HYPEROXIA AND RESPIRATORY DYSFUNCTION AT ICU ADMISSION ARE ASSOCIATED WITH POOR OUTCOMES IN MECHANICALLY VENTILATED PATIENTS

Objective: to identify the association of hyperoxia at the time of Intensive Care Unit (ICU) admission with clinical outcomes in adult patients on invasive mechanical ventilation and with respiratory dysfunction defined by a PaO2/ FIO2 ratio (P/F) below 300. Methods : retrospective cohort observational study with data from adult patients admitted to a general ICU, with 8 beds from a university hospital. Hyperoxia was defined as PaO2 > 120mmHg and patients were classified in 4 subgroups: 1.hyperoxia and P/F > 300, 2. hyperoxia and P/F ≤ 300, 3.no hyperoxia and P/F > 300, and 4. no hyperoxia and P/F ≤ 300. Results: a total of 129 patients were included. Hyperoxia was present in one third (43, 33.3%) of all patients. It was more frequent in patients without respiratory dysfunction (P/F ratio > 300, 30 of 54 individuals 55.6%) in comparison to those with respiratory dysfunction (P/F ratio 13 of 88, 14.7%), p=0,044. The ICU mortality was not different among the subgroups; however, the ICU length of stay was greater in the subgroup of patients with hyperoxia and P/F < 300. Conclusion : at ICU admission, hyperoxia was more frequent in mechanically ventilated patients without respiratory dysfunction and associated with greater ICU length of stay in those with worse P/F ratio.


INTRODUCTION
Tissue hypoxia results in a frequent condition that motivates Intensive Care Unit (ICU) admission, causing impairment of aerobic metabolism, lactic acid generation, and cellular and organ dysfunction 1 . In contrast, hyperoxia, defined by some authors as an arterial oxygen pressure (PaO2) higher than 120mmHg 2,3 , has been associated with a bactericidal effect and a decrease in the incidence of operative wound infections 4 . However, the elevation of PaO2 to supraphysiological levels causes greater exposure of lung tissues to Reactive Oxygen Species (ROS), with a consequent increase in pulmonary inflammation, atelectasis, alveolar edema, and adult respiratory distress syndrome 5 .
The lung is the first organ involved by the excessive use of O2. Pulmonary capillary endothelial and alveolar epithelial cells are targets for ROS, resulting in injury-induced lung edema, alveolar flooding, hemorrhage, and collagen, elastin, and hyaline membrane deposits 5,6 . Oxygen toxicity caused by ROS progresses in overlapping phases based on the degree of severity and reversibility of injury. However, there are studies that show that even with the improvement of pulmonary exchange, there is a tendency to maintain a high fraction of inspired oxygen (FIO2) in patients undergoing mechanical ventilation (MV) 7,8 .
Increased oxidative stress can also have a systemic effect. In an animal model of sepsis, the exposure to hyperoxia was associated with a higher increase in serum ROS and inflammatory cytokines, greater spread of infection, and multiple organ dysfunction. The impairment in microvascular perfusion can induce a paradoxical reduction in regional O2 delivery, although other reports have suggested a beneficial role of hyperoxia in hemodynamic stabilization and redistribution of blood flow to splanchnic organs in animal shock models 9 .
Hyperoxia may be present in 1 in 4 critically-ill mechanically ventilated patients. In fact, it is particularly frequent in patients requiring FIO2 less than 50%, as intensivists usually accept higher levels of SpO2 and PaO2 in patients requiring lower FIO2 during invasive MV3. In mechanically ventilated ICU patients, hyperoxia has been associated with a higher mortality rate, a decrease of ventilator free days, an increased rate of nosocomial infections, and worsening of organ dysfunction 10. The frontier between the therapeutic and deleterious effects of high FIO2 use seems slight. Moreover, the consequences of supra-normal PaO2 exposure in critically ill patients are still unclear 11 .
We hypothesize that hyperoxia is frequent in mechanically ventilated patients admitted to the ICU, and it is associated with poor clinical outcomes. This study aims to evaluate the association between PaO2 > 120mmHg and the presence of respiratory dysfunction at the time of ICU admission with clinical outcomes in patients needing invasive MV.

METHODS
This is a retrospective observational cohort study.

Population and design
Data from all patients admitted to the ICU for one year, from March 2017 to March 2018, were evaluated. Inclusion criteria were admission to a clinical general ICU with 8 beds in a university hospital. Patients aged older than 18 years receiving invasive MV at the time of ICU admission were included. Patients under noninvasive ventilation (NIV), with an ICU length of stay less than 3 days or under palliative care were excluded from the study.

Data collection and variables
The data used for the analysis were recorded in a database routinely used for documentation and patient care from the day of admission to the ICU, and throughout the stay in the unit.
Demographic variables (age and sex), Acute Physiology and Chronic Health Evaluation II severity score (APACHE II) 12 , and Sequential Organ Failure Assessment score (SOFA) 13 at ICU admission were collected from the database.
Arterial blood gases were collected in the first 24 hours after ICU admission. The FIO2 at the time arterial blood gases analysis was recorded and the PaO2/ FIO2 ratio (P/F) was calculated.
We assessed the frequency of hyperoxia according to the presence or absence of respiratory dysfunction and the following clinical outcomes: new organic dysfunctions (neurological, cardiologic, hepatic, renal, and hematologic) on the third day of hospitalization in the ICU, need for tracheostomy, MV need at the time of discharge, ICUs length of stay, and mortality.

Statistical Analysis
Data were subjected to descriptive statistical analyses of univariate nature for continuous variables (measures of central tendency and dispersion) and categorical (relative frequencies). For bivariate analyses, Chi-square test and Fisher's exact test were used for the analysis of categorical variables, while ANOVA and Kruskal-Wallis test were used to compare continuous variables with normal and non-normal distribution, respectively. A generalized linear model was performed to assess whether the relationship between ICU length of stay and hyperoxia was independent of patient severity by the Wald chi-squared test. The p-value ≤ 0.05 was adopted as an index of significance

Ethics
The study was approved by the institutional Research Ethics Committee (CAAE: 64529817.1.0000.5045). Because it is a retrospective study with the collection of data recorded in a unit database, the completion of the Informed Consent Form was waived. The study followed the 466/12 resolution of the National Health Council and the Helsinki Declaration.

RESULTS
A total of 186 patients were admitted to the ICU during the study period, of which 132 required the use of invasive MV. A total of 3 patients were excluded because they had a length of ICU stay of less than 12 hours. The remaining 129 patients were divided into the 4 subgroups according to the occurrence of hyperoxia and respiratory dysfunction (Figure 1). Hyperoxia was present in one third (43, 33.3%) of all patients. It was more frequent in patients without respiratory dysfunction (30 of 54, 55.6%) in comparison to those with lower P/F ratio (13 of 88, 14.7%), p=0,044.

Vasconcelos et al
The mean age of the patients was 57.8 ± 17.3 years-old and the majority was female (60.5%). At admission to the ICU, the mean APACHE II score was 20.4 ± 6.6 points and the SOFA score was 8.2 ± 4.3 points.
There was no statistical difference between the subgroups concerning age, sex, APACHE II score, and SOFA score. Regarding outcomes, ICU mortality was similar between the subgroups, however, patients with hyperoxia and respiratory dysfunction at ICU admission remained longer in the unit (Table 1 and Figure 2

DISCUSSION
The main findings of the present investigation were the following: unplanned exposure to hyperoxia was more frequent in patients admitted with respiratory dysfunction and is associated with increased ICU length of stay in patients who had both hyperoxia and P/F ratio below 300. Hyperoxia was related to an increase in ICU length of stay, regardless of the severity of the patients, calculated by the APA-CHE II score.
An elevation of oxygen partial pressure increases the amount of ROS, generates an inflammatory component, and causes extensive cellular damage [18][19][20] . Previous studies have shown that an FIO2 > 0.6 for 48 hours or more seems to induce lung injury mechanical with an inflammatory component similar to that found in Adult Respiratory Distress Syndrome, increasing morbidity and mortality, hospitalization time, and costs [21][22][23][24] .
Some authors failed to relate hyperoxia to in-hospital mortality. However, these results were interpreted as influenced by the method applied in the investigation 25,26 . Meanwhile, other studies have found that longer exposures to hyperoxia is associated with an increased risk for unfavorable outcomes, such as on mortality and subsequent organ dysfunction 27,28 .
In 2016, the World Health Organization (WHO) strongly recommended the use of a high FIO2 in adult patients undergoing general anesthesia to reduce the risk of surgical site infection 4 . Although this exposure to hyperoxia for a short period may be beneficial in surgical patients 27,28 , there is no evidence to support the extension of this recommendation to critically ill patients, especially those with pulmonary compromise.
Previous studies have shown an association between increased mortality and hyperoxia in patients with traumatic brain injury, cardiac arrest, and cerebrovascular events 11,29 . In this study, the association between ICU mortality and hyperoxia at admission was not identified. One possible explanation is the heterogeneity of diseases and the low number of patients evaluated.
In the same way as this study, others authors have identified the relationship between hyperoxia and organic dysfunctions, including respiratory dysfunction 24,25 . However, these are observational studies and point to the need for randomized clinical trials.
This study has limitations. It was carried out in a single-center university and tertiary hospital, a reference unit for the regional public health system. It assists patients with complex diseases with multiple comorbidities. In addition, as this was a retrospective study, other data on MV, which could further clarify the size of the impact of hyperoxia in the prognosis of patients, were not analyzed.
Despite these limitations, the present investigation was able to show the more frequency of hyperoxia in patients without respiratory dysfunction and its association with an increase in the length of ICU stay and worsening of gas exchange in patients admitted with respiratory dysfunction assessed by the P/F ratio. This data is of clinical relevance, since it is an alert to the benefit of greater vigilance to avoid hyperoxia in all patients, but with more attention to those admitted with respiratory dysfunction. Future clinical trials should address strategies to decrease the prevalence of hyperoxia in mechanically ventilated patients with respiratory compromise and to assess its impacts on relevant clinical outcomes.
In conclusion, inadvertent hyperoxia at ICU admission is more frequent in patients without respiratory dysfunction and associated with greater ICU length of stay regardless of the clinical severity of the patient. This study warns about the need for rational use of supplemental oxygen in critically ill patients. It also points to the need for multicenter, prospective, and randomized studies to evaluate the benefits of careful titration of FIO2 to avoid hyperoxia during invasive ventilatory support.