Journal Information
Vol. 30. Issue 3.
Pages 272-281 (May - June 2024)
Share
Share
Download PDF
More article options
Visits
3038
Vol. 30. Issue 3.
Pages 272-281 (May - June 2024)
Original article
Full text access
Hyperoxemia in invasively ventilated COVID–19 patients–Insights from the PRoVENT–COVID study
Visits
3038
A.M. Tsonasa,
Corresponding author
a.m.tsonas@amsterdamumc.nl

Corresponding author at: Department of Intensive Care, G3–228, Amsterdam UMC, location ‘AMC’, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands.
, D.M. van Meenena, M. Bottaa, G.S. Shresthab, O. Rocac,d, F. Paulusa,e, A.S. Netoa,f,g,h, M.J. Schultza,i,j, PRoVENT–COVID Collaborative Group 1
a Department of Intensive Care, Amsterdam UMC, location ‘AMC’, Amsterdam, The Netherlands
b Department of Critical Care Medicine, Tribhuvan University Teaching Hospital, Maharajgunj, Kathmandu, Nepal
c Department of Intensive Care, Vall d'Hebron Univerity Hospital, Barcelona, Spain
d Ciber Enfermedades Respiratorias (CibeRes), Instituto de Salud Carlos III, Madrid, Spain
e ACHIEVE, Centre of Applied Research, Amsterdam University of Applied Sciences, Faculty of Health, Amsterdam, The Netherlands
f Australian and New Zealand Intensive Care Research Centre (ANZIC–RC), Monash University, Melbourne, Australia
g Data Analytics Research and Evaluation (DARE) Centre, Austin Hospital, Melbourne, Australia
h Department of Critical Care Medicine, Hospital Israelita Albert Einstein, Sao Paulo, Brazil
i Department of Critical Care Medicine, Mahidol University, Bangkok, Thailand
j Nuffield Department of Medicine, University of Oxford, Oxford, United Kingdom
Ver más
This item has received
Article information
Abstract
Full Text
Bibliography
Download PDF
Statistics
Figures (3)
Show moreShow less
Tables (2)
Table 1. Baseline Characteristics of the Patients According to the Groups In the Unmatched and Matched Cohort.
Table 2. Clinical Outcomes According to Groups In the Unmatched and Matched Cohort.
Show moreShow less
Additional material (1)
Abstract
Objective

We determined the prevalences of hyperoxemia and excessive oxygen use, and the epidemiology, ventilation characteristics and outcomes associated with hyperoxemia in invasively ventilated patients with coronavirus disease 2019 (COVID–19).

Methods

Post hoc analysis of a national, multicentre, observational study in 22 ICUs. Patients were classified in the first two days of invasive ventilation as ‘hyperoxemic’ or ‘normoxemic’. The co–primary endpoints were prevalence of hyperoxemia (PaO2 > 90 mmHg) and prevalence of excessive oxygen use (FiO2 ≥ 60% while PaO2 > 90 mmHg or SpO2 > 92%). Secondary endpoints included ventilator settings and ventilation parameters, duration of ventilation, length of stay (LOS) in ICU and hospital, and mortality in ICU, hospital, and at day 28 and 90. We used propensity matching to control for observed confounding factors that may influence endpoints.

Results

Of 851 COVID–19 patients, 225 (26.4%) were classified as hyperoxemic. Excessive oxygen use occurred in 385 (45.2%) patients. Acute respiratory distress syndrome (ARDS) severity was lowest in hyperoxemic patients. Hyperoxemic patients were ventilated with higher positive end–expiratory pressure (PEEP), while rescue therapies for hypoxemia were applied more often in normoxemic patients. Neither in the unmatched nor in the matched analysis were there differences between hyperoxemic and normoxemic patients with regard to any of the clinical outcomes.

Conclusion

In this cohort of invasively ventilated COVID–19 patients, hyperoxemia occurred often and so did excessive oxygen use. The main differences between hyperoxemic and normoxemic patients were ARDS severity and use of PEEP. Clinical outcomes were not different between hyperoxemic and normoxemic patients.

Keywords:
Coronavirus disease 2019
COVID–19
ARDS
Ventilation
Invasive ventilation
Hyperoxemia
Normoxemia
Oxygen management
Mortality
Full Text
Introduction

Both severe and moderate hyperoxemia have been reported to be associated with worse outcomes in critically ill patients,1,2 and high levels of fraction of inspired oxygen (FiO2) have detrimental effects on lung tissue, causing damage comparable to that seen in acute respiratory distress syndrome (ARDS).3,4 However, the exact targets of arterial oxygen tension (PaO2) and FiO2 remain debated, especially in ARDS patients where the relationship between oxygenation and outcome is complex. One seminal study, named ‘OXYGEN–ICU’, showed a conservative oxygen strategy targeting PaO2 of 70–100 mmHg compared with a liberal oxygenation therapy targeting PaO2 > 150 mmHg to improve survival.5 More recent studies comparing a conservative oxygen strategy with less liberal oxygen strategies, however, failed to show benefit.6-11

Coronavirus disease 2019 (COVID–19) is currently the most common form of ARDS, and patients with COVID–19 ARDS almost always experience profound impairments in gas exchange.12-14 To determine the exact prevalence of hyperoxemia and of excessive oxygen use in invasively ventilated COVID–19 patients, we performed a secondary analysis of a conveniently–sized multicentre observational study, named the ‘PRactice of VENTilation in COVID–19’ (PRoVENT–COVID).15 We compared the epidemiology, ventilation characteristics and outcomes in hyperoxemic versus normoxemic patients. We used propensity matching to control for observed confounding factors. The hypothesis was that hyperoxemia and excessive oxygen use occur often in COVID–19 patients under invasive ventilation.

MethodsStudy design

Secondary analysis of PRoVENT–COVID, an investigator–initiated, national, multicentre, observational cohort study undertaken at 22 ICUs in the Netherlands. The study protocol of PRoVENT–COVID and the analysis plan for the current analysis have been prepublished16,17 and the study is registered at clinicaltrials.gov (NCT04346342). Other post–hoc evaluations of PRoVENT–COVID regarding ventilation characteristics and strategies,15,18,19 and gas exchange,20,21 were reported earlier.

Patients

Consecutive patients were eligible for participation if they were > 18 years of age, admitted to one of the participating ICUs, and had received invasive ventilation for acute hypoxemic respiratory failure related to COVID–19. COVID–19 was to be confirmed by RT–PCR. For the current analysis, we excluded patients that were transferred from or to a non–participating hospital in the first two days of invasive ventilation, as we could not collect data on gas exchange during these days in those patients, and patients without PaO2 data on the first two days of invasive ventilation.

Collected data

Patient demographics, medical history, presence and severity of ARDS, and extent of infiltrates on the chest radiography or computed tomography scan was collected at baseline.

Since the first day of ventilation had a flexible length and could range from one minute to 24 hours duration depending on the timing of start of invasive ventilation in the ICU, we merged this day with the second day and named it ‘day 1’. The following calendar day was named ‘day 2’.

We collected detailed ventilation data at one hour after start of invasive ventilation in the ICU, which could be at arrival if the patients started with invasive ventilation in the normal ward or in the emergency room, or after intubation in the ICU after ICU admission. Thereafter, we collected ventilation data at 08:00, 16:00 and 24:00 hours over the first four days of ventilation. Ventilation data included ventilator settings and ventilation parameters, arterial blood gas analyses results and use of adjunctive therapies for refractory hypoxemia.

We also collected typical aspects of ICU monitoring and care, and common ICU complications. Patients were followed until day 90 for intubation status, ICU– and hospital–discharge, and death.

Exposures

The primary exposure of interest was hyperoxemia on day 1 or day 2 of invasive ventilation, defined as PaO2 > 90 mmHg. Patients were categorized as ‘hyperoxemic’ if the daily mean PaO2 was > 90 mmHg, or ‘hypoxemic’ if the daily mean PaO2 was ≤ 55 mmHg, on either day 1 or day 2; all other patients were classified as ‘normoxemic’. The first PaO2 value was ignored, because it is plausible that this value could not be affected by FiO2 titrations by ICU team members.

The secondary exposure of interest was excessive use of oxygen. At each time–point on day 1 and day 2, oxygen use was classified as ‘excessive’ if FiO2 was ≥ 60% following a previous blood gas analysis showing PaO2 > 90 mmHg or recorded SpO2 > 92%.

Outcomes

The co–primary outcomes were the prevalence of hyperoxemia and the prevalence of excessive oxygen use. Secondary outcomes included key ventilator settings and parameters, including tidal volume (VT), positive end–expiratory pressure (PEEP), driving pressure (ΔP) and respiratory system compliance (Crs), and typical clinical outcomes, including duration of ventilation, length of stay (LOS) in hospital and ICU, and mortality in ICU, hospital, and at day 28 and day 90.

Statistical analysis

Due to the very small number of hypoxemic patients, i.e., 8 out of 851 patients, hypoxemic patients were added to the cohort of normoxemic patients in all analyses.

To assess differences between hyperoxemic and normoxemic patients, Wilcoxon–Mann–Whitney test for continuous data and Fisher exact test for categorical data were used. Ventilation data was reported at three specific moments: 1) at start of ventilation, 2) on day 1, and 3) on day 2. Start of ventilation was based on the measurements collected within the first hour after start of ventilation. As previously mentioned, the measurements of the first flexible calendar day and first full calendar day were merged, and day 1 was based on the means of these measurements. Day 2 was based on the means of the measurements of the following calendar day. Cumulative frequency distributions of VT, PEEP, ΔP, and Crs are shown for patients categorized as hyperoxemic versus patients that are categorized as normoxemic, at day 1 and at day 2. Locally estimated scatterplot smoothing (LOESS) method was used to inspect the relationship between 28–day mortality and PaO2 and FiO2 at day 1 and at day 2.

To further evaluate the associations of outcome with occurrence of hyperoxemia, a propensity matched analysis was performed. For each patient, a propensity score was estimated with logistic regression and used to match hyperoxemic patients to normoxemic patients (1:1) using a caliper of 0.05 standard deviation of the logit of the propensity score and applying nearest matching without replacement. Based on clinical relevance and one previous analysis,22 the following variables were selected a priori: age, sex, BMI, chronic diseases including heart failure, diabetes mellitus, chronic renal failure, chronic liver failure, chronic pulmonary obstructive disease, active or hematologic neoplasm and immunosuppression, and PaO2/FiO2, Crs, total respiratory rate (RR) and bicarbonate at baseline.

Time until extubation is shown in a cumulative distribution plot with death as a competing risk and compared with a Fine–Gray competing risk model. Probability of survival at day 28 and 90 was estimated using Kaplan–Meier curves and compared with a log–rank test.

In a sensitivity analysis, we excluded the hypoxemic patients to check whether there were differences in clinical outcomes.

All analyses were conducted in R v.3.6.1 (R Foundation, Vienna, Austria) and significance level was set at 0.05.

ResultsPatients

1122 patients were included in PRoVENT–COVID (Fig. 1). We excluded 271 patients, mainly because of early transfer from or to a non–participating ICU. The remaining 851 patients most often were male (73%) with a median age of 66 [58–72] years, and the majority of patients had moderate to severe ARDS (Table 1).

Fig. 1.

Flowchart of inclusions.

(0.5MB).
Table 1.

Baseline Characteristics of the Patients According to the Groups In the Unmatched and Matched Cohort.

  Unmatched Cohort (n = 851)Matched Cohort (n = 346)
  Hyperoxemic(n = 225)  Normoxemic(n = 626)  p  Hyperoxemic (n = 173)Normoxemic(n = 173)  p 
Age, years  67.0 (58.0–73.0)  65.0 (58.0–72.0)  0.315  67.0 (59.0–73.0)66.0 (59.0–73.0)  0.763 
Male gender–no (%)  157 (69.8)  462 (73.8)  0.257  125 (72.3)126 (72.8)  0.999 
Body mass index, kg/m2  27.7 (25.5–30.5)  27.8 (25.4–30.8)  0.606  27.7 (25.4–30.7)27.5 (24.9–30.4)  0.711 
Use of non-invasive ventilation–no (%)  17 (7.8)  58 (9.5)  0.495  11 (6.6)20 (11.6)  0.132 
Duration of non-invasive ventilation, hours  4.0 (2.0–11.0)  8.0 (2.0–21.0)  0.320  3.0 (1.5–14.0)13.0 (2.5–36.0)  0.214 
Chest CT scan performed–no (%)  82 (36.4)  224 (35.8)  0.872  55 (31.8)62 (35.8)  0.495 
Lung parenchyma affected–no (%)  0.693    0.042 
0%  3 (3.6)  8 (3.5)    0 (0.0)2 (3.2)   
25%  24 (28.9)  74 (32.7)    15 (26.8)19 (30.2)   
50%  22 (26.5)  71 (31.4)    14 (25.0)26 (41.3)   
75%  28 (33.7)  61 (27.0)    22 (39.3)15 (23.8)   
100%  6 (7.2)  12 (5.3)    5 (8.9)1 (1.6)   
Chest X-ray performed–no (%)  136 (91.9)  366 (90.4)  0.740  111 (91.7)102 (91.1)  0.999 
Quadrants affected–no (%)      0.306    0.152 
6 (4.4)  29 (7.9)    5 (4.5)8 (7.9)   
27 (20.0)  90 (24.7)    22 (20.0)25 (24.8)   
41 (30.4)  94 (25.8)    35 (31.8)19 (18.8)   
61 (45.2)  152 (41.6)    48 (43.6)49 (48.5)   
Severity of ARDS–no (%)      <0.001    0.753 
Mild  61 (30.3)  50 (9.1)    29 (18.8)34 (22.2)   
Moderate  103 (51.5)  317 (57.5)    92 (59.7)86 (56.2)   
Severe  36 (18.0)  184 (33.4)    33 (21.4)33 (21.6)   
Co-existing disorders–no (%)           
Hypertension  72 (32.0)  210 (33.5)  0.741  64 (37.0)62 (35.8)  0.911 
Heart failure  12 (5.3)  24 (3.8)  0.338  10 (5.8)7 (4.0)  0.620 
Diabetes  42 (18.7)  153 (24.4)  0.080  36 (20.8)33 (19.1)  0.788 
Chronic kidney disease  7 (3.1)  28 (4.5)  0.439  5 (2.9)2 (1.2)  0.448 
Baseline creatinine, µmol/L*  77.0 (64.0–96.2)  76.0 (61.0–96.0)  0.263  80.0 (65.0–98.0)74.5 (61.0–96.0)  0.153 
Liver cirrhosis  1 (0.4)  2 (0.3)  0.999  0 (0.0)0 (0.0)  NA 
Chronic obstructive pulmonary disease  17 (7.6)  52 (8.3)  0.778  14 (8.1)15 (8.7)  0.999 
Active hematological neoplasia  5 (2.2)  9 (1.4)  0.540  2 (1.2)2 (1.2)  0.999 
Active solid neoplasia  8 (3.6)  13 (2.1)  0.218  6 (3.5)2 (1.2)  0.283 
Neuromuscular disease  1 (0.4)  5 (0.8)  0.999  1 (0.6)0 (0.0)  0.999 
Immunosuppression  8 (3.6)  14 (2.2)  0.327  4 (2.3)5 (2.9)  0.999 
Previous medication–no (%)           
Systemic steroids  11 (4.9)  24 (3.8)  0.557  9 (5.2)10 (5.8)  0.999 
Inhalation steroids  29 (12.9)  70 (11.2)  0.544  21 (12.1)24 (13.9)  0.750 
Angiotensin converting enzyme inhibitor  45 (20.0)  107 (17.1)  0.361  38 (22.0)37 (21.4)  0.999 
Angiotensin II receptor blocker  29 (12.9)  65 (10.4)  0.322  23 (13.3)23 (13.3)  0.999 
Beta-blockers  45 (20.0)  124 (19.8)  0.999  37 (21.4)32 (18.5)  0.591 
Insulin  13 (5.8)  46 (7.3)  0.540  9 (5.2)10 (5.8)  0.999 
Metformin  31 (13.8)  107 (17.1)  0.292  26 (15.0)27 (15.6)  0.999 
Statins  72 (32.0)  196 (31.3)  0.867  59 (34.1)55 (31.8)  0.732 
Calcium channel blockers  33 (14.7)  117 (18.7)  0.186  30 (17.3)35 (20.2)  0.582 
Vital signs           
Heart rate, bpm  91.0 (78.5–104.0)  92.0 (79.0–107.0)  0.480  91.0 (79.5–104.0)90.0 (78.0–105.0)  0.890 
Mean arterial pressure, mmHg  85.0 (73.0–99.5)  86.0 (75.0–102.0)  0.531  86.0 (75.8–100.3)83.0 (72.0–99.0)  0.176 
Organ support–no (%)           
Continuous sedation  214 (95.1)  605 (97.0)  0.209  163 (94.2)168 (97.7)  0.171 
Inotropic or vasopressor  184 (81.8)  484 (77.6)  0.217  139 (80.3)142 (82.6)  0.678 
Vasopressor  183 (81.3)  484 (77.6)  0.256  138 (79.8)142 (82.6)  0.582 
Inotropic  13 (5.8)  22 (3.5)  0.170  11 (6.4)9 (5.2)  0.818 
Fluid balance, mL  739.0 (124.2–1528.0)  634.0 (17.4–1427.7)  0.113  759.0 (112.0–1573.5)716.0 (104.0–1555.0)  0.798 
Urine output, mL  698.5 (352.5–1090.0)  760.0 (420.0–1215.0)  0.032  655.0 (350.0–1055.0)750.0 (395.0–1165.0)  0.163 

Data are median (quartile 25% - quartile 75%) or No (%). Percentages may not total 100 because of rounding.

Prevalence of hyperoxemia and excessive oxygen use

Of 851 patients, 182 (21.4%) patients were hyperoxemic on day 1 and 77 (9.0%) patients were hyperoxemic on day 2. Only 34 (4%) patients were hyperoxemic on both days, but 225 (26.4%) patients were hyperoxemic on either day 1 or day 2 (Fig. 1). eTable 1, eFigure 1 and eResults show group assignments, and daily mean PaO2 and SpO2.

Excessive oxygen use occurred at least once in 385 (45.2%) patients. The prevalence was not different between hyperoxemic and normoxemic patients on day 1 but occurred more often in normoxemic patients on day 2 (eTable 2). eTable 1, eFigure 1 and 2, and eResults show group assignments, and daily mean FiO2. Median daily FiO2 was slightly lower in hyperoxemic patients on both days.

Patient demographics and ventilation parameters

Moderate to severe ARDS was less often seen in hyperoxemic patients than in normoxemic patients, and hyperoxemic patients had a lower urinary output and a slightly higher plasma lactate at baseline (Table 1).

At start of ventilation, hyperoxemic patients received a similar VT at a comparable ΔP, and consequently had comparable Crs to normoxemic patients (eTable 1). At start of ventilation, hyperoxemic patients received ventilation with higher PEEP than normoxemic patients. VT was slightly higher in hyperoxemic patients, and the difference in PEEP persisted over the successive days (Fig. 2, eTable 2 and eFigure 3). Rescue therapies for hypoxemia were applied more often in normoxemic patients (Table 2 and eTable 2).

Fig. 2.

Cumulative frequency distributions of ventilation variables on day 1 of ventilation in the hyperoxemic (purple) and normoxemic (green) group, in the unmatched (left panels) and matched (right panels) cohorts. Horizontal dotted lines represent 50% of the patients and vertical dotted lines represent the median of the variable at the start of ventilation. All measurements are the means of a maximum of six measurements. P–values from Wilcoxon–Mann–Whitney test.

VT: tidal volume; PBW: predicted body weight; MP: mechanical power; VR: ventilator ratio; ΔP: driving pressure, RR: respiratory rate; MV: minute ventilation.

(0.39MB).
Table 2.

Clinical Outcomes According to Groups In the Unmatched and Matched Cohort.

  Unmatched Cohort (n = 851)  Matched Cohort (n = 346) 
  Hyperoxemic(n = 225)  Normoxemic(n = 626)  p  Hyperoxemic(n = 173)  Normoxemic(n = 173)  p 
Duration of ventilation, days  12.0 (7.0 – 22.0)  14.0 (8.0 – 23.0)  0.043  12.0 (7.0 – 21.0)  13.0 (8.0 – 22.5)  0.219 
In survivors at day 28, days  14.0 (8.0 – 25.0)  16.0 (10.0 – 28.3)  0.178  14.0 (9.0 – 25.0)  17.0 (10.0 – 26.0)  0.514 
Tracheostomy – no (%)  38 (17.0)  106 (17.0)  0.999  29 (17.0)  25 (14.5)  0.557 
Reintubation – no (%)  27 (12.1)  81 (13.0)  0.815  20 (11.7)  24 (14.0)  0.628 
Pneumothorax – no (%)  6 (2.7)  30 (4.8)  0.246  4 (2.3)  10 (5.8)  0.171 
Thromboembolic complications – no (%)  61 (27.1)  192 (30.7)  0.350  44 (25.4)  48 (27.7)  0.715 
Pulmonary embolism  42 (18.7)  153 (24.4)  0.080  32 (18.5)  42 (24.3)  0.238 
Deep vein thrombosis  10 (4.4)  34 (5.4)  0.726  7 (4.0)  4 (2.3)  0.542 
Ischemic stroke  7 (3.1)  19 (3.0)  0.999  5 (2.9)  7 (4.0)  0.770 
Myocardial infarction  7 (3.1)  8 (1.3)  0.082  5 (2.9)  2 (1.2)  0.448 
Systemic arterial embolism  1 (0.4)  3 (0.5)  0.999  0 (0.0)  1 (0.6)  0.999 
Acute kidney injury – no (%)  105 (46.7)  293 (47.1)  0.938  82 (47.4)  83 (48.3)  0.914 
Use of RRT – no (%)  43 (19.1)  116 (18.5)  0.842  33 (19.1)  28 (16.2)  0.573 
Use of rescue therapy – no (%)*  153 (68.9)  508 (81.5)  <0.001  121 (71.2)  133 (77.3)  0.217 
Prone positioning  115 (51.6)  400 (64.2)  0.001  90 (52.6)  96 (55.8)  0.589 
Recruitment maneuver  14 (7.4)  39 (7.8)  0.999  11 (7.6)  10 (7.0)  0.999 
Use of NMBA  92 (40.9)  328 (52.4)  0.003  75 (43.4)  90 (52.0)  0.132 
ECMO  1 (0.4)  7 (1.1)  0.689  1 (0.6)  1 (0.6)  0.999 
Use of continuous sedation – no (%)*  224 (99.6)  622 (99.4)  0.999  172 (99.4)  172 (99.4)  0.999 
Use of inotropic or vasopressor –no (%)*  220 (97.8)  590 (94.2)  0.044  168 (97.1)  166 (96.0)  0.770 
Use of vasopressor  219 (97.3)  590 (94.2)  0.073  167 (96.5)  166 (96.0)  0.999 
Use of inotropic  34 (15.1)  54 (8.6)  0.010  27 (15.6)  20 (11.6)  0.347 
ICU length of stay, days  14.0 (8.0 – 24.0)  17.0 (10.0 – 27.0)  0.020  14.0 (8.0 – 24.0)  16.0 (9.3 – 27.8)  0.152 
In survivors, days  16.0 (10.0 – 32.5)  19.0 (12.0 – 30.0)  0.214  17.0 (10.0 – 32.8)  19.0 (12.0 – 30.5)  0.642 
Hospital length of stay, days  21.0 (12.0 – 32.0)  25.0 (16.0 – 39.0)  0.008  21.0 (12.0 – 31.8)  23.0 (14.0 – 38.0)  0.220 
In survivors, days  28.0 (20.0 – 42.0)  32.0 (22.0 – 47.0)  0.041  27.5 (20.0 – 42.0)  32.0 (19.0 – 42.3)  0.702 
ICU mortality – no (%)  76 (34.1)  221 (35.6)  0.744  65 (38.0)  64 (37.2)  0.911 
Hospital mortality – no (%)  79 (36.4)  224 (37.0)  0.935  66 (39.8)  65 (39.4)  0.999 
28-day mortality – no (%)  75 (33.6)  193 (31.0)  0.502  63 (36.8)  56 (32.7)  0.496 
90-day mortality – no (%)  80 (37.6)  234 (39.5)  0.624  66 (40.7)  66 (41.5)  0.910 

Data are median (quartile 25% - quartile 75%) or No (%). Percentages may not total 100 because of rounding.

RRT: renal replacement therapy; NMBA: neuromuscular blocking agent; ECMO: extracorporeal membrane oxygenation.

assessed in the first four days of ventilation.

Outcomes

A flat relationship was seen between 28–day mortality and PaO2 at day 1; a U–shape relationship was seen between 28–day mortality and PaO2 at day 2, with a nadir PaO2 of 75 mmHg (eFigure 4). Mortality rates increased with higher FiO2 (eFigure 5).

Duration of ventilation and length of stay in hospital and ICU were shorter in hyperoxemic patients (Table 2), but not when death was treated as a competing risk (Fig. 3, eFigure 6 and 7). Mortality was not different between the two groups (Table 2, eFigure 8 and 9).

Fig. 3.

Mortality and pattern of extubation in the hyperoxemic (pink) and normoxemic (green) groups, before (left panels) and after (right panels) matching. SHR and p–value are from Fine–Gray competing risk model.

SHR: subdistribution hazard ratio.

(0.25MB).
Matched analysis

We matched 346 patients, resulting in fairly comparable groups, with persisting differences in PaO2 (Table 1, eTable 1 and 3, and eFigure 1, 10 and 11). In the matched analysis, there were no differences in any of the clinical outcomes (Table 2, Fig. 3, eFigures 6 to 9).

Sensitivity analysis

The sensitivity analysis in which we excluded hypoxemic patients did not show differences in clinical outcomes (eTable 4 and 5; eFigure 12 to 14).

Discussion

In this cohort of intubated patients with COVID–19 ARDS (1) the prevalence of hyperoxemia was high, and (2) many patients experienced excessive oxygen use, albeit that the prevalence per patient was low; in addition, (3) hyperoxemic patients received ventilation with a slightly higher VT, higher PEEP but a lower FiO2; and (4) there were no differences in clinical outcomes between hyperoxemic and normoxemic patients.

This analysis is one of the first to investigate the prevalence of hyperoxemia and excessive use of oxygen in a large cohort of invasively ventilated COVID–19 patients. Granular ventilation data was collected over the first days by investigators that were trained in data collection to ensure good quality of the data. Patients were included within a relatively short timeframe and therefore unlikely subjected to changes in the local protocols. We recruited patients in different types of hospitals, increasing the generalizability of the findings. This planned analysis was unknown to the caregivers at the time of data collection, minimizing the risk of observation bias. Finally, we had a sophisticated pre–published statistical analysis plan, including a propensity matched analysis to control for confounding factors.

The prevalence of hyperoxemia in our study was comparable,23,24 but mostly lower12,25,26 than in other cohorts of COVID–19 patients. One study that specifically examined hyperoxemia in invasively ventilated COVID–19 patients reported a prevalence threefold higher than in our cohort.26 Patients in that study, alike the other studies that reported a higher prevalence of hyperoxemia under invasive ventilation, had a higher PaO2/FiO2 at start of ventilation, suggesting that patients in those studies had less severe ARDS. It cannot be excluded, however, that the caregivers involved in care for patients in our cohort targeted lower oxygen levels, either because of the local protocols that were being used, or because they are more aware of the potential risks of hyperoxemia.27-29

The prevalence of hyperoxemia was remarkably lower than that seen in a large international cohort of ARDS patients from 2014.22 This difference may be explained in several ways. First, it is possible that ventilation strategies have changed over recent years. Lower tidal volumes are increasingly used, and VT reduction can result in lower oxygen levels, as also seen in the seminal ARDS Network trial named ‘ARMA’.30 Second, and in line with the suggestion above on oxygen targets, the findings of several studies in this topic5,29,31 may have resulted in lower oxygen targets. Third, hyperoxemia could be more difficult to achieve in patients with COVID–19 ARDS, due to extensive pulmonary infiltrates or sometimes the presence of pulmonary embolism.

One important finding of our study is that outcomes were not different between hyperoxemic versus normoxemic patients. However, this may not be too surprising since the differences in oxygen levels between the two patient groups were not as large as in the initial investigations that studied the effects of hyperoxemia in critically ill patients––of note, this was also the case in the recent randomized clinical trials that all showed no benefit of a low oxygen versus a high oxygen strategy.8-11 However, we noted an increasing mortality beyond a PaO2 of 75 mmHg in the LOESS curve.

Differences in ventilation between hyperoxemic and normoxemic patients were minimal. PEEP, however, was consistently higher in the hyperoxemic patients, even after matching. This may not be unexpected, as higher PEEP can result in more lung recruitment and thus improve oxygenation, as for instance also seen in patients in one study that was performed before just before the COVID–19 pandemic.32 Alike in patients with ARDS due to another cause, PEEP can also reaerate consolidated regions of the lungs in COVID–19 patients,33 thereby improving oxygenation.34 However, we do not suggest that high PEEP should be applied in all COVID–19 patients to improve oxygenation, as improved oxygenation does not automatically result in better outcome and non–recruitable patients are at high risk of overdistension and hemodynamic impairment.32 Interestingly, prone positioning was used more often in the normoxemic group. This finding suggests that prone positioning was adequately used as a rescue therapy for refractory hypoxemia.

Excessive oxygen use was seen in almost half of the patients in this cohort. However, only 13.6% of all the observed time–points showed excessive use of oxygen, which at least suggests that clinicians responded adequately to hyperoxemia with a reduction in FiO2. Interestingly, use of excessive oxygen was much lower than in previous cohorts of patients with ARDS due to another cause.22,27,35 Of note, the LOESS curve demonstrated increasing mortality with increasing levels of FiO2 regardless of the presence of hyperoxemia, and although this is partially linked to severity of disease, it could also represent the deleterious effects that high levels of FiO2 can have on lung tissue.36

Limitations of this analysis are as follows. We restricted the analysis to the first two days of oxygenation, and therefore cannot be sure whether dysoxemia at later time points had detrimental effects. The definitions of hyperoxemia and excessive oxygen were arbitrary, as there are no definitive targets for PaO2 or FiO2. Higher cut-offs for PaO2 and FiO2 could have resulted in other prevalences of hyperoxemia and excessive oxygen use, and maybe even associations of dysoxemia with outcome. Due to the low number of hypoxemic patients in this cohort, we were not able to compare outcomes in these patients. However, the sensitivity analysis in which we excluded hypoxemic patients showed that there were no differences in clinical outcomes. We were not able to collect data to gain insight in oxygen management per se, and we did also not collect the protocols in place at the participating centres. Unfortunately, we did not acquire ventilation data before intubation and therefore were not able to calculate the ROX index in patients previously on high flow nasal oxygen (HFNO). This index (ratio of SpO2/FiO2 to respiratory rate (RR)) predicts whether patients on HFNO will be in need of intubation, and it would have been interesting to see whether hyperoxemic patients had a lower chance of HFNO failure based on the ROX index.

Conclusions

In this cohort of COVID–19 ARDS patients, hyperoxemia and excessive oxygen use occurred often, but prevalences were lower than in previous studies in patients with ARDS due to another cause. The main difference between hyperoxemic and normoxemic patients was ARDS severity, use of PEEP and FiO2 and prone positioning. We found no effect of hyperoxemia on outcomes.

References
[1]
E Damiani, E Adrario, M Girardis, R Romano, P Pelaia, M Singer, et al.
Arterial hyperoxia and mortality in critically ill patients: a systematic review and meta-analysis.
[2]
NR Aggarwal, RG Brower, DN Hager, BT Thompson, G Netzer, C Shanholtz, et al.
Oxygen exposure resulting in arterial oxygen tensions above the protocol goal was associated with worse clinical outcomes in acute respiratory distress syndrome.
Crit Care Med, 46 (2018), pp. 517-524
[3]
RH Kallet, MA. Matthay.
Hyperoxic acute lung injury.
Respir Care, 58 (2013), pp. 123-141
[4]
M Singer, PJ Young, JG Laffey, P Asfar, FS Taccone, MB Skrifvars, et al.
Dangers of hyperoxia.
[5]
M Girardis, S Busani, E Damiani, A Donati, L Rinaldi, A Marudi, et al.
Effect of Conservative vs conventional oxygen therapy on mortality among patients in an intensive care unit: the oxygen-ICU randomized clinical trial.
JAMA, 316 (2016), pp. 1583-1589
[6]
R Panwar, M Hardie, R Bellomo, L Barrot, GM Eastwood, PJ Young, et al.
Conservative versus liberal oxygenation targets for mechanically ventilated patients. A pilot multicenter randomized controlled trial.
Am J Respir Crit Care Med, 193 (2016), pp. 43-51
[7]
P Asfar, F Schortgen, J Boisrame-Helms, J Charpentier, E Guerot, B Megarbane, et al.
Hyperoxia and hypertonic saline in patients with septic shock (HYPERS2S): a two-by-two factorial, multicentre, randomised, clinical trial.
Lancet Respir Med, 5 (2017), pp. 180-190
[8]
D Mackle, R Bellomo, M Bailey, R Beasley, A Deane, G Eastwood, et al.
Conservative oxygen therapy during mechanical ventilation in the ICU.
N Engl J Med, 382 (2020), pp. 989-998
[9]
L Barrot, P Asfar, F Mauny, H Winiszewski, F Montini, J Badie, et al.
Liberal or conservative oxygen therapy for acute respiratory distress syndrome.
N Engl J Med, 382 (2020), pp. 999-1008
[10]
OL Schjorring, TL Klitgaard, A Perner, J Wetterslev, T Lange, M Siegemund, et al.
Lower or higher oxygenation targets for acute hypoxemic respiratory failure.
N Engl J Med, 384 (2021), pp. 1301-1311
[11]
H Gelissen, HJ de Grooth, Y Smulders, EJ Wils, W de Ruijter, R Vink, et al.
Effect of low-normal vs high-normal oxygenation targets on organ dysfunction in critically ill patients: a randomized clinical trial.
JAMA, 326 (2021), pp. 940-948
[12]
G Grasselli, M Greco, A Zanella, G Albano, M Antonelli, G Bellani, et al.
Risk factors associated with mortality among patients with COVID-19 in intensive care units in Lombardy, Italy.
JAMA Intern Med, 180 (2020), pp. 1345-1355
[13]
MJ Cummings, MR Baldwin, D Abrams, SD Jacobson, BJ Meyer, EM Balough, et al.
Epidemiology, clinical course, and outcomes of critically ill adults with COVID-19 in New York City: a prospective cohort study.
Lancet, 395 (2020), pp. 1763-1770
[14]
C Ferrando, F Suarez-Sipmann, R Mellado-Artigas, M Hernandez, A Gea, E Arruti, et al.
Clinical features, ventilatory management, and outcome of ARDS caused by COVID-19 are similar to other causes of ARDS.
Intensive Care Med, 46 (2020), pp. 2200-2211
[15]
M Botta, AM Tsonas, J Pillay, LS Boers, AG Algera, LDJ Bos, et al.
Ventilation management and clinical outcomes in invasively ventilated patients with COVID-19 (PRoVENT-COVID): a national, multicentre, observational cohort study.
Lancet Respir Med, 9 (2021), pp. 139-148
[16]
LS Boers, M Botta, AM Tsonas, AG Algera, J Pillay, DA Dongelmans, et al.
PRactice of VENTilation in Patients with Novel Coronavirus Disease (PRoVENT-COVID): rationale and protocol for a national multicenter observational study in The Netherland.
Annals of Translational Medicine, 8 (2020),
[17]
PRoVENT-COVID collaborators. Prevalence of Hyperoxia and Excess Oxygen, and Associations with Outcomes in Patients with COVID–19 ARDS – insights from the PRoVENT–COVID study. 2021. https://sites.google.com/view/provent-covid/effects-of-dysoxia-sap?authuser=0 (accessed 05 January, 2021).
[18]
W Stilma, DMP van Meenen, CMA Valk, H de Bruin, F Paulus, A Serpa Neto, et al.
Incidence and practice of early prone positioning in invasively ventilated COVID-19 patients-insights from the PRoVENT-COVID observational study.
J Clin Med, 10 (2021), pp. 4783
[19]
AM Tsonas, M Botta, J Horn, MJ Brenner, MS Teng, BA McGrath, et al.
Practice of tracheostomy in patients with acute respiratory failure related to COVID-19 - insights from the PRoVENT-COVID study.
Pulmonology, 28 (2022), pp. 18-27
[20]
JP Roozeman, G Mazzinari, A Serpa Neto, MW Hollmann, F Paulus, MJ Schultz, et al.
Prognostication using SpO2/FiO2 in invasively ventilated ICU patients with ARDS due to COVID-19 - insights from the PRoVENT-COVID study.
J Crit Care, 68 (2022), pp. 31-37
[21]
AM Tsonas, M Botta, J Horn, L Morales-Quinteros, A Artigas, MJ Schultz, et al.
Clinical characteristics, physiological features, and outcomes associated with hypercapnia in patients with acute hypoxemic respiratory failure due to COVID-19—insights from the PRoVENT-COVID study.
J Crit Care, 69 (2022),
[22]
F Madotto, E Rezoagli, T Pham, M Schmidt, B McNicholas, A Protti, et al.
Hyperoxemia and excess oxygen use in early acute respiratory distress syndrome: insights from the LUNG SAFE study.
[23]
K Roedl, D Jarczak, L Thasler, M Bachmann, F Schulte, B Bein, et al.
Mechanical ventilation and mortality among 223 critically ill patients with coronavirus disease 2019: A multicentric study in Germany.
Aust Crit Care, (2020),
[24]
F Fusina, F Albani, M Bertelli, E Cavallo, S Crisci, R Caserta, et al.
Corrected minute ventilation is associated with mortality in ards caused by COVID-19.
Respir Care, (2020),
[25]
E Estenssoro, CI Loudet, FG Rios, VS Kanoore Edul, G Plotnikow, M Andrian, et al.
Clinical characteristics and outcomes of invasively ventilated patients with COVID-19 in Argentina (SATICOVID): a prospective, multicentre cohort study.
Lancet Respir Med, 9 (2021), pp. 989-998
[26]
EP Gomes, MM Reboredo, GB Costa, EV Carvalho, BV. Pinheiro.
Hyperoxemia and excessive oxygen use in COVID-19-related ARDS: preliminary results of a prospective cohort study.
J Bras Pneumol, 47 (2021),
[27]
AE de Graaff, DA Dongelmans, JM Binnekade, E. de Jonge.
Clinicians' response to hyperoxia in ventilated patients in a Dutch ICU depends on the level of FiO2.
Intensive Care Med, 37 (2011), pp. 46-51
[28]
E de Jonge, L Peelen, PJ Keijzers, H Joore, D de Lange, PH van der Voort, et al.
Association between administered oxygen, arterial partial oxygen pressure and mortality in mechanically ventilated intensive care unit patients.
Crit Care, 12 (2008), pp. R156
[29]
HJ Helmerhorst, DL Arts, MJ Schultz, PH van der Voort, A Abu-Hanna, E de Jonge, et al.
Metrics of arterial hyperoxia and associated outcomes in critical care.
Crit Care Med, 45 (2017), pp. 187-195
[30]
RG Brower, MA Matthay, A Morris, D Schoenfeld, BT Thompson, et al.
Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome.
N Engl J Med, 342 (2000), pp. 1301-1308
[31]
DK Chu, LH Kim, PJ Young, N Zamiri, SA Almenawer, R Jaeschke, et al.
Mortality and morbidity in acutely ill adults treated with liberal versus conservative oxygen therapy (IOTA): a systematic review and meta-analysis.
Lancet, 391 (2018), pp. 1693-1705
[32]
AB Cavalcanti, EA Suzumura, LN Laranjeira, DM Paisani, LP Damiani, et al.
Effect of lung recruitment and titrated positive end-expiratory pressure (PEEP) vs low PEEP on mortality in patients with acute respiratory distress syndrome: a randomized clinical trial.
JAMA, 318 (2017), pp. 1335-1345
[33]
MR Smit, LFM Beenen, CMA Valk, MM de Boer, MJ Scheerder, JT Annema, et al.
Assessment of lung reaeration at 2 levels of positive end-expiratory pressure in patients with early and late COVID-19-related acute respiratory distress syndrome.
J Thorac Imaging, 36 (2021), pp. 286-293
[34]
D Chiumello, M Bonifazi, T Pozzi, P Formenti, GFS Papa, G Zuanetti, et al.
Positive end-expiratory pressure in COVID-19 acute respiratory distress syndrome: the heterogeneous effects.
[35]
S Rachmale, G Li, G Wilson, M Malinchoc, O. Gajic.
Practice of excessive FiO2 and effect on pulmonary outcomes in mechanically ventilated patients with acute lung injury.
Respir Care, 57 (2012), pp. 1887-1893
[36]
HJF Helmerhorst, LRA Schouten, GTM Wagenaar, NP Juffermans, J Roelofs, MJ Schultz, et al.
Hyperoxia provokes a time- and dose-dependent inflammatory response in mechanically ventilated mice, irrespective of tidal volumes.
Intensive Care Med Exp, 5 (2017), pp. 27

Ary Serpa Neto discloses personal fees received from Drager, outside the submitted work. All other authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

PRoVENT–COVID Collaborative Group: (in alphabetic order): S. Ahuja; J.P. van Akkeren; A.G. Algera; C.K. Algoe; R.B. van Amstel; P. van de Berg; D.C. Bergmans; D.I. van den Bersselaar; F.A. Bertens; A.J. Bindels; J.S. Breel; C.L. Bruna; M.M. de Boer; S. den Boer; L.S. Boers; M. Bogerd; L.D. Bos; M. Botta; O.L. Baur; H. de Bruin; L.A. Buiteman–Kruizinga; O. Cremer; R.M. Determann; W. Dieperink; J. v. Dijk; D.A. Dongelmans; M.J. de Graaff; M.S. Galekaldridge; L.A. Hagens; J.J. Haringman; S.T. van der Heide; P.L. van der Heiden; L.L. Hoeijmakers; L. Hol; M. W. Hollmann; J. Horn; R. van der Horst; E.L. Ie; D. Ivanov; N.P. Juffermans; E. Kho; E.S. de Klerk; A.W. Koopman; M. Koopmans; S. Kucukcelebi; M.A. Kuiper; D.W. de Lange; I. Martin–Loeches; G. Mazzinari; D.M. van Meenen; N. van Mourik; S.G. Nijbroek; E.A. Oostdijk; F. Paulus; C.J. Pennartz; J. Pillay; I.M. Purmer; T.C. Rettig; O. Roca; J.P. Roozeman; M.J. Schultz; A. Serpa Neto; G.S. Shrestha; M.E. Sleeswijk; P.E. Spronk; A.C. Strang; W. Stilma; P. Swart; A.M. Tsonas; C.M.A. Valk; A.P. Vlaar; L.I. Veldhuis; W.H. van der Ven; P. van Velzen; P. van Vliet; P. van der Voort; L. van Welie; B. van Wijk; T. Winters; W.Y. Wong; A.R. van Zanten.

Copyright © 2022. Sociedade Portuguesa de Pneumologia
Pulmonology
Article options
Tools
Supplemental materials

Are you a health professional able to prescribe or dispense drugs?