a b s t r a c t

Objectives: This study aimed to test the association between hypoxia level and outcomes according to shock sta- tus in Traumatic brain injury patients.

Methods: Adult TBI patients transported by emergency medical services in 10 provinces were enrolled. Hypoxia was a main exposure; three groups by oxygen saturation (SaO2, non-hypoxia (>=94%), mild hypoxia (90 <= SaO2 b 94%)), and severe hypoxia (b90%). Shock status (bsystolic blood pressure 90 mmHg) was an interactive expo- sure. The outcomes were hospital mortality and worsened disability (a 2-point increase of Glasgow Outcome

Scale). Multivariable logistic regression was used to calculate the adjusted odds (AORs) with 95% Confidence in- tervals (CIs).

Results: Of the 6125 patients, the mortality/disability rates were 49.4%/69.0% in severe hypoxia, 30.7%/46.9% in mild hypoxia, and 18.5%/27.5% in normoxia (p b 0.0001). Mortality/disability rates were 47.1%/57.1% in shock status and 20.5%/31.4% in non-shock status (p b 0.0001). AORs (95% CIs) for worsened disability/mortality com- pared with normoxia (reference) were 3.23 (2.47-4.21)/2.24 (1.70-2.96) in patients with severe hypoxia and

2.11 (1.63-2.74)/1.84 (1.39-2.45) in those with mild hypoxia. AORs (95% CIs) for worsened disability/mortality was 1.58 (1.20-2.09)/1.33 (1.01-1.76) by severe hypoxia than normoxia in patient with only non-shock status in the interaction analysis.

Conclusions: There was a trend toward worsened outcomes with mild and severe hypoxia in patient with and without shock, however, the only met statistical significance for patients with both severe hypoxia and non- shock status.

(C) 2018

Introduction

Traumatic brain injury is one of the most serious health prob- lems worldwide. The most recent estimates indicate that 1.1 million Americans are treated in emergency departments. About 2.5-6.5 mil- lion people currently live with physical, cognitive, or psychological im- pairment, and 50,000 individuals die due to TBI each year [1-4]. Similar patterns of an increased burden of TBI resulting from its high fa- tality have also been reported in many other countries [5-7].

The extent of neurological injury after TBI is not determined solely by the traumatic impact itself, but by time. Secondary brain injury

* Corresponding author at: Department of Emergency Medicine, Seoul National University College of Medicine, 101 Daehak-Ro, Jongno-Gu, Seoul 110-744, Republic of Korea.

E-mail addresses: [email protected] (D.E. Seo), [email protected] (S.D. Shin).

occurs as a result of a complex process that began with primary injury and is characterized by neuroinflammation, ischemia/reperfusion in- jury, cerebral edema, intracranial hemorrhage, and intracranial hyper- tension. Patients who survive early TBI are very vulnerable to secondary insult to the damaged brain, primarily due to hypoxia and hypotension during the initial period of resuscitation [2]. Cerebral hyp- oxia and hypotension are known to cause adverse outcomes in TBI pa- tients [8-14]. In major TBIs, mortality is much more important when two parameters are combined. The adjusted mortality probability for both hypotension and hypoxia were two times higher than for patients with hypotension or hypoxia only [15].

To prevent brain damage caused by hypoxia, sufficient oxygen must be supplied through the blood and sufficient blood pressure must be maintained for optimal brain perfusion. Optimal cerebral perfusion is determined by the difference between mean arterial pressure (MAP) and intracranial pressure . However, ICP cannot be accurately

https://doi.org/10.1016/j.ajem.2018.12.022

0735-6757/(C) 2018

measured in the pre-hospital environment. Rather, it can be estimated based on brain damage. Oxygen saturation and MAP can determine ox- ygen delivery to the brain and the amount of oxygen delivered to the brain is directly related to Hypoxic brain damage.

The current definition of hypoxia and protocols for oxygen delivery suggests 90% or 94% oxygen saturation as a uniform standard of oxygen supply without taking into account these complex factors affecting cere- bral perfusion [16-18]. In previous studies involving non-Traumatic cases, 90% oxygen saturation was used to define hypoxia. On the other hand, 94% was recommended as the optimal level of oxygen supply for long-term preservation in other studies on trauma patients. In par- ticular, hypoxia in shock has not been thoroughly investigated, particu- larly with respect to optimal oxygen levels that do not require oxygen supply. It is unclear how many percent oxygen saturation is required for oxygenation in patients with TBI. We hypothesized that hypoxic ef- fects are dependent on the presence of shock and the extent of intracra- nial injury.

Methods

This study was approved by the Institutional Review Board at the re- search site. This material has been approved for use in Korea Centers for Disease Control and Prevention (CDC). The patient’s informational con- sent was waivered because the data did not contain any personal infor- mation and caused minimal risk to the subject.

Study design and data source

This study is a cross-sectional observational study. Data from the Emergency Medical Services-treated Severe Trauma Registry (EMS- STR) database were used and collected from 10 provinces in Korea from 2012 to 2013. The EMS-STR was developed to monitor the inci- dence and outcomes of Severe trauma patients transported by EMS in Korea by the Korea CDC and the Central Fire Services (CFSs). The index cases with an abnormal Revised Trauma Score , which was defined as shock (systolic blood pressure b90 mmHg), abnormal respi- ration rate (b10 or N29 respirations/min), or abnormal mental status (non-alert response using the AVPU scale), were extracted from the EMS run sheet All cases meeting the criteria for abnormal RTSs were reviewed for the individuals’ hospital medical records by the expert re- viewers of the Korea CDC.

Data variables included demographic and socioeconomic variables, administrative EMS variables, prehospital clinical variables, emergency care-related variables in the hospital, injury variables, and hospital out- comes. A Data quality management (DQM) committee was formed to maintain the data quality. The DQM committee consisted of emergency physicians, injury epidemiologists, trauma surgeons and biostatistical experts. During the medical record review, the DQM committee offered advice regarding the questions and provided feedback about the col- lected data using a quality assurance protocol.

Study setting

Emergency medical services (EMS) in this study setting were similar to those provided by emergency medical technicians (EMT) – basic or intermediate. EMTs were able to supply oxygen through either basic air- way, back-vale mask ventilation, or intubation under direct medical control [19]. EMS medical directors in all EMS agencies that were over- seen by the fire department of each county were employed to provide medical oversight for trauma care as well as other serious emergency conditions. The directors usually visited the fire department to review the EMS run sheet and severe trauma registry and score the quality of the prehospital care provided by the EMTs of the EMS agency. Trauma care equipment in most ambulances included basic airways (oropha- ryngeal airway, nasopharyngeal airway), advanced airways (supraglottic airway, endotracheal tube), ventilation devices (portable

oxygen tank, facial or bag-valve mask), circulation devices and materials (intravenous set), vital sign monitors (non-invasive blood pressure monitor, oxygen saturation monitor), and a defibrillator. The country has approximately 1400 ambulances that are operated by a province’s fire department headquarters on a tax-based budget [20]. Level 1 emer- gency medical technicians are the top providers in this setting, and their service level is equivalent to an intermediate level EMT in the USA. The EMS level 1 technician can provide basic life support and limited ad- vanced life support including intravenous fluid resuscitation and ad- vanced airway under direct medical supervision. They are capable of transporting all patients to the ED and providing CPR during Ambulance transport unless the patients achieve a return of spontaneous circula- tion at the scene. All EMTs are required to fulfill 40 h of continuing med- ical education to maintain relevant medical skills and knowledge in accordance with the Rescue and Fire EMS Act [19]. The national trauma protocol requires EMTs to minimize patients’ movement to prevent fur- ther injury, stay and treat within 10 min in the field, and transport pa- tients who meet the criteria of the trauma triage scheme of the US CDC to a higher level of emergency department or regional trauma cen- ter. The recording of details of prehospital trauma care and relevant in- formation in the EMS trauma registry is encouraged [20].

Approximately 400 EDs are designated by the Ministry of Health and Welfare as levels 1 through 3 according to the emergency care capacity and resource measures including staffing, equipment, and size of the de- partment space. Level 1 and 2 EDs have more resources and better facil- ities for emergency care and must be staffed by emergency physicians 24 h for 365 days a year to provide high-quality care, whereas level 3 EDs can be staffed by general physicians [20]. Additionally, there are EDs in small hospitals that provide lower levels of services that are not formally designated by the government as EDs [19]. Those hospitals manage the primary care of minimally injured patients.

Study subjects

The study subjects were TBI patients who were 15 years old or above and admitted to the study from EDs across the 10 provinces from 2012 to 2013. TBI patients were defined according to the International Classi- fication of Disease (ICD)-10th version, including all patients with diag- nosis codes of S06.0-S06.9 (concussion, traumatic cerebral edema, diffuse brain injury, focal brain injury, epidural hemorrhage, traumatic subdural hemorrhage, traumatic subarachnoid hemorrhage, intracra- nial injury with prolonged coma, other intracranial injuries, and unspec- ified Intracranial injury). If a patient had another injury code as well as the TBI code, the patient was considered a TBI patient (see http://apps. who.int/classifications/icd10/browse/2016/en#/S00-S09). Patients whose information about oxygen saturation, blood pressure, injury se- verity or hospital outcomes (mortality and disability) was unknown or unmeasured were excluded.

Variables

Study variables included demographic factors (age, gender, and ur- banization level), injury factors (mechanism, intent, and event date and time), EMS elapsed time intervals (response time, scene time, and transport time), prehospital clinical parameters (blood pressure, respi- ratory rate, pulse rate, oxygen saturation, AVPU scale score in the field), prehospital care (airway management, oxygen supply, fluid ther- apy, and CPR), and hospital care factors (level of ED, systolic blood pres- sure, respiratory rate and heart rate at ED, AVPU scale score in the ED, intervention or operation, transfusion, CPR, and intensive care), hospital outcomes (final diagnosis, mortality, and disability measured using the Glasgow Outcome Scale at pre-event and at hospital discharge). Hypoxia as a main exposure was defined as a decrease below normal levels of oxygen in inspired gases, arterial blood, or tissues without reaching anoxia. This definition did not include the concept of an oxy- gen level that adversely affected the brain. In this study, the hypoxia

state was divided into three levels of oxygen saturation (SaO2) mea- sured in the field as follows: non-hypoxia (94% or higher SaO2), mild hypoxia (90-93% SaO2), and severe hypoxia (b90% SaO2). Another ex- posure, shock, was defined as an interaction term as a systolic blood pressure lower than 90 mmHg in the field.

Outcome measures

The primary outcome was hospital mortality. The secondary out- come was worsened disability, which was defined as a case in which the difference between GOS score at discharge and pre-event GOS score was 2 points or more or the patient died at hospital discharge. The outcomes were measured by the medical record reviewers of the Korea CDC based on medical records written by duty physicians, sur- geons, or registered nurses.

Statistical analysis

Demographic findings were compared among the three hypoxia groups and shock vs. non-shock groups for the distribution of risk fac- tors and outcomes. The continuous variables were compared using the Wilcoxon sum-rank test, and categorical variables were compared using a chi-square test.

We determined the association between SaO2 and hospital out- comes using restricted cubic spline analysis for calculating the log odds of outcomes according to the change in oxygen saturation level. The method was also used for systolic blood pressure and outcomes.

A multivariable logistic analysis was used to determine the associa- tion between the main exposure and outcomes for the strata of injury severity and hospital care groups. The crude odds ratios (ORs) with 95% confidence intervals (CIs) for hospital outcomes of the hypoxia group in Model 1 were calculated without adjustment. AORs with 95% CIs adjusted for age and gender (Model 2) and adjusted for age, gender, metropolis, mechanism of injury, season and week of injury, time of in- jury, response time interval, scene time interval, advanced airway, fluid therapy, and shock status (Model 3) were calculated for the outcomes. To compare the effect size, a final interaction analysis was performed to calculate the AORs (95% CIs) of the hypoxia groups according to shock

status for hospital outcomes.

Results

Demographic findings

Of 8306 adult TBI patients, 6125 patients were analyzed; exclusions included 793 patients without SaO2 values, 658 without SBP values, 224 with Traumatic cardiac arrest, 150 without NISS (New Injury Severity Scale) scores, and 772 without GOS scores (Fig. 1).

The demographic characteristics of the patients according to hyp- oxia status were summarized in Table 1 and Supplementary Table 1. Among 6125 patients, 636 patients (10.4%) had severe hypoxia, 554 pa- tients (9.0%) had mild hypoxia, and 4935 patients (80.6%) were in normoxia status. Of these patients, 49.4%, 30.7%, and 18.5% of them with severe hypoxia, mild hypoxia and normoxia died, respectively. To- tals of 69.0%, 46.9%, and 27.5% of patients with severe hypoxia, mild hypoxia, and normoxia, respectively, resulted in worsened disability (each p b 0.0001).

The demographic findings of the patients according to shock status were compared in Table 2 and Supplementary Table 2. Of the total num- ber of patients, 8.5% of them suffered from shock, and 91.5% were clas- sified as having non-shock status. Hospital mortality and worsened disability were 47.1% and 57.1% in the shock status group, respectively, and 20.5% and 31.4% in the non-shock group, respectively (p b 0.0001).

Main analysis

The main results of the multivariable logistic regression analysis are shown in Table 3. In the full model (Model 3), AORs (95% CIs) for wors- ened disability/mortality compared with normoxia (reference) were

3.23 (2.47-4.21)/2.24 (1.70-2.96) in severe hypoxia and 2.11 (1.63-2.74)/1.84 (1.39-2.45) in mild hypoxia. AORs (95% CIs) by shock status were 2.07 (1.45-2.96) for worsened disability and 2.14 (1.48-3.10) for mortality.

Interaction analysis

AORs (95% CIs) for worsened disability/mortality were significantly different according to shock status as follows: 1.35 (0.73-2.52)/1.26 (0.68-2.32) in patients with mild hypoxia and shock status, 1.49 (0.88-2.51)/1.31 (0.79-2.20) in patients with severe hypoxia and

Fig. 1. Study population. SPO, saturation of peripheral oxygenation SBP, systolic blood pressure NISS, new injury severity GOS, Glasgow Outcome Scale.

Table 1

Demographic findings of the study population among exposure groups.

Variable All Severe Mild Normoxia

p-Value

Variable

All

Shock

Normal

p-Value

Hypoxia Hypoxia

N

%

N

%

N

%

Table 2

Demographic findings of the study population between shock status groups.

All	N 6125	% 100.0		N 636	% 100.0		N 554	% 100.0		N 4935	% 100.0			All Gender			6125	100.0	518		100.0	5607		100.0	0.5889
Gender												b0.0001		Male			4431	72.3	380		73.4	4051		72.2
Male	4431	72.3	491		77.2	433		78.2	3507		71.1			Female			1694	27.7	138		26.6	1556		27.8
Female	1694	27.7	145		22.8	121		21.8	1428		28.9	Age group, years 0.1906
Age group,												0.0001		15 ~ 24			659	10.8	65		12.5	594		10.6
years														25 ~ 34			642	10.5	66		12.7	576		10.3
15 ~ 24	659	10.8	57		9.0	34		6.1	568		11.5			35 ~ 44			720	11.8	62		12.0	658		11.7
25 ~ 34	642	10.5	73		11.5	44		7.9	525		10.6			45 ~ 54			1250	20.4	94		18.1	1156		20.6
35 ~ 44	720	11.8	68		10.7	56		10.1	596		12.1			55 ~ 64			1178	19.2	86		16.6	1092		19.5
45 ~ 54	1250	20.4	124		19.5	116		20.9	1010		20.5			65 ~ 74			968	15.8	79		15.3	889		15.9
55 ~ 64	1178	19.2	119		18.7	118		21.3	941		19.1			75~			708	11.6	66		12.7	642		11.4
65 ~ 74	968	15.8	107		16.8	116		20.9	745		15.1	Metropolis 0.3525
75~ 708		11.6	88		13.8	70		12.6	550		11.1			No			3559	58.1	291		56.2	3268		58.3
Metropolis												0.2745		Yes			2566	41.9	227		43.8	2339		41.7
No	3559	58.1	382		60.1	334		60.3	2843		57.6			Injury mechanism											0.0008
Yes	2566	41.9	254		39.9	220		39.7	2092		42.4			Traffic accident			3317	54.2	315		60.8	3002		53.5
Injury												b0.0001		Fall or slip down			2233	36.5	146		28.2	2087		37.2
mechanism														Collision			331	5.4	33		6.4	298		5.3
Traffic	3317	54.2	377		59.3	289		52.2	2651		53.7			Other mechanism			244	4.0	24		4.6	220		3.9
accident														Mental status at the field											b0.0001
Fall or slip	2233	36.5	197		31.0	219		39.5	1817		36.8			Alert			1603	26.2	175		33.8	1428		25.5
Collision	331	5.4	26		4.1	17		3.1	288		5.8			Verbal response			2426	39.6	49		9.5	2377		42.4
Other	244	4.0	36		5.7	29		5.2	179		3.6			Pain response			1607	26.2	102		19.7	1505		26.8
mechanism														No response			489	8.0	192		37.1	297		5.3
Mental status in												b0.0001		Hypoxia status											b0.0001
the field														Severe Hypoxia			636	10.4	241		46.5	395		7.0
Alert	1603	26.2	56		8.8	82		14.8	1465		29.7			Mild Hypoxia			554	9.0	51		9.8	503		9.0
Verbal	2426	39.6	109		17.1	210		37.9	2107		42.7			Normoxia			4935	80.6	226		43.6	4709		84.0
response																Treatment at EMS location
Pain response	1607	26.2	223		35.1	214		38.6	1170		23.7			Bleeding control			2879	47.0	240		46.3	2639		47.1	0.7487
No response	489	8.0	248		39.0	48		8.7	193		3.9			Wound care			2495	40.7	191		36.9	2304		41.1	0.0615
Shock status												b0.0001		Intravenous fluid			295	4.8	58		11.2	237		4.2	b0.0001
Yes (SBP b 90	518	8.5	241		37.9	51		9.2	226		4.6			Charlson comorbidity											0.0144
mmHg)														None			5663	92.5	493		95.2	5170		92.2
No (SBP >= 90	5607	91.5	395		62.1	503		90.8	4709		95.4			1 or more			462	7.5	25		4.8	437		7.8
mmHg)														Mental status at the ED											b0.0001
Treatment at														Alert			2517	41.1	160		30.9	2357		42.0
EMS location														Verbal response			1307	21.3	53		10.2	1254		22.4
Bleeding	2879	47.0	269		42.3	301		54.3	2309		46.8	0.0001		Pain response			1299	21.2	69		13.3	1230		21.9
control														No response			1002	16.4	236		45.6	766		13.7
Wound care	2495	40.7	201		31.6	244		44.0	2050		41.5	b0.0001		AIS >= 3
Intravenous	295	4.8	45		7.1	38		6.9	212		4.3	0.0005		Head			3291	53.7	321		62.0	2970		53.0	b0.0001
fluid														Chest			832	13.6	153		29.5	679		12.1	b0.0001
Charlson												0.1044		Abdomen			44	0.7	8		1.5	36		0.6	0.02
comorbidity														Spine			84	1.4	13		2.5	71		1.3	0.0199
None	5663	92.5	597		93.9	502		90.6	4564		92.5			Extremity			317	5.2	57		11.0	260		4.6	b0.0001
1 or more	462	7.5		39	6.1		52	9.4		371	7.5					Outcomes

Mental status at

b0.0001	Hospital death	1396	22.8	244	47.1	1152	20.5	b0.0001
	Disability, (GOS <= 3)	2054	33.5	296	57.1	1758	31.4	b0.0001

ED
Alert	2517	41.1	75	11.8	151	27.3	2291	46.4
Verbal	1307	21.3	87	13.7	121	21.8	1099	22.3
response
Pain response	1299	21.2	141	22.2	161	29.1	997	20.2
No response	1002	16.4	333	52.4	121	21.8	548	11.1
AIS >= 3
Head	3291	53.7	496	78.0	402	72.6	2393	48.5	b0.0001
Chest	832	13.6	219	34.4	140	25.3	473	9.6	b0.0001
Abdomen	44	0.7	16	2.5	4	0.7	24	0.5	b0.0001
Spine	84	1.4	14	2.2	8	1.4	62	1.3	0.1539
Extremity	317	5.2	60	9.4	43	7.8	214	4.3	b0.0001
Outcomes
Hospital 1396		22.8	314	49.4	170	30.7	912	18.5	b0.0001
Disability, 2054		33.5	439	69.0	260	46.9	1355	27.5	b0.0001
(GOS <= 3)

death

SBP, systolic blood pressure; EMS, emergency medical service; ED, emergency depart- ment; AIS, Abbreviate injury scale; GOS, Glasgow Outcome Scale.

shock status, 1.22 (0.88-1.69)/1.20 (0.87-1.66) in patients with mild hypoxia and non-shock status, and 1.58 (1.20-2.09)/1.33 (1.01-1.76) in those with severe hypoxia and non-shock status (Table 4).

EMS, emergency medical service; ED, emergency department; AIS, Abbreviate injury scale; GOS, Glasgow Outcome Scale.

Discussion

Both exposures, hypoxia and shock status, revealed significant in- creases in Disability and mortality rates. Severe hypoxia followed by mild hypoxia was significantly associated with poor hospital outcomes. However, hypoxia groups with shock status were not significantly asso- ciated with poor outcomes. Only severe hypoxia with non-shock status was associated with an increase in disability and mortality in the inter- action models.

From this study, we found that the effect size of hypoxia for out- comes in TBI patients was different according to shock status. In patients already suffering from shock, hypoxia did not add any significant effect. However, if TBI patients were not in shock, severe hypoxia b90% SaO2 significantly affected brain damage to result in disability and mortality. Spaite DW et al. studied the associations between mortality and out-

of-Hospital hypotension and hypoxia separately and in combination [15]. Mortality for the four study groups (neither hypotension nor

Table 3

Multivariable logistic regression analysis for outcomes by hypoxia and shock status.

Exposure	Outcomes	Group	Total		Outcomes		Model 1			Model 2			Model 3
			N		n %		AOR 95% CI			AOR 95% CI			AOR 95% CI
Hypoxia	Disability	Total	6125		2054 33.5
		Normoxia	4935		1355 27.5		1.00			1.00			1.00
		Severe hypoxia	554		260 46.9		3.45 2.66	4.47		3.38 2.60	4.39		3.23 2.47	4.21
		Mild hypoxia	636		439 69.0		2.29 1.78	2.94		2.12 1.64	2.73		2.11 1.63	2.74
	Mortality	Total	6125		1396 22.8
		Normoxia	4935		912 18.5		1.00			1.00			1.00
		Severe hypoxia	554		170 30.7		2.43 1.86	3.18		2.35 1.79	3.08		2.24 1.70	2.96
Shock status		Mild hypoxia	636		314 49.4		1.92 1.46	2.52		1.82 1.38	2.39		1.84 1.39	2.45
	Disability	Total	6125		2054 33.5
		Normal	5607		1758 31.4		1.00			1.00			1.00
		Shock	518		296 57.1		2.10 1.49	2.96		2.19 1.54	3.11		2.07 1.45	2.96
	Mortality	Total	6125		1396 22.8
		Normal	5607		1152 20.5		1.00			1.00			1.00
		Shock	518		244 47.1		2.29 1.60	3.27		2.32 1.62	3.33		2.14 1.48	3.10

AOR, adjusted odds ratio.

95% CI, 95% confidence interval.

Model 1; not adjusted (crude).

Model 2; adjusted for age and gender.

Model 3; adjusted for shock status, gender, age, injury mechanism and intent, event time – season, weekday, and hour – and response time interval.

hypoxia, hypotension only, hypoxia only, and both hypotension and hypoxia) was 5.6%, 20.7%, 28.1%, and 43.9%, respectively. The results that combined out-of-hospital hypotension and hypoxia were associ- ated with significantly increased mortality, which was consistent with the findings of this study. Their exclusion criteria (exclusions:

b10 years of age, out-of-hospital oxygen saturation <=10%, and out-of- hospital systolic blood pressure b40 or N200 mmHg) and definition of

exposure (systolic blood pressure b90 mmHg and SaO2 b90% each) were different from those in our study. The overall effects of hypoxia or hypotension were similar with those of our study. However, they did not compare the effect size of hypoxia under hypotension status.

Zebrack M et al. studied severe traumatic brain injury in children (N

= 299) [21]. They found that untreated hypoxia was not significantly associated with death or disability, except in the setting of hypotension. These findings may be contrary to our study findings. They used the ex- posure (hypoxia and hypotension) measured at the time when the pa- tients visited the level 1 trauma center. However, blood pressure (31%) and oxygenation (34%) were not recorded during some portion of “early care.” Documented hypotension occurred in 118 children (39%). An at- tempt to treat documented hypotension was made in 48% of cases (57 of 118 children). Documented hypoxia occurred in 131 children (44%). Untreated hypotension was associated with increased mortality, but untreated hypoxia without hypotension was not significantly associ- ated with outcomes. This study had a significant limitation because more than half of the patients had no information on hypoxia and

Table 4

Interaction effect of shock with hypoxia status on outcomes in traumatic brain injury.

Outcomes Shock status Hypoxia

Mild Severe

AOR 95% CI AOR 95% CI

hypotension, which could have resulted in selection bias. A number of patients might have had significant hypoxia status that was not mea- sured in the clinical setting. Even though this study was performed at a single center retrospectively, the effect size of hypoxia under normal blood pressure in TBI patients may be controversial.

Fuller G et al. aimed to fully characterize the association between ad- mission SBP and mortality (N = 5057) [10]. Admission SBP demon- strated a smooth u-shaped association with outcomes in a bivariate analysis, with increasing mortality at both lower and higher values and no evidence of any threshold effect. Adjustment for confounding factors slightly attenuated the association between mortality and SBP at levels b120 mmHg and abolished the relationship for higher SBP values. Case-mix adjusted odds of death were 1.5 times greater at b120 mmHg, doubled at b100 mmHg, tripled at b90 mmHg and was six times greater at SBP b 70 mmHg, p b 0.01. We used the cut-off value of 90 mmHg for shock status. If we used different cut-off values such as 120 mmHg, 100 mmHg, or 70 mmHg, the effect might have been different.

We used the parameters for oxygen saturation and blood pressure that were measured by EMS personnel in the field. Numerous studies involving trauma care for patients suffering from shock or hypoxia were compared between physician-staffed and paramedic-staffed ser- vices because the providers offer a different level of services when the patients are suffering from shock or hypoxia or both. In studies favoring physician-staffed ambulance or helicopter services, the paramedic- staffed services resulted in better outcomes [22,23]. From the observa- tional study during the 6-year study period, 458 total patients showed that one-year mortality was higher in the paramedic-staffed EMS group compared to the physician-staffed group, 57% vs. 42%, respec- tively. This system difference affected the outcomes of the study. Our study had lower levels of service providers, and the mortality and dis- ability due to secondary injury followed by hypoxia or hypotension

Disability Shock	1.35	0.73	2.52	1.49	0.88	2.51	might have been more significant. Prehospital hypoxia can be prevented by appropriate management
Normal	1.22	0.88	1.69	1.58	1.20	2.09	such as providing a basic airway and oxygen therapy or advanced air-

Mortality
	Shock	1.26	0.68	2.32	1.31	0.79	2.20
	Normal	1.20	0.87	1.66	1.33	1.01	1.76

AOR, adjusted odds ratio.

95% CI, 95% confidence interval.

AORs and 95% CIs were calculated from models with adjustment for shock status, gender, age, injury mechanism and intent, event time – season, weekday, and hour – and response time interval.

way for patients with severe hypoxia status. A secondary analysis from the ProTECT III trial (a multicenter randomized, double-blind, placebo-controlled trial of early administration of progesterone in 882 patients with acute moderate-to-severe nonPenetrating TBI) showed better outcomes in the advanced airway management group than the non-advanced airway group in a prehospital setting (mortality 13.8% v. 19.5%, respectively, p = 0.03) [24]. In our study setting, the placement

of a prehospital advanced airway by an EMT was allowed for only trau- matic cardiac arrest patients who were excluded from the study. There- fore, most patients did not receive advanced airway management. This study setting will be considered when interpreting the findings.

This study recommends a change in the hospital protocol for TBI pa-

tients with hypoxia or shock. If the patient is suffering from hypoxia without shock, it is recommended that the SaO2 of the patient be main- tained above the level of 94% through vigorous treatment using basic or advanced airway management for adequate oxygen supply. In many EMS systems in Asia, advanced airway protocols are limited to patients in cardiac arrest and not for non-arrest traumas [25,26]. To prevent hyp- oxia and maintain a normal oxygen level in TBI patients, the protocol would need to be changed. Further studies are recommended to search for the exact cut-off value for hypoxia (90% v. 94%) or to determine the effect of consecutive episodes of hypoxia in TBI patients with shock or non-shock status.

Limitations

As this study excluded patients whose medical records on oxygen saturation, blood pressure, injury severity or hospital outcomes were unknown or unmeasured, it is possible that their inclusion might have altered the results. Moreover, as the study only included those who were 15 years old or above, the results might not be applicable to pedi- atric patients. Those exclusions would affect the outcomes and contrib- ute to selection bias. Data about SaO2, shock status, mental status, and of TBI patients were collected by EMTs, duty physicians, surgeons, and reg- istered nurses using the EMS run sheet. Despite the existence of a stan- dard measurement manual, the devices differ by hospital, and the measurer’s judgment might differ from person to person. This measure- ment bias was not considered in the study. As we mentioned, the study was performed at the low-to-intermediate level of EMS, which is differ- ent from the EMS level in Western countries or North America. Caution should be taken when generalizing these findings to different settings. Finally, we should consider the limited sample size to interpret the re- sults. It appears that the hypoxia was associated with higher mortal- ity/disability in the shock group than the non-shock group. However, the confidence intervals are larger in the shock group because there are fewer patients.

Conclusion

The study found that both prehospital shock (SBP b90 mmHg) and hypoxia (SaO2 b90% or b94%) consistently had a non-favorable effect on outcomes after TBI. Hypoxia resulted in different effect sizes between shock and non-shock status. Hypoxia without shock status was signifi- cantly associated with poor outcomes, and hypoxia with shock was not associated with poor outcomes in the interaction model.

Author contributions

Ms. Seo and Dr. Shin had full access to all of the data in the study and assumed responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Dr. Shin.

Acquisition, analysis, or interpretation of the data: Dr. Shin and Ms.

Seo.

Drafting of the manuscript: Ms. Seo. Critical revision of the manuscript for important intellectual content:

Drs. Song, Ro, and Park.

Statistical analysis: Dr. Shin. Obtainment of funding: Dr. Shin. Manuscript approval: all authors.

Acknowledgements

This study was supported by the Seoul Metropolitan Fire Depart- ment and Seoul Metropolitan Health Department of Korea. The study was funded by the Korea Centers for Disease Control & Prevention (CDC) (2013-2014) (Grant for Private Support Program). The Korea CDC approved the use of the database in this study.

Conflict of interest

The authors declare no conflicts of interest relevant to this paper.

Data sharing statement

The Korea Centers for Disease Control and Prevention (CDC) ap- proved the use of the database in this study. Data sharing should be ap- proved by the Korea CDC

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi. org/10.1016/j.ajem.2018.12.022.

References

1. Corrigan JD, Selassie AW, Orman JA. The epidemiology of traumatic brain injury. J Head Trauma Rehabil 2010;25(2):72-80.
2. Stahel PF, Smith WR, Moore EE. Hypoxia and hypotension, the “lethal duo” in trau- matic brain injury: implications for prehospital care. Intensive Care Med 2008;34 (3):402-4.
3. Ghajar J. Traumatic brain injury. Lancet 2000;356(9233):923-9.
4. McArthur DL, Chute DJ, Villablanca JP. Moderate and severe traumatic brain injury: epidemiologic, imaging and neuropathologic perspectives. Brain Pathol 2004;14 (2):185-94.
5. Rubiano AM, Carney N, Chesnut R, Puyana JC. Global neurotrauma research chal- lenges and opportunities. Nature 2015;527(7578):S193-7.
6. Leo P, McCrea M. Epidemiology. In: Laskowitz D, Grant G, editors. Translational Re- search in Traumatic Brain Injury. Frontiers in Neuroscience; 2016 Boca Raton (FL).
7. Pimentel MA, Brennan T, Lehman LW, King NK, Ang BT, Feng M. Outcome prediction for patients with traumatic brain injury with dynamic features from intracranial pressure and arterial blood pressure signals: a Gaussian process approach. Acta Neurochir Suppl 2016;122:85-91.
8. Thelin EP. Experimental models combining traumatic brain injury and hypoxia. Methods Mol Biol 2016;1462:459-79.
9. Hellewell SC, Yan EB, Agyapomaa DA, Bye N, Morganti-Kossmann MC. Post- traumatic hypoxia exacerbates brain tissue damage: analysis of Axonal injury and glial responses. J Neurotrauma 2010;27(11):1997-2010.
10. Fuller G, Hasler RM, Mealing N, Lawrence T, Woodford M, Juni P, et al. The associa- tion between admission systolic blood pressure and mortality in significant trau- matic brain injury: a multi-centre cohort study. Injury 2014;45(3):612-7.
11. Miller JD, Becker DP. Secondary insults to the injured brain. J R Coll Surg Edinb 1982;

    27(5):292-8.

    Struchen MA, Hannay HJ, Contant CF, Robertson CS. The relation between acute physiological variables and outcome on the Glasgow Outcome Scale and Disability Rating Scale following severe traumatic brain injury. J Neurotrauma 2001;18(2): 115-25.

12. Jones PA, Andrews PJ, Midgley S, Anderson SI, Piper IR, Tocher JL, et al. Measuring the burden of secondary insults in head-injured patients during intensive care. J Neurosurg Anesthesiol 1994;6(1):4-14.
13. Chesnut R, Marshall S, Piek J, Blunt B, Klauber M, Marshall L. Early and late systemic hypotension as a frequent and fundamental source of cerebral ischemia following severe brain injury in the Traumatic Coma Data Bank. Monitoring of Cerebral Blood Flow and Metabolism in Intensive Care. Springer; 1993. p. 121-5.
14. Spaite DW, Hu C, Bobrow BJ, Chikani V, Barnhart B, Gaither JB, et al. The effect of combined out-of-hospital hypotension and hypoxia on mortality in major traumatic brain injury. Ann Emerg Med 2017;69(1):62-72.
15. Johansson P, Svensson E, Alehagen U, Jaarsma T, Brostrom A. The contribution of hypoxia to the association between sleep apnoea, insomnia, and cardiovascular mortality in community-dwelling elderly with and without cardiovascular disease. Eur J Cardiovasc Nurs 2015;14(3):222-31.
16. Dodson SR, Hensley Jr FA, Martin DE, Larach DR, Morris DL. Continuous oxygen sat- uration monitoring during cardiac catheterization in adults. Chest 1988;94(1): 28-31.
17. Yilmaz Avci A, Avci S, Lakadamyali H, Can U. Hypoxia and inflammation indicate sig- nificant differences in the severity of obstructive Sleep apnea within similar apnea- hypopnea index groups. Sleep Breath 2017;21(3):703-11.
18. Ro YS, Shin SD, Lee YJ, Lee SC, Song KJ, Ryoo HW, et al. Effect of dispatcher-assisted cardiopulmonary resuscitation program and location of out-of-hospital cardiac ar- rest on survival and neurologic outcome. Ann Emerg Med 2017;69(1):52-61 [e1].
19. Cho EJ, Shin SD, Jeong S, Kwak YH, Suh GJ. The effect of atmosphere temperature on out-of-hospital Cardiac arrest outcomes. Resuscitation 2016;109:64-70.
20. Zebrack M, Dandoy C, Hansen K, Scaife E, Mann NC, Bratton SL. Early resuscitation of children with moderate-to-severe traumatic brain injury. Pediatrics 2009;124(1): 56-64.
21. Pakkanen T, Virkkunen I, Kamarainen A, Huhtala H, Silfvast T, Virta J, et al. Pre- hospital severe traumatic brain injury – comparison of outcome in paramedic versus physician staffed emergency medical services. Scand J Trauma Resusc Emerg Med 2016;24:62.
22. Pakkanen T, Kamarainen A, Huhtala H, Silfvast T, Nurmi J, Virkkunen I, et al. Physician-staffed helicopter emergency medical service has a beneficial impact on the incidence of prehospital hypoxia and secured airways on patients with severe traumatic brain injury. Scand J Trauma Resusc Emerg Med 2017;25(1):94.
23. Denninghoff KR, Nuno T, Pauls Q, Yeatts SD, Silbergleit R, Palesch YY, et al. prehospital intubation is associated with favorable outcomes and lower mortality in ProTECT III. Prehosp Emerg Care 2017;21(5):539-44.
24. Sun KM, Song KJ, Shin SD, Tanaka H, Shaun GE, Chiang WC, et al. Comparison of emergency medical services and trauma care systems among pan-Asian countries: an international, multicenter, population-based survey. Prehosp Emerg Care 2017; 21(2):242-51.
25. Kong SY, Shin SD, Tanaka H, Kimura A, Song KJ, Shaun GE, et al. Pan-Asian trauma outcomes study (PATOS): rationale and methodology of an international and multi- center trauma registry. Prehosp Emerg Care 2017:1-26.