• Users Online: 18
  • Home
  • Print this page
  • Email this page
Home About us Editorial board Ahead of print Current issue Search Archives Submit article Instructions Subscribe Contacts Login 


 
 Table of Contents  
ORIGINAL ARTICLE
Year : 2020  |  Volume : 7  |  Issue : 2  |  Page : 197-204

Oxygen exposure as quantified by time-weighted area under curve for arterial oxygen content is associated with mortality in mechanically ventilated critically ill patients


1 Department of Intensive Care, Caboolture Hospital, Caboolture; School of Medicine, University of Queensland, St Lucia, Queensland, Australia
2 Department of Intensive Care, Redcliffe Hospital, Redcliffe; School of Medicine, University of Queensland, St Lucia, Queensland, Australia
3 Department of Intensive Care, Caboolture Hospital, Caboolture; Department of Intensive Care, Redcliffe Hospital, Redcliffe; Department of Intensive Care, The Prince Charles Hospital, Chermside; School of Medicine, University of Queensland, St Lucia, Queensland; The George Institute for Global Health, Sydney, New South Wales, Australia

Date of Submission01-Oct-2019
Date of Acceptance19-Apr-2020
Date of Web Publication27-Jun-2020

Correspondence Address:
BSc, MBBS James P Harvey
Royal Brisbane and Women’s Hospital, Herston, Queensland, 4006; Townsville Hospital, Townsville, Queensland
Australia
Login to access the Email id

Source of Support: None, Conflict of Interest: None


DOI: 10.4103/roaic.roaic_84_19

Rights and Permissions
  Abstract 

Background Oxygen is frequently administered to intensive care patients, for both treatment and prophylaxis. Arterial oxygen content (CaO2) represents the total amount of oxygen in arterial blood, both bound to hemoglobin and dissolved. CaO2 could be a useful marker of tissue oxygen levels and oxygen exposure.
Aims We undertook this study to determine the relationship between CaO2 and mortality in mechanically ventilated critically ill patients.
Settings and design A retrospective cohort study of all mechanically ventilated adult patients in the Multiparameter Intelligent Monitoring in Intensive Care III database was conducted. Patients with less than three arterial blood gases were excluded. The primary exposure variable was time-weighted CaO2 (TWCaO2) over the course of the entire ICU admission. The primary outcome was 6-month mortality. Multivariate logistic regression analysis was used to assess the relationship between CaO2 and mortality.
Results A total of 7452 patients were identified who satisfied all inclusion and exclusion criteria. In the multivariate analysis, higher CaO2 was significantly associated with increased mortality. After adjustment for age, sex, transfusion, admission type, Elixhauser Comorbidity Index, Simplified Acute Physiology Score II, and time-weighted fraction of inspired oxygen, the highest quartile had an odds ratio (mortality) of 1.22 (95% confidence interval, 1.03–1.46; P=0.02). The second (TWCaO2, 12.2–13.3 ml/100 ml) and third (TWCaO2, 13.4–14.4 ml/100 ml) quartiles had odds ratio (mortality) of 1.19. Postestimation analysis revealed good model discrimination with a c-statistic of 0.80 for the final model.
Conclusion In mechanically ventilated patients, after adjusting for disease severity and comorbidities, higher oxygen exposure as indicated by TWCaO2 over the entire ICU admission was associated with increased mortality.

Keywords: arterial oxygen content, intensive care, mechanical ventilation, oxygen saturation, partial pressure of oxygen in arterial blood


How to cite this article:
Harvey JP, Jayawardena DG, Ramanan M. Oxygen exposure as quantified by time-weighted area under curve for arterial oxygen content is associated with mortality in mechanically ventilated critically ill patients. Res Opin Anesth Intensive Care 2020;7:197-204

How to cite this URL:
Harvey JP, Jayawardena DG, Ramanan M. Oxygen exposure as quantified by time-weighted area under curve for arterial oxygen content is associated with mortality in mechanically ventilated critically ill patients. Res Opin Anesth Intensive Care [serial online] 2020 [cited 2020 Oct 28];7:197-204. Available from: http://www.roaic.eg.net/text.asp?2020/7/2/197/288002


  Introduction Top


Oxygen is frequently administered to intensive care patients to treat or prevent hypoxemia [1],[2]. The goal of oxygen therapy is to prevent tissue hypoxia and thereby reduce the risk of organ injury or cell death [2]. Oxygen delivery is proportional to cardiac output and arterial oxygen content (CaO2). The CaO2 is the total amount of oxygen in arterial blood both bound to hemoglobin and dissolved in blood. It can be calculated as follows: CaO2=(Hb×1.36×SaO2)+(PaO2×0.0031), where 0.0031 is the solubility coefficient of oxygen in human plasma [3]. Each gram of hemoglobin is theoretically capable of carrying 1.39 ml of oxygen, though in vivo, this value is slightly reduced. As the solubility of oxygen in the blood is low, CaO2 is largely determined by the hemoglobin concentration and the percentage saturated with oxygen. Thus, when hemoglobin is fully saturated, additional oxygen only marginally increases CaO2 [4]. The effects of oxygen therapy are frequently assessed by measuring arterial oxygen saturation (SaO2) and partial pressure of oxygen in arterial blood (PaO2) [5],[6]. Normal values vary with age and chronic respiratory illnesses, and explicit guidelines for PaO2 and SaO2 targets do not exist. Currently, guidelines from the British Thoracic Society recommend maintaining SaO2 94–98% for seriously ill, nonventilated patients [2]. The harms of hypoxemia are well established. In addition to predicting inpatient mortality, lower PaO2 levels in acute respiratory distress syndrome have been identified as a risk factor for long-term neuropsychological impairment [7]. The risks associated with supranormal oxygen levels in critically ill patients are less clear. At a cellular level, direct pulmonary toxicity of hyperoxia leads to histopathological changes similar to acute respiratory distress syndrome and ventilator-associated lung injury [8]. Hyperoxemia may also impair the innate immune response, reduce cardiac output, reduce mucociliary clearance, and result in the production of reactive oxygen species [9],[10],[11],[12]. The AVOID trial demonstrated that in ST elevation myocardial infarction, supplemental oxygen increased myocardial injury, recurrent infarction, major cardiac arrhythmia, and infarct size [13]. The vasoconstrictor effects of oxygen on coronary, cerebral, and other peripheral vasculature may also decrease the perfusion to organs to a greater extent than oxygen supplementation increases the CaO2, thereby reducing oxygen delivery to the tissue [14]. In neonatal resuscitation, use of 100% oxygen has been associated with increased mortality compared against 21% oxygen, resulting in a dramatic change in practice [15]. CaO2 is potentially a better indicator of oxygen delivery to the tissues than either SaO2 or PaO2, yet specific evidence to support this is currently lacking [16]. Lower CaO2 has been associated with lower rates of inpatient survival and higher rates of acute renal failure in patients with community-acquired pneumonia [17]. No previous studies have investigated the association between CaO2 and survival in ICU patients. We hypothesized that lower CaO2 over the course of an ICU admission would be associated with increased 6-month mortality in mechanically ventilated critically ill patients.


  Patients and methods Top


Selection and design of participants

Multiparameter Intelligent Monitoring in Intensive Care III (MIMIC-III) is an open-access research database that includes all patients admitted to ICUs at Beth Israel Deaconess Medical Center in Boston, Massachusetts, USA [18]. The MIMIC III database was approved by institutional review boards of Beth Israel Deaconess Medical Center and Massachusetts Institute of Technology (IRB protocol 2001-P-001699/3). The database contains 46 520 individual patients and 58 976 hospital admissions between 2001 and 2012. Establishment of the database for research purposes was approved by institutional review boards of Beth Israel Deaconess Medical Center and Massachusetts Institute of Technology (IRB protocol 2001-P-001699/3).

Data extraction

All authors completed the Data or Specimens Only Research course provided by Collaborative Institute Training Initiative before accessing the MIMIC-III database. We conducted a retrospective cohort study using the MIMIC-III database (version 1.4). The first ICU admission for all adult patients (age≥16) requiring mechanical ventilation was included. Data were extracted for patient age, admission type, length of ICU stay, duration of mechanical ventilation, transfusion requirement, illness severity scores, comorbidities, and mortality at 6 months. Data from every arterial blood gas (ABG) during the ICU admission were extracted. Patients who had two or fewer ABGs were excluded. The most recent recorded fraction of inspired oxygen (FiO2) before the ABG was used. CaO2 was calculated using most recent hemoglobin available within 24 h of arterial sampling. Predicted hospital mortality was assessed using the Simplified Acute Physiology Score II (SAPS II) risk of death. The SAPS, Oxford Acute Severity of Illness Score (OASIS), Logistic Organ Dysfunction Score (LODS), and Sequential Organ Failure Score (SOFA) were also extracted. Comorbidities were categorized using the Elixhauser Comorbidity Index (ECI) [19]. The primary outcome was 6-month mortality.

Statistical analysis

Data were extracted from the MIMIC-III database using PostgreSQL 10.3, and all statistical analyses were performed using Stata 13.0 (StatsCorp LP, College Station, Texas, USA). Continuous data were summarized as means (SD) and medians (interquartile range, IQR) for approximately normally distributed and skewed data, respectively. Categorical data were summarized as proportions. The Mann–Whitney U test and c2 test were used in bivariate analyses for continuous and categorical variables, respectively. CaO2 was calculated for each ABG using the formula outlined in Introduction. The time-weighted area under the curve was calculated by using the trapezoidal method for each patient using all ABGs, with complete data performed during the patient’s ICU stay. The same method was used to calculate time-weighted FiO2 (TWFiO2). The primary outcome measure was odds ratio (OR) with 95% confidence intervals (CI) of mortality at 6 months. A P value of 0.05 was considered statistically significant. Univariate logistic regression was used to test for the effect of prespecified confounders and to test for the functional form of continuous variables. Multivariate logistic regression analysis was performed to determine the effect of CaO2 on mortality after adjustment for confounders. In the multivariate model, we adjusted for emergency versus elective admission, SAPS II risk of death, TWFiO2, age, sex, transfusion, and ECI. The c-statistic for area under receiver operating characteristic curve was calculated to assess model discrimination for the multivariate model. The Hosmer–Lemeshow goodness-of-fit test to assess model calibration was not used as our large sample size was likely to guarantee statistical significance. SAPS II was substituted for the other available risk of death prediction scores, OASIS, LODS, SOFA, and SAPS to test whether they improved the discrimination of the multivariate model. The c-statistics thus generated were compared using a nonparametric method proposed by DeLong et al. [20].


  Results Top


Of 46 520 patients, 7452 satisfied all inclusion and exclusion criteria and were included in the final analysis ([Figure 1]). The overall 6-month mortality rate was 26% (1933 out of 7452). There were some significant differences between the baseline characteristics of survivors and nonsurvivors ([Table 1]).
Figure 1 Diagram of inclusion and exclusion of the study population. ABG, arterial blood gas; Hb, hemoglobin; FiO2, fraction of inspired oxygen; PaO2, partial pressure of oxygen in arterial blood; SaO2, oxygen saturation.

Click here to view
Table 1 Comparison of the characteristics of survivors and nonsurvivors

Click here to view


The median age was 68 years (IQR, 57–77), with survivors being younger (median, 66; IQR, 56–76) than nonsurvivors (median, 72; IQR, 62–81). Overall, 61% (4532) were male, but only 56% (1074) of nonsurvivors were male. Nonsurvivors were more likely to have an emergency ICU admission (92 vs. 67%) and to be transfused with packed red blood cells (56 vs. 50%). Nonsurvivors had longer ICU (9.1 vs. 4.2) and hospital (13.3 vs. 10.4) lengths of stay, longer mechanical ventilation hours (140 vs. 21), higher median SAPS II scores (48 vs. 37), and higher median ECI (21 vs. 5). The median number of ABGs per patient was 6 (IQR, 4–10) with a long right-sided tail (range, 3–350). Nonsurvivors had more ABGs (median, 7 vs. 5). The median TWFiO2 over the course of the entire ICU admission was 50% in both groups; however, the distributions between survivors and nonsurvivors were different, and hence, the U test was significant during bivariate analysis. The median time-weighted CaO2 (TWCaO2) was similar between survivors (median, 13.3 ml/100 ml; IQR, 12.2–14.6) and nonsurvivors (median, 13.3 ml/100 ml; IQR, 12.3–14.3). In the univariate analyses ([Table 2] and [Figure 2]), there was no association between CaO2 and mortality regardless of whether CaO2 was included as a continuous or categorical (divided into quartiles or deciles) variable. All the other tested variables were significantly associated with mortality. Higher age, illness severity scores (SAPS, SAPS II, OASIS, LODS, and SOFA), ECI, and TWFiO2 were all associated with significantly increased OR of mortality, as was female sex and receipt of transfusion. All variables were retained for the multivariate phase of statistical analysis, though illness severity scores were only used one at a time.
Table 2 Univariate analysis

Click here to view
Figure 2 Univariate analysis of mortality by time-weighted arterial oxygen content.

Click here to view


In the multivariate analysis, higher CaO2 was significantly associated with increased mortality ([Table 3] and [Figure 3]). After adjustment for age, sex, transfusion, admission type, ECI, SAPS II, and TWFiO2, the highest quartile (TWCaO2>14.4 ml/100 ml) had an OR (mortality) of 1.22 (95% CI, 1.03–1.46; P=0.02). The second (TWCaO2, 12.2–13.3 ml/100 ml) and third (TWCaO2, 13.4–14.4 ml/100 ml) quartiles had OR (mortality) of 1.19 (95% CI, 1.01–1.40; P=0.04) and 1.19 (95% CI, 1.01–1.41; P=0.04), respectively, compared with the lowest quartile (TWCaO2, <12.2 ml/100 ml). The other included variables all remained highly significant, with p values less than or equal to 0.001 ([Table 3]). Postestimation analysis revealed good model discrimination (c-statistic=0.80) for the final model.
Table 3 Risk-adjusted multivariate model

Click here to view
Figure 3 Multivariate analysis of mortality by time-weighted arterial oxygen content adjusted for age, sex, transfusion, Elixhauser comorbidity index, Simplified Acute Physiology Score II, admission type and time-weighted FiO2. FiO2, fraction of inspired oxygen.

Click here to view


The effect of replacing SAPS II with the other available illness severity scores was tested by comparing c-statistics using a nonparametric method proposed by DeLong et al. [20] ([Table 4]). SAPS II had a significantly better discrimination than SAPS (P<0.001), LODS (P=0.001), and SOFA (P<0.001) but was not significantly different to OASIS (P=0.62).
Table 4 Comparison of risk adjustor variables

Click here to view



  Discussion Top


Key findings

In this large retrospective cohort study, we found that in mechanically ventilated critically ill patients, after adjusting for illness severity and comorbidities, higher oxygen exposure, as quantified by TWCaO2 over the entire course of ICU admission, was associated with increased mortality.

Relationship to previous findings

Both hypoxemia and anemia occur commonly in critically ill patients. Hyperoxemia in ventilated patients is associated with morbidity, and relative hypoxemia is known to be beneficial in some conditions [21],[22]. Recently, randomized controlled trials have compared conservative versus liberal oxygen strategies. A meta-analysis of 16 000 patients concluded that liberal oxygen therapy increases mortality without improving other patient-important outcomes [23]. Despite the apparent risks, studies show that ICU patients are subjected to substantive periods of hyperoxia, with this exposure predominately occurring at a low FiO2 [24],[25]. Retrospective studies involving heterogeneous patient groups have also provided inconsistent results on the risks of hyperoxia in intensive care patients. Girardis et al. [26] demonstrated a relationship between in-hospital mortality and a high FiO2 and both low and high PaO2 within the first 24 h of admission. This U-shaped relationship between PaO2 values and mortality has been previously reported after adjusting for SAPS II scores, with the highest mortality observed in patients exposed to the highest time-weighted PaO2 [24],[26]. Further retrospective observational studies have failed to demonstrate this U-shaped relationship between hyperoxia and mortality when baseline patient characteristics were adjusted for [27],[28]. In mechanically ventilated patients, the optimal oxygenation targets, whether it be SaO2, PaO2, or CaO2, remain unknown [29],[30]. Our study adds further evidence to the deleterious effects of supplemental oxygen in critically ill patients. The relationship between CaO2 and mortality in our analysis is nonlinear, with the association between CaO2 and mortality most pronounced at a high CaO2. This supports the possibility that harm associated with hyperoxia may exist only at high levels. This hypothesis is partially supported in animal models, where prolonged exposure to a FiO2 up to 60% does not appear to cause pulmonary damage, even after prolonged exposure [8],[31],[32]. Higher CaO2 levels may also reflect presumed clinical risk, acting as a marker of illness severity, with supplemental oxygen provided as a ‘safety-net’ for higher risk patients [27]. The same may potentially be true for transfusion thresholds. It remains unclear whether hyperoxia confers an independent mortality risk or whether the increased risk in the higher CaO2 exposure group may have arisen from inadequate adjustment for illness severity group. In one retrospective study, the use of Acute Physiology and Chronic Health Evaluation (APACHE) III illness severity score instead of SAPS II attenuated the increase in standardized mortality ratio with increasing PaO2 [27].

Clinical implications

Hemoglobin, PaO2, and SaO2 are frequently performed investigations in intensive care patients, and CaO2 can be easily calculated. Despite this, oxygen therapy is often titrated to SaO2, PaO2, or SpO2 [2],[22]. Routine use of CaO2 as a measure of oxygenation may offer advantages over these markers. Given that modest changes in hemoglobin can substantially modify the CaO2, and anemia is a common finding in critically ill patients, the use of CaO2 may also be a more reliable predictor of tissue oxygen levels and hence oxygen exposure.

Strengths and limitations

This is the largest study to investigate the relationship between CaO2 and mortality and the first such study in a critically ill population. The use of MIMIC-III data, collected from medical and surgical ICU patients from a single hospital center, is a high-quality database that has been used extensively for research purposes. The inclusion of all ABGs over the duration of admission in our study is advantageous over previous studies, which have relied upon ‘best’ and ‘worst’ PaO2 levels [24],[27]. Oxygen toxicity in animal models is known to be proportional to exposure duration [8]. We have not only achieved better quantification of arterial oxygen levels using CaO2 but have also measured oxygen exposure over the entire ICU admission, not just the first 24 h. The retrospective, observational nature of this study means that it can only be used to identify associations and cannot infer causality. We were unable to identify a floor level below which reduced CaO2 increased mortality. Such a level must exist because we know that significant hypoxemia leads to tissue injury, organ failure, and death. Our data set was too small to detect this level. We used SAPS II for risk adjustment, as it was the best risk adjustor available. Better model performance has been demonstrated with the use of alternative illness severity scores, such as the APACHE III and IV compared with SAPS II [33]. It was not possible to calculate APACHE III or IV risk of death using the MIMIC-III database. With better risk adjustment, it may be possible that the observed association between CaO2 and mortality may become insignificant. Diagnostic groups were not adjusted for in this model. It is possible that optimal CaO2 may vary between surgical and medical diagnoses. Owing to the risk of introducing collinearity, the mixed cohort of patients, and nonuniform transfusion thresholds, we did not adjust for hemoglobin as part of the multivariate analysis. Furthermore, we did not investigate the number or volume of transfusions. The oxygen targets, ventilation modes, and strategies for patients were also not identified and may introduce bias. De-identification of MIMIC database meant mortality-adjustment by year over the 11-year period was not possible. This may be important as several studies have demonstrated the falling mortality of critically ill patients over time [34],[35].

Future research

Further research should focus on replicating this study with prospective ICU data using a superior risk adjustment tool such as APACHE IV, SAPS III, or Australian and New Zealand Risk of Death (ANZROD) may provide better risk adjustment [36]. The interaction between admission diagnosis and CaO2 should be studied.


  Conclusion Top


In mechanically ventilated critically ill patients, higher oxygen exposure during ICU admissions as measured by TWCaO2 is significantly associated with increased mortality at 6 months. These results add further weight to the accumulating evidence supporting the toxicity of oxygen exposure.

Acknowledgements

Presentation at a meeting: Australian and New Zealand Intensive Care Society (ANZICS) annual scientific meeting, Adelaide, South Australia, October 13, 2018.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.



 
  References Top

1.
Eastwood GM, Peck L, Young H, Suzuki S, Garcia M, Bellomo R. Intensive care clinicians’ opinion of conservative oxygen therapy (SpO(2) 90-92%) for mechanically ventilated patients. Aust Crit Care 2014; 27:120–125.  Back to cited text no. 1
    
2.
O’Driscoll BR, Howard LS, Davison AG. BTS guideline for emergency oxygen use in adult patients. Thorax 2008; 63 (Suppl 6):vi1–vi68.  Back to cited text no. 2
    
3.
McLellan SA, Walsh TS. Oxygen delivery and haemoglobin. Contin Educ Anaesth Crit Care Pain 2004; 4:123–126.  Back to cited text no. 3
    
4.
Collins JA, Rudenski A, Gibson J, Howard L, O’Driscoll R. Relating oxygen partial pressure, saturation and content: the haemoglobin-oxygen dissociation curve. Breathe (Sheff) 2015; 11:194–201.  Back to cited text no. 4
    
5.
Slutsky AS. Consensus conference on mechanical ventilation − January 28-30, 1993 at Northbrook, Illinois, USA. Part 2. Intensive Care Med 1994; 20:150–162.  Back to cited text no. 5
    
6.
Panwar R, Capellier G, Schmutz N, Davies A, Cooper DJ, Bailey M et al. Current oxygenation practice in ventilated patients-an observational cohort study. Anaesth Intensive Care 2013; 41:505–514.  Back to cited text no. 6
    
7.
Mikkelsen ME, Christie JD, Lanken PN, Biester RC, Thompson BT, Bellamy SL et al. The adult respiratory distress syndrome cognitive outcomes study: long-term neuropsychological function in survivors of acute lung injury. Am J Respir Crit Care Med 2012; 185:1307–1315.  Back to cited text no. 7
    
8.
Kallet RH, Matthay MA. Hyperoxic acute lung injury. Respir Care 2013; 58:123–141.  Back to cited text no. 8
    
9.
Baleeiro CE, Wilcoxen SE, Morris SB, Standiford TJ, Paine IIIR. Sublethal hyperoxia impairs pulmonary innate immunity. J Immunol 2003; 171:955–963.  Back to cited text no. 9
    
10.
Brueckl C, Kaestle S, Kerem A, Habazettl H, Krombach F, Kuppe H et al. Hyperoxia-induced reactive oxygen species formation in pulmonary capillary endothelial cells in situ. Am J Respir Cell Mol Biol 2006; 34:453–463.  Back to cited text no. 10
    
11.
Tashiro N. Influence of hyperoxia and hypoxia on bronchial mucociliary clearance. Hokkaido Igaku Zasshi 1988; 63:597–606.  Back to cited text no. 11
    
12.
Lodato RF. Decreased O2 consumption and cardiac output during normobaric hyperoxia in conscious dogs. J Appl Physiol 1989; 67:1551–1559.  Back to cited text no. 12
    
13.
Stub D, Smith K, Bernard S, Bray JE, Stephenson M, Cameron P et al. A randomized controlled trial of oxygen therapy in acute myocardial infarction air verses oxygen in myocardial infarction study (AVOID Study). Am Hear J 2012; 163:339–345 e1.  Back to cited text no. 13
    
14.
Iscoe S, Beasley R, Fisher JA. Supplementary oxygen for nonhypoxemic patients: O2 much of a good thing? Crit Care 2011; 15(3 LB-Iscoe2011):305.  Back to cited text no. 14
    
15.
Saugstad OD, Ramji S, Soll RF, Vento M. Resuscitation of newborn infants with 21% or 100% oxygen: an updated systematic review and meta-analysis. Neonatology 2008; 94:176–182.  Back to cited text no. 15
    
16.
Rizvi A, Macedo P, Babawale L, Tighe HC, Hughes JMB, Jackson JE et al. Hemoglobin is a vital determinant of arterial oxygen content in hypoxemic patients with pulmonary arteriovenous malformations. Ann Am Thorac Soc 2017; 14:903–911.  Back to cited text no. 16
    
17.
Sanz F, Novella L, Fernandez-Fabrellas E, Briones ML, Chiner E, Lera R et al. Arterial oxygen content could reflect acute organ damage in community-acquired pneumonia. Eur Respir J 2014; 44(Suppl 58):4747.  Back to cited text no. 17
    
18.
Johnson AE, Pollard TJ, Shen L, Lehman LW, Feng M, Ghassemi M et al. MIMIC-III, a freely accessible critical care database. Sci Data 2016; 3:160035.  Back to cited text no. 18
    
19.
van Walraven C, Austin PC, Jennings A, Quan H, Forster AJ. A modification of the Elixhauser comorbidity measures into a point system for hospital death using administrative data. Med Care 2009; 47:626–633.  Back to cited text no. 19
    
20.
DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44:837–845.  Back to cited text no. 20
    
21.
Bone RC, Pierce AK, Johnson JrRL. Controlled oxygen administration in acute respiratory failure in chronic obstructive pulmonary disease: a reappraisal. Am J Med 1978; 65:896–902.  Back to cited text no. 21
    
22.
Beasley R, Chien J, Douglas J, Eastlake L, Farah C, King G et al. Thoracic Society of Australia and New Zealand oxygen guidelines for acute oxygen use in adults: ‘swimming between the flags. Respirology 2015; 20:1182–1191.  Back to cited text no. 22
    
23.
Chu DK, Kim LHY, 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; 391:1693–1705.  Back to cited text no. 23
    
24.
de Jonge E, Peelen L, Keijzers PJ, Joore H, de Lange D, van der Voort PH et al. Association between administered oxygen, arterial partial oxygen pressure and mortality in mechanically ventilated intensive care unit patients. Crit Care 2008; 12:R156.  Back to cited text no. 24
    
25.
de Graaff 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.  Back to cited text no. 25
    
26.
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; 316:1583–1589.  Back to cited text no. 26
    
27.
Eastwood G, Bellomo R, Bailey M, Taori G, Pilcher D, Young P et al. Arterial oxygen tension and mortality in mechanically ventilated patients. Intensive Care Med 2012; 38:91–98.  Back to cited text no. 27
    
28.
Sutton AD, Bailey M, Bellomo R, Eastwood GM, Pilcher DV. The association between early arterial oxygenation in the ICU and mortality following cardiac surgery. Anaesth Intensive Care 2014; 42:730–735.  Back to cited text no. 28
    
29.
Gilbert-Kawai ET, Mitchell K, Martin D, Carlisle J, Grocott MP. Permissive hypoxaemia versus normoxaemia for mechanically ventilated critically ill patients. Cochrane Database Syst Rev 2014; 2014:CD009931.  Back to cited text no. 29
    
30.
Hafner S, Beloncle F, Koch A, Radermacher P, Asfar P. Hyperoxia in intensive care, emergency, and peri-operative medicine: Dr. Jekyll or Mr. Hyde? A 2015 update. Ann Intensive Care 2015; 5:42.  Back to cited text no. 30
    
31.
Clark JM, Lambertsen CJ. Pulmonary oxygen toxicity: a review. Pharmacol Rev 1971; 23:37–133.  Back to cited text no. 31
    
32.
McIntyre WF, Um KJ, Alhazzani W, Lengyel AP, Hajjar L, Gordon AC et al. Association of vasopressin plus catecholamine vasopressors vs catecholamines alone with atrial fibrillation in patients with distributive shock a systematic review and meta-analysis. JAMA 2018; 319:1889–1900.  Back to cited text no. 32
    
33.
Keegan MT, Gajic O, Afessa B. Comparison of APACHE III, APACHE IV, SAPS 3, and MPM0III and influence of resuscitation status on model performance. Chest 2012; 142:851–858.  Back to cited text no. 33
    
34.
Kaukonen KM, Bailey M, Suzuki S, Pilcher D, Bellomo R. Mortality related to severe sepsis and septic shock among critically ill patients in Australia and New Zealand, 2000-2012. JAMA 2014; 311:1308–1316.  Back to cited text no. 34
    
35.
ANZICS Centre for Outcome and Resource Evaluation. Annual Report 2017. Melbourne. 2017.  Back to cited text no. 35
    
36.
Keegan MT, Soares M. What every intensivist should know about prognostic scoring systems and risk-adjusted mortality. Rev Bras Ter Intensiva 2016; 28:264–269.  Back to cited text no. 36
    


    Figures

  [Figure 1], [Figure 2], [Figure 3]
 
 
    Tables

  [Table 1], [Table 2], [Table 3], [Table 4]



 

Top
 
 
  Search
 
Similar in PUBMED
 Related articles
Access Statistics
Email Alert *
Add to My List *
* Registration required (free)

 
  In this article
Abstract
Introduction
Patients and methods
Results
Discussion
Conclusion
References
Article Figures
Article Tables

 Article Access Statistics
    Viewed568    
    Printed31    
    Emailed0    
    PDF Downloaded37    
    Comments [Add]    

Recommend this journal


[TAG2]
[TAG3]
[TAG4]