Hyperoxia is associated with a greater risk for mortality in critically ill traumatic brain injury patients than in critically ill trauma patients without brain injury

Abstract

Objective:

The role of hyperoxia in patients with traumatic brain injury (TBI) remains controversial. The objective of this study was to determine the association between hyperoxia and mortality in critically ill TBI patients compared to critically ill trauma patients without TBI.

Design:

Secondary analysis of a multicenter retrospective cohort study.

Setting:

Three regional trauma centers in Colorado, USA, between October 1, 2015, and June 30, 2018.

Patients:

We included 3464 critically injured adults who were admitted to an intensive care unit (ICU) within 24 h of arrival and qualified for inclusion into the state trauma registry. We analyzed all available SpO2 values during the first seven ICU days. The primary outcome was in-hospital mortality. Secondary outcomes included the proportion of time spent in hyperoxia (defined as SpO₂ > 96%) and ventilator-free days.

Interventions:

None.

Measurements and Main Results:

In-hospital mortality occurred in 163 patients (10.7%) in the TBI group and 101 patients (5.2%) in the non-TBI group. After adjusting for ICU length of stay, TBI patients spent a significantly greater amount of time in hyperoxia versus non-TBI patients (p = 0.024). TBI status significantly modified the effect of hyperoxia on mortality. At each specific SpO₂ level, the risk of mortality increases with increasing FiO₂ for both patients with and without TBI. This trend was more pronounced at lower FiO₂ and higher SpO₂ values, where a greater number of patient observations were obtained. Among patients who required invasive mechanical ventilation, TBI patients required significantly more days of ventilation to day 28 than non-TBI patients.

Conclusions:

Critically ill trauma patients with a TBI spend a greater proportion of time in hyperoxia compared to those without a TBI. TBI status significantly modified the effect of hyperoxia on mortality. Prospective clinical trials are needed to better assess a possible causal relationship.

Keywords

Critical care combat hyperoxia intensive care unit supplemental oxygen trauma traumatic brain injury

Introduction

Clinical variation in oxygenation for traumatic brain injury (TBI) protection may be associated with both protection and exacerbation of neuronal injury and associated neurological outcomes. TBI is characterized by initial local and/or regional trauma with edema followed by secondary injuries resulting from subsequent inflammatory reactions.¹ Hypoxia, hypotension, and hypothermia are key drivers of such secondary injury and are the targets of many therapeutic interventions.² Oxygen administration can be a lifesaving maneuver with undisputed importance to minimize the risk of hypoxia and associated secondary injury in TBI and other non-TBI critically injured patients.^3,4 However, the administration of oxygen in excess of physiologic needs is common and frequently leads to hyperoxia (oxygen saturation [SpO₂] > 96% or arterial partial pressure of oxygen [Pao₂] > 100 mmHg).^5–9 Supraphysiologic levels of oxygen in medical intensive care unit (ICU) patients may be associated with increased morbidity and mortality.^7,8,10,11 Our group has demonstrated that hyperoxia is both common and associated with a greater risk for mortality in critically ill trauma patients.¹²

The effects of hyperoxia in TBI patients, however, remain controversial. Recent studies have yielded conflicting conclusions. Although the primary outcomes of interest varied widely (neurological outcomes, mortality, or acute lung injury), some studies supported the avoidance of hypoxia, some supported hyperoxia, and others were neutral.¹³ For example, early hyperoxia in severe trauma patients, including those with TBI, may be associated with reduced in-hospital mortality.¹⁴ Hyperbaric hyperoxia may also decrease intracranial pressure, improve cerebral oxygenation, and reduce mortality in severe TBI patients.¹⁵ Similar improvements in 6-month neurological function have been observed with normobaric hyperoxia.¹⁶ However, other studies suggest early hyperoxia may increase mortality in TBI patients depending on the threshold chosen for hyperoxia (Pao₂> 150 mmHg to Pao₂> 487 mmHg).^17–20 These studies are problematic due to a lack of a standard definition for hyperoxia and a focus on only mechanically ventilated patients. Further, none of these studies accounted for cumulative oxygen exposure.¹² Therefore, limited evidence exists regarding hyperoxia in TBI patients.

In this study, we report the results of a secondary analysis of a large multicenter cohort of critically injured patients¹² to compare critically ill trauma patients with and without TBI. The objective was to compare time spent in hyperoxia and to determine whether hyperoxia is more strongly associated with mortality in critically ill trauma patients with and without TBI. We hypothesized that supplemental oxygen use and associated hyperoxia would be higher in critically ill trauma patients with TBI compared to non-TBI trauma patients. Furthermore, we hypothesized that time in hyperoxia would be associated with increased mortality in TBI patients to a greater extent than in trauma patients without TBI.

Materials and methods

Study design

We performed a secondary analysis of a large multicenter cohort study of critically injured trauma patients¹² to evaluate the association between hyperoxia and supplemental oxygen use and in-hospital mortality in critically injured patients both with and without TBI. Subjects were identified from three regional trauma centers affiliated with the University of Colorado Health (UCHealth) system. These included two level I trauma centers (University of Colorado Hospital, Aurora, CO, and Memorial Hospital Center, Colorado Springs, CO) and one level II trauma center (Medical Center of the Rockies, Loveland, CO). All share a common reporting structure to the Colorado Trauma Registry and use the same electronic health record system (Epic, Verona, WI). Our study was reviewed and approved by the Colorado Multiple Institutional Review Board (COMIRB) #17-1359.

Study sample

Critically ill trauma patients admitted to the ICU within 24 h of admission at a participating hospital between October 1, 2015, and June 30, 2018, were included in this study if they also had an ICD-10 code for acute injury/trauma and met criteria for entry into the Colorado state trauma registry. Patients without an injury code from the trauma registry definitions were excluded. Subjects were further divided into those with a TBI and those without a TBI according to ICD-10 codes (Supplementary Table 1). ICD-10 codes for TBI screenings (Z13.850) or personal history of TBI (Z87.820) were not included in the definition of TBI.

Outcomes and variables collected

Details of individual variable definitions and data collection methods are described in our previously published cohort study.¹² Briefly, we collected all measurements of fraction of inspired oxygen (FiO₂) and SpO₂ from the time of presentation in the emergency department through the first 7 days in the ICU. Measured covariates included: age, race/ethnicity, sex, smoking status, Acute Physiology and Chronic Health Evaluation (APACHE) II score, mechanism of injury, patient comorbidities (presence of a cardiovascular comorbidity, presence of a chronic pulmonary comorbidity, and the number of other comorbidities), and time of intubation and extubation (if applicable). Cardiovascular comorbidities included congestive heart failure, valvular disorders, pulmonary circulation disorders, peripheral vascular disease, and coronary artery disease. Chronic pulmonary disorders included chronic obstructive pulmonary disorder, asthma, interstitial lung disease, and fibrotic lung disease (Supplemental Table 2). We categorized oxygenation status based on a modified Delphi approach agreed upon by a panel of 31 international military and civilian experts in trauma surgery, emergency medicine, critical care, and military operational medicine as hypoxia (SpO₂< 88%), borderline (SpO₂ 88–89%), normoxia (SpO₂ 90–96%), and hyperoxia (SpO₂> 96%).²¹ The primary outcome was in-hospital mortality. Secondary outcomes included: the proportion of time spent in hyperoxia (defined as SpO₂ > 96% and either on supplemental oxygen or with FiO₂> 21%) and ventilator-free days (VFD) to day 28.

Statistical analysis

We calculated summary statistics for all variables overall and stratified by TBI status. Differences in patient characteristics were analyzed using chi-squared tests for categorical data and either unequal variance t-tests or Mann–Whitney tests for continuous data, as appropriate. The primary outcome, in-hospital mortality, was analyzed using a Cox proportional hazards regression model with time-dependent covariates. TBI status was the primary predictor of interest. All available oxygenation variables (SpO₂ and FiO₂) were evaluated longitudinally for the first seven days of the ICU stay. Linear and quadratic terms were used for both SpO₂ and FiO₂, along with their interaction, to accommodate a nonlinear relationship between these variables and the risk of mortality.¹² Because of the difficulty of interpreting coefficients in a nonlinear interaction model, graphical summaries of the relationship between the oxygenation variables and risk of mortality were assessed using risk scores. These risk scores are calculated as the ratio of the exponentiated linear predictor with specified values of the oxygenation variables and reference values for all other covariates to the exponentiated linear predictor for a hypothetical patient with mean values for all predictors included in the model.

We conducted a secondary analysis of VFD. Subjects who died were assigned to a different category than those with zero VFD. Our first analysis employed a Wilcoxon test to determine whether a significant difference in the distribution of VFD exists between TBI and non-TBI patients due to the highly non-normal nature of this outcome variable. For covariate-adjusted modeling, we analyzed ventilator days (i.e. number of days on a ventilator while in hospital for each patient who survived hospital discharge) among survivors using a hurdle model with negative binomial outcome distribution. Hurdle models treat zero responses differently from nonzero responses and involve specifying a logistic model for the probability of response exceeding zero and a separate truncated count model for those responses that do exceed zero.²² TBI status was included as a predictor in these models, as well as primary exposure variables, percentage of time in hyperoxia (SpO₂> 96%), and percentage of time on room air (FiO₂ = 21%). Interactions between TBI status and these exposure variables were included to test for differential effects between TBI and non-TBI patients of oxygenation on mean ventilator days.

Another secondary analysis was performed to assess how the percentage of time spent in hyperoxia in the first 7 days of hospitalization might depend on TBI status. For these models, the data were reformatted by ventilation status so that each patient contributed up to two rows of data, one for time spent mechanically ventilated and one for time spent not mechanically ventilated. However, if a patient spent the entirety of their first 7 days in the hospital in one or the other of these states, there would only be a single row of data for that patient. We used a generalized estimating equations approach to fit regression models for the dependence of the proportion of time spent in hyperoxia on explanatory variables, including TBI status and ventilation status. This approach was adopted to address issues with the outcome distribution, which is a proportion (bounded between 0 and 1), and with correlation within patients (due to having repeated observations for those who spent time both on and off of mechanical ventilation). The identity link function was used in order to maintain interpretations of regression coefficients as mean differences in the proportion of time spent hyperoxic. TBI status was again the primary predictor of interest, while ventilation status was included in the model as a covariate.

We adjusted all regression models for age, race/ethnicity, sex, smoking status, APACHE II score, presence of cardiopulmonary comorbidities, number of other comorbidities, and MOI based on the results of univariate analyses (Supplemental Tables 1–3). Analyses were performed in R (Vienna, Austria)²³ with the exception of the proportion of time hyperoxia analysis, with was performed with SAS software (v 9.4 SAS Institute, Cary, NC) and PROC GENMOD. Estimation of all Cox models was performed with the survival package v.2.44.1.1 in R v3.5.2.²⁴ Marginal risk scores were calculated using the ggeffects package v.0.9.0 in R v3.5.2.²⁵ Hurdle models were estimated using the pscl package (v.1.5.5 (32,33).

Results

Patient characteristics

Out of 3464 individuals, 1524 (44%) were admitted with either a primary or secondary diagnosis of TBI (Table 1). No statistically significant differences in age, sex, race, or ethnicity were noted between TBI and non-TBI patients. The median APACHE II scores were similar between TBI and non-TBI groups. Patients with a TBI were more likely to have a fall (49.6% vs. 24.8%) or motor vehicle collision (36.9% vs. 26.1%) as the mechanism of injury and less likely to have penetrating trauma (3.9% vs. 10.2%). The proportion of patients requiring mechanical ventilation outside of the operating room at any point during their hospitalization (40.9% vs. 41.2%) was similar between groups.

Table 1.

Characteristics of critically ill trauma patients stratified by TBI status.

Characteristics	Non-TBIn = 1940 (56.0%)	TBIn = 1524 (44.0%)	Overalln = 3464	p-Value
Demographics, n (%)
Age, years; mean (SD)	51.7 (20.4)	53.7 (21.7)	52.6 (21.0)	<0.01
Male	1237 (63.8)	1026 (67.3)	2263 (65.3)	0.03
Race/ethnicity
Non-Hispanic white	1278 (65.9)	1068 (70.1)	2346 (67.7)	<0.01
Hispanic	314 (16.2)	233 (15.3)	547 (15.8)
Non-Hispanic black	187 (9.6)	93 (6.1)	280 (8.1)
Other	161 (8.3)	130 (8.5)	291 (8.4)
Insurance status
Private	621 (32.0)	534 (35.0)	1155 (33.3)	<0.01
Medicare	603 (31.1)	519 (34.1)	1122 (32.4)
Medicaid	618 (31.9)	386 (25.3)	1004 (29.0)
Other	18 (0.9)	13 (0.9)	31 (0.9)
Missing	80 (4.1)	72 (4.7)	152 (4.4)
Medical History, n (%)
Comorbidities
Cardiovascular	477 (24.6)	378 (24.8)	855 (24.7)	0.88
Pulmonary	403 (20.8)	208 (13.6)	611 (17.6)	<0.01
Amount of other—median (IQR)	2 (1–4)	3 (1–4)	3 (1–4)	<0.01
Tobacco Use
Current or former smoker	947 (48.8)	542 (35.6)	1489 (43.0)	<0.01
Never smoker	666 (34.3)	560 (35.6)	1226 (35.4)
Not reported/missing	327 (16.9)	422 (27.7)	749 (21.6)
Alcohol use	840 (43.3)	635 (41.7)	1475 (42.6)	0.33
Other substance use	440 (22.7)	202 (13.3)	642 (18.5)	<0.01
Clinical
EMS arrival	1304 (67.2)	1221 (80.1)	2525 (72.9)	<0.01
Mechanism of injury
Fall	482 (24.8)	753 (49.6)	1235 (35.7)	<0.01
Motor vehicle	507 (26.1)	562 (36.9)	1069 (30.9)
Penetrating	198 (10.2)	60 (3.9)	258 (7.4)
Other	753 (38.8)	149 (9.8)	902 (26.0)
Ever mechanically ventilated	793 (40.9)	628 (41.2)	1421 (41.0)	0.84
APACHE II score—median (IQR)	7 (5–13)	8 (6–16)	8 (5–14)	<0.01
GCS score
3–8	340 (17.5)	436 (28.6)	776 (20.4)	<0.01
9–12	204 (10.5)	153 (10.0)	357 (10.3)
13–5	1371 (70.7)	930 (61.0)	2301 (66.4)
Not reported/missing	25 (1.3)	5 (0.3)	30 (0.9)
LOS—median (IQR)	6 (3–12)	5 (3–10)	6 (3–11)	<0.01

APACHE II, Acute Physiology and Chronic Health Evaluation II; IQR, interquartile range; LOS, length of stay in days; SD, standard deviation; TBI, traumatic brain injury.

Primary outcome

The primary outcome (in-hospital mortality) occurred in 163 patients (10.7%) in the TBI group and 101 patients (5.2%) in the non-TBI group. Cox regression modeling revealed a significant effect modification of TBI status for the nonlinear interaction between SpO₂ and FiO₂ (p = 0.005) (Figure 1). The risk for mortality was higher for TBI patients with hyperoxia than non-TBI patients with hyperoxia, particularly at lower FiO₂. At each specific SpO₂ level, both for patients with and without TBI, the risk of mortality increases with increasing FiO₂. This trend was observed at all FiO₂ and SpO₂ levels but was more pronounced at lower FiO₂ and higher SpO₂ values where a greater number of patient observations were available.

Figure 1.

Mortality risk at given SpO₂ and FiO₂ by TBI status. (A) Heatmap of risk scores for mortality at each pulse oximetry (SpO₂) and fraction of inspired oxygen (FiO₂) level stratified by TBI status. For each SpO₂ level, greater FiO₂ increases risk of mortality. Greater FiO₂ increased the risk for mortality in TBI patients to a greater extent than in non-TBI trauma patients. (B) Heatmap of risk differences between patients with and without TBI. Circles in the bottom right panel represent the number of patient-hour observations associated with each SpO₂ and FiO₂ level. The greatest proportion of observations was obtained at higher SpO₂ and lower FiO₂ values. All risk scores are relative to a patient with average values of all covariates. Therefore, a score of 1 implies an equal risk to the average, a score greater than 1 implies elevated risk, and a score less than 1 implies reduced risk.

Secondary outcomes

Of the 171,295 patient-hours recorded for TBI patients, 84,277 (49.2%) were spent in hyperoxia (SpO₂ > 96%) and 84,812 (49.5%) in normoxia (SpO₂ 90–96%) (Table 2). Non-TBI patients spent 101,139 h (44.0%) in hyperoxia and 124,814 h (54.3%) in normoxia (Table 2). Very little time was spent in either borderline (SpO₂ 88–89%) or hypoxia (SpO₂ < 88%) categories for both groups. After adjusting for ICU length of stay and mechanical ventilation status, critically ill TBI patients spent significantly more time in hyperoxia than critically ill non-TBI patients (p = 0.024).

Table 2.

Outcomes by TBI status.^a

Outcome	Non-TBI	TBI	p-Value
Clinical outcomes:
In-hospital mortality, n (%)	101 (5.2)	163 (10.7)	<0.01
Ever mechanically ventilated, n (%)	793 (40.9)	628 (41.2)	0.84
Ventilator-free days to day 28—median (IQR)	28 (26–28)	28 (24–28)	0.04
Oxygenation outcomes:
Time spent at oxygenation levels, patient-hours (%)
Hyperoxia (SpO₂ > 96%)	101,139 (44.0)	84,277 (49.2)	<0.01
Normoxia (SpO₂ 90–96%)	124,814 (54.3)	84,812 (49.5)
Borderline hypoxia (SpO₂ 88–89%)	1563 (0.7)	911 (0.5)
Hypoxia (SpO₂< 88%)	2370 (1.0)	1484 (0.9)
Time spent at room air, patient-hours (%)	76,515 (33.3)	65,229 (38.0)	<0.01
Time spent in hyperoxia and receiving supplemental oxygen, hours (%)	78,114 (34.0)	64,863 (37.8)	<0.01

IQR, interquartile range; SpO₂, oxygen saturation; TBI, traumatic brain injury.

P-values in this table represent unadjusted tests of proportions or Mann–Whitney tests. These may differ from p-values of the adjusted analyses presented in the main text.

Critically ill TBI patients experienced a median of 28 VFD (IQR 24–28) to day 28, while critically ill non-TBI trauma patients experienced a median of 28 VFD (IQR 26–28) to day 28 (Table 2). Unadjusted analysis indicated the distribution of VFD differed between TBI and non-TBI patients (p = 0.04) (Figure 2). Adjusted analyses with the hurdle model revealed patients with TBI were not significantly more likely than patients without TBI ever to require invasive mechanical ventilation (p = 0.086) (Supplemental Table 3). However, among patients who did require invasive mechanical ventilation, TBI patients experienced a significantly greater duration of ventilation to day 28 than non-TBI patients (p < 0.001).

Figure 2.

Proportion of patients with a given number of VFD. Proportion of patients with a given number of VFD among TBI (blue) and non-TBI (red) patients. VFDs for TBI and non-TBI patients were significantly different (p = 0.039).

Discussion

In this analysis of a large multicenter cohort of critically injured patients, hyperoxia was more common in TBI patients, and the association between hyperoxia and risk for mortality is significantly modified by TBI status. After adjusting for time spent in hyperoxia, critically ill TBI patients also experienced a greater duration of invasive mechanical ventilation to day 28 than their non-TBI counterparts.

Exposure to supraphysiologic levels of oxygen remains common for both non-TBI and TBI patients. However, in our cohort, patients with TBI spent a greater proportion of their time in hyperoxia when compared to patients without TBI. The clinical reasoning and decision-making processes that lead to these differences remain unclear. Clinicians may wish to minimize exposure to hypoxia in TBI patients and thereby mitigate the potentially detrimental effects of cerebral ischemia on brain recovery.²⁶ The use of cerebral oxygenation monitoring, such as near-infrared spectroscopy (NIRS) and continuous brain tissue oxygenation (PbtO2), may guide clinical decisions regarding oxygen titration. However, cerebral oxygenation monitoring was used infrequently at the three included institutions during the study period. Additionally, clinicians may be actively targeting higher SpO₂ based on ongoing clinical trials evaluating the role of hyperbaric hyperoxia.²⁷ Further data are required to elucidate the source of the observed difference in the proportion of time spent in hyperoxia.

Of note, the risk for mortality was greater in non-TBI patients than in TBI patients at high FiO₂ and low SpO₂ values (Figure 1B). However, very few patient observations were recorded within these parameters (FiO₂> 80% and SpO₂< 90%), likely because such a clinical scenario occurs infrequently. We, therefore, caution against overinterpretation and dissemination of the risk difference between TBI and non-TBI patients in this setting (i.e. the right lower corner of Figure 1B).

One distinguishing feature of our study is the continuous assessment of oxygen in terms of both dose and duration of oxygen exposure. Prior studies have sought to establish a safe “upper limit” of hyperoxia for TBI patients. Rincon et al. demonstrated that mechanically ventilated TBI patients with a Pao₂≥ 300 mmHg experienced a 50% increase in in-hospital mortality compared to normoxic TBI patients (odds ratio [OR] 1.5, 95% confidence interval [CI] 1.02–2.4).²⁰ Davis et al. found mortality for TBI patients increased significantly above a Pao₂ of 487 mmHg (OR 0.5, 95% CI 0.36–0.71).¹⁸ Notably, neither of these studies assessed cumulative oxygen exposure but instead chose to measure a single arterial blood gas in the first 24 h of ICU admission²⁰ or Pao₂ on arrival.²⁸ These isolated measurements of oxygenation may inaccurately represent the cumulative oxygenation exposure. Additionally, choosing one “upper limit” threshold oversimplifies oxygen titration in clinical practice. Patients rarely, if ever, spend their time in the ICU with a consistent oxygen saturation or partial pressure of oxygen. Rather, oxygen is titrated throughout the hospital stay, depending upon clinical fluctuations. Assessing the time spent at a given oxygen exposure, therefore, represents a key component of both this study and our previous work.¹² Our study also included both mechanically ventilated and nonmechanically ventilated TBI patients. This is important given the inherent resistance at many institutions to down-titrating mechanically ventilated patients to FiO₂ of 21% due to weaning protocols that often stop down-titration at 30–40%.

Duration of exposure likely modifies the effects of hyperoxia in TBI patients, as it does in other populations.¹² However, hypoxia in TBI is also associated with increased morbidity and mortality.^1,2 The results of our study and others may leave clinicians feeling a need to “thread the needle” with regard to oxygenation in TBI patients. That is, strictly avoid both hypoxia and hyperoxia simultaneously. Our group demonstrated in a pilot study that targeted education and best practice advisories could significantly improve supplemental oxygen titration in critically ill trauma patients.²⁹ Critical care providers were responsive to these changes and readily accepted the down-titration of oxygen in patient care. Recent awareness of the potential harms associated with hyperoxia in critically ill patient populations made this paradigm shift easier to attain. Future work may focus on provider views on oxygenation in TBI patients and the potential factors that may limit willingness to reduce supplemental oxygen use in these patients.

This study should be interpreted in the context of several important limitations due to the nature of the disease of interest and the selected primary outcome. First, the study was retrospective in nature. The primary outcome was uncommon—only 7.6% of patients died during hospitalization. Our large cohort, almost 3500 patients, partially mitigates this limitation, as power in survival analyses of the type we performed derives from the number of deaths rather than the proportion. We did not control for prehospital care and oxygenation practices which may have influenced the primary outcome. Establishing a causal connection between FiO₂/SO₂ and mortality is statistically challenging because they both influence each other via different pathways. We accounted for this by employing well-validated methods for FiO₂/SO₂ previously used by our group.^12,29,30 We used a previously identified, optimal range of SpO₂ (90–96%) to define normoxia, with hyperoxia defined as SpO₂ > 96%. However, the precise definition of normoxia and hyperoxia in critically ill trauma patients and, even more specifically, TBI patients is unknown. This study provides a basis to suggest that targeting SpO₂ 90–96% may improve patient outcomes in the TBI population. We are evaluating this approach in an ongoing multicenter trial to identify precise optimal oxygenation targets in critically ill trauma patients (NCT #04534959).³⁰ Finally, higher elevation may affect oxygen delivery and oxygenation targets. All three of the included hospitals are located more than 1500 m above sea level. Therefore, it is possible our findings may not be replicated at lower elevations.

In conclusion, this large multicenter cohort study demonstrated that hyperoxia is both more common and associated with a greater risk for mortality in critically ill TBI patients compared to critically ill non-TBI trauma patients. Critically ill patients with TBI spent a significantly greater proportion of time in hyperoxia compared to critically ill non-TBI trauma patients. Prospective studies are needed to elucidate the mechanisms underlying these differences, inform the clinical decision-making processes driving the increased time TBI patients spend in hyperoxia, and create interventions to target normoxia in this patient population.

Supplemental Material

sj-docx-1-sci-10.1177_00368504231160416 - Supplemental material for Hyperoxia is associated with a greater risk for mortality in critically ill traumatic brain injury patients than in critically ill trauma patients without brain injury

Supplemental material, sj-docx-1-sci-10.1177_00368504231160416 for Hyperoxia is associated with a greater risk for mortality in critically ill traumatic brain injury patients than in critically ill trauma patients without brain injury by David J Douin, Layne Dylla, Erin L Anderson, John D Rice, Conner L Jackson, Vikhyat S Bebarta, Robert T Neumann, Steven G Schauer and Adit A Ginde in Science Progress

Footnotes

Acknowledgments

The authors thank Caroline M. Ledbetter, MPH for her work with preliminary data analysis.

Author contributions

Study concept and design: DJD, LD, and AAG; Acquisition of data: ELA, AAG; Statistical analysis: JDR, CLJ; Interpretation of data: All authors; Drafting of the manuscript: DJD, LD; Critical revision of the manuscript for important intellectual content: All authors.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. This manuscript expresses the authors’ opinions and does not reflect the policy or opinions of the Department of the Army, Department of the Air Force, Department of Defense, or US Government.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the U.S. Department of Defense/SOCOM (grant number W81XWH-17-C-0241).

ORCID iDs

David J Douin

Conner L Jackson

Supplemental material

Supplemental material for this article is available online.

Author biographies

David J Douin, MD is an assistant professor of anesthesiology at the University of Colorado School of Medicine. His research focuses on epidemiological and clinical trials of critically ill patients, particularly on oxygen delivery, transfusion, and injured/trauma patients. He has been the site Co-PI of two trials assessing the effects of hyperoxemia in critically injured trauma (NCT04534959) and burn (NCT04534972) patients. National leadership in clinical trials networks includes the NIH Accelerating COVID-19 Therapeutic Interventions and Vaccines (ACTIV)-3/4 and the CDC-funded Investigating Respiratory Viruses in the Acutely Ill (IVY). Funding for his research includes NIH (NIGMS), DoD, and CDC, and he has published over 45 original research articles in leading medical journals.

Layne Dylla, MD, PhD is an assistant professor of emergency medicine at the University of Colorado School of Medicine. Her research interests include the emergency and pre-hospital treatment of acute ischemic stroke. She has published over 15 original research articles in leading medical journals.

Erin L Anderson, RN is a research services program manager in the department of emergency medicine at the University of Colorado School of Medicine. She is the lead program manager for the SAVE-O2 clinical trials and related projects.

John D Rice, PhD is a research assistant professor in the Department of Biostatistics in the University of Michigan School of Public Health. He received his MSPH in Biostatistics from Emory University in 2010, and his PhD in Biostatistics from the University of Michigan in 2015, where his dissertation focused on statistical methods for cancer research. He completed postdoctoral training at the University of Rochester in 2017, where he worked in the areas of HIV testing behavior and cardiovascular outcomes. Prior to joining the University of Michigan, Dr. Rice was an assistant professor in the Department of Biostatistics and Informatics in the Colorado School of Public Health from 2018–2022. His research interests include longitudinal data analysis, recurrent events, and semiparametric regression methods for binary and semicontinuous outcomes data.

Conner L Jackson, MS is an instructor of biostatistics and informatics at the University of Colorado Center for Innovative Design and Analysis (CIDA)

Vikhyat S Bebarta, MD is a professor of emergency medicine at the University of Colorado School of Medicine and director of the CU Anschutz Center for Medicine and Battlefield (COMBAT) Research.

Robert T Neumann, MD is an associate professor of neurosurgery and director of the Neuro-Intensive Care Unit at the University of Colorado School of Medicine. His research interests include traumatic brain injury, and the care of critically ill patients with neurologic injuries or disorders.

Steven G Schauer, DO, MS is an emergency medicine physician at the US Army Institute of Surgical Research, JBSA Fort Sam Houston, in San Antonio TX. His research interests include traumatic injury, critical illness, and oxygenation targets.

Adit A Ginde, MD, MPH is a professor of emergency medicine at the University of Colorado School of Medicine. His research focuses on epidemiological and clinical trials in the acute care setting, with a particularly focus on early care of critically ill and injured patients. Specifically, he has extensive experience with designing, overseeing, and implementing clinical trials in a variety of clinical environments including emergency department, inpatient (including intensive care unit), perioperative, and outpatient settings, particularly in the areas of acute respiratory failure and sepsis/infection (including COVID-19). He has been the lead investigator for 6 large multicenter clinical trials and participated in the design (as protocol committee member and co-investigator) and implementation (as site PI) of over a dozen other clinical trials. National leadership in clinical trials networks include the NHLBI Prevention and Early Treatment of Acute Lung Injury (PETAL) network executive and steering committees and steering committees for the NHLBI Collaborating Network of Networks for Evaluating COVID-19 and Therapeutic Strategies (CONNECTS), the trans-NIH Accelerating COVID-19 Therapeutic Interventions and Vaccines (ACTIV)-3, the CDC-funded Influenza and Other Viruses in the Acutely Ill (IVY), the DoD-funded Pragmatic Critical Care Research Group (PCCRG), and the NIAID Strategies and Treatments for Respiratory & Viral Diseases (STRIVE). Funding for his research includes NIH (including NHLBI, NCATS, NIAID, NINDS, NIA), DoD, CDC, foundation, and industry sponsors, and I have published over 240 original research articles in leading medical journals.

References

Pélieu

Kull

Walder

. Prehospital and emergency care in adult patients with acute traumatic brain injury. Med Sci (Basel) 2019; 7. doi: 10.3390/medsci7010012

Tohme

Delhumeau

Zuercher

, et al. Prehospital risk factors of mortality and impaired consciousness after severe traumatic brain injury: an epidemiological study. Scand J Trauma Resusc Emerg Med 2014; 22: 1.

McMullan

Rodriquez

Hart

, et al. Prevalence of prehospital hypoxemia and oxygen use in trauma patients. Mil Med 2013; 178: 1121–1125.

Wetterslev

Meyhoff

Jørgensen

, et al. The effects of high perioperative inspiratory oxygen fraction for adult surgical patients. Cochrane Database Syst Rev 2015. doi: 10.1002/14651858.CD008884.pub2

Kilgannon

Jones

Shapiro

, et al. Association between arterial hyperoxia following resuscitation from cardiac arrest and in-hospital mortality. JAMA 2010; 303: 2165–2171.

Smit

Smulders

de Waard

, et al. Moderate hyperoxic versus near-physiological oxygen targets during and after coronary artery bypass surgery: a randomised controlled trial. Crit Care 2016; 20: 55.

Panwar

Hardie

Bellomo

, et al. Conservative versus liberal oxygenation targets for mechanically ventilated patients. A pilot multicenter randomized controlled trial. Am J Respir Crit Care Med 2016; 193: 43–51.

Girardis

Busani

Damiani

, 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.

Suzuki

Eastwood

Peck

, et al. Current oxygen management in mechanically ventilated patients: a prospective observational cohort study. J Crit Care 2013; 28: 647–654.

10.

Suzuki

Eastwood

Glassford

, et al. Conservative oxygen therapy in mechanically ventilated patients: a pilot before-and-after trial. Crit Care Med 2014; 42: 1414–1422.

11.

Johnson

Dodampahala

Rosselot

, et al. The association between arterial oxygen tension and neurological outcome after cardiac arrest. Ther Hypothermia Temp Manag 2017; 7: 36–41.

12.

Douin

Anderson

Dylla

, et al. Association between hyperoxia, supplemental oxygen, and mortality in critically injured patients. Crit Care Explor 2021; 3: e0418.

13.

Douin

Schauer

Anderson

, et al. Systematic review of oxygenation and clinical outcomes to inform Oxygen targets in critically Ill trauma patients. J Trauma Acute Care Surg 2019. doi: 10.1097/TA.0000000000002392

14.

Baekgaard

Abback

Boubaya

, et al. Early hyperoxemia is associated with lower adjusted mortality after severe trauma: results from a French registry. Crit Care 2020; 24: 604.

15.

Rockswold

Zaun

, et al. A prospective, randomized phase II clinical trial to evaluate the effect of combined hyperbaric and normobaric hyperoxia on cerebral metabolism, intracranial pressure, oxygen toxicity, and clinical outcome in severe traumatic brain injury. J Neurosurg 2013; 118: 1317–1328.

16.

Taher

Pilehvari

Poorolajal

, et al. Effects of normobaric hyperoxia in traumatic brain injury: a randomized controlled clinical trial. Trauma Mon 2016; 21: e26772.

17.

Raj

Bendel

Reinikainen

, et al. Hyperoxemia and long-term outcome after traumatic brain injury. Crit Care 2013; 17: R177.

18.

Davis

Meade

Sise

, et al. Both hypoxemia and hyperoxemia may be detrimental in patients with severe traumatic brain injury. J Neurotrauma 2009; 26: 2217–2223.

19.

Brenner

Stein

, et al. Association between early hyperoxia and worse outcomes after traumatic brain injury. Arch Surg 2012; 147: 1042–1046.

20.

Rincon

Kang

Vibbert

, et al. Significance of arterial hyperoxia and relationship with case fatality in traumatic brain injury: a multicentre cohort study. J Neurol Neurosurg Psychiatry 2014; 85: 799–805.

21.

Schauer

Anderson

Cunningham

, et al. Consensus-based recommendation for oxygenation targets in critically injured patients. Presented at Special Operations Medical Association Scientific Assembly; May 9, 2019, http://wwwspecialoperationsmedicineorg/Documents/2019%20SOMSA/Handouts/THURS-Research%20Schauer%20-%20Oxygen%20consensuspdf. 2021.

22.

Zeileis

Kleiber

Jackman

. Regression models for count data in R. J Stat Softw 2008; 27. doi: 10.18637/jss.v027.i08

23.

RCoreTeam. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. https://www.R-project.org/.

24.

Therneau

. A Package for Survival Analysis in R. R package version 3.2-3. https://CRAN.R-project.org/package = survival.

25.

Ludecke

. ggeffects: tidy data frames of marginal effects from regression models. J Open Source Softw 2018; 3(26): 772. doi: 10.21105/joss.00772

26.

Chi

Knudson

Vassar

, et al. Prehospital hypoxia affects outcome in patients with traumatic brain injury: a prospective multicenter study. J Trauma Acute Care Surg 2006; 61: 1134–1141.

27.

Daly

Thorpe

Rockswold

, et al. Hyperbaric oxygen therapy in the treatment of acute severe traumatic brain injury: a systematic review. J Neurotrauma 2018; 35: 623–629.

28.

Davis

Meade

Sise

, et al. Both hypoxemia and extreme hyperoxemia may be detrimental in patients with severe traumatic brain injury. J Neurotrauma 2009; 26: 2217–2223.

29.

Dylla

Anderson

Douin

, et al. A quasi-experimental study of targeted normoxia in critically ill trauma patients. J Trauma Acute Care Surg 2021; 91:S169–S175.

30.

Dylla

Douin

Anderson

, et al. A multicenter cluster randomized, stepped wedge implementation trial for targeted normoxia in critically Ill trauma patients: study protocol and statistical analysis plan for strategy to avoid excessive oxygen (SAVE-O2) trial. Trials 2021; 22:784.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.05 MB