You are here

Article Text

Article menu
Critical care
Hyperoxia for sepsis and development of acute lung injury with increased mortality
  1. Ryo Yamamoto1,
  2. Seitaro Fujishima2,
  3. Kazuma Yamakawa3,
  4. Toshikazu Abe4,5,
  5. Hiroshi Ogura6,
  6. Daizoh Saitoh7,
  7. Satoshi Gando8,9 and
  8. Junichi Sasaki1
  1. 1Department of Emergency and Critical Care Medicine, Keio University School of Medicine, Tokyo, Japan
  2. 2Center for Preventive Medicine, Keio University Hospital, Tokyo, Japan
  3. 3Division of Trauma and Surgical Critical Care, Osaka General Medical Center, Osaka, Japan
  4. 4Department of Health Services Research, Faculty of Medicine, University of Tsukuba, Ibaraki, Japan
  5. 5Health Services Research and Development Center, University of Tsukuba, Tsukuba, Japan
  6. 6Department of Traumatology and Acute Critical Medicine, Osaka University Graduate School of Medicine, Osaka, Japan
  7. 7Division of Traumatology, Research Institute, National Defense Medical College, Tokorozawa, Japan
  8. 8Division of Acute and Critical Care Medicine, Department of Anesthesiology and Critical Care Medicine, Hokkaido University Faculty of Medicine, Sapporo, Japan
  9. 9Department of Acute and Critical Care Medicine, Sapporo Higashi Tokushukai Hospital, Sapporo, Japan
  1. Correspondence to Dr Ryo Yamamoto; ryo.yamamoto{at}gmail.com

Abstract

Background Supraphysiological oxygen administration causes unfavourable clinical outcomes in various diseases. This study aimed to determine whether hyperoxia would be associated with increased mortality in patients with severe infection.

Methods A post-hoc analysis of a nationwide multicentre prospective observational study on sepsis (SPICE Study) was conducted, including adult patients admitted to the intensive care unit with available arterial partial pressure of oxygen (PaO2) at the treatment initiation for severe infection. Hyperoxia was defined as a PaO2 level of ≥300 mm Hg and in-hospital mortality was compared between patients with and without hyperoxia.

Results Of the 563 patients eligible for the study, 49 had hyperoxia at treatment initiation for severe infection. The in-hospital all-cause mortality rates of patients with and without hyperoxia were 14 (29.2%) and 90 (17.6%), respectively. Inverse probability weighting analyses with propensity scores revealed the association between hyperoxia and increased in-hospital mortality rate (28.8% vs 18.8%; adjusted OR 1.75 (1.03 to 2.97); p=0.038), adjusting for patient demographics, comorbidities, site of infection, severity of infection, haemodynamic and respiratory status, laboratory data and location of patient at infection development. Acute lung injury developed more frequently in patients with hyperoxia on the following days after infection treatment, whereas sepsis-related mortality was comparable regardless of hyperoxia exposure.

Conclusion Hyperoxia with PaO2 ≥300 mm Hg at treatment initiation of severe infection was associated with an increased in-hospital mortality rate in patients requiring intensive care. The amount of oxygen to administer to patients with severe infection should be carefully determined.

Trial registration number University Hospital Medical Information Network Clinical Trial Registry (UMIN000027452).

  • Critical Care
  • Oxidative Stress
  • ARDS

Data availability statement

Data are available upon reasonable request. The study data are available from the SPICE-ICU Study group; however, restrictions apply to the availability of these data, which were used under licence for the current study and are not publicly available. However, data are available from the authors upon reasonable request and with the permission of the SPICE-ICU Study group.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/bmjresp-2023-001968

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • Hyperoxia exposure is associated with adverse clinical events in various diseases.

    WHAT THIS STUDY ADDS

    • A nationwide retrospective observational study on 563 patients with severe infection requiring intensive care showed the association between hyperoxia (arterial partial pressure of oxygen (PaO2) ≥300 mm Hg) at treatment initiation and higher in-hospital mortality, as well as higher incidence of acute lung injury, compared with no hyperoxia.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • Restriction of oxygen administration to avoid hyperoxia during resuscitation should be carefully considered in daily practice, and the appropriate PaO2 for patients with sepsis should be validated in future studies.

    Introduction

    Oxygen administration is a fundamental treatment for patients with critical illnesses.1 2 However, a supraphysiological amount of oxygen in the blood and/or tissue has been related to unfavourable clinical outcomes in various diseases, including postcardiac arrest syndrome, accidental hypothermia, traumatic brain injury and postcardiac surgery.3–6 A recent randomised controlled trial (RCT) on patients receiving mechanical ventilation showed that oxygen titration targeting at 70–100 mm Hg of arterial partial pressure of oxygen (PaO2) reduced the mortality rate in the intensive care unit (ICU).7

    The pathophysiology behind the harmful effects of hyperoxia has been investigated, and oxygen toxicity in the brain and pulmonary tissues affects critically ill patients.8–11 Cerebral vasoconstriction and mitochondrial dysfunction were reportedly due to hyperoxia in the injured brain, which paradoxically decreases oxygen delivery to the cerebral tissues.8 10 Additionally, pulmonary vasoconstriction with alveolar damages by unnecessary reactive oxygen species was observed, particularly in patients experiencing hyperoxia under mechanical ventilation.9 11 Furthermore, hyperoxia-induced acute lung injury (ALI) was another adverse effect caused by excessive oxygen, which leads to unfavourable clinical outcomes.9 11

    Under critical illness such as sepsis, impaired oxygen utilisation at the intracellular mitochondria and reactive oxygen species production were observed.12–14 Although such toxic oxygen derivatives were considered to lead to several organ dysfunctions, appropriate tissue oxygen tension during such cytopathic hypoxia remains unclear. Subgroup analyses of RCTs on conservative oxygen usage during mechanical ventilation reported conflicting results on the harmful effects of hyperoxia among patients with sepsis.15 16 Moreover, several post-hoc analyses targeting on the association between PaO2 and mortality due to sepsis also showed various effects of hyperoxia from benefits to harms.17 18

    Accordingly, to eventually elucidate an optimal PaO2 target for patients with sepsis, this study conducted a post-hoc analysis on a multicentre prospective observational study on ICU patients with severe infections. We aimed to determine whether hyperoxia would be associated with unfavourable clinical outcomes in these patients. We hypothesise that hyperoxia at treatment initiation for severe infection in the ICU was associated with increased in-hospital mortality rate post-ICU admission.

    Methods

    Study design

    This is a post-hoc analysis of a nationwide multicentre prospective observational study that was conducted by the Japanese Association for Acute Medicine Sepsis Prognostication in Intensive Care Unit and Emergency Room (SPICE-ICU) from December 2017 to May 2018.19 The SPICE-ICU Study included patients with newly developed infectious diseases requiring ICU admission at 22 participating tertiary care centres and was registered at the University Hospital Medical Information Network Clinical Trial Registry on 22 May 2017 (UMIN-CTR ID: UMIN000027452) prior to study initiation. Patients or the public were not involved in the design, or conduct, or reporting, or dissemination plans of our research.

    The SPICE-ICU Study enrolled consecutive patients aged ≥16 years, with newly developed severe infectious diseases requiring intensive care with the aim to describe the characteristics of patients fulfilling the existing sepsis definitions (eg, sepsis-3 and sepsis-2).20 21 Although the SPICE-ICU Study screened all patients admitted to or stayed in the ICU with newly developed suspected infections rather than only those who met any existing criteria for sepsis, 619 of 652 (94.9%) met either sepsis-3 or severe sepsis at sepsis-2 criteria. Suspected infection at screening was defined as receiving newly initiated antibiotics with obtainment of culture specimen or imaging tests for infection identification, and data retrieval was initiated immediately after the screening. At screening, patients could be located at the emergency department (ED), regular ward or ICU, but those transferred from other hospitals were not included. Patient inclusion was confirmed when any infectious diseases were finally diagnosed regardless of presence of bacteraemia or organs infected.

    Study subjects

    Data from the SPICE-ICU Study (2017–2018) were reviewed retrospectively. Patients with severe infections were included if they (1) were ≥16 years old, (2) had any infectious diseases, (3) were admitted to the ICU and (4) had available PaO2 data obtained at the time of study inclusion (treatment initiation for severe infection). Patients diagnosed with acute respiratory distress syndrome (ARDS) during ICU admission were excluded because previous studies have already suggested that hyperoxia can be harmful and should be avoided in patients with ARDS.22

    Methods

    Patient data for the SPICE-ICU Study were prospectively collected and entered into an online data collection portal at each hospital. Age, sex, body mass index, Charlson Comorbidity Index, Clinical Frailty Scale (CFS), infection site, place of patient before ICU admission, vital signs on ICU admission, laboratory investigations and arterial blood gas assay with fraction of inspired oxygen (FiO2) obtained at the diagnosis of severe infection (treatment initiation) and on following days until day 4, presence of bacteraemia, Sequential Organ Failure Assessment (SOFA) score during study inclusion and mechanical ventilation use were all recorded. The timing of diagnosis of severe infection was clinically determined by a treating physician. Additionally, data on length of ICU and hospital stay, duration of ventilator use, survival status at discharge, sepsis-related death and new diagnosis of ALI after ICU admission were obtained.

    Severe sepsis at sepsis-2 was defined as having a suspected infection site, ≥2 systemic inflammatory response syndrome criteria and ≥1 organ dysfunction. Sepsis-3 was defined as having a suspected infection site and organ dysfunction (an acute change in the total SOFA score of ≥2 points consequent to the infection). Septic shock was defined according to the sepsis-3 or sepsis-2 definitions. High frailty was defined as CFS ≥5.23 Suspected infection source was recorded during treatment initiation for severe infection and the diagnosis of the infection site was recorded at discharge. Based on previous literature on hyperoxia in other diseases, hyperoxia was defined as PaO2 of ≥300 mm Hg.2 4 24 Hyperoxia during treatment initiation for severe infection was defined as hyperoxia at ICU admission. The database lacked detailed indications for high amount of oxygen administration and haemodynamic status before, during and after oxygen administration.

    The primary outcome was in-hospital all-cause mortality after inclusion. Secondary outcomes included sepsis-related mortality (failure to resuscitation from sepsis), ICU-free, hospital-free and ventilator-free days to 28 days after inclusion, and incidence of newly developed ALI after inclusion.

    Analysis

    Inverse probability weighting (IPW) using propensity scores was conducted to adjust for background characteristics between patients with and without hyperoxia.25 26 The propensity score for weighting was developed using a logistic regression model to estimate the probability of hyperoxia exposure. Based on previous studies, relevant covariates were carefully selected from known or potential predictors for receiving supraphysiological amounts of oxygen and predicting clinical outcomes in patients with sepsis or severe infections.19 21 27–29 These covariates included age, sex, Charlson Comorbidity Index, high frailty, patient location during infection development (ED, ward or ICU), and vital signs (Glasgow Coma Scale (GCS), systolic blood pressure (SBP) and respiratory rate), lactate and invasive mechanical ventilation use during treatment initiation for severe infection. SOFA score and PaO2/FiO2 ratio (PF ratio) were also included as covariates because they are considered survival predictors in sepsis and would be related to the amount of oxygen administered.25 29 Patients with missing covariates were excluded from the propensity score calculation. The discrimination ability of the propensity score was evaluated using the c-statistic.26 The weight was calculated as the inverse of the propensity score of hyperoxia exposure. To avoid extreme weight, patients with a propensity score of ≤0.05 or ≥0.95 were excluded from the IPW analyses. The primary outcome was compared using the Χ2 test, and secondary outcomes were compared using ORs or Hodges-Lehmann estimator.26

    Four sensitivity analyses were performed to validate the primary results. First, to confirm that results were not dependent on the propensity score calculation, generalised estimating equation analysis with the logit link function was used to adjust for patient backgrounds and differences in pre-sepsis conditions among locations at infection development.30 Second, multivariate logistic regression was conducted with covariates selected from those for propensity score calculation. Third, IPW was conducted with further restriction on the propensity score to minimise the effects of overweighting.25 26 Finally, the primary analysis was repeated only on patients meeting either the sepsis-3 or severe sepsis at sepsis-2 criteria to validate that the results were consistent across pre-existing criteria.

    Additionally, restricted cubic spline curves for estimating in-hospital mortality by PaO2 at treatment initiation for severe infection in the ICU were drawn, in which a generalised additive model was adopted using the same covariates for propensity score calculation. Then, any PaO2 thresholds affecting the clinical outcomes of patients with sepsis were explored.

    Subgroup analyses were performed to analyse the relationships between hyperoxia, clinical characteristics and in-hospital mortality. Targeted subgroups were selected based on previous research on the clinical outcomes of patients with sepsis. The IPW analyses on the primary outcome were repeated in patient subgroups stratified by age (<65 vs ≥65 years); presence of chronic cardiopulmonary diseases including congestive heart failure, coronary artery disease and chronic lung disease; confirmed infection site (pulmonary vs extrapulmonary) and presence of septic shock. Subgroup analyses were also conducted on patients without hypoxia, defined as having PaO2 of <60 mm Hg.

    Descriptive statistics are presented as a median (IQR) or a number (percentage). The results were presented with standardised difference and 95% CI. The balance of covariates before and after weighting was evaluated with a standardised difference, with <0.1 considered insignificant.25 The hypothesis was tested only on the primary outcome, with a two-sided alpha threshold of 0.05 considered significant. All statistical analyses were conducted using the IBM SPSS for Windows V.29.0 (IBM Corp) and R V.4.0.2 (R Foundation for Statistical Computing, Vienna, Austria).

    Results

    Patient characteristics

    Of the 652 patients with severe infection in the SPICE-ICU Study, 563 had available PaO2 data at treatment initiation for severe infection and were not diagnosed with ARDS at infection development; therefore, they were eligible for this study (online supplemental figure 1).

    Altogether, 49 patients (8.7%) had hyperoxia with PaO2 of ≥300 mm Hg at treatment initiation for severe infection in the ICU. The patient characteristics are shown in table 1. The median PaO2 levels were 356 and 95 mm Hg in patients with and without hyperoxia, respectively. Patients with hyperoxia were older and had higher FiO2, PF ratio and SOFA score at treatment initiation for severe infection, as well as lower GCS and SBP at infection development than those without hyperoxia. Furthermore, a higher proportion of patients with hyperoxia had gastrointestinal/hepatobiliary/pancreatic infection, developed septic shock and received invasive mechanical ventilation.

    View this table:
    Table 1

    Characteristics of patients with sepsis

    A propensity model for hyperoxia exposure was developed, and a discrimination power was calculated, yielding a c-statistic of 0.813 (0.743–0.883). No patient with hyperoxia and 12 patients without hyperoxia were excluded from the IPW analyses due to missing covariates for propensity score calculation. Table 1 shows the patient characteristics after IPW with standardised differences, where differences in covariates such as patient demographics, comorbidities, severity of sepsis, respiratory status, vital signs and laboratory data at treatment initiation for severe infection were successfully attenuated using the propensity score (standardised difference <0.1).

    PaO2 and FiO2 were higher in patients with hyperoxia than in those without on day 1 (356 (322–441) vs 95 (74–140) mm Hg and 1.0 (0.9–1.0) vs 0.4 (0.3–0.6), respectively), but they had comparable data on the following days. PF ratio was also similar between patients with and without hyperoxia (figure 1).

    Figure 1
    Figure 1

    Daily arterial partial pressure of oxygen (PaO2) and fraction of inspired oxygen (FiO2). Daily PaO2 (A), FiO2 (B) and PF ratio (C) are shown with median and IQR. PaO2 and FiO2 were higher in patients with hyperoxia than in those without on day 1 (356 (322–441) vs 95 (74–140) mm Hg and 1.0 (0.9–1.0) vs 0.4 (0.3–0.6), respectively), but they had comparable data on the following days. PF ratio was also similar between patients with and without hyperoxia. PF ratio, PaO2/FiO2 ratio.

    In-hospital mortality and secondary outcomes

    The in-hospital all-cause mortality rates of patients with and without hyperoxia were 14 (29.2%) and 90 (17.6%), respectively, and the IPW analyses revealed the association between hyperoxia and increased in-hospital mortality (28.8% vs 18.8%; adjusted OR 1.75 (1.03 to 2.97); p=0.038; table 2). The four sensitivity analyses also showed a relationship between hyperoxia and increased in-hospital mortality (online supplemental table 1).

    View this table:
    Table 2

    Hyperoxia and clinical outcomes

    Furthermore, the restricted cubic spline curve of mortality prediction by PaO2 was shown as a convex downward curve of mortality odds, with the PaO2 levels of approximately 60–240 mm Hg indicating a lower in-hospital mortality risk in patients with severe infections (figure 2).

    The secondary outcomes are summarised in table 2. Hyperoxia was associated with more frequent ALI developments on days 2 and 3 after treatment initiation for severe infection (figure 3), whereas the incidence of sepsis-related mortality was comparable between patients with and without hyperoxia. Additionally, ICU-free and ventilator-free days were relatively shorter in patients with hyperoxia, with the median differences between them being not significant. Hospital-free days were similar between patients with and without hyperoxia.

    Figure 2
    Figure 2

    Restricted cubic spline curve of mortality prediction by arterial PaO2. The restricted cubic spline curve of mortality prediction by PaO2 revealed a convex downward curve of mortality odds as PaO2 increased, with PaO2 levels of approximately 60–240 mm Hg indicating a lower in-hospital mortality risk among patients with severe infection. PaO2, arterial partial pressure of oxygen.

    Figure 3
    Figure 3

    Daily risks for development of ALI. Daily risks of ALI development are shown with ORs and 95% CI. Hyperoxia was associated with more frequent ALI developments on days 2 and 3 after treatment initiation for severe infection. ALI, acute lung injury.

    Subgroup analysis

    In the subgroup analyses (table 3), a relationship between higher in-hospital mortality and hyperoxia was observed in several subgroups, including the elderly aged >65 years and patients without chronic cardiopulmonary diseases, pulmonary infection and septic shock (OR 2.25 (1.25 to 4.04), 2.49 (1.35 to 4.60), 2.24 (1.26 to 3.99) and 1.96 (1.05 to 3.68), respectively).

    View this table:
    Table 3

    28-day mortality in the subgroup analyses

    Contrarily, younger patients (<65 years) and those with chronic cardiopulmonary diseases, pulmonary infection or septic shock had comparable mortality regardless of hyperoxia exposure.

    Furthermore, in the subgroup excluding patients with hypoxia (PaO2<60 mm Hg), hyperoxia was also associated with a higher in-hospital mortality rate (OR 1.75 (1.03 to 2.97)).

    Discussion

    This study showed that the exposure to hyperoxia at PaO2≥300 mm Hg at treatment initiation for severe infection was associated with an increased in-hospital mortality risk among patients requiring intensive care. The results remained after adjustment for background characteristics, sepsis severity, respiratory function and mechanical ventilation use, and were also validated through several sensitivity analyses.

    One of the pathophysiological mechanisms underlying the harmful effects of hyperoxia on severe infection is lung injury developed due to supraphysiological FiO2. Hyperoxia-induced ALI should be considered when the FiO2 exceeds 0.6–0.7 and becomes clinically manifested when it is >0.8.31 32 In our study, the median FiO2 at treatment initiation for severe infection was 1.0 and 0.4 in patients with and without hyperoxia, respectively. Although FiO2 was decreased to 0.3–0.4 on the next day and the duration of inhaling higher PaO2 was unclear, incidences of ALI in patients with hyperoxia were higher on following days after hyperoxia exposure. Notably, patients with hyperoxia had relatively longer ventilator use and ICU stay, although the median differences were not significant.

    Another possible rationale for the unfavourable outcomes due to hyperoxia is synergistic tissue toxicity caused by systemic inflammation and hyperoxia.9 Several studies on traumatic brain injury reported that supranormal oxygen suppressed cell metabolism, resulting in neuronal death.33 Additionally, hyperoxia in the early resuscitation phase was related to increased 28-day mortality risk and/or length of ICU stay among patients with severe injury or accidental hypothermia,6 34 suggesting that redundant oxygen by hyperoxia may affect tissue or cell metabolism particularly in patients with systemic inflammation, which becomes problematic as organ dysfunction can occur days later rather than during resuscitation. Of note, the incidence of sepsis-related mortality (failure to resuscitation from sepsis) was comparable regardless of hyperoxia exposure in this study.

    Subgroup analyses suggested that hyperoxia should be avoided particularly in the elderly and patients without chronic cardiopulmonary diseases, pulmonary infections and septic shock. Considering that the elderly are vulnerable to suboptimal tissue/organ oxygenation,35 the adverse effects of hyperoxia would have emerged in such populations. Contrarily, as patients with cardiopulmonary diseases, pulmonary infections or septic shock would have high baseline risks for the development of pulmonary dysfunction, additional harms due to hyperoxia would not become clinically obvious. Safe oxygen ranges in sepsis should be further examined in these populations because the results need to be carefully interpreted because the sample size of each subgroup was limited.

    This study does not recommend invariably administering restricted oxygen for patients with severe infection. Although higher risks of in-hospital mortality with approximately PaO2 >240 mm Hg were observed in the restricted spline curve for mortality prediction, hypoxia at approximately <60 mm Hg was also inappropriate. Therefore, it would be recommended adjusting the target PaO2 between 60 and 240 mm Hg or at least to avoid PaO2 ≥300 mm Hg once severe infection is suspected. However, as other potential diagnoses usually exist even when sepsis is sufficiently suspected, a higher oxygen amount will be needed depending on other possible aetiologies for critical illness.

    The duration of hyperoxia exposure during resuscitation was unknown in this study, which is a major limitation to validate the association between hyperoxia (PaO2 ≥300 mm Hg) and increased in-hospital mortality. In addition, as PaO2 would have been adjusted to <300 mm Hg following a standard practice in ICU in most patients, the generalisability of the current results for oxygen treatment with longer exposure of hyperoxia would be limited. Moreover, other thresholds for inappropriate PaO2 may exist depending on the timing of hyperoxia exposure, infection severity and patient characteristics.

    The results must be interpreted in the context of the study’s design. We retrospectively retrieved data from the SPICE-ICU Study that did not record the indications for administering high oxygen amounts instead of low to moderate one. Therefore, the findings may differ if the reasons for oxygen therapy with high FiO2 are dependent on unrecorded, strong prognostic factors, such as the intensive care quality, intubation indication/requirement and peripheral oxygen saturation measurement reliability. Another limitation was the lack of detailed clinical information on pulmonary and other organ functions before, during and after hyperoxia exposure. Although supraphysiological oxygen tension would cause tissue toxicity, they could not be objectively evaluated. Finally, various subsets of population with severe infection exist, depending on infection site, types of bacteria, critical care practices and degree of disturbed organ function, which were not analysed in this study because of limited sample size. Therefore, the association between hyperoxia and higher mortality would be different in these subgroups and should be further revealed in other studies.

    In conclusion, hyperoxia defined as PaO2 ≥300 mm Hg at treatment initiation for severe infection was associated with increased in-hospital mortality risk in patients requiring intensive care. Restriction of oxygen administration to avoid hyperoxia during resuscitation should be carefully considered in daily practice, and the appropriate arterial oxygen tension for patients with sepsis should be validated in future studies.

    Data availability statement

    Data are available upon reasonable request. The study data are available from the SPICE-ICU Study group; however, restrictions apply to the availability of these data, which were used under licence for the current study and are not publicly available. However, data are available from the authors upon reasonable request and with the permission of the SPICE-ICU Study group.

    Ethics statements

    Patient consent for publication

    Ethics approval

    This study involves human participants. The SPICE-ICU Study was supported by the Japanese Association for Acute Medicine (approval no. 2014-01) andwas approved by the institutional review board for conducting research with human participants in accordance with the Helsinki Declaration at the head institute of the SPICE-ICU Study group (approval no. 016-0386 from Hokkaido University on 8 February 2017; study title, 'Japanese Association for Acute Medicine Sepsis Prognostication in Intensive Care Unit and Emergency Room'). The institutional review board at the head institute of the study group waived informed consent because of the anonymous nature of the data.

    References

    1. ↵
      1. Angus DC
      . Oxygen therapy for the critically ill. N Engl J Med 2020;382:1054–6. doi:10.1056/NEJMe2000800
      OpenUrl
    2. ↵
      1. Helmerhorst HJF,
      2. Roos-Blom M-J,
      3. van Westerloo DJ, et al
      . Association between arterial Hyperoxia and outcome in Subsets of critical illness: A systematic review, meta-analysis, and meta-regression of cohort studies. Crit Care Med 2015;43:1508–19. doi:10.1097/CCM.0000000000000998
      OpenUrlCrossRefPubMed
    3. ↵
      1. Barbateskovic M,
      2. Schjørring OL,
      3. Russo Krauss S, 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. doi:10.1002/14651858.CD012631.pub2
    4. ↵
      1. Damiani E,
      2. Adrario E,
      3. Girardis M, et al
      . Arterial Hyperoxia and mortality in critically ill patients: a systematic review and meta-analysis. Crit Care 2014;18:711. doi:10.1186/s13054-014-0711-x
      OpenUrlPubMed
    5. ↵
      1. Palmer E,
      2. Post B,
      3. Klapaukh R, et al
      . The association between supraphysiologic arterial oxygen levels and mortality in critically ill patients. a multicenter observational cohort study. Am J Respir Crit Care Med 2019;200:1373–80. doi:10.1164/rccm.201904-0849OC
      OpenUrlCrossRefPubMed
    6. ↵
      1. Yamamoto R,
      2. Yoshizawa J,
      3. Takauji S, et al
      . Hyperoxia for accidental hypothermia and increased mortality: a post-hoc analysis of a multicenter prospective observational study. Crit Care 2023;27:131. doi:10.1186/s13054-023-04407-8
      OpenUrl
    7. ↵
      1. Mackle D,
      2. Bellomo R,
      3. Bailey M, et al
      . Conservative oxygen therapy during mechanical ventilation in the ICU. N Engl J Med 2020;382:989–98. doi:10.1056/NEJMoa1903297
      OpenUrlCrossRefPubMed
    8. ↵
      1. Angelos MG,
      2. Yeh ST,
      3. Aune SE
      . Post-cardiac arrest hyperoxia and mitochondrial function. Resuscitation 2011:S48–51. doi:10.1016/S0300-9572(11)70151-4
    9. ↵
      1. Dias-Freitas F,
      2. Metelo-Coimbra C,
      3. Roncon-Albuquerque R
      . Molecular mechanisms underlying hyperoxia acute lung injury. Respir Med 2016;119:23–8. doi:10.1016/j.rmed.2016.08.010
      OpenUrl
    10. ↵
      1. Raj R,
      2. Bendel S,
      3. Reinikainen M, et al
      . Hyperoxemia and long-term outcome after traumatic brain injury. Crit Care 2013;17:R177. doi:10.1186/cc12856
    11. ↵
      1. Sinclair SE,
      2. Altemeier WA,
      3. Matute-Bello G, et al
      . Augmented lung injury due to interaction between Hyperoxia and mechanical ventilation. Crit Care Med 2004;32:2496–501. doi:10.1097/01.ccm.0000148231.04642.8d
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Cohen ES,
      2. Hollenberg SM
      . Tissue oxygenation and sepsis. Crit Care Med 2001;29:1479–80. doi:10.1097/00003246-200107000-00032
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Ince C,
      2. Sinaasappel M
      . Microcirculatory oxygenation and Shunting in sepsis and shock. Crit Care Med 1999;27:1369–77. doi:10.1097/00003246-199907000-00031
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Sair M,
      2. Etherington PJ,
      3. Curzen NP, et al
      . Tissue oxygenation and perfusion in endotoxemia. Am J Physiol 1996;271:H1620–5. doi:10.1152/ajpheart.1996.271.4.H1620
      OpenUrl
    15. ↵
      1. Semler MW,
      2. Casey JD,
      3. Lloyd BD, et al
      . Oxygen-saturation targets for critically ill adults receiving mechanical ventilation. N Engl J Med 2022;387:1759–69. doi:10.1056/NEJMoa2208415
      OpenUrlCrossRefPubMed
    16. ↵
      1. Young P,
      2. Mackle D,
      3. Bellomo R, et al
      . Conservative oxygen therapy for mechanically ventilated adults with sepsis: a post hoc analysis of data from the intensive care unit randomized trial comparing two approaches to oxygen therapy (ICU-ROX). Intensive Care Med 2020;46:17–26. doi:10.1007/s00134-019-05857-x
      OpenUrl
    17. ↵
      1. Demiselle J,
      2. Wepler M,
      3. Hartmann C, et al
      . Hyperoxia toxicity in septic shock patients according to the Sepsis-3 criteria: a post hoc analysis of the Hyper2S trial. Ann Intensive Care 2018;8:90. doi:10.1186/s13613-018-0435-1
      OpenUrl
    18. ↵
      1. Martín-Fernández M,
      2. Heredia-Rodríguez M,
      3. Tamayo E
      . “Reply to "Hyperoxemia in postsurgical sepsis/septic shock patients is associated with reduced mortality"” Crit Care 2022;26:177. doi:10.1186/s13054-022-04045-6
      OpenUrl
    19. ↵
      1. Abe T,
      2. Yamakawa K,
      3. Ogura H, et al
      . Epidemiology of sepsis and septic shock in intensive care units between Sepsis-2 and Sepsis-3 populations: sepsis prognostication in intensive care unit and emergency room (SPICE-ICU). J Intensive Care 2020;8:44. doi:10.1186/s40560-020-00465-0
      OpenUrl
    20. ↵
      1. Levy MM,
      2. Fink MP,
      3. Marshall JC, et al
      . SCCM/ESICM/ACCP/ATS/SIS International sepsis definitions conference. Intensive Care Med 2003;29:530–8. doi:10.1007/s00134-003-1662-x
      OpenUrlCrossRefPubMed
    21. ↵
      1. Singer M,
      2. Deutschman CS,
      3. Seymour CW, et al
      . The third International consensus definitions for sepsis and septic shock (Sepsis-3). JAMA 2016;315:801–10. doi:10.1001/jama.2016.0287
      OpenUrlCrossRefPubMed
    22. ↵
      1. Fan E,
      2. Del Sorbo L,
      3. Goligher EC, et al
      . An official American thoracic society/European society of intensive care medicine/society of critical care medicine clinical practice guideline: mechanical ventilation in adult patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2017;195:1253–63. doi:10.1164/rccm.201703-0548ST
      OpenUrlCrossRefPubMed
    23. ↵
      1. Rockwood K,
      2. Theou O
      . Using the clinical frailty scale in allocating scarce health care resources. Can Geriatr J 2020;23:210–5. doi:10.5770/cgj.23.463
      OpenUrlCrossRefPubMed
    24. ↵
      1. Wang C-H,
      2. Chang W-T,
      3. Huang C-H, et al
      . The effect of hyperoxia on survival following adult cardiac arrest: a systematic review and meta-analysis of observational studies. Resuscitation 2014;85:1142–8. doi:10.1016/j.resuscitation.2014.05.021
      OpenUrlCrossRefPubMed
    25. ↵
      1. Austin PC
      . An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011;46:399–424. doi:10.1080/00273171.2011.568786
      OpenUrlCrossRefPubMedWeb of Science
    26. ↵
      1. Austin PC,
      2. Stuart EA
      . The performance of inverse probability of treatment weighting and full matching on the propensity score in the presence of model misspecification when estimating the effect of treatment on survival outcomes. Stat Methods Med Res 2017;26:1654–70. doi:10.1177/0962280215584401
      OpenUrlCrossRefPubMed
    27. ↵
      1. Abe T,
      2. Ogura H,
      3. Shiraishi A, et al
      . Characteristics, management, and in-hospital mortality among patients with severe sepsis in intensive care units in Japan: the FORECAST study. Crit Care 2018;22:322. doi:10.1186/s13054-018-2186-7
      OpenUrl
    28. ↵
      1. Angus DC,
      2. van der Poll T
      . Severe sepsis and septic shock. N Engl J Med 2013;369:2063. doi:10.1056/NEJMc1312359
    29. ↵
      1. Brookhart MA,
      2. Schneeweiss S,
      3. Rothman KJ, et al
      . Variable selection for propensity score models. Am J Epidemiol 2006;163:1149–56. doi:10.1093/aje/kwj149
      OpenUrlCrossRefPubMedWeb of Science
    30. ↵
      1. Bible J,
      2. Albert PS,
      3. Simons-Morton BG, et al
      . Practical issues in using generalized estimating equations for inference on transitions in longitudinal data: what is being estimated Stat Med 2019;38:903–16. doi:10.1002/sim.8014
      OpenUrl
    31. ↵
      1. Barrot L,
      2. Asfar P,
      3. Mauny F, et al
      . Liberal or conservative oxygen therapy for acute respiratory distress syndrome. N Engl J Med 2020;382:999–1008. doi:10.1056/NEJMoa1916431
      OpenUrlCrossRefPubMed
    32. ↵
      1. Fisher AB,
      2. Beers MF
      . Hyperoxia and acute lung injury. Am J Physiol Lung Cell Mol Physiol 2008;295:L1066. doi:10.1152/ajplung.90486.2008
    33. ↵
      1. Davis DP,
      2. Meade W,
      3. Sise MJ, et al
      . Both hypoxemia and extreme hyperoxemia may be detrimental in patients with severe traumatic brain injury. J Neurotrauma 2009;26:2217–23. doi:10.1089/neu.2009.0940
      OpenUrlCrossRefPubMedWeb of Science
    34. ↵
      1. Yamamoto R,
      2. Fujishima S,
      3. Sasaki J, et al
      . Hyperoxemia during resuscitation of trauma patients and increased intensive care unit length of stay: inverse probability of treatment weighting analysis. World J Emerg Surg 2021;16:19.:19. doi:10.1186/s13017-021-00363-2
      OpenUrl
    35. ↵
      1. Steptoe A,
      2. Deaton A,
      3. Stone AA
      . Subjective wellbeing, health, and ageing. Lancet 2015;385:640–8. doi:10.1016/S0140-6736(13)61489-0
      OpenUrlCrossRefPubMed

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Contributors RY and SF designed the study. KY, TA, HO, DS, SG and JS performed data collection. SF managed quality control. RY and KY performed data analysis and interpretation. RY, SF and JS performed writing and critical revision. RY is the guarantor. All authors revised the article.

    • Funding This research was supported by Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, and Technology (KAKENHI, 22K16631).

    • Competing interests None declared.

    • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.