Risk Factors on Hospital Arrival for Acute Respiratory Distress Syndrome following Pediatric Trauma

Abstract

Objective

To determine risk factors identifiable at hospital arrival associated with acute respiratory distress syndrome (ARDS) development among critically injured children

Design

Retrospective cohort study

Setting

Level I or II adult or pediatric trauma centers contributing to the National Trauma Data Bank from 2007–2016

Patients

Patients <18 years admitted to an intensive care unit with traumatic injury

Interventions

None

Measurements and Main Results

We determined associations between patient, injury, and clinical characteristics present at hospital arrival with development of ARDS recorded as a hospital complication. ARDS occurred in 1.8% (n=2590) of 146,058 critically injured children. The only demographic factor associated with higher risk of developing ARDS on multivariable analysis was African American race (Relative Risk (RR) 1.42 vs white, 95% CI 1.13–1.78). Injury characteristics included firearm injuries (RR 1.93, 1.50–2.48) and motor vehicle crashes (RR 1.91, 1.57–2.31) relative to falls; spine (RR 1.39, 1.20–1.60), chest (RR 1.36, 1.22–1.52), or lower extremity injuries (RR 1.26, 1.10–1.44); amputations (RR 2.10, 1.51–2.91), and more severe injury (RR 3.69 for ISS 40–75 vs. 1–8, 2.50–5.44). Clinical parameters included abnormal respiratory status (intubated RR 1.67, 1.23–2.26; hypopnea RR 1.23, 1.05–1.45; tachypnea RR 1.26, 1.10–1.44), and lower Glasgow Coma Scale score (RR 5.61 for GCS 3 vs 15, 4.44–7.07).

Conclusions

We provide the first description of the incidence of and risk factors for ARDS among pediatric trauma patients. Improved understanding of the risk factors associated with ARDS following pediatric trauma may help providers anticipate its development and intervene early to improve outcomes for severely injured children.

Introduction

Acute respiratory distress syndrome (ARDS) is a recognized complication of severe traumatic injury in both adults^1–11 and children.^12–14 Despite improvements in ventilator strategy and fluid management,⁴ the incidence of post-traumatic ARDS among adults has not decreased¹⁰ and remains a frequent contributor to morbidity and mortality following trauma.^3–7,15 ARDS develops in 5–10% of adults following traumatic injury.^3–5,9–10

Many studies have identified risk factors for post-traumatic ARDS in adults, including blunt trauma, chest injury, femoral fractures, higher injury severity, shock, and blood transfusion.^3–9,11 Little is known, however, about post-traumatic ARDS in children. Studies of pediatric ARDS have incorporated trauma patients within larger heterogeneous ARDS cohorts, with fewer than 10 children with trauma included in most studies.^12–14,16 There are no studies evaluating the incidence of ARDS following trauma in the pediatric population or which children are at risk for post-traumatic ARDS.

Early application of therapeutic interventions such as lung protective ventilation is associated with lower mortality in patients with ARDS¹⁷ and may reduce progression to ARDS in at-risk patients.¹⁸ ARDS is substantially under-recognized in children,¹⁹ however, suggesting that there may be missed opportunities to intervene early to improve outcomes. The aim of this study was therefore to determine the risk factors identifiable at the time of hospital arrival associated with ARDS development among critically injured children. Identification of such factors may help enable early recognition of high-risk patients and facilitate timely implementation of appropriate treatment and ultimately prevention strategies.

Materials and Methods

We conducted a retrospective cohort study of pediatric trauma patients included in the National Trauma Data Bank (NTDB)²⁰ from 2007–2016. This study was exempt from review by the University of Washington Institutional Review Board as data were publicly available and de-identified.

National Trauma Data Bank

The NTDB is the largest aggregate of trauma registry data worldwide, containing over 7 million records from over 1000 facilities.²¹ Patients with traumatic injury who are transferred via Emergency Medical Services or sustain injuries resulting in hospital admission or death are eligible for inclusion. Data are collected by trained registrars with common field definitions across centers and a robust error validation system.

Participants

We queried the 2007–2016 NTDB research datasets for patients <18 years old. We included patients with ≥1 ICU day at a Level I or II adult or pediatric trauma center. Patients were excluded if they sustained burn- or drowning-related injuries, given a different underlying physiologic mechanism for ARDS development than other trauma patients. Patients from nine facilities that did not record hospital complications were excluded (n=1805).

Exposures

We extracted demographic, injury, and clinical data for each encounter. Demographic data included age, gender, race/ethnicity, and type of insurance. Injury data included type (e.g.. blunt, penetrating), mechanism (e.g. fall, firearm), nature of injury (e.g. fracture, amputation), body region, and Injury Severity Score (ISS). Only injuries with an Abbreviated Injury Scale (AIS) score ≥3 were included. Clinical data included Emergency Department (ED) Glasgow Coma Scale (GCS) score and vital signs, which were categorized using age-adjusted normative values.^22,23

Outcome

The primary outcome was ARDS recorded as a complication of the initial post-injury hospitalization. NTDB registrars recorded ARDS if patients met American-European Consensus Conference criteria²⁴ through 2011, modified Berlin criteria²⁵ from 2012–2014, and full Berlin criteria from 2015–2016 (Supplemental Digital Content 1).

Statistical Analysis

We determined ARDS incidence for each risk factor, and used 10-iteration multiple imputation to account for missing data (detailed in Supplemental Digital Content 2). We estimated associations between demographic, injury, and clinical variables with development of ARDS in bivariate analyses using generalized linear Poisson regression models, clustered by facility. We included all clinically important variables with bivariate p<0.005 in a backwards selection model-building process to develop a multivariable generalized linear Poisson regression model, using p<0.005 for variables to remain in the model. We developed a second multivariable model for examining associations between chest injury characteristics and ARDS. Multivariable models were adjusted for admission year and transfer status. We conducted all analyses using Stata/SE 14.2 statistical software (StataCorp LP, College Station, TX).

Results

ARDS occurred in 1.8% (n=2590) of 146,058 pediatric trauma patients admitted to the ICU across 460 facilities. ARDS incidence was higher among subsets of the cohort with ISS ≥16 (3.1%), ICU stay ≥3 days (3.7%), and patients requiring mechanical ventilation (3.8%). Among patients who required mechanical ventilation who also had ISS ≥16 and ICU stay ≥3 days, ARDS incidence was 6.6%. In-hospital mortality was 20.0% among patients with ARDS and 4.3% among patients without ARDS.

Risk factors for development of ARDS were assessed in bivariate analyses across three domains of patient demographic, injury, and clinical characteristics.

Demographic factors (Table 1)

Table 1.

Demographic characteristics of patients who developed ARDS, with relative risks for ARDS development on bivariate Poisson regression

Risk Factor	Incidence of ARDS	Bivariate Analysis
	No. (%)	RR	95% CI	p-value
	(n = 2590, 1.8%)
Age group				<0.001
< 1 year	342 (1.6)	1.11	0.91 – 1.35
≥ 1–5 years	325 (1.6)	1.15	0.83 – 1.60
≥ 5–12 years	482 (1.4)	Ref
≥ 12–17 years	1441 (2.1)	145	1.22 – 1.71
Race/ethnicity				0.002
Non-Hispanic White	1466 (1.8)	Ref
Non-Hispanic African American	539 (2.5)	1.43	1.12 – 1.83
Hispanic	363 (1.5)	0.85	0.51 – 1.41
Asian/Pacific Islander	48 (1.5)	0.84	0.42 – 1.70
Other/Unknown/Biracial	174 (1.2)	0.68	0.47 – 0.98
Male Gender	1688 (1.7)	0.94	0.86 – 1.02	0.15
Type of Insurance				0.001
Private	859 (1.5)	Ref
Public	1038 (1.9)	1.30	1.12 – 1.50
Other	254 (1.8)	1.20	0.95 – 1.52

ARDS, Acute respiratory distress syndrome; RR, relative risk; CI, confidence interval

Teenagers had the highest risk of ARDS on bivariate regression relative to 5–12-year-olds (relative risk (RR) 1.45, 95% confidence interval (CI) 1.22–1.71). African Americans had higher risk of ARDS than white patients (RR 1.43, 1.12–1.83), and patients with public insurance had a higher risk than patients with private insurance (RR 1.30, 1.12–1.50).

Injury factors (Table 2)

Table 2.

Injury characteristics of patients who developed ARDS, with relative risks for ARDS development on bivariate Poisson regression

Risk Factor	Incidence of ARDS	Bivariate Analysis
	No. (%)	RR	95% CI	p-value
	(n = 2590, 1.8%)
Mechanism of injury				<0.001
Fall	202 (0.5)	Ref
Motor vehicle crash	1095 (3.1)	5.53	4.45 – 6.87
Firearm	195 (2.6)	4.74	3.57 – 6.28
Pedestrian/cyclist	370 (1.8)	3.27	2.72 – 3.93
Struck by/against	102 (0.7)	1.32	1.02 – 1.69
Other	591 (2.1)	3.84	3.07 – 4.81
Injury type				<0.001
Blunt	1979 (1.6)	Ref
Penetrating	225 (2.2)	1.32	1.10 – 1.58
Other / Unspecified	345 (2.6)	1.58	1.17 – 2.15
Injury to multiple body regions	2209 (3.0)	5.59	4.45 – 7.01	<0.001
Body region of injuries with AIS ≥3a
Brain injury	1701 (2.4)	2.05	1.76 – 2.38	<0.001
Chest	1269 (4.2)	3.70	3.28 – 4.17	<0.001
Upper Extremity	125 (4.6)	2.64	1.83 – 3.81	<0.001
Lower Extremity	516 (3.8)	2.39	2.00 – 2.86	<0.001
Abdomen	483 (2.6)	1.57	1.37 – 1.80	<0.001
Spine	222 (4.1)	2.42	2.12 – 2.77	<0.001
Pelvis	285 (4.4)	2.51	2.10 – 3.00	<0.001
Nature of injuries with AIS ≥3a
Internal	1936 (2.6)	2.87	2.58 – 3.20	<0.001
Fracture	1812 (2.7)	2.67	2.38 – 3.01	<0.001
Open wound	965 (3.5)	2.54	2.20 – 2.94	<0.001
Vascular	216 (4.5)	2.68	2.26 – 3.19	<0.001
Nerves	71 (4.0)	2.30	1.82 – 2.91	<0.001
Amputation	27 (5.3)	2.99	2.01 – 4.46	<0.001
Crush	9 (3.9)	2.22	1.19 – 4.16	0.013
Injury Severity Score				<0.001
1 – 8	133 (0.4)	Ref
9 – 15	342 (0.8)	1.91	1.55 – 2.35
16 – 24	551 (1.6)	3.93	3.08 – 5.01
25 – 39	1096 (4.6)	11.63	8.89 – 15.21
40 – 75	411 (9.1)	22.80	16.99 – 30.60

ARDS, Acute respiratory distress syndrome; RR, relative risk; CI, confidence interval; AIS, Abbreviated Injury Score

^aCategories not exclusive; patients may have multiple body regions affected or injury natures, with each treated as a distinct risk factors. Relative risks represent association with ARDS compared with individuals who did not have that type of injury.

Falls were the most common injury mechanism in the cohort as a whole, but motor vehicle crashes (MVCs) represented the most common mechanism among those with ARDS (42.9%), and were associated with the highest risk of ARDS development relative to falls (RR 5.53, 4.45–6.87). Firearm injuries were also associated with higher risk of ARDS relative to falls (RR 4.74, 3.57–6.28), and penetrating injuries in general were associated with higher risk of ARDS compared to blunt injuries (RR 1.32, 1.10–1.58).

Patients with injuries to multiple body regions had 5.59 times the risk (95% CI 4.45–7.01) of developing ARDS compared to patients with isolated injuries. Brain injuries were the most common injury region among both those with ARDS (66.6%) and without ARDS (49.1%), while chest injuries had the highest risk of ARDS development (RR 3.70, 3.28–4.17). Across all body regions, amputations were the nature of injury associated with the highest risk of ARDS (RR 2.99, 2.01–4.46). Injury severity was strongly associated with ARDS development, with higher risk of ARDS with each sequentially higher ISS category.

Clinical factors (Table 3)

Table 3.

Clinical characteristics in the Emergency Department of patients who developed ARDS, with relative risks for ARDS development on bivariate Poisson regression

Risk Factor	Incidence of ARDS	Bivariate Analysis
	No. (%)	RR	95% CI	p-value
	(n = 2590, 1.8%)
Respiratory Rate				<0.001
Intubated	1263 (6.6)	9.33	7.36 – 11.82
Not breathing	108 (5.0)	6.95	4.76 – 10.16
Hypopnea	247 (1.1)	1.69	1.43 – 2.01
Normal	388 (0.7)	Ref
Tachypnea	486 (1.2)	1.69	1.46 – 1.96
Oxygen Saturation				<0.001
≥ 97%	1341 (1.4)	Ref
90 – 96%	273 (2.5)	1.57	1.35 – 1.83
< 90%	173 (5.9)	2.45	2.03 – 2.96
Heart Rate				<0.001
Pulseless	24 (6.0)	4.92	2.93 – 8.27
Bradycardic	190 (2.6)	2.08	1.76 – 2.46
Normal	781 (1.2)	Ref
Tachycardic	1504 (2.2)	1.80	1.61 – 2.00
Blood Pressure				<0.001
Not Hypotensive	2214 (1.7)	Ref
Hypotensive	253 (5.0)	2.85	2.50 – 3.24
Temperature				<0.001
≥ 36.0° C	1363 (1.3)	Ref
< 36.0° C	586 (3.8)	2.85	2.53 – 3.20
Glasgow Coma Scale				<0.001
15	393 (0.5)	Ref
13 – 14	142 (1.0)	2.00	1.67 – 2.41
9 – 12	207 (2.4)	4.52	3.71 – 5.52
6 – 8	318 (4.4)	9.21	7.28 – 11.66
4 – 5	153 (6.8)	14.50	11.80 – 17.81
3	1264 (6.3)	13.76	11.70 – 16.19

ARDS, Acute respiratory distress syndrome; RR, relative risk; CI, confidence interval; C, Celsius

Only 15% of ARDS patients had a normal respiratory rate in the ED, and 25% were hypoxemic. Over half of ARDS patients had been intubated prior to their arrival or in the ED compared to only 13% of those without ARDS (RR 9.33, 7.36–11.82). Among unintubated patients, hypopnea and tachypnea were both associated with higher risk of ARDS. Both mild (oxygen saturation 90–96%) and more severe (<90%) hypoxemia in the ED were associated with ARDS development. Nearly 70% of patients with ARDS had an abnormal heart rate in the ED, with pulselessness, bradycardia, and tachycardia all associated with higher risk of ARDS. Hypotension was associated with 2.85 times higher risk of ARDS relative to those who were not hypotensive (2.50–3.24). Risk of ARDS was higher with each sequentially decreasing GCS category.

Multivariable analysis (Table 4)

Table 4.

Relative risk for development of ARDS on multivariable logistic regression, adjusted for year and transfer status

Risk Factor	RR	95% CI	p-value
Race/ethnicity			0.003
Non-Hispanic White	Ref
Non-Hispanic African American	1.42	1.13 – 1.78
Hispanic	0.93	0.58 – 1.50
Asian/Pacific Islander	1.05	0.57 – 1.93
Other/Unknown/Biracial	0.69	0.49 – 0.97
Mechanism of injury			<0.001
Fall	Ref
Motor vehicle crash	1.91	1.57 – 2.31
Pedestrian/cyclist	1.41	1.17 – 1.70
Firearm	1.93	1.50 – 2.48
Struck by/against	1.19	0.93 – 1.52
Other	1.93	1.58 – 2.36
Chest injurya	1.36	1.22 – 1.52	<0.001
Lower extremity injurya	1.26	1.10 – 1.44	0.001
Spinal injurya	1.39	1.20 – 1.60	<0.001
Nature of Injury: Amputationa	2.10	1.51 – 2.91	<0.001
Injury Severity Score			<0.001
1 – 8	Ref
9 – 15	1.60	1.31 – 1.95
16 – 24	2.15	1.68 – 2.76
25 – 39	3.38	2.46 – 4.64
40 – 75	3.69	2.50 – 5.44
Respiratory Rate			<0.001
Intubated	1.67	1.23 – 2.26
Not breathing	1.09	0.70 – 1.71
Hypopnea	1.23	1.05 – 1.45
Normal	Ref
Tachypnea	1.26	1.10 – 1.44
Glasgow Coma Scale			<0.001
15	Ref
13 – 14	1.76	1.47 – 2.12
9 – 12	3.44	2.82 – 4.21
6 – 8	5.37	4.15 – 6.95
4 – 5	6.76	5.39 – 8.49
3	5.61	4.44 – 7.07

ARDS, Acute respiratory distress syndrome; RR, relative risk; CI, confidence interval

^aAbbreviated Injury Score ≥3

ARDS risk factors representing all three domains of demographic, injury, and clinical characteristics remained in our final multivariable model. African American race (RR 1.42, 1.13–1.78) was the only demographic factor that remained in the model. Among the injury characteristics, firearm injuries (RR 1.93, 1.58–2.36) and MVCs (RR 1.91, 1.57–2.31) remained the injury mechanisms most strongly associated with ARDS relative to falls. Injuries to the spine (RR 1.39, 1.20–1.60), chest (RR 1.36, 1.22–1.52), and lower extremities (RR 1.26, 1.10–1.44) were all associated with ARDS development, as were injuries involving amputations (RR 2.10, 1.51–2.91). Compared to ISS<8, each increasing ISS category was associated with progressively higher risk of ARDS.

Of the clinical factors, the ED vital signs that remained associated with ARDS were abnormal respiratory rate (intubation [RR 1.67, 1.23–2.26], hypopnea [RR 1.23, 1.05–1.45], or tachypnea [RR 1.26, 1.10–1.44]) and lower GCS (RR 5.61, 4.44–7.07 for GCS 3 vs. 15). The association between ARDS and GCS did not change when controlling for intubation, chemical paralysis, or eye injury.

Chest injury characteristics (Table 5)

Table 5.

Characteristics of chest injuries with Abbreviated Injury Scale ≥3 among patients who developed ARDS, with relative risks for ARDS development on bivariate and multivariable Poisson regression

Risk Factor	Incidence of ARDS	Bivariate Analysis			Multivariable Analysisa
	No. (%)	RR	95% CI	p-value	RR	95% CI	p-value
	(n = 1269, 4.2%)
Lung contusion	1011 (4.6)	3.74	3.37 – 4.16	<0.001	1.58	1.31 – 1.90	<0.001
Pneumo/Hemothorax	675 (4.2)	2.92	2.58 – 3.30	<0.001
Rib/sternum fracture	341 (4.3)	2.67	2.29 – 3.12	<0.001
Chest wall injury	39 (4.1)	2.36	1.52 – 3.66	<0.001
Diaphragm injury	47 (3.9)	2.22	1.69 – 2.93	<0.001
Cardiac injury	15 (5.4)	3.12	1.61 – 6.04	0.001
Vascular injury	42 (7.1)	4.21	3.11 – 5.71	<0.001	1.62	1.23 – 2.12	0.001
Type of injury				<0.001			<0.001
Penetrating	89 (2.7)	Ref			Ref
Blunt	1095 (4.4)	1.65	1.33 – 2.05		1.54	1.21–1.95
AIS of chest injury				<0.001			<0.001
3 “Serious”	995 (3.8)	Ref			Ref
4 “Severe”	227 (6.8)	1.82	1.35 – 2.47		1.58	1.20 – 2.08
5 – 6 “Critical /Maximum”	47 (6.3)	1.69	1.30 – 2.21		2.24	1.66 – 3.03

ARDS, Acute respiratory distress syndrome; RR, relative risk; CI, confidence interval; AIS, Abbreviated Injury Scale

^aReported variables are those that remained in final model with p<0.005

Half of patients with ARDS sustained a chest injury with an AIS ≥3. Chest injuries associated with ARDS on multivariable analysis included intrathoracic vascular injuries (RR 1.62, 1.23–2.12) and lung contusions (RR 1.58, 1.31–1.90). Blunt chest injuries were associated with higher risk of ARDS relative to penetrating injuries (RR 1.54, 1.21–1.95). Higher chest AIS was strongly associated with ARDS (RR 2.24, 1.66–3.03 for AIS 5–6 relative to AIS 3). In patients with ARDS without a serious chest injury, sustaining a spinal injury was associated with the highest risk of ARDS development (RR 2.42, 1.90–3.07) among the body regions.

Discussion

In this study, we provide the first description of the incidence of and risk factors for development of ARDS associated with pediatric trauma. In this ten-year cohort of nearly 150,000 children admitted to the ICU following a traumatic injury, 1.8% developed ARDS. Risk factors identifiable at the time of hospital arrival spanned all domains evaluated, including patient demographics, injury characteristics, and clinical parameters, with ISS and presenting GCS the factors associated with the highest risk of ARDS development.

The 1.8% incidence of ARDS among pediatric ICU patients is lower than the ARDS incidence among adults with traumatic injuries.^{3–5,7,10–11} This is consistent with previous findings that children have lower incidence of ARDS from all etiologies compared to adults.^13,26 While this may in part be due to differences in lung and chest wall development, pulmonary physiology, biomechanics, and mechanisms of injury, there may also be under-recognition of the problem.^27–28 Children have historically been classified as having ARDS based on diagnostic criteria developed for use in adults with no specific pediatric considerations,^24–25 which likely has underestimated the true incidence of ARDS in the pediatric population.

Our data do demonstrate, however, that ARDS is a common complication among the most critically injured children. While our cohort of all post-traumatic pediatric ICU patients represents a wide range of injury severity, nearly 7% of patients with ISS ≥16 who were hospitalized ≥3 days and required mechanical ventilation developed ARDS. This highlights the need to better understand the patient factors, injury types and mechanisms, and early clinical signs that are most commonly associated with pediatric post-traumatic ARDS in order to better anticipate its development and minimize its impact on injury recovery.

The only patient factor that was independently associated with higher risk of ARDS development was African American race relative to white race. This association remained significant even after analyses controlling for age, insurance, type of trauma center, injury mechanism, blunt vs. penetrating injury, and injury severity. We presume, however, that race functions in our regression model as a surrogate for a variety of injury, socioeconomic, and quality of care factors that we were not able to control for, as there is no physiologic basis for this association and previous evaluations have not found an association between race and risk of ARDS.^8–9

While thoracic trauma is associated with high mortality in children and frequently results in lung contusions due to a compliant chest wall,²⁹ we found that overall injury severity was much more strongly associated with ARDS development than whether or not a chest injury was present. ISS remained very highly associated with ARDS even after controlling for confounding factors such as shock that likely contribute greatly to a systemic inflammatory response. While chest trauma occurred in half of ARDS patients, injuries to the spine and lower extremities conferred a similarly elevated risk of ARDS as direct chest injuries. This supports the concept of post-traumatic ARDS as a result not just of direct pulmonary insult, but of overwhelming systemic inflammation due to remote as well as local tissue injury.

The risk of pulmonary complications following spinal cord injury is well known.^30–31 Our data demonstrate that children are at high risk for developing ARDS after spinal injuries. Children who sustain spinal trauma likely experience prolonged mechanical ventilation, impaired airway clearance, and risk for aspiration, and warrant attention to their potential for ARDS development similarly to patients with chest trauma.

Physiologic factors present on ED arrival may also help identify patients at highest risk for ARDS development. While abnormalities across all vital signs on bivariate analyses were likely confounded by injury severity, respiratory status and GCS remained risk factors for ARDS development on multivariable analysis. With fewer than 15% of patients with ARDS having had a normal respiratory rate on ED presentation and only 16% with a normal GCS, these markers, especially when combined with a high ISS, are potential early and sensitive indicators of risk for ARDS.

There were several limitations to this study. The NTDB is not population-based, and thus the reported ARDS incidence is only for NTDB facilities meeting study criteria. The NTDB represents nearly all Level I or II adult or pediatric trauma centers in the U.S., however, and thus likely represents the majority of children at risk to develop ARDS following trauma.

Additionally, adult ARDS definitions have limitations when applied to children.²⁶ New pediatric consensus criteria (PALICC) were published in 2015,²⁷ however the NTDB had not yet adopted these criteria as of 2016. Recent studies have demonstrated that PALICC criteria identify more patients compared to AECC or Berlin criteria, and that patients with more severe disease are identified by both PALICC and AECC/Berlin criteria.^32–33 We thus believe that our cohort as identified by AECC/Berlin criteria represents a more severely ill subpopulation of the patients who would have been identified by PALICC criteria.

The timing of ARDS onset is not included in the NTDB dataset, and thus we were unable to evaluate whether events such as aspiration or pneumonia occurred prior to ARDS onset. Our findings are thus inclusive of the full scope of ARDS triggers that patients are susceptible to following traumatic injury, including direct chest trauma, severe systemic inflammation, transfusion-associated lung injury, aspiration, and pneumonia.

Finally, there were >10% missing data for insurance type, oxygen saturation, and temperature; we relied on multiple imputation methods to estimate the associations between imputed variables and ARDS. The only variable that differed between models was oxygen saturation, which was present in the complete case analysis (Supplemental Digital Content 3) but not our imputed model. The estimated relative risks for the other variables remained similar between the two models.

Conclusions

Multiple demographic, injury, and clinical characteristics identify children at highest risk for ARDS development at the time of hospital arrival. Early recognition and diagnosis of ARDS allows for timely implementation of therapies such as lung protective ventilation and conservative fluid administration that have been associated with reduced ARDS mortality after trauma,⁸ and that may help limit ARDS development.¹⁸ Improved understanding of the patient characteristics and patterns of injury associated with ARDS development after trauma allows the opportunity for early intervention in high-risk patients to diagnose ARDS earlier in its clinical course and potentially alter clinical practice that may minimize progression of lung disease and ultimately improve outcomes for critically injured children.