Effects of Conservative Oxygen Therapy versus Conventional Oxygen Therapy on the Mortality in ICU Patients: A Meta-Analysis

Objective To compare the effects of conservative oxygen therapy and conventional oxygen therapy on the mortality of critically ill patients in ICU. Methods Searching for randomized controlled clinical trials (RCT) on the effect of conservative oxygen therapy and conventional oxygen therapy on the mortality of critically ill patients in computer databases, including PubMed, Embase, Cochrane Library, CNKI, VIP, and Wanfang, with postdate before August 2022. We have two researchers evaluating the quality of the literature included and extracting data as per the inclusion and exclusion criteria and then analyzed it with RevMan 5.4 statistical software. Primary outcome included short-term mortality (28-day mortality or ICU mortality); secondary outcome included 90-day mortality, ICU length of stay, hospital length of stay, incidence of new organ dysfunction in ICU, incidence of new infection in ICU, and incidence of ICUAW. Results A total of 5779 subjects were included in 10 articles, including 2886 in the conservative oxygen therapy group and 2893 in the conventional oxygen therapy group. The meta-analysis showed that conservative oxygen therapy had an advantage over conventional oxygen therapy in terms of short-term mortality (P=0.03). Subgroup analysis based on different conservative oxygen targets showed that this advantage was statistically significant when the target is set above 90% (RR = 0.76, 95% CI = 0.62∼0.94, P=0.01), while there was no significant difference between conservative oxygen therapy and conventional oxygen therapy when the target is set below 90% (RR = 0.95, 95% CI = 0.79∼1.16, P=0.63). In addition, in terms of the incidence of new infections in the ICU (P=0.03) and the incidence of ICUAW (P=0.03), conservative oxygen therapy also had advantages over conventional oxygen therapy, and the difference was statistically significant. But in terms of 90-day mortality (P=0.61), ICU length of stay (P=0.96), hospital length of stay (P=0.47), and incidence of new organ dysfunction in ICU (P=0.61), there was no significant difference between conservative oxygen therapy and conventional oxygen therapy. Conclusion Compared with conventional oxygen therapy, conservative oxygen therapy can reduce the short-term mortality of severe patients, especially when the conservative oxygen therapy target is set above 90%. And it can also reduce the incidence of ICU new infections and ICUAW, while having no effect on 90-day mortality, ICU length of stay, and hospital length of stay.


Introduction
Oxygen therapy is a widely used treatment method in critically ill patients.Due to the patients' inspired oxygen concentration (FiO 2 ) often exceeding ambient oxygen concentration during ICU stay, they often reach an excessive arterial partial pressure of oxygen (PaO 2 ) level within 24 hours of admission [1,2].In this case, the hyperoxia can compensate and prevent tissue hypoxia by promoting oxygen delivery to afected organs [3].However, studies have shown that prolonged exposure to hyperoxemia can also be harmful.Hyperoxia may result in acute lung injury, atelectasis, or increased risk of infection due to oxidative stress and infammation.In addition, hyperoxia can result in vasoconstriction, reducing coronary blood fow and cardiac output, and alter microvascular perfusion [4,5].
In an observational study from the Netherlands, the inhospital mortality was found to be linearly associated with FiO 2 among ICU patients and U-shaped with PaO 2 (i.e., lower and higher PaO 2 were both associated with higher mortality), with both independent of each other [6].Terefore, the concept of conservative oxygen therapy strategy has been proposed.However, the opinions on the setting of oxygen therapy standards and optimal oxygenation goals are still inconsistent in various clinical guidelines [7][8][9].
In a randomized clinical trial of optimal oxygenation in ICU patients published in 2016, patients receiving oxygen therapy according to a conservative strategy (PaO 2 of 70-100 mmHg or arterial oxygen saturation (SpO 2 ) of 94-98%) have an improved ICU mortality compared with the conventional control group (PaO 2 up to 150 mmHg or SpO 2 97-100%) [10].Tis trial is the frst RCT to demonstrate the potential harm of conventional oxygen therapy.Some earlier observational studies also proved this point [11][12][13].
However, some similar randomized controlled studies afterwards found that conservative oxygen therapy did not improve patient survival [14,15].Terefore, uncertainty remains about optimal oxygenation goals for ICU patients.To address the limitations of previous analyses, we attempted to perform a meta-analysis by searching existing RCT studies to compare the impact of these two oxygen therapy strategies on mortality in critically ill patients.

Inclusion and Exclusion Criteria
2.1.1.Types of Studies.Te types of studies included the published RCT studies at home and abroad on the efect of conservative oxygen therapy and conventional oxygen therapy on the mortality of critically ill patients.Language is limited to Chinese and English.Wanfang) and RCT studies were collected that compared efects of conservative oxygen therapy and conventional oxygen therapy on the mortality of critically ill patients, with postdate before August 2022.Te search uses a combination of subject headings and free words and traces the references included in the literature to supplement the acquisition of the relevant literature.Chinese search terms include "氧疗," "病死率," and "重症患者"; English search terms include conservative oxygen, liberal oxygen, conventional oxygen, hypoxia, hyperoxia, oxygen defciencies, hypoxemia, anoxia, critical Illness, critical Care, intensive care units, and randomized controlled trial.
2.3.Literature Screening and Data Extraction.Two researchers independently screened the literature, extracted data, and cross-checked.If there were any diferences, they were resolved through discussion.For the literature lacking information, try to get in touch with the original author to supplement it.Te extracted data included ① basic information of included studies, including author's name and publication year; ② basic characteristics of research subjects, including sample size and patient type; ③ intervention measures including SpO 2 and PaO 2 levels of conservative oxygen therapy and conventional oxygen therapy; ④ key elements of risk of bias assessment; and ⑤ main data of outcome indicators concerned.

Risk of Bias Assessment of Included Studies.
Te risk of bias assessment of the included studies was assessed using the risk of bias assessment tool for RCTs recommended by the Cochrane Handbook version 5.1.0:① whether the randomization method was correct; ② whether the allocation was concealed; ③ whether subjects and investigators were blinded; ④ completeness of outcome data; ⑤ whether the results of the study were selectively reported; and ⑥ other sources of bias.Te risk of bias was assessed independently by 2 reviewers, and the results were crosschecked.In case of disagreements, they were resolved through discussion.
2.5.Statistical Analysis.RevMan 5.4 statistical software was used for meta-analysis.Te relative risk (RR) was used for enumeration data, and the standardized mean diference (SMD) was used for measurement data as efcacy analysis statistics.P < 0.05 was considered to be statistically signifcant.Te heterogeneity of the included studies was analyzed by the X 2 test (the test level was α � 0.1), and the I 2 statistic was used for evaluation.If the heterogeneity test result I 2 < 50%, a fxed-efects model is used for meta-analysis; if the heterogeneity test result I 2 ≥ 50%, it indicates that there is statistical heterogeneity among the results of each study, thus further analysis of heterogeneity is required, and metaanalysis was performed using a random-efects model after excluding signifcant clinical and methodological heterogeneity.Publication bias was assessed by drawing a funnel plot.

Search Result.
A total of 2099 related literature studies were retrieved.After reading the literature titles and abstracts, according to the inclusion and exclusion criteria, 10 RCT studies with a total of 5779 patients were fnally included.Te screening process is shown in Figure 1, and the basic characteristics of the included studies are shown in Table 1.

Quality Evaluation of Included
Literature.Te 10 included studies were all RCT studies, of which 7 were randomly generated by computer randomization scheme and 3 were generated by a random list.9 articles described allocation concealment, 2 were double-blind, and another 2 were single-blind.Te results of literature quality evaluation are shown in Sup 1.

Efects on Hospital Length of Stay.
Two studies [10,16] described the hospital length of stay, and there was no heterogeneity among the studies (P � 0.21, I 2 � 37%), and thus, a fxed-efects model was used for meta-analysis.Te results showed that there was no statistical signifcance regarding the diference in hospital length of stay between conservative oxygen therapy and conventional oxygen therapy groups (SMD � 0.05, 95% CI � −0.12-0.22,P � 0.54).Te results are shown in Sup 4.

Efect on Incidence of New ICU Organ Dysfunction.
Six studies [10,14,[17][18][19][20] provided data on incidence of new ICU organ dysfunction, including myocardial infarction, shock, liver and kidney failure, and intestinal ischemia.Heterogeneity was found among the studies (P � 0.005, I 2 � 70%), and thus, a random-efects model was used for meta-analysis.Te results showed that there was no statistical signifcance regarding the diference in incidence of new ICU organ dysfunction between conservative oxygen therapy and conventional oxygen (RR � 0.96, 95% CI � 0.83-1.12,P � 0.61).Te results are shown in Sup 5.

Canadian Respiratory Journal
Canadian Respiratory Journal for meta-analysis.Te results showed that conservative oxygen therapy had an advantage over conventional oxygen therapy in terms of the incidence of new ICU infections, and the diference was statistically signifcant (RR � 0.8, 95% CI � 0.66-0.98,P � 0.03).Te results are shown in Sup 6.

Efect on Incidence of ICUAW.
Two studies [18,19] described ICUAW, and there was no heterogeneity between studies (P � 0.84, I 2 � 0%); therefore, a fxed-efects model was used for meta-analysis.Te results showed that conservative oxygen therapy had an advantage over conventional oxygen therapy in terms of the incidence of ICUAW, and the diference was statistically signifcant (RR � 0.53, 95% CI � 0.29-0.94,P � 0.03).Te results are shown in Sup 7.

Publication Bias Results.
Te funnel plots of studies on short-term mortality in the included literature were asymmetric, indicating publication bias.Te results are shown in Sup 8.

Discussion
Tis study is a meta-analysis of the efect of conservative oxygen therapy and conventional oxygen therapy on the mortality of critically ill patients.Te results show that the conservative oxygen therapy can reduce the short-term mortality rate of critically ill patients compared with conventional oxygen therapy.Te advantage of conservative oxygen therapy over conventional oxygen therapy in terms of short-term mortality is statistically signifcant, especially when the conservative oxygen therapy target is set above 90%.In addition, the conservative oxygen therapy can also reduce the incidence of new ICU infections as well as the incidence of ICUAW, but there was no statistical signifcance regarding the diference in 90-day mortality, ICU length of stay, hospital length of stay, and incidence of new ICU organ dysfunction between the two groups.
Our fndings suggest an association between hyperoxemia and increased mortality in critically ill patients, consistent with the fndings of an observational study published in 2017, which found that patients with PaO 2 between 120 and 200 mmHg had lower mortality than patients with PaO 2 ≥ 200 mmHg, and the duration of hyperoxemia was positively correlated with in-hospital mortality [23].Now, a number of clinical studies have confrmed that hyperoxia can cause damage to the body, such as atelectasis and pulmonary interstitial fbrosis, increase the risk of lower respiratory tract infection, leading to lung damage [24], or cause coronary artery contraction, excite the vagus nerve, reduce cardiac output and myocardial blood supply, leading to myocardial damage [25], or induce apoptosis of normal brain tissue cells, causing repeated cerebral ischemia, resulting in brain tissue damage [26].Two recently published meta-analyses [2,27] showed that compared with open and conservative oxygen therapy, conservative oxygen therapy strategies can reduce mortality, which is consistent with the results of this meta-analysis.
In this meta-analysis, conservative oxygen therapy can improve short-term mortality in critically ill patients.However, comparing the 90-day mortality rate, the diference between the two was not statistically signifcant.Tis may be related to the fact that only one study included in this outcome indicator excluded patients with severe hypoxic respiratory failure.Our fndings do not support the use of conservative oxygen therapy in ICU patients with severe hypoxic respiratory failure.Teoretically, these patients have more severe gas exchange disturbances and refractory hypoxemia, requiring more intensive respiratory support [28].In addition, our study showed that there was no statistically signifcant diference between the two groups in terms of ICU length of stay, hospital length of stay, and incidence of new ICU organ dysfunction, and conservative   Canadian Respiratory Journal oxygen therapy did not signifcantly improve the overall prognosis of critically ill patients.Considering that the condition of ICU patients is critical and complex and the prognosis and outcome of patients are afected by the severity of the disease and various treatment methods, a single conservative oxygen therapy strategy has limited impact on the prognosis of patients.
Limitations of this study are as follows: ① the target value of SpO 2 set by conservative oxygen therapy is not uniform, and there is still a lack of high-quality evidence to defne conservative oxygen therapy strategies.② Te included populations are diferent, such as severe pneumonia, septic shock, and ARDS.③ Te treatment levels of the included literature studies vary.
In conclusion, this meta-analysis included a large number of domestic and foreign literature studies and a large number of cases, and the heterogeneity of each literature is low.Te analysis results show that compared with conventional oxygen therapy, the conservative oxygen therapy can reduce the short-term mortality rate of critically ill patients, as well as the incidence of ICU new infections and incidence of ICUAW.Of course, a large number of highquality RCT studies are still needed to be further confrmed in the future to provide more evidence-based medical evidence for clinical practice.

Figure 1 :
Figure 1: Literature screening process and result.

Figure 2 :
Figure 2: Meta-analysis of short-term mortality in diferent groups.
Primary outcome indicators included short-term mortality rate (28-day fatality rate or ICU fatality rate).Secondary outcome indicators included 90-day fatality rate, ICU length of stay, hospital length of stay, incidence of new organ dysfunction in ICU, incidence of new infections in ICU, and ICUAW incidence.

Table 1 :
Basic characteristics of included studies.