Survival rates after pediatric traumatic out-of-hospital cardiac arrest suggest an underappreciated therapeutic opportunity

Abstract

Objectives.

Children with traumatic arrests represent almost one third of annual pediatric out-of-hospital cardiac arrests (OHCA). However, traumatic arrests are often excluded from study populations since survival post traumatic arrest is thought to be negligible. We hypothesized children treated and transported by Emergency Medical Services (EMS) personnel after traumatic OHCA would have lower survival compared to children treated after medical OHCA.

Methods.

We performed a secondary, observational study of children <18 years of age treated and transported by 78 EMS agencies in southwestern Pennsylvania after OHCA from 2010–2014. Etiology was determined as trauma or medical by EMS services. We analyzed patient, cardiac arrest, and resuscitation characteristics and ascertained vital status using the National Death Index. We used multivariable logistic regression to test the association of etiology with mortality after covariate adjustment.

Results.

Forty-eight (23%) of 209 children had traumatic OHCA. Children with trauma were older than those with medical OHCA (13.2 (3.8 – 15.9) vs. 0.5 (0.2 – 2.4) years, p<0.001). Prehospital return of spontaneous circulation (ROSC) frequency for trauma vs. medical etiology was similar (90% vs 87%, p=0.84). Patients with trauma had higher mortality (69% vs. 45% p=0.004).

Conclusion.

More than 8 of 10 children with EMS treated and transported OHCA achieved ROSC. Despite lower survival rates than medical OHCA patients, almost one third of children with a traumatic etiology survived throughout the study period. Future research programs warrant inclusion of children with traumatic OHCA to improve outcomes.

Introduction

The incidence of pediatric out-of-hospital cardiac arrest (OHCA) in the US is 8–20 per 100,000 children with one-third of these events due to traumatic etiology.¹ Return of spontaneous circulation (ROSC) following pediatric traumatic arrest has been reported to be between 20–41% while hospital survival ranges from 1 to 24%.^1,2,3,4 Mortality following pediatric traumatic OHCA is associated with lack of bystander cardiopulmonary resuscitation (CPR) and head and neck trauma.^5,6 In children with ROSC, most deaths occurred within one day, but there is a paucity of long-term, post-discharge outcome data.^5,7

Despite the frequency of pediatric traumatic OHCA, few studies have specifically evaluated its outcome and children with this mechanism are commonly excluded from research trials.^{5,6, 8–11} Comparative evaluation of medical and traumatic OHCA characteristics and long-term outcomes may inform research study design and clinical care that leads to better outcomes for this understudied population.

We sought to analyze an existing regional OHCA database to compare rates of ROSC and vital status of children after an OHCA of traumatic versus medical etiology. We hypothesized that children with a traumatic OHCA will have worse vital status than children with medical OHCA.

Materials & Methods

Design

This is a secondary observational study of pediatric OHCA patients from an existing regional OHCA database. The database, previously described,¹² collected data from prehospital records of responding Emergency Medical Services (EMS) (N=78) in Southwestern Pennsylvania for patients of all ages with OHCA from 2010–2014. A common National EMS Information Systems (NEMSIS)-compliant web-based prehospital electronic health record, called emsCharts, was used to evaluate these pre-hospital records. This system has custom reporting which makes it possible to identify cases and report data based on multiple fields in the record. An electronic query was built so that certain requested content was pulled automatically, and manual chart abstraction was not needed for data collection. Using this system, EMS records were queried for sudden cardiac arrest and included patients with a medical category of “cardiac arrest” or documentation of any of the following in the EMS record: chest compressions, defibrillation, or automated defibrillator use. Patient information including name, sex, date of birth, race, status of last known residence and social security number were extracted and used to query the National Death Index for vital status until the end of December 2014. Data from the prehospital EMS records and the National Death Index were de-identified and organized into a collective database for analysis.

This study was approved by the University of Pittsburgh Human Research Protection Office and through the Institutional Review Boards (IRB) at Allegheny Health Network, Excela Health hospitals and St. Clair hospital.

Prehospital EMS Treatment Standards

Statewide pediatric ALS protocols outlined by the Pennsylvania Department of Health detail that EMS personnel are required to contact medical command for orders to terminate resuscitation in pediatric patients with OHCA, except where death on arrival criteria are met.¹³ In addition, statewide BLS protocol states that any infant less than 1 year old is to be immediately transported once CPR is started without further analyzing vital status in the field.¹³

Study Population

Patients included in this analysis were aged 0–17 years and were treated and transported by an EMS agency for an OHCA. A patient was defined as treated if chest compressions were administered, or if defibrillation, or automated external defibrillator (AED) was used. Children that were not treated for a cardiac arrest at the scene by EMS were not included in the original study database.

Data collection

Patient demographics included age, sex, and race. Prehospital OHCA event characteristics included prehospital ROSC, rhythm (shockable vs. non-shockable), bystander CPR, witnessed status, epinephrine given, transport mode (e.g., ground or helicopter), and transport times (e.g., dispatch to scene, scene, scene to hospital arrival). Prehospital ROSC was assigned as yes if, upon arriving in the ED, the patient was not pulseless, was not receiving ongoing CPR and was not displaying obvious signs of death. A shockable rhythm was defined as a rhythm of ventricular tachycardia (VT) or ventricular fibrillation (VF) or if prehospital defibrillation was provided. Epinephrine was designated as not given if it was documented as zero doses or if there was no documentation of the medication being given. Mortality was recorded as yes if the patient had died or no if the patient survived until the end of December 31^st, 2014.

Outcome Measures

The primary outcome was mortality from January 2010 until the end of December 2014 for traumatic vs. medical pediatric OHCA patients. Secondary outcomes were rates of ROSC and clinical variables associated with death.

Statistical Analysis

We compared clinical and outcomes data by traumatic vs. medical etiology using descriptive statistics. Data were nonparametric and reported as median and interquartile range (IQR) values. The traumatic cohort was compared to the medical cohort using a Fisher Exact test for categorical data and Wilcoxon rank sum for continuous variables. Mortality in the traumatic vs. medical cohorts was evaluated with a survival time variable. This variable was created by taking the death date on the death certificate minus the day EMS was dispatched for patients that died. For patients with no death certificate in the database, survival time was calculated using the date of last observation minus the day EMS was dispatched for that patient. The last date of observation for patients with no death certificate was the end of the research period on December 31^st, 2014. A chi-square test was then used to evaluate mortality by etiology at different time points (same day, 1-day post arrest, 10-days post arrest, 1-month post arrest and overall mortality). Variables with a p-value of less than 0.1 in univariable regressions were included in a backwards stepwise multivariable logistic regression including the primary outcome of mortality. Significance was determined as p<0.05 with a 95% confidence interval (CI). Statistical analyses were performed in STATA, version 13.

Results

Patients

There were 213 cases of EMS transported pediatric OHCA during the study period. Of these cases, four were excluded due to lack of National Death Index data (Figure 1), providing 209 patients for analysis (Table 1). The median (interquartile range [IQR]) age of patients in the study was 1.1 (0.2–8.7) and 37% were female. Forty-eight patients had a traumatic arrest (23%). Traumatic arrest patients were older than medical arrest patients (13.2 [3.8–15.9] vs. 0.5 [0.2–2.4] years, p<.001).

Figure 1.

Patient cohort consort diagram. The survival data for pediatric patients with OHCA were included (n =213). Four patients were excluded after a non-successful query in the National Death Index. The resulting patients were included in the final analysis (n = 209) and encompassed two groups, medical (n=161) and trauma (n=48).

Table 1.

Baseline patient, resuscitation, and transport characteristics: overall and by etiology

Median (IQR), n(%)	Total N=209	Medical N=161	Trauma N=48	P - Value
Age, years	1.1 (0.2 – 8.7)	0.5 (0.2 – 2.4)	13.2 (3.8 – 15.9)	<0.001
Female, sex (N = 208)	77 (37)	58 (36)	19 (40)	0.734
Race (N = 209)				0.224
Caucasian	115 (55)	83 (51)	32 (66)
Black	39 (19)	32 (20)	7 (15)
Other	8 (4)	8 (5)	0
Unknown	47 (22)	39 (24)	9 (19)
Witnessed Arrest (N = 168)	64 (38)	42 (34)	22 (49)	0.106
Bystander CPR (N = 101)				0.051
Layperson	61 (60)	53 (65)	8 (40)
Professional	32 (32)	21 (26)	11 (55)
None	8 (8)	7 (9)	1 (5)
Shockable Rhythm (N = 209)1	67 (32)	50 (31)	17 (35)	0.599
Epinephrine Given (N = 209)2	56 (27)	32 (20)	24 (50)	<0.001
Prehospital ROSC (N = 209)	183 (88)	140 (87)	43 (90)	0.804
Survival to censored period (N = 209)	108 (51)	88 (55)	15 (31)	0.005
Days of survival during research period3	704 (235–1237)	709 (395 −1214)	463 (3 – 1306)	0.353
Transport Mode (N = 204)				<0.001
EMS4 Transport	184 (92)	154 (98)	30 (68)
Helicopter Transport	17 (8)	3 (2)	14 (32)
Time from dispatch to scene, minutes	7 (5 – 12)	6 (5 – 9)	12 (6 – 17)	< 0.001
Time on scene, minutes	9.5 (5 – 17)	8 (4 – 14)	17 (8 – 24)	<0.001
Time from scene to hospital arrival, minutes	9 (5 – 14)	9 (5 – 13)	10 (8 – 15)	0.069
Total EMS time, minutes	27 (19 – 40)	26 (18 – 35)	42 (25 – 57)	< 0.001

¹Shockable rhythm designated if shockable rhythm, (e.g.VF/VT) was recorded any time during treatment or transport, or if patient received shock during transport.

²Epinephrine designated as zero if it is documented as zero or if there is no documentation showing that was given.

³Days of survival ended if patient died or if patient reached end of December 2014

OHCA, out-of-hospital cardiac arrest. CPR, cardiopulmonary resuscitation. ROSC, return of spontaneous circulation.

⁴EMS, emergency medical services.

Event Characteristics

The frequency of prehospital ROSC was similar for trauma (43 [90%] vs. medical (140 [87%], p=0.804) OHCA patients. 49% of patients in the trauma cohort had a witnessed arrest compared to the medical cohort (34%), p=0.106. Bystander CPR was initiated by a professional responder 55% of the time in the trauma cohort vs. 26% of the time in the medical cohort, p=0.051. The frequency of shockable rhythms was similar for each cohort, but more children received epinephrine in the trauma cohort (50%) when compared to the medical cohort (20%), p<0.001.

Transport details

More patients in the trauma vs. medical cohort were transported to the hospital in a helicopter (32% vs. 2%, p<0.001) while 98% of patients in the medical cohort were transported by ambulance vs. 68% of the trauma cohort. Total EMS transport times were longer for traumatic vs. medical arrest patients (42 [25–57] vs. 26 [18–35] minutes, p<0.001). Time from dispatch to scene and time on scene were significantly longer for trauma vs. medical patients (Table 1). The time from scene to hospital was longer for trauma vs. medical patients (10 [8–15] vs. 9 [5–13] minutes, p<0.001).

Univariable and Multivariable Regression for Mortality

Univariable associations with mortality are reported in Table 2. Multivariable analysis revealed that traumatic etiology (OR 4.84, 95% CI:1.77 – 13.24) was associated with mortality in the censored period. In addition, witnessed arrest (OR 0.25, 95% CI: 0.11 – 0.54), prehospital ROSC (OR 0.05, 95% CI: 0.006 – 0.39), and longer EMS time (OR 0.98, 95% CI: 0.95–1.00), in minutes, were associated with decreased mortality.

Table 2.

Univariable and multivariable regression for mortality within the study period

	Univariable			Multivariable
Variable	OR	95% CI	p-value	OR	95% CI	p-value
Witnessed	0.29	0.15–0.56	<0.001	0.25	0.11 – 0.54	0.001
Prehospital ROSC	0.03	0.004-.24	0.001	0.05	0.006 – 0.39	0.004
Traumatic etiology1	2.65	1.34–5.26	0.005	4.84	1.77–13.24	0.002
Age, years	1.02	0.98–1.06	0.417		-
Shockable rhythm	2.05	1.13–3.71	0.018	1.88	0.85–4.14	0.117
Bystander CPR	1.13	0.61–2.09	0.688		-
Total EMS time, min2	0.98	0.97–1.00	0.046	0.98	0.95–1.00	0.022
Ambulance vs. helicopter	0.87	0.32–2.35	0.783		-

¹Etiology was designated as either trauma or medical by EMS services.

²Total EMS time was the amount of time (in minutes) from dispatch to arrival at initial facility

OR, odds ratio; CI, confidence interval; ROSC, return of spontaneous circulation; CPR, cardiopulmonary resuscitation; EMS, emergency medical system

Mortality through the end of December 2014

Children with a traumatic etiology of OHCA had higher mortality at 1-day, 10-days, and 1-month post-event compared to children with medical etiology, all p<0.05. (Figure 2). Same day mortality was similar between trauma (48%) and medical (36%), p=0.138. More children died by the end of the research period in the trauma cohort (69%) compared to the medical cohort (45%), p=0.004.

Figure 2.

Mortality endpoints during censored period by etiology. A survival time variable was created for deceased patients (death certificate date minus date of EMS dispatch) and those alive at end of research period (date of last observation minus date of EMS dispatch) A chi-square test evaluated different mortality time points (same day, 1-day post arrest, 10-days post arrest, 1-month post arrest and overall mortality), p<0.05 with a 95% confidence interval. Higher mortality for trauma cohort found at every time point. Significant differences between the two groups were seen for every time point except prehospital ROSC and same day mortality.

Discussion

In this study of pediatric patients with OHCA, we identified a very high frequency of prehospital ROSC rates for children with traumatic and medical etiology. Although mortality was higher for children with traumatic vs. medical OHCA, survival from traumatic OHCA was substantially better than described in previous reports.^1–4 However, traumatic etiology of OHCA remained associated with mortality after covariate adjustment.

Traumatic arrests are frequently excluded from study populations involving OHCA, particularly in children. The exclusion of traumatic cases may result from prior studies suggesting futility in these cases.^1,3 That approach critically limits the ability to both inform care and direct future research necessary to improve outcomes. A key finding of our study, thus, is that despite the fact that the traumatic cohort had a higher rate of mortality than the medical cohort, one third of victims of pediatric traumatic OHCA exhibited long term survival. Our findings strongly support both the need and value of additional investigation of pediatric traumatic OHCA to optimize outcomes.

Patients with traumatic OHCA represented a quarter of the events in our database in whom resuscitation was attempted. Prehospital ROSC rates were higher than previously reported in other studies, where they ranged between 20–41%.^1–4,14 The multivariable regression showing an association between prehospital ROSC and decreased mortality aligns with previous reports.^5,6,14,15 Despite having similar prehospital ROSC rates, mortality through the research period was higher for the traumatic cohort compared to the medical cohort. Furthermore, the mortality data displayed a distinct pattern for each etiology. The trajectory of mortality diverged and became different after the same day mortality end point. No child with traumatic etiology died after 10 days post arrest, while mortality after a medical arrest continued to increase steadily over the research period. Early death after pediatric trauma has been described previously in McLaughlin et. al,⁷ but their categorization of early vs. late death was < 24 hr or > 24 hr during hospital admission. Also, late complications were less frequent in their pediatric population, which would align with our trajectory of survival which was 100% in the trauma cohort after 10 days post arrest. The 2015 Utstein guidelines encourage longer or more detailed tracking up to 12 months, where feasible.¹⁶ Our use of the National Death Index to track mortality end points provided the opportunity to monitor patients for deaths (up to 2 years post arrest) for most patients in our dataset, longer than most reports in pediatric cardiac arrest. The National Death Index could, therefore, serve to enhance future studies on pediatric OHCA.

When analyzing the mortality trajectory between etiologies, we also observed that overall mortality was significantly higher in the trauma cohort vs. the medical cohort, despite similar prehospital ROSC rates. Furthermore, if a child who suffered a traumatic arrest survived to the 10-day mortality end point then they went on to survive until the end of the research period in December 2014. This identifies the need to examine the early post resuscitation period after traumatic OHCA in children in a more granular fashion to identify potential factors that contributed to this mortality difference vs. the medical cohort. Although the specific etiology of traumatic arrest was not included in this study, other research has found survival to be higher in blunt force trauma as opposed to penetrating injuries,^17,18,19 which could have inadvertently influenced our mortality data. Our prehospital records were not linked to hospital data, limiting our ability to address this issue. Lott et al, in the 2021 European Resuscitation Council guidelines, recently described traumatic arrest as a special circumstance with specific considerations regarding an algorithm for resuscitation practices in this population.²⁰ Further research evaluating the in-hospital mortality trajectory and clinical practice of the special considerations noted above are indicated to identify possible areas to improve post-arrest management and outcomes for traumatic arrest patients.

Similar to previous literature, witnessed arrests were associated with decreased mortality in our study.^21,22 Also, witnessed arrests were found to be more common in our traumatic vs. medical cohorts, even though trauma had a higher overall mortality. A possible reason for this discrepancy could be the etiology and nature of an arrest in the medical cohort. Sudden infant death syndrome (SIDS) contributes importantly to the prevalence of medical arrests in infants.^23,24 In our study, of the 73 patients who died in the medical cohort by the end of the censored period, almost 50% of those were labeled as SIDS or unknown infant mortality on their death certificates. This group of infants also reduced the median age of patients in our medical group compared to the traumatic cohort. Atkins et al. found that infants were the least likely to have a witnessed arrest and represented the group with the highest incidence and lowest survival rates in their study population.²⁴ This may contribute to the incongruity between the percentage of witnessed arrest by etiology and the results of our multivariable model. Bystander CPR has been favorably associated with better outcomes in prior studies.^1,5 We were limited by high rates of missing data for reporting bystander CPR, and thus further research is necessary to define its impact particularly in the setting of traumatic OHCA, where data are limited.

Transport data for the trauma cohort showed significantly longer total EMS time, time on scene, and time from dispatch to the scene. Chen et. al showed that the time from scene to hospital was shorter in their traumatic vs medical cohorts. However, in their medical cohort, 13.7% were transported via private vehicle as opposed to EMS.⁶ This contrasts with our emergency response system in the United States and may contribute to the divergence with our data, which showed an insignificant, yet slightly longer scene to hospital time in the trauma cohort. Also, total EMS time, time from dispatch to scene, and scene time were not evaluated in the study aforementioned. Discrepancies for the longer times found in the trauma cohort during our study could be related to the higher number of patients transported by air in the trauma group, a greater need for interventions in regards to trauma care provided at the scene and longer distances traveled to arrive at a level I pediatric trauma center instead of the nearest hospital.

Our study was limited by the fact that the database did not include hospital data and was also limited to children who were treated at the scene and transported to the hospital. Further, etiology of cardiac arrest is often not known or precisely determined in the prehospital phase, thus not reported in this study. Thus, children with cardiac arrest due to abusive head trauma (or other abuse), for example, whose diagnosis may not be available until investigation post-hospital admission, may have been misclassified as medical arrests in this analysis. Although field termination of a pediatric patient is exceedingly rare, the incidence of non-transports was not quantified in our study population, which could have had an unintentional effect on the data involving older traumatic cohort patients where scene termination may be more likely. Also, the completeness of documentation varied considerably between EMS services. Therefore, all arrest variables and characteristics were not available through documentation for every patient in the database. Same day mortality was determined by dates of arrest and thus was not synonymous with 24-hour mortality. Therefore, these time points in the analysis could be affected by the time of day that the arrest took place. Prehospital records were not linked to hospital records; thus, survival to hospital discharge was not assessable. We were unable to report functional outcomes in this study, which are emerging as a key determinant of quality of survival after pediatric arrests.^25,26,27 Finally, these data were collected from EMS agencies in Southwestern Pennsylvania, which may not be representative of all regions in the US or North America.

Conclusions

Prehospital ROSC was similarly high for pediatric patients with traumatic vs medical OHCA. Despite lower survival rates, one-third of children in the trauma OHCA cohort survived through the study period, which is higher than previously reported.^1,3,4 Our findings suggest that future research on pediatric traumatic OHCA is warranted and represents a largely overlooked opportunity to improve resuscitation outcomes.

Abbreviations:

OHCA

Out-of-hospital cardiac arrest

ROSC

return of spontaneous circulation

CPR

cardiopulmonary resuscitation

EMS

emergency medical services

SIDS

sudden infant death syndrome

OR

odds ratio

CI

confidence interval