homogenous cohort with similarities in their degree of hypoxia. We presented all available blood samples to minimise any selection bias. We have modified the sentence in the limitations to address this comment. “Our data is clearly susceptible to sampling bias, as the sickest patients tend to get more frequent blood gas sampling than more stable patients. However, our results show that those who died were not sampled excessively when compared with non-survivors”.

Minor points:

2) P21, L353-357: Discussion about mitochondrial dysfunction looks too strong and too assertive, even though I agree with the opinion that mitochondrial function should affect oxygen utilization and clinical outcome of this patient population.

Response: We have modified this sentence to reflect the reviewer’s comments. “There is increasing evidence that upon cell entry the SARS-CoV-2 virus hijacks host's mitochondria which may contribute to mitochondrial dysfunction and cellular death”.

3) P22, L381-382: The same as above.

Response: We have modified the sentence to reflect the reviewer’s comments. “Urgent further evaluation of the relationship between impaired oxygen extraction and survival in COVID-19 is justified”.

The changes within the manuscript are highlighted in yellow. We are looking forward to hearing from you soon.

Thank you for considering this manuscript for publication in PLOS one.

Yours sincerely,

Professor Mike Grocott

BSc MBBS MD FRCA FRCP FFICM

Professor of Anaesthesia and Critical Care Medicine, Head, Integrative Physiology and Critical Illness Group, CES Lead, Critical Care Research Area, Southampton NIHR Respiratory BRC

Submitted filename: Response to reviewers comments.docx

Click here for additional data file.

23 May 2022

Dynamic blood oxygen indices in mechanically ventilated COVID-19 patients with acute hypoxic respiratory failure: a cohort study

PONE-D-21-30166R1

Dear Dr. Grocott,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Antonino Salvatore Rubino, M.D., Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

The manuscript has improved its quality after revision

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: All questions and comments have been answered. I would like extend my gratitude to the authors for taking the time to respond to the feedback and assimilate them in their work for maximal scientific impact of the paper.

Reviewer #2: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

1 Jun 2022

PONE-D-21-30166R1

Dynamic blood oxygen indices in mechanically ventilated COVID-19 patients with acute hypoxic respiratory failure: a cohort study

Dear Dr. Grocott:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Antonino Salvatore Rubino

Academic Editor

PLOS ONE