Too much tolerance for hyperoxemia in mechanically ventilated patients with SARS-CoV-2 pneumonia? Report from an Italian intensive care unit

Background: In COVID-19 patients requiring mechanical ventilation, the administration of high oxygen (O₂) doses for prolonged time periods may be necessary. Although life-saving in most cases, O₂ may exert deleterious effects if administered in excessive concentrations. We aimed to describe the prevalence of hyperoxemia and excessive O₂ administration in mechanically ventilated patients with SARS-CoV-2 pneumonia and determine whether hyperoxemia is associated with mortality in the Intensive Care Unit (ICU) or the onset of ventilator-associated pneumonia (VAP).

Materials and methods: Retrospective single-center study on adult patients with SARS-CoV-2 pneumonia requiring invasive mechanical ventilation for ≥48 h. Patients undergoing extracorporeal respiratory support were excluded. We calculated the excess O₂ administered based on the ideal arterial O₂ tension (PaO₂) target of 55–80 mmHg. We defined hyperoxemia as PaO₂ > 100 mmHg and hyperoxia + hyperoxemia as an inspired O₂ fraction (FiO₂) > 60% + PaO₂ > 100 mmHg. Risk factors for ICU-mortality and VAP were assessed through multivariate analyses.

Results: One hundred thirty-four patients were included. For each day of mechanical ventilation, each patient received a median excess O₂ of 1,121 [829–1,449] L. Hyperoxemia was found in 38 [27–55]% of arterial blood gases, hyperoxia + hyperoxemia in 11 [5–18]% of cases. The FiO₂ was not reduced in 69 [62–76]% of cases of hyperoxemia. Adjustments were made more frequently with higher PaO₂ or initial FiO₂ levels. ICU-mortality was 32%. VAP was diagnosed in 48.5% of patients. Hyperoxemia (OR 1.300 95% CI [1.097–1.542]), time of exposure to hyperoxemia (OR 2.758 [1.406–5.411]), hyperoxia + hyperoxemia (OR 1.144 [1.008–1.298]), and daily excess O₂ (OR 1.003 [1.001–1.005]) were associated with higher risk for ICU-mortality, independently of age, Sequential Organ failure Assessment score at ICU-admission and mean PaO₂/FiO₂. Hyperoxemia (OR 1.033 [1.006–1.061]), time of exposure to hyperoxemia (OR 1.108 [1.018–1.206]), hyperoxia + hyperoxemia (OR 1.038 [1.003–1.075]), and daily excess O₂ (OR 1.001 [1.000–1.001]) were identified as risk factors for VAP, independently of body mass index, blood transfusions, days of neuromuscular blocking agents (before VAP), prolonged prone positioning and mean PaO₂/FiO₂ before VAP.

Conclusion: Excess O₂ administration and hyperoxemia were common in mechanically ventilated patients with SARS-CoV-2 pneumonia. The exposure to hyperoxemia may be associated with ICU-mortality and greater risk for VAP.

Introduction

Supplemental oxygen (O₂) is a life-saving therapy in hypoxemic patients in order to guarantee adequate tissue O₂ delivery. Nonetheless, excessive O₂ administration may also exert deleterious effects (1). In recent years, several studies supported the use of more conservative oxygenation strategies in Intensive Care Units (ICUs), whereas liberal O₂ therapy and the exposure to arterial hyperoxia in critically ill patients were associated with adverse outcomes (2–4). The lung is the first organ affected by O₂ toxicity. Hyperoxia induces oxidative stress and inflammation in the lung (1) and may impair the surfactant system, thus causing alveolar collapse and the reduction in pulmonary compliance (5). Excess O₂ administration may also compromise muco-ciliary clearance and the anti-microbial capacity of the immune cells, thus contributing to the development of ventilator-associated pneumonia (VAP) (6).

In moderate or severe acute respiratory distress syndrome (ARDS), the administration of high inspired O₂ fractions (FiO₂) is frequently required to maintain normoxemia (arterial O₂ tension [PaO₂] 80–100 mmHg) and this may predispose to additional hyperoxia-induced lung injury. In order to limit the exposure to hyperoxia, the ARDS Network recommends using a PaO₂ target of 55–80 mmHg in mechanically ventilated patients (7). Nonetheless, ARDS patients are frequently exposed to excessive FiO₂ levels (8) and even undergo a condition of hyperoxemia (PaO₂ > 100 mmHg) in a substantial number of cases (9). In a meta-analysis of randomized controlled trials (RCTs), excessive O₂ administration (FiO₂ > 50%) resulting in PaO₂ levels above the protocol goal (>80 mmHg) was associated with mortality and lower ventilator- and hospital-free days (10). A recent multicentre RCT was aimed to compare a conservative (target PaO₂ 55–70 mmHg, SpO₂ 88–92%) with a liberal oxygenation strategy (target PaO₂ 90–105 mmHg, peripheral O₂ saturation [SpO₂] ≥ 96%) in ARDS patients, however, this study was prematurely stopped due to safety concerns (higher mortality and five mesenteric ischemic events in the conservative O₂ group) (11). Therefore, which is the safest oxygenation target for mechanically ventilated patients with ARDS remains an open question.

This is a problem of major importance for patients with severe acute respiratory syndrome due to novel Coronavirus (SARS-CoV-2). A large Italian cohort study showed that 12% of these patients received FiO₂ up to 100% and an FiO₂ ≥ 50% was necessary in 89% of total (12). Hyperoxia-induced lung injury may add to the inflammatory process caused by the viral infection. Moreover, these patients often require long periods of mechanical ventilation and are at risk of exposure to high O₂ concentrations for several days (13). Despite the key role of O₂ in the treatment of SARS-CoV-2, the potential adverse effects of a prolonged exposure to hyperoxia in this patient category remain unexplored.

The primary aim of this study was to explore whether mechanically ventilated patients with SARS-CoV-2 pneumonia in our ICU received an excessive amount of O₂ and were exposed to hyperoxemia. In addition, we evaluated if hyperoxemia was associated with mortality or the onset of VAP.

Materials and methods

This retrospective observational study was conducted in the “General, respiratory and major trauma Intensive care Unit” of the Azienda Ospedaliera Universitaria “Ospedali Riuniti Umberto I-Lancisi-Salesi” of Ancona, Italy. During the pandemic phase, this ICU provided 18 beds for COVID-19 patients. The study protocol was approved by the local Ethics Committee (Comitato Etico Regionale Marche). Written informed consent was not requested due to the retrospective study design. This study included all consecutive adult (>18 year old) patients with SARS-CoV-2 pneumonia admitted to the ICU between February 2020 and May 2021, who required endotracheal intubation and invasive mechanical ventilation for at least 48 consecutive hours. COVID-19 infection was confirmed by means of real time polymerase chain reaction on nasopharyngeal swab or bronchoalveolar lavage. Exclusion criteria were: duration of mechanical ventilation < 48 h; ICU-discharge or death within 48 h; use of extracorporeal membrane oxygenation (ECMO) or extracorporeal carbon dioxide removal (ECCO₂R); admission from another ICU with a length of stay > 48 h; re-admissions after a previous ICU admission for SARS-CoV-2 pneumonia; COVID-19+ patients without pneumonia admitted to the ICU for different reasons.

Patient management

According to the most recent guidelines for the management of ARDS and COVID-19 (7, 14, 15), all patients received a lung protective ventilation strategy with a tidal volume (TV) of 4–6 ml/kg of ideal body weight, while maintaining a plateau pressure (Pplat) ≤ 30 cmH₂O and a driving pressure ≤ 15 cmH₂O. A positive end-expiratory pressure (PEEP) ≥ 5 cmH₂O was applied in all patients, using an open-lung strategy. Neuromuscular blocking agents (NMBA) (continuous infusion of cisatracurium or rocuronium) were used in the early phase of mechanical ventilation in cases of refractory hypoxemia despite deep sedation to facilitate lung protective ventilation, during prone positioning, in cases of patient-ventilator dyssynchrony and/or in presence of high respiratory drive despite optimal sedation (15). Prone positioning was used in patients with a PaO₂/FiO₂ < 150 mmHg for a duration of at least 16 h per session (15). During the COVID-19 pandemic, we implemented and applied a protocol of prolonged prone positioning with a duration of 36 consecutive hours per session, which showed to be feasible and potentially more effective in improving oxygenation (16). Inhaled nitric oxide was administered as a rescue therapy in cases of refractory hypoxemia despite the use of NMBA and prone ventilation. Veno-venous ECMO was used in patients with refractory hypoxemia despite all other therapies (17).

A VAP prevention strategy was routinely applied, including: oral care with chlorhexidine three times a day; head of bed elevation (30° whenever possible); check (3 times a day) and maintenance of the cuff pressure of the endotracheal tube (25 cmH₂O); routine aspiration of tracheal secretions by means of a closed suction system; frequent check of the correct positioning and function of the nasogastric tube; weaning from mechanical ventilation as soon as possible; tracheostomy in cases of anticipated difficult weaning (18). During prone ventilation, the administration of enteral nutrition was discontinued or, alternatively, continued at low dosage (10–20 ml/h) with frequent check of the gastric residual volume and possible episodes of regurgitation. Stress ulcer prophylaxis (pantoprazole 40 mg/day) was administered in patients with risk factors for gastro-intestinal bleeding (including coagulopathy, mechanical ventilation >48 h, hypotension), who represented the vast majority of the patients enrolled. Deep venous thrombosis prophylaxis was applied in all patients with low molecular weight heparin (at least 4,000 UI/day). In patients with VAP, antibiotic therapy was based on local written protocols in accordance with the most recent guidelines (19).

Data collection

For each included patient we collected: age, gender, body mass index (BMI), comorbidities (obesity, arterial hypertension, diabetes mellitus, chronic obstructive pulmonary disease, ischemic cardiomyopathy), the Sequential Organ Failure Assessment (SOFA) score at ICU admission, arterial lactate levels at ICU admission, PaO₂/FiO₂ at ICU admission, ICU length-of-stay, duration of invasive mechanical ventilation, ICU mortality, antibiotics prior to ICU admission, steroids prior to ICU admission, total steroid dose administered in the ICU (adjusted for body weight, mg/kg/day), episodes of VAP (early/late onset, causal pathogen, antibiotic resistance), use of NMBA, prone positioning (number of sessions, prolonged prone ventilation), use of proton pump inhibitors, blood transfusions (number of units).

The onset of VAP was defined by the detection of a causative agent in a respiratory sample (tracheal aspirate or bronchoalveolar lavage) associated with a new or progressive lung infiltrate, plus at least two clinical criteria (fever or hypothermia, leukocytosis or leukopenia, purulent secretions), after at least 48 h of invasive mechanical ventilation (18). In cases of multiple VAP episodes, only the first episode was considered. An episode of VAP occurring within 4 days after endotracheal intubation was defined as “early onset VAP” (18).

Starting from endotracheal intubation and for the whole duration of invasive mechanical ventilation, we collected data from all arterial blood gas (ABG) analyses, including PaO₂, SaO₂, pH, PaCO₂ with date and time. In our ICU, ABG are generally performed at least every 8 h in all patients. For each ABG we recorded data on mechanical ventilation: TV (ml), respiratory rate (RR), minute ventilation (l/min), PEEP, FiO₂, and PaO₂/FiO₂. In addition, for each ABG we noted if the patient was on NMBA, prone ventilation, and measures of respiratory mechanics if available (Pplat, driving pressure, static compliance [Cstat]).

Based on the ARDS Network protocol, we considered a PaO₂ of 80 mmHg as the upper limit of the oxygenation target. For each ABG showing a PaO₂ > 80 mmHg, we estimated the “ideal FiO₂” that would be sufficient to obtain a PaO₂ of 80 mmHg based on the PaO₂/FiO₂ ratio, according to the following formula:

${\mathrm{FiO}}_{\mathrm{2ideal}}=(80*{\mathrm{FiO}}_{\mathrm{2actual}})/{\mathrm{PaO}}_{\mathrm{2actual}}$

We then estimated the excess O₂ being administered as follows:

$\begin{array}{c}\mathrm{Excess}\mathrm{O}{}_2(\mathrm{L}/\min )=\mathrm{Minute}\mathrm{ventilation}(\mathrm{L}/\min )*({\mathrm{FiO}}_{\mathrm{2 actual}}-{\mathrm{FiO}}_{\mathrm{2 ideal}})\hfill \end{array}$

We then multiplied the value obtained for the time (minutes) elapsed since the previous ABG and calculated:

- Total excess O₂ (L): total cumulative amount of excess O₂ administered during the whole duration of invasive mechanical ventilation;

- Daily excess O₂ (L/day): average amount of excess O₂ administered in each day of invasive mechanical ventilation;

- Three-days excess O₂ (L): cumulative amount of excess O₂ administered during the first 3 days of invasive mechanical ventilation;

- Daily excess O₂ before VAP (L/day): average amount of excess O₂ administered in each day of invasive mechanical ventilation before the first diagnosis of VAP.

Episodes of “hyperoxemia” were defined by a PaO₂ > 100 mmHg. We defined as “uncorrected hyperoxemia” any episode of hyperoxemia that was not followed by a reduction in the FiO₂. In addition, we defined as “hyperoxia + hyperoxemia” any episode of FiO₂ > 60% in presence of a PaO₂ > 100 mmHg. Herein, the term “hyperoxia” (generally defined as the administration of any O2 dose > 21%) was used to indicate a high FiO2 (>60%). We then calculated the prevalence of hyperoxemia and hyperoxia + hyperoxemia for the whole duration of invasive mechanical ventilation, in the first 3 days of mechanical ventilation and for the days prior to VAP diagnosis. The duration of exposure to hyperoxemia was estimated as follows:

1. Daily time of exposure to hyperoxemia (hours per day of MV): whenever an ABG showed a PaO2 > 100 mmHg, we considered as time of exposure to hyperoxemia the time elapsed since the previous ABG; the total sum of hours was then divided by the number of days of MV.

2. Time of exposure to hyperoxemia in the first 3 days of MV: total sum of hours of hyperoxemia in the first 3 days of MV.

3. Daily time of exposure to hyperoxemia before VAP: average amount of hours of hyperoxemia in each day of invasive mechanical ventilation before the first diagnosis of VAP.

In addition, we calculated: mean FiO₂, mean PaO₂, mean PaO₂/FiO₂, highest FiO₂, highest PaO₂.

Statistical analysis

Statistics was performed with GraphPad Prism version 6 (GraphPad Software, La Jolla, CA, United States) and Statistical Package for Social Science software, version 17.0 (SPSS Inc., Chicago, IL, United States). Normality of distribution was checked with the Shapiro–Wilk test. Continuous variables were expressed as mean ± standard deviation or median [1^st–3^rd quartile], as appropriate. Unpaired t-test or Mann Whitney U-test were used for comparisons of two groups. The chi-square test was used for nominal variables. We constructed multivariate binary logistic regression models in order to evaluate the independent association between the exposure to hyperoxemia (prevalence of hyperoxemia, prevalence of hyperoxia + hyperoxemia, daily excess O₂) and the outcomes of interest (ICU mortality, diagnosis of VAP). Separate models were constructed for each index of exposure to hyperoxemia in order to avoid multi-collinearity. The basic assumptions for conducting logistic regression analyses were verified, including the absence of multi-collinearity and the linearity of the logit for each continuous independent variable (20). Covariates included in the models were selected based on their well-established association with the outcome of interest (6, 20). A p-value < 0.05 was used to indicate statistical significance.

Results

Between February 27, 2020 and May 12, 2021, a total of 207 COVID-19 patients was admitted to our ICU. Of these, exclusions were: 43 patients who underwent ECMO or ECCO₂R; 7 patients who had an ICU length of stay < 48 h; 1 patient who was extubated before 48 h; 1 non-intubated patient; 13 patients transferred from a different ICU in which they stayed for more than 48 h; 6 re-admissions; 2 COVID-19+ patients without pneumonia. Therefore, we included 134 patients in total. The vast majority of patients was male, and the most frequent comorbities were obesity and arterial hypertension (Table 1). ICU-mortality was 32%. ICU Non-survivors were older, had a higher prevalence of ischemic cardiomyopathy and worse PaO₂/FiO₂ and SOFA score at ICU-admission (Table 1).

TABLE 1

Table 1. General characteristics of the included patients.

Excess O₂ and hyperoxemia

We analyzed 9,583 ABGs in total. A PaO₂ > 80 mmHg was found in 68.8% of cases. All the analyzed patients received an excessive amount of O₂ in relation to the oxygenation target indicated by the guidelines (55–80 mmHg). We estimated that, for the whole duration of invasive mechanical ventilation, each patient received an average excess O₂ of 17,741 [8,950–27,248] L, corresponding to a daily excess O₂ of 1,121 [829–1,449] L (minimum: 319 L; maximum: 3,818 L). In each patient, hyperoxemia (PaO₂ > 100 mmHg) was present in 38 [27–55]% of ABGs, and 11 [5–18]% of ABGs showed a condition of hyperoxia + hyperoxemia (FiO₂ > 60% + PaO₂ > 100 mmHg). Hyperoxemia was more frequently associated with higher PaO₂/FiO₂ and possible indicators of less severe pulmonary dysfunction (lower Pplat and driving pressure) (Table 2). Figure 1 show the prevalence of hyperoxemia stratified by the FiO₂ and PaO₂/FiO₂.

TABLE 2

Table 2. Comparison of all arterial blood gases with or without hyperoxemia (PaO₂ > 100 mmHg).

FIGURE 1

Figure 1. Number and percentage of arterial blood gases showing hyperoxemia stratified by PaO₂ and FiO₂ levels.

In 69 [62–76]% of cases, an ABG showing hyperoxemia was not followed by a reduction in the FiO₂ (uncorrected hyperoxemia). The FiO₂ was more likely to be reduced in presence of higher PaO₂ and initial FiO₂ levels, although even in presence of a PaO₂ ≥ 150 mmHg or an initial FiO₂ ≥ 80% no change was made in about 40% of cases (Figure 2). The choice of reducing the FiO₂ was not influenced by PaO₂/FiO₂ or PEEP levels (Figure 2).

FIGURE 2

Figure 2. Episodes of uncorrected hyperoxemia stratified by PaO₂, FiO₂, PaO₂/FiO₂, and PEEP levels.

Hyperoxemia and intensive care unit-mortality

Intensive care unit-survivors showed higher mean PaO₂/FiO₂, mean PaO₂ and maximum PaO₂ as compared to Non-survivors, however mean and maximum FiO₂ were significantly lower (Table 3). Hyperoxemia was more frequent among ICU-survivors, and the daily time of exposure was higher, even if Non-survivors showed a higher prevalence of hyperoxia + hyperoxemia (both for the whole duration of mechanical ventilation and in the first 3 days) (Table 3). After adjusting for age, SOFA score at ICU-admission and mean PaO₂/FiO₂, the prevalence of hyperoxemia, the duration of exposure, the prevalence of hyperoxia + hyperoxemia and daily excess O₂ were independently associated with ICU-mortality (Table 4). The total excess O₂ and the excess O₂ in the first 3 days were not significantly associated with mortality in the logistic regression analysis.

TABLE 3

Table 3. Comparison of oxygenation variables between ICU-survivors and Non-survivors.

TABLE 4

Table 4. Binomial logistic regression models for the association between hyperoxemia and ICU-mortality.

Hyperoxemia and ventilator-associated pneumonia

Sixty-five patients (48.5% of total) had at least one episode of VAP during the ICU-stay. The vast majority (91%) of VAP episodes was late-onset. The most frequently isolated pathogens were: Staphylococcus Aureus (29 cases); Acinetobacter Baumanii (20 cases); Enterobacteriaceae (Klebsiella spp., Escherichia Coli, Serratia spp., Enterobacter spp., 34 cases); Pseudomonas Aeruginosa (14 cases); Corynebacterium Striatum (10 cases); Proteus Mirabilis (10 cases); Streptococcus Pneumoniae (3 cases); Aspergillus spp. (7 cases). Multi-drug resistant pathogens were isolated in 33% of cases.

Patients with VAP had longer ICU-stay, more frequently underwent prolonged prone positioning, and showed lower PaO₂/FiO₂ than those with no episodes of VAP (Table 5). Patients with at least one episode of VAP showed higher prevalence of hyperoxia + hyperoxemia (before VAP and in the first 3 days of mechanical ventilation) and received a higher excess O₂ (daily and in the first 3 days of mechanical ventilation) (Table 5).

TABLE 5

Table 5. Comparison between patients whit at least one episode of VAP and those without any episode of VAP.

After adjusting for BMI, blood transfusions, days of NMBA (before VAP), prolonged prone positioning and mean PaO₂/FiO₂ before VAP, a higher prevalence of hyperoxemia was associated with a higher risk of VAP (Table 6): the adjusted risk for VAP increased by 3.3% for each unitary increase in the percentage of ABGs with hyperoxemia. Similarly, the time of exposure to hyperoxemia before the diagnosis of VAP, the prevalence of hyperoxia + hyperoxemia, as well as the daily excess O₂, were independently associated with the risk of VAP (Table 6). Moreover, patients in the highest tertile of daily excess O₂ showed a 4.3 times greater adjusted risk of developing VAP as compared to those in the lowest tertile. The total amount of excess O₂ received before the diagnosis of VAP was not significantly associated with the onset of VAP, nor were the prevalence of hyperoxemia or hyperoxia + hyperoxemia in the first 3 days of mechanical ventilation and the duration of hyperoxemia in the first 3 days.

TABLE 6

Table 6. Binomial logistic regression models for the association between hyperoxemia and VAP.

Discussion

This retrospective single-centre study on 134 mechanically ventilated patients with SARS-CoV-2 pneumonia showed that: first, the dose of supplemental O₂ administered was often excessive in comparison to the oxygenation target indicated by current guidelines and the exposure to hyperoxemia was frequent; second, in most cases of hyperoxemia the FiO₂ was not varied, adjustments in O₂ dose were made more frequently in presence of higher PaO₂ or higher initial FiO₂; third, the exposure to hyperoxemia was independently associated with higher risks of ICU-mortality and VAP.

Our data are consistent with those of previous studies showing that clinicians generally tolerate higher PaO₂ values than those commonly recommended. In a retrospective study in critically ill mechanically ventilated patients, de Graaf et al. showed a prevalence of hyperoxemia (PaO₂ > 120 mmHg) of 22% and adjustments in the FiO₂ were rarely made, especially if this was ≤40% (21). Suzuki et al. calculated that an excess O₂ dose of 3,472 L per patient was administered on average during mechanical ventilation and no change in the FiO₂ was made in most cases of hyperoxemia if the initial level was 30–40% (22). In a recent study in COVID-19 mechanically ventilated patients, the prevalence of hyperoxemia in the ICU using a standard oxygenation protocol was 75.9% in the first day (23). Similarly in our study, the FiO₂ was more likely to be reduced in presence of more severe hyperoxemia, whereas PaO₂ values of 100–119 mmHg were accepted in more than 80% of cases. The initial FiO₂ also determined the clinicians’ behavior against hyperoxemia: if this was <40% no change was made in almost 90% of cases, while an initial FiO₂ ≥ 80% was corrected in more than 60% of cases.

A possible explanation of this too liberal attitude toward O₂ therapy is a lack of perception of the risks associated with hyperoxemia. A survey by Helmerhorst et al. showed that most clinicians recognize the potential deleterious effects of a prolonged exposure to excessive O₂ concentrations, including hyperoxia-induced lung injury, and show little tolerance toward even mild hyperoxemia (24). Nonetheless, a large proportion of their ICU patients was exposed to higher arterial O₂ levels than self-reported target ranges (24). These data suggest that, in actual clinical practice, clinicians tend to tolerate a certain degree of hyperoxia, as perceived as a safety buffer against hypoxemia. In the context of the COVID-19 pandemic, such a liberal attitude may be partly justified by a greater fear against episodes of hypoxemia in a clinical scenario in which a ready accessibility to the patient is not always guaranteed due to the isolation precautions, and an SpO₂ of 98–100% on the monitor could appear more reassuring than an SpO₂ of 90–92%.

As a matter of fact, O₂ toxicity and hyperoxia-induced lung injury are relatively slow processes in comparison to more acute conditions that can rapidly induce pulmonary oedema, respiratory failure and shock (such as aspiration pneumonia or sepsis-induced ARDS). A continuous exposure to hyperoxia for days/weeks can lead to diffuse alveolar damage, respiratory failure and high risk of death, while prolonged exposures to sub-lethal O₂ doses generally induce pulmonary fibrosis (1). In healthy volunteers, breathing 95% O₂ for 17 h caused a significant alveolar-capillary leak, due to a progressive destruction of the alveolar-capillary membrane, alveolar hemorrhage, formation of microthrombi, and intrapulmonary shunt (25). Moreover, oxidative stress compromises the surfactant system, thus causing atelectasis and reduction in lung compliance (26). Patients with ARDS may be particularly susceptible to O₂ toxicity, since the ongoing inflammatory process may already compromise their adaptive and anti-oxidant capacities. Moreover, hyperoxia may aggravate and predispose the lung to the deleterious effects of positive pressure ventilation (5).

A recent meta-analysis of RCTs showed that the use of higher oxygenation targets in the critically ill may increase mortality, even if with a very-low level of evidence (27). The LOCO₂ trial failed to prove the safety and efficacy of a more restrictive oxygen therapy in patients with ARDS (11), however, these results cannot be conclusive. Similarly, the multi-center HOT-ICU trial did not show any difference in 90-day mortality with a lower oxygenation target as compared to higher PaO2 target in patients with acute hypoxemic respiratory failure (28). Our study showed that the exposure to hyperoxemia in mechanically ventilated patients with SARS-CoV-2 pneumonia may be associated with higher ICU mortality, independently of other risk factors such as the mean PaO₂/FiO₂, SOFA score at admission and age. Moreover, a higher prevalence of hyperoxemia, as well as an excess O₂ administration, was independently associated with a greater risk of VAP.

Ventilator-associated pneumonia is the most common ICU-acquired infection among mechanically ventilated patients, leading to higher mortality, longer need for mechanical ventilation, and an increase in healthcare cost (18). For the implementation of effective preventive strategies, it is crucial to gain a deep understanding of the pathophysiology and risk factors for VAP. There is ample evidence that a long-term exposure to hyeroxia can impair pulmonary innate immunity and bacterial phagocytosis capacity (29). In a retrospective study on mechanically ventilated critically ill patients, the presence of hyperoxemia (defined as PaO₂ > 120 mmHg) at ICU-admission and the number of days of exposure to hyperoxemia were independently associated with the onset of VAP (6). Preclinical studies in animal models also supported a role of hyperoxia in the pathogenesis of VAP. Entezari et al. showed that a prolonged exposure to hyperoxia can compromise the ability of alveolar macrophages to phagocytose Pseudomonas Aeruginosa (30) and increased mortality in infected mice (31). Hyperoxia increased mortality in mice with Acinetobacter pneumonia, in which the administration of procysteine was able to improve survival by increasing the phagocytic activity of alveolar macrophages under hyperoxic conditions (32). By applying a too liberal O₂ therapy, we may be losing the opportunity to control for a significant risk factor for VAP, a complications that occurred in almost 50% of patients in our cohort.

In our study, the use of prolonged prone positioning (sessions of up to 36 consecutive hours) was also associated with an increased risk of VAP. From a theoretical point of view, prone positioning could prevent the onset of VAP by facilitating the drain of respiratory secretions and limiting ventilator-induced lung injury, although most clinical studies showed no significant impact on the actual incidence of VAP (33). Several factors may explain the association observed, including the need for prolonged administration of NMBA and deep sedation, the increase in abdominal pressure with a higher risk of aspiration of gastric content, the need to limit the dose of enteral nutrition. ICU-acquired weakness is a frequent complication in COVID-19 patients, especially in those who received higher doses of NMBA and sedatives and lower caloric/protein intake, and may be responsible for a longer duration of mechanical ventilation and prolonged exposure to the risk of VAP (34).

Our study has several limitations. First, the retrospective design that does not allow to define a cause-effect relationship between the exposure to hyperoxia and mortality or VAP, but only enables to describe associations. Second, the relatively low sample size and the involvement of a single center, which limits the generalizability of our results. The sample size limited the number of confounders that could be included in multivariate regression models, since the inclusion of too many independent variables would lead to a mathematically unstable outcome (20). Third, we based our analysis on ABG data that can only provide a partial picture of the exposure to hyperoxemia, limited to the moment in which the ABG was made. Unfortunately, we could not collect SpO₂ data, which could have been useful for a more continuative evaluation of the oxygenation status and the responses of clinicians to SpO₂ values above the target. Fourth, the calculation of the “ideal FiO2” and “excess O2” can only provide an imprecise estimate of the amount of O2 administered in excess. In fact, factors such as PEEP, prone positioning or use of iNO will determine the most appropriate FiO2 of the patient by influencing gas exchange: therefore, calculating the ideal FiO2 merely on the basis of the PaO2/FiO2 ratio may be reductive. Despite its limitations, to the best of our knowledge, this is the first study that describes the prevalence of hyperoxemia in mechanically ventilated COVID-19 patients and explores the potential effects of excess O₂ doses on outcome. Future larger studies are needed to confirm our findings.

Conclusion

In mechanically ventilated patients with SARS-CoV-2 pneumonia admitted to our ICU, the administration of O₂ was often excessive in comparison to the PaO₂ target indicated by the guidelines, and the exposure to hyperoxemia was frequent. In addition, most episodes of hyperoxemia were not followed by a reduction in the FiO₂; changes were made more frequently in the presence of higher PaO₂ or higher initial FiO₂. The prevalence of hyperoxemia was independently associated with a greater risk of ICU-mortality, as well as with a greater risk of developing VAP. The retrospective nature of our study does not allow to draw conclusions on a possible cause-effect relationship between the exposure to excessive amounts of O₂ and outcome. However, these data add to the ample literature that warns against the possible deleterious effects of a too liberal O₂ therapy. In the absence of strong evidence of the safety of hyperoxemia in critically ill patients (in particular in those with ARDS and SARS-CoV-2 pneumonia), more efforts should be made to avoid the exposure to excessive amounts of supplemental O₂. Further studies are needed to define the best oxygenation target for this patient category.

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethics statement

The studies involving human participants were reviewed and approved by Comitato Etico Regionale Marche. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

Author contributions

ED contributed to the study design, analysis, and interpretation of data and drafted the manuscript. EC, AC, EA, and AD contributed to the study design and interpretation of the data and revised the manuscript critically. GM, SV, RG, RD, and CS participated in the acquisition and analysis of the data. All authors have agreed both to be personally accountable for the author’s own contributions and to ensure that questions related to the accuracy or integrity of any part of the work, even ones in which the author was not personally involved, are appropriately investigated, resolved, and the resolution documented in the literature and approved the submitted version of the manuscript.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Acknowledgments

We wish to thank the medical and nurse staff for their help in the realization of this study and their end-less efforts during the COVID-19 pandemic.

References

1. Damiani E, Donati A, Girardis M. Oxygen in the critically ill: friend or foe? Curr Opin Anaesthesiol. (2018) 31:129–35. doi: 10.1097/ACO.0000000000000559

PubMed Abstract | CrossRef Full Text | Google Scholar

2. Chu DK, Kim LH, Young PJ, Zamiri N, Almenawer SA, Jaeschke R, et al. Mortality and morbidity in acutely ill adults treated with liberal versus conservative oxygen therapy (IOTA): a systematic review and meta-analysis. Lancet. (2018) 391:1693–705. doi: 10.1016/S0140-6736(18)30479-3

CrossRef Full Text | Google Scholar

3. Girardis M, Busani S, Damiani E, Donati A, Rinaldi L, Marudi A, 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. (2016) 316:1583–9. doi: 10.1001/jama.2016.11993

PubMed Abstract | CrossRef Full Text | Google Scholar

4. Panwar R, Hardie M, Bellomo R, Barrot L, Eastwood GM, Young PJ, et al. Conservative versus liberal oxygenation targets for mechanically ventilated patients – a pilot multicenter randomized controlled trial. Am J Resp Crit Care Med. (2016) 193:43–51. doi: 10.1164/rccm.201505-1019OC

PubMed Abstract | CrossRef Full Text | Google Scholar

5. Bailey TC, Martin EL, Zhao L, Veldhuizen RAW. High oxygen concentrations predispose mouse lungs to the deleterious effects of high stretch ventilation. J Appl Physiol. (2003) 94:975–82. doi: 10.1152/japplphysiol.00619.2002

PubMed Abstract | CrossRef Full Text | Google Scholar

6. Six S, Jaffal K, Ledoux G, Jaillette E, Wallet F, Nseir S. Hyperoxemia as a risk factor for ventilator-associated pneumonia. Crit Care. (2016) 20:195. doi: 10.1186/s13054-016-1368-4

PubMed Abstract | CrossRef Full Text | Google Scholar

7. ARDSnet. NIH NHLBI ARDS Clinical Network Mechanical Ventilation Protocol Summary (2008). Available online at: http://www.ardsnet.org/files/ventilator_protocol_2008-07.pdf (accessed March 2022).

Google Scholar

8. Rachmale S, Li G, Wilson G, Malinchoc M, Gajic O. Practice of excessive F(IO(2)) and effect on pulmonary outcomes in mechanically ventilated patients with acute lung injury. Respir Care. (2012) 57:1887–93. doi: 10.4187/respcare.01696

PubMed Abstract | CrossRef Full Text | Google Scholar

9. Madotto F, Rezoagli E, Pham T, Schmidt M, McNicholas B, Protti A, et al. Hyperoxemia and excess oxygen use in early acute respiratory distress syndrome: insights from the LUNG SAFE study. Crit Care. (2020) 24:125. doi: 10.1186/s13054-020-2826-6

PubMed Abstract | CrossRef Full Text | Google Scholar

10. Aggarwal NR, Brower RG, Hager DN, Thompson BT, Netzer G, Shanholtz G, 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. (2018) 46:517–24. doi: 10.1097/CCM.0000000000002886

PubMed Abstract | CrossRef Full Text | Google Scholar

11. Barrot L, Asfar P, Mauny F, Winiszewski H, Montini F, Badie J, et al. Liberal or conservative oxygen therapy for acute respiratory distress syndrome. N Engl J Med. (2020) 382:999–1008. doi: 10.1056/NEJMoa1916431

PubMed Abstract | CrossRef Full Text | Google Scholar

12. Grasselli G, Zangrillo A, Zanella A, Antonelli M, Cabrini L, Castelli A, et al. Baseline characteristics and outcomes of 1591 patients infected with SARS-CoV-2 admitted to ICUs of the Lombardy region, Italy. JAMA. (2020) 323:1574–81. doi: 10.1001/jama.2020.5394

PubMed Abstract | CrossRef Full Text | Google Scholar

13. Ziehr DR, Alladina J, Petri CR, Maley JH, Moskowitz A, Medoff BD, et al. Respiratory pathophysiology of mechanically ventilated patients with COVID-19: a cohort study. Am J Respir Crit Care Med. (2020) 201:1560–4. doi: 10.1164/rccm.202004-1163LE

PubMed Abstract | CrossRef Full Text | Google Scholar

14. Fan E, Brodie D, Slutsky AS. Acute respiratory distress syndrome: advances in diagnosis and treatment. JAMA. (2018) 319:698–710. doi: 10.1001/jama.2017.21907

PubMed Abstract | CrossRef Full Text | Google Scholar

15. Nasa P, Azoulay E, Khanna AK, Jain R, Gupta S, Javeri Y, et al. Expert consensus statements for the management of COVID-19-related acute respiratory failure using a Delphi method. Crit Care. (2021) 25:106. doi: 10.1186/s13054-021-03491-y

PubMed Abstract | CrossRef Full Text | Google Scholar

16. Carsetti A, Damia Paciarini A, Marini B, Pantanetti S, Adrario E, Donati A. Prolonged prone position ventilation for SARS-CoV-2 patients is feasible and effective. Crit Care. (2020) 24:225. doi: 10.1186/s13054-020-02956-w

PubMed Abstract | CrossRef Full Text | Google Scholar

17. Elso Guidelines for Cardiopulmonary Extracorporeal Life Support. Extracorporeal Life Support Organization, Version 1.4 August 2017 (2017). Available online at: www.elso.org (accessed March 2022).

Google Scholar

18. Kalanuria AA, Zai W, Mirski M. Ventilator-associated pneumonia in the ICU. Crit Care. (2014) 18:208. doi: 10.1186/cc13775

PubMed Abstract | CrossRef Full Text | Google Scholar

19. American Thoracic Society, Infectious Diseases Society of America. Guidelines for the management of adults with hospital-acquired, ventilator-associated, and healthcare-associated pneumonia. Am J Respir Crit Care Med. (2005) 171:388–416. doi: 10.1164/rccm.200405-644ST

PubMed Abstract | CrossRef Full Text | Google Scholar

20. Stoltzfus JC. Logistic regression: a brief primer. Acad Emerg Med. (2011) 18:1099–104. doi: 10.1111/j.1553-2712.2011.01185.x

PubMed Abstract | CrossRef Full Text | Google Scholar

21. De Graaf AE, Dongelmans DA, Binnekade JM, de Jonge E. Clinicians’ response to hyperoxia in ventilated patients in a Dutch ICU depends on the level of FiO2. Intensive Care Med. (2011) 37:46–51. doi: 10.1007/s00134-010-2025-z

PubMed Abstract | CrossRef Full Text | Google Scholar

22. Suzuki S, Eastwood GM, Peck L, Glassford NJ, Bellomo R. Current oxygen management in mechanically ventilated patients: a prospective observational cohort study. J Crit Care. (2013) 28:647–54. doi: 10.1016/j.jcrc.2013.03.010

PubMed Abstract | CrossRef Full Text | Google Scholar

23. Gomes EP, Reboredo MM, Costa GB, Barros FS, Carvalho EV, Pinheiro BV. Impacts of a fraction of inspired oxygen adjustment protocol in covid-19 patients under mechanical ventilation: a prospective cohort study. Med Intensiva. (2022). doi: 10.1016/j.medin.2022.04.004

PubMed Abstract | CrossRef Full Text | Google Scholar

24. Helmerhorst HJF, Schultz MJ, van der Voort PHJ, Bosman RJ, Juffermans RP, de Jonge E, et al. Self-reported attitudes versus actual practice of oxygen therapy by ICU physicians and nurses. Ann Intensive Care. (2014) 4:23. doi: 10.1186/s13613-014-0023-y

PubMed Abstract | CrossRef Full Text | Google Scholar

25. Davis WB, Rennard SI, Bitterman PB, Crystal RG. Pulmonary oxygen toxicity. Early reversible changes in human alveolar structures induced by hyperoxia. N Engl J Med. (1983) 309:878–83. doi: 10.1056/NEJM198310133091502

PubMed Abstract | CrossRef Full Text | Google Scholar

26. Zenri H, Rodriquez-Capote K, McCaig L, Yao LJ, Brackenbury A, Possmayer F, et al. Hyperoxia exposure impairs surfactant function and metabolism. Crit Care Med. (2004) 32:1155–60.

Google Scholar

27. Barbateskovic M, Schjørring OL, Russo Krauss S, Jakobsen JC, Meyhoff CS, Dahl RM, et al. Higher versus lower fraction of inspired oxygen or targets of arterial oxygenation for adults admitted to the intensive care unit. Cochrane Database Syst Rev. (2019) 2019:CD012631.

Google Scholar

28. Schjørring OL, Klitgaard TL, Perner A, Wetterslev J, Lange T, Siegemund M, et al. Lower or higher oxygenation targets for acute hypoxemic respiratory failure. N Engl J Med. (2021) 384:1301–11.

Google Scholar

29. Nußbaum B, Radermacher P, Asfar P, Hartmann C. Does hyperoxia enhance susceptibility to secondary pulmonary infection in the ICU? Crit Care. (2016) 20:239.

Google Scholar

30. Entezari M, Weiss DJ, Sitapara R, Whittaker L, Wargo MJ, Li J, et al. Inhibition of high-mobility group box 1 protein (HMGB1) enhances bacterial clearance and protects against Pseudomonas aeruginosa pneumonia in cystic fibrosis. Mol Med. (2012) 18:477–85. doi: 10.2119/molmed.2012.00024

PubMed Abstract | CrossRef Full Text | Google Scholar

31. Patel VS, Sitapara RA, Gore A, Phan B, Sharma L, Sampat V, et al. High mobility group box-1 mediates hyperoxia-induced impairment of Pseudomonas aeruginosa clearance and inflammatory lung injury in mice. Am J Respir Cell Mol Biol. (2013) 48:280–7. doi: 10.1165/rcmb.2012-0279OC

PubMed Abstract | CrossRef Full Text | Google Scholar

32. Saito K, Kimura S, Saga T, Misonoo Y, Yoshizawa S, Akasaka Y, et al. Protective effect of procysteine on Acinetobacter pneumonia in hyperoxic conditions. J Antimicrob Chemother. (2013) 68:2305–10. doi: 10.1093/jac/dkt192

PubMed Abstract | CrossRef Full Text | Google Scholar

33. Alexiou VG, Ierodiakonou V, Dimopoulos G, Falagas ME. Impact of patient position on the incidence of ventilator-associated pneumonia: a meta-analysis of randomized controlled trials. J Crit Care. (2009) 24:515–22. doi: 10.1016/j.jcrc.2008.09.003

PubMed Abstract | CrossRef Full Text | Google Scholar

34. Schweickert WD, Hall J. ICU-acquired weakness. Chest. (2007) 131:1541–9. doi: 10.1378/chest.06-2065

PubMed Abstract | CrossRef Full Text | Google Scholar