Zi-wei Ke, Yue Jiang, Ya-ping Bao, Ye-qin Yang, Xiao-mei Zong, Min Liu, Xiang-yun Guan, Zhong-qiu Lu

1 Department of Emergency Medicine, the First Affi liated Hospital of Wenzhou Medical University, Wenzhou 325000, China

2 Department of Nursing, Taizhou Hospital of Zhejiang Province, Taizhou 317000, China

3 Operating Room, the First Affi liated Hospital of Wenzhou Medical University, Wenzhou 325000, China

4 Faculty of Nursing, Wenzhou Medical University, Wenzhou 325000, China

5 Yiwu Hospital Affi liated to Wenzhou Medical University, Yiwu 322000, China

KEYWORDS: Mechanical ventilation; Hyperoxemia; Fraction of inspired oxygen; Arterial blood gas;Intensive care unit

INTRODUCTION

Oxygen (O2) is the basic element that maintains life and organ functions.Since its application in the 18thcentury,it has been one of the most commonly used therapeutic drugs in intensive care unit (ICU) patients, especially in mechanical ventilation (MV) patients.Adequate oxygen supplementation helps to avoid hypoxemia and improves patients’ clinical outcomes.However, excessive oxygen supplementation can induce hyperoxemia, which may cause potentially deleterious effects.[1,2]In a high-oxygen environment, the delicate balance between the production and counteraction of reactive oxygen species (ROS) is disturbed, and large amounts of ROS accumulate in cells,which may harm proteins, nucleic acids, and lipids, leading to cell death.[3,4]Ultimately, it may cause damage to organs,such as absorption atelectasis, pulmonary interstitial f ibrosis,respiratory infection, myocardial injury, and cerebral ischemia, and may even increase the in-hospital mortality of critically ill patients.[3,5-9]

Despite the potentially harmful eff ects of hyperoxemia,several previous studies[10-14]conducted in the Netherlands,Japan, and Australia showed that hyperoxemia was common in ICU MV patients.Due to the still sparse literature in China, the investigation of hyperoxemia management is required.Thus, we aim to conduct a retrospective study to provide more information about hyperoxemia management in ICU patients.

METHODS

We retrospectively screened the medical records of MV patients admitted to the two ICUs from January 1,2018, to December 31, 2018.Patients aged ≥18 years who required invasive MV for more than 24 hours were eligible for analysis.We excluded patients who met one of the following conditions: having a risk of imminent death by the attending medical team; undergoing cardiopulmonary bypass; requiring extracorporeal membrane oxygenation(ECMO); hyperbaric oxygen therapy; chronic obstructive pulmonary disease; requiring hyperoxia treatment (such as CO poisoning); paraquat poisoning; and transferred from another ICU with MV.For patients who were readmitted to the ICU and required MV for more than 24 hours, we considered the f irst admission only.

Data collection

We designed a case record form and collected data from each patient, including sex, age, reasons for ICU admission, types of admission (medical, unplanned surgery,planned surgery), and Acute Physiology and Chronic Health Evaluation II (APACHE II) score.All arterial blood gas(ABG) tested during MV was retrieved, and the following variables were recorded: arterial partial pressure of oxygen(PaO2), arterial oxygen saturation (SaO2), arterial carbon dioxide tension (PaCO2), and arterial blood pH.All MV settings, including positive end expiratory pressure (PEEP),the fraction of inspired oxygen (FiO2), were recorded as well.We also recorded ABG test time points, including the day hyperoxemia occurred after initiation of MV, the day hyperoxemia occurred after admission to the ICU, as well as working shifts (day shift 8:00-16:30, middle-night shift 16:30-23:00, night shift 23:00-8:00).In these two ICUs,ABG was routinely determined at 7:00 every morning, and when bedside clinicians thought that it was necessary to test.All ABG test results were reported to clinicians, and all adjustments of ventilator settings were recorded by nurses.

The incidence of hyperoxemia and intensivists’ response to hyperoxemia were the outcomes of interest.According to previous studies,[10-15]PaO2＞120 mmHg (1 mmHg=0.133 kPa) was defined as mild to moderate hyperoxemia,and ＞300 mmHg was defined as extreme hyperoxemia.Intensivists’ response to hyperoxemia was assessed based on the reduction of FiO2within one hour after hyperoxemia was recorded,[11]which was def ined as a positive response to hyperoxemia.Meanwhile, the increase in FiO2or the lack of FiO2adjustment within one hour was def ined as a negative response to hyperoxemia.

Statistical analysis

The data were analyzed by Statistical Package for the Social Sciences (SPSS) (version 14.0).Continuous and normally distributed variables were reported as mean (standard deviation [SD]), while Student’st-tests and analysis of variance were performed to compare the differences between groups.Non-normally distributed data were reported as median (interquartile range [IQR]).Mann-WhitneyU-tests were performed to compare the diff erences between the groups.Categorical variables were described asn(%).Chi-square test or Fisher’s exact test was performed to compare the differences between groups.Multivariable logistic regression analysis was performed to determine the independent factors associated with the intensivists’ response to hyperoxemia.Differences were considered statistically signif icant at a two-sidedP-value of 0.05.

RESULTS

Study population

Our study collected data from two general ICUs, which had 68 beds in total.During the study period, we screened the medical records of 2,742 patients who were admitted to these two general ICUs.Of the 606 eligible patients, 14 were excluded from analysis due to incomplete data, and thus 592 patients were finally analyzed.The baseline characteristics and oxygenation-related variables of the study patients are shown in Table 1.

Incidence of hyperoxemia

We retrospectively collected data on 592 patients’ABG during MV, and a total of 4,083 ABG values were retrieved.The median interval between two successive ABG values was 19.33 (7.75-24.25) hours.The incidence of hyperoxemia based on ABG results was as follows: 368(62.16%) patients had PaO2between 60 and 120 mmHg,50 (8.45%) patients had PaO2＜60 mmHg, 174 (29.39%)patients had PaO2＞120 mmHg, and 19 (3.21%) patients had extreme hyperoxemia at PaO2＞300 mmHg.

Intensivists’ response to hyperoxemia

Overall, under conditions of hyperoxemia, FiO2＜0.4 was found in 383 (64.70%) patients.When severe hyperoxemia occurred, FiO2was decreased in 228 (38.51%)patients.When mild to moderate hyperoxemia occurred,FiO2decreased only in 37 (6.25%) patients (P＜0.001).In particular, in cases of mild to moderate hyperoxemia with FiO2≤0.4, only 13 (2.20%) patients had a decrease in FiO2within one hour.

Factors related to intensivists’ response to hyperoxemia

A total of 393 (66.39%) patients experienced hyperoxemia.To determine the independent factors that may aff ect intensivists’ response to hyperoxemia, they were divided into two groups according to whether FiO2decreased within one hour when hyperoxemia was first observed.The first group was the positive-response group (FiO2was adjusted within one hour after hyperoxemia occurred),and the other was the negative-response group (FiO2increased or was not adjusted within one hour).Univariate analysis was performed to compare characteristics between the positive- and negative-response groups.The result showed that a positive response correlated with PaO2(191[152-243] mmHg vs.146 [130-170] mmHg,P＜0.001),FiO2(0.6 [0.5-0.9] vs.0.4 [0.4-0.5],P＜0.001), PEEP (5[5-6] cmH2O vs.5 [3-5] cmH2O,P=0.010), reasons for admission (P=0.080), working shifts (P=0.010), the day hyperoxemia occurred after initiation of MV (1 [1-1] day vs.2 [1-3] days,P＜0.001), and the day hyperoxemia occurred after admission to the ICU (1 [1-2] days vs.2 [1-4] days,P＜0.001) (Table 2).A multivariable logistic regression analysis was performed with variables that reached a signif icance threshold of 0.05 in univariate analysis.Among them, multivariable logistic regression analysis showed that a positive response was independently associated with FiO2(odds ratio [OR] 1.09, 95% conf idence interval [95%CI]1.06-1.12,P＜0.001), PaO2(OR1.01, 95%CI1.00-1.01,P=0.002), and working shifts (OR5.09, 95%CI1.87-13.80,P=0.001) (Table 3).

Table 1.Baseline characteristics and oxygenation-related variables of study patients

Table 2.Comparison of characteristics between the positive- and negative-response groups

Table 3.Multivariable logistic regression analysis of appropriate response

DISCUSSION

Key f indings

Hyperoxemia was frequently observed in MV patients in the ICUs.We found that most cases were not rectified by appropriate adjustment of FiO2, especially in the case of mild to moderate hyperoxemia, which indicated that a liberal oxygen strategy was prevalent in this cohort.In addition,the intensivists’ response to hyperoxemia was influenced by the FiO2level, PaO2level, and working shifts.Mild to moderate hyperoxemia, hyperoxemia with lower FiO2, and hyperoxemia occurring during the night and middle-night shifts were more likely to be ignored.

Relationship to previous f indings

A large retrospective study[10]in a Dutch ICU found that hyperoxemia (PaO2＞120 mmHg) occurred frequently(22%) in ICU MV patients, but FiO2was decreased in only 25% of the cases.When FiO2was ＜0.40, the adjustment rate of FiO2was lower (22%).Another observational study[16]reported that 74% of MV patients in ICUs were exposed to excessive FiO2for the median duration of 17 hours.Furthermore, a multicenter observational study[11]in Japan also reported that hyperoxemia was common in MV patients in the ICU, yet FiO2adjustments were infrequent in hyperoxemia cases.Our study revealed that hyperoxemia was commonly seen in the ICUs and neglected in most observations, which conf irmed the above-mentioned reports.All of these f indings consistently showed that liberal oxygen therapy was prevalent in ICU patients with MV.In our study,hyperoxemia with lower FiO2was more likely to be ignored,and this was consistent with the previous studies.[10,11]The possible reason may be that some intensivists think that low FiO2is safe.[14,17]Mild to moderate hyperoxemia was more likely to be ignored than extreme hyperoxemia because the harm caused by mild to moderate hyperoxia is controversial while the harm caused by extreme hyperoxemia has been confirmed.In addition, hyperoxemia occurring during the night and middle-night shifts was more likely to be ignored.The reason may be that intensivists working at night are more prone to fatigue and lack of work enthusiasm.[18]

Clinical implications

In 2008, the British Thoracic Society (BTS) published guidelines for oxygen use, insisting that hyperoxemia will cause potential harm to patients; therefore, hyperoxemia should also be avoided.In addition, they recommended a target oxygen saturation range of 94%-98% for acutely ill patients, which can help them avoid both hyperoxemia and hypoxemia.[19]In 2017, a clinical randomized controlled trial confirmed that a conservative protocol for oxygen therapy (maintain 70 mmHg≤PaO2≤100 mmHg or 94%≤SpO2≤98%), compared with conventional therapy(maintain PaO2≤150 mmHg or 97%≤SpO2≤100%), resulted in lower ICU mortality and lower occurrence of shock, liver failure, and bacteremia.[20]A meta-analysis published in 2018 by Chu et al[21]concluded that patients who received conservative oxygen administration had lower in-hospital mortality (relative risk [RR] 1.21, 95%CI1.03-1.43), 30-day mortality (RR1.14, 95%CI1.01-1.29), and had the longest follow-up mortality (RR1.10, 95%CI1.00-1.20).However,not all studies[22-23]have reached the same conclusion.Thus,effects of hyperoxemia are still controversial, and further research is needed in the future.

Limitations

Our study had several limitations.First, considering the fact that the research only included two ICUs, we did not know if our findings could be generated in other ICUs.However, it had a certain degree of external validity because our ICUs had all the typical characteristics of an ICU at a tertiary level—a general hospital in China.Second, the interval between each ABG analysis was too long.More frequent measurements might reflect current practices in more detail.Third, our study only included patients with invasive MV longer than 24 hours, which may result in a selective bias towards a cohort of patients with more severe sickness.However, optimal oxygen therapy procedures were particularly important in this cohort.Lastly, we didn’t get information on PaO2obtained prior to ICU admission.

CONCLUSIONS

Hyperoxemia occurs frequently and is neglected in most cases, particularly when mild to moderate hyperoxemia,hyperoxemia with lower FiO2, hyperoxemia during the night and middle-night shifts, or FiO2less likely to be decreased.Patients may be at a risk of oxygen toxicity because of the liberal oxygen strategy.Therefore, further research is needed to improve oxygen management for patients with MV in the ICUs.

ACKNOWLEDGMENT

The authors are gratefully thankful to Xiao-quan Xu,Chun-lei He, Xiao-dan Cai, Qiang-li Xie, Li-mei Li, Xin-yi Zhang for their support throughout the process of conducting this research study.

Funding:None.

Ethical approval:This study was approved by the hospital ethics committee, and the requirement for informed consent was waived.

Conf licts of interests:There is no conf lict of interest.

Contributors:All authors made an individual contribution to the writing of the article, including design, literature search, data acquisition, data analysis, statistical analysis, manuscript preparation,and manuscript editing.