FOCUS ON CARDIAC ARREST Pediatric Out-of-Hospital Cardiac Arrests: An Epidemiological Study

Objective: To identify the epidemiological patterns of pediatric out-of-hospital cardiac arrests (OHCA) in Queensland, Australia and to investigate associations between patient variables and prehospital outcome. Methods: Included were pediatric ( > 4days – 18years) OHCA patients attended by paramedics in the state of Queensland (Australia) between January 2009 and December 2019. Patient and arrest characteristics were described. Factors associated with return of spontaneous circulation (ROSC) on hospital arrival were investigated. Results: A total of 1,612 pediatric patients were included; 611 were deceased prior to paramedic arrival and 1,001 received resuscitation attempts by paramedics. Approximately one quarter (26.8%) of resuscitation-attempted patients achieved ROSC on hospital arrival. Most arrests (49.7%) were due to medical causes. Arrests due to trauma had the lowest rate of ROSC on hospital arrival (9.6%), whereas those due to drug overdose had the highest rate (40%). Patients in rural areas had a lower rate of ROSC on hospital arrival than those in metropolitan areas (20.7% vs 32.5%, p < 0.001). The median response interval to all OHCA patients was 8minutes. Trauma was considerably more prevalent in rural areas than in metropolitan areas, while all other etiologies were comparable. Older pediatric age groups had higher rates of ROSC on hospital arrival than infants, particularly early adolescents (39.4% vs. 14.9%, p ¼ 0.001). Etiology, age, bystander witness, shock-able initial rhythm, and geographic locality factors were independently associated with ROSC on hospital arrival. Conclusions: Approximately a quarter of pediatric prehospital OHCA achieved ROSC on hospital arrival. Prehospital outcome differs according to patient cohort and is associated with diverse patient demographic variables.


Introduction
Pediatrics are a patient group that require special physical assessments and treatments as well as specific communication techniques and emotional supports (1). In medicine, the pediatric population is typically defined as patients up to 18 years of age, although a globally accepted standard definition does not exist (2). They are a complex cohort, and elicit a significant emotional response in health care professionals. Pediatric patients are complex for several reasons; they tend to deteriorate or decompensate faster, have smaller bodies with higher physiological instability, treatment is complicated by the presence of parents with heightened emotions, and they cannot always communicate as easily as adults, making key history unreliable or often impossible to ascertain (3).
Children account for approximately 3-29% of prehospital fatalities (4)(5)(6)(7)(8). As a minority patient group, paramedics are infrequently exposed to managing these cases, making skills maintenance challenging (1). Paramedics are required to accommodate for the care of different demographics and illnesses and retain knowledge to attend multiple subsets of age groups. The anatomy and therefore medical needs of the infant group are disparate to that of the early adolescent age group, which necessitates individual needs (2). This has resulted in paramedics reporting that exposure to this lowfrequency, high-pressure scenario has caused hesitance, anxiety, or feelings of unpreparedness (1). Practitioners rely on research to guide evidence-based best practice to improve the outcomes of these patients; however, research in this arena is scant.
Pediatric out-of-hospital cardiac arrest (OHCA) requires investigation, and of particular importance is assessment of the underlying causes and the identification of survivability trends. Inconsistency of research reporting has made it difficult to determine patterns in causes of prehospital mortality specific to the pediatric cohort (4,9). Current literature highlights that administration of epinephrine (10), shorter response times (11), presenting shockable rhythm, and witnessed arrest appear to be some defining variables that make a difference in pediatric survival rates (12). Research clearly identifying and exploring survival trends of OHCA can be used to pinpoint areas of academic priority to inform educational approaches. The absence of a standard definition for this population adds to the inconsistency in research (4,9). This suggests an opportunity to exclusively research paramedic experiences within a defined group of children from 4 days to 18 years old among all etiologies causing OHCA.
The purpose of this study is to describe the epidemiology and characteristics of pediatric OHCA attended in the prehospital setting by paramedics in Queensland, Australia, and associations with prehospital outcome.

Study Design, Setting, and Data Source
The Queensland Ambulance Service (QAS) is a state-wide emergency ambulance and patient transport service that provides prehospital health care to 5.19 million residents plus visitors to the state of Queensland (Australia) across 1.7 million square kilometers (13,14). Queensland's pediatric population makes up 25.6% of its diverse population (14). QAS clinical guidelines for pediatric cardiac arrests are informed by Australian Resuscitation Council guidelines (15).
When attending an OHCA, the QAS uses a two-tiered response model that consists of two sub-levels of advanced care paramedics (ACP 1 and ACP 2) and critical care paramedics (CCP). ACPs are trained to provide advanced life support, including airway management, respiratory procedures, cardiac management, drug and fluid administration, the administration of basic and advanced pharmacological agents, and the recognition of life extinct. ACP 1 s can, among other drugs, administer midazolam, morphine, naloxone, and adrenaline through intramuscular or intranasal access and use semi-automatic defibrillation, bag-valve-mask ventilation, and insertion of laryngeal and nasopharyngeal airways. ACP 2 s have the additional capacity to administer a wider range of drugs including tenecteplase and enoxaparin, gain peripheral intravenous access, use manual defibrillation, and identify STelevation myocardial infarctions on 12-lead electrocardiograms. CCPs are skilled in all ACP procedures, as well as more advanced procedures and administration of additional drugs such as amiodarone, ketamine, and propofol with further access to external jugular intravenous cannulation and intraosseous access. The detailed scope of practice of ACPs and CCPs can be found in Supplementary Material 1 (16). The QAS response to OHCA typically involves concurrent deployment of ACPs and CCPs (where available).
The QAS transports medical etiology OHCAs to the closest local hospital. For trauma OHCAs, the patient will be transported directly to a major trauma service if there is one within 15 minutes of road transport time (17). If a major trauma service is longer than 15 minutes of road transport time, the patient will be transported to the closest appropriate hospital  (17). Pediatric patients who are deceased on arrival (where they are exhibiting obvious signs of death or sustained injuries that are incompatible with life) or resuscitation was attempted but death is declared at the scene are not transported by paramedics. Death is declared in the field after 20 minutes of resuscitation when all of the Recognition of Life Extinct (RoLE) criteria are met (18).
This study was a retrospective analysis of the QAS OHCA database. The database, established in 2000, records all OHCA cases attended by QAS paramedics via a standardized electronic data collection form. Over 140 variables are collected for each patient, capturing patient demographics, prehospital times, and clinical and procedural data relating to prehospital management (19). Detailed descriptions of the QAS OHCA database can be found in previous publications (19,20).

Study Cohort and Measurements
OHCA patients of all etiologies, aged between 4 days and 18 years old, attended by QAS paramedics between January 1, 2009 and December 31, 2019 were included. Neonatal patients under 4 days old were excluded to avoid an obstetric bias. All OHCA patients who QAS paramedics were called to, were included in the study. This includes patients who were deceased prior to paramedic arrival or received resuscitation attempts on scene or during transport by paramedics. Patient characteristics were collected and assessed using the variables seen in Table 1.

Definitions
Although most demographic measures were extracted directly from the database, some measures were collapsed to aid in analysis. Geographic locality ("rurality") was classified into metropolitan (remoteness score 0 to 0.2), regional (remoteness score >0.2 and 2.4), and remote areas (remoteness score >2.4 and 5.92) based on incident address postcode according to the Australian Statistical Geography Standard, where areas are given a remoteness score from 0 to 15 and the lower the value, the greater the access to services (21). Patients in remote and regional areas were referred to as "rural", as this allows for more robust analysis given small numbers in remote areas (19,20).
The primary outcome measure was return of spontaneous circulation (ROSC) that sustained to hospital arrival (hereafter referred to as ROSC on hospital arrival) (22). Any ROSC induced in-hospital after paramedic treatment was ceased is not recorded in this data set.

Statistical Analysis
The primary aim of this study was to describe the epidemiology and characteristics of pediatric OHCA attended by paramedics in Queensland, Australia. The secondary aims were to investigate the prevalence and prehospital outcomes of etiologies within different patient groups such as rurality and age groups, and to discover any associations between resuscitation patient characteristics and prehospital outcome.
Data were first analyzed with standard descriptive analysis. Any continuous variables not normally distributed via visual inspection of histogram data were described with medians and interquartile ranges, while categorical variables were expressed with percentages. Age was initially presented as both a continuous and categorical variable, with all analysis then completed on the categorized groups using the Eunice Kennedy Shriver National Institute of Child Health and Human Development age stages (infant <1 year, toddler 1 to <2 years, early childhood 2 to <6 years, middle childhood 6 to <12 years, and early adolescence 12 to 18 years of age) to accurately designate medical need by age in childhood (2). This definition was widely cited and endorsed by a peer-reviewed systematic review of Australian pediatric health terminology (2).
Analysis was conducted in four stages. In the first stage, the characteristics of the entire sample population were described. To understand if there was any location effect on etiological presentations and outcomes, etiology prevalence analysis was conducted between rural and metropolitan patients. The second stage assessed the broad etiology groups (medical versus non-medical causes) for prehospital outcome while prevalence of each individual etiology was described within three cohorts: deceased prior to paramedic arrival (DOA), resuscitation attempted without attaining ROSC, and resuscitation attempted with ROSC attained.
Further analysis was undertaken on the patient group that received resuscitation attempts (with or without ROSC) in the third stage, specifying the epidemiology and rates of ROSC on hospital arrival of those patients treated by paramedics. The clinical characteristics of this group were described by age group and then by resuscitation outcome. Finally, characteristics of resuscitation attempted patients, grouped by attaining or not attaining ROSC on hospital arrival, were then compared in a univariate analysis. Variables with between-group p-values <0.2 in the univariate analysis were then included in a multivariate logistic regression to assess their association with prehospital outcome (20). A two-sided p-value <0.05 was considered statistically significant (20). The findings of the multivariate logistic regression were reported as adjusted odds ratios (AOR) and 95% confidence intervals (CI).

Research Ethics Statement
This study was approved by the Monash University Human Research Ethics Committee [approval number #25288] and complied with the Australian National Health and Medical Research Council research standards. QAS data are collected in compliance with the Information Privacy Act 2009 (Qld), and participants are considered to give implied consent.

Patient Characteristics
One thousand six hundred twelve children were identified and included in this study; Figure 1 shows the final sample.
The characteristics of the included patients are described in Table 1. The majority (60.4%) of patients were male, 58.6% of patients received bystander CPR, and four patients had public access defibrillation performed. Most arrests occurred in private residences (73.8%) and were unwitnessed (78.6%). Across all patients, early adolescence was the most prevalent age group (42.4%), followed by infants (28.9%). The median age of the DOA patient cohort was 14 years, whilst the median age of the resuscitation cohort was 3 years. The median response interval to all OHCA patients was 8 minutes.

Comparison of Rural and Metropolitan Cohorts
Comparison by rurality (Supplementary Figure 1) identified that most etiologies had similar prevalence between rural and metropolitan patients, aside from trauma, which had higher prevalence in the rural population (26.3% of rural OHCA compared to 16.2% of metropolitan, p < 0.001). Rural patients had similar median response intervals to metropolitan of 8 minutes for both cohorts. Of the rural resuscitation attempted cohort, 446 were regional (median response interval 8 minutes; with an interquartile range of 6-13), and 45 were remote (median response interval 7 minutes; with an interquartile range of 4-18). Bystanderwitnessed arrest rates were comparable between cohorts, with a slightly higher rate in rural (17.5% rural and 16.1% metropolitan; 22.7% rural and 22.4% metropolitan, for resuscitation attempted cohort only). For all etiologies combined, ROSC on hospital arrival in metropolitan was 32.5%, where rural was 20.7% (p < 0.001). The DOA to resuscitation attempt ratios showed disparity in age groups when comparing rural to metropolitan, with the exception of the infant age group, where 32% of both rural and metropolitan infants were DOA. 31.1% of rural toddlers were DOA compared to metropolitans 22.6%, early childhood 34.5% to metropolitans 21.1%, middle childhood 32.1% to metropolitans 17.9% and early adolescent 52.0% to metropolitans 45.5%. CCP attendance in rural settings was 63.1%, where 92.2% of metropolitan incidence had CCP attendance. A map of Queensland with a designation of metropolitan, regional, and remote areas can be found through the Australian Bureau of Statistics (23).

Etiology Comparison of DOA vs Resuscitation Attempt Cohorts
As described in Table 1, the medical etiology category was found to be the most prevalent cause of OHCA across both DOA (43.7%) and resuscitation-attempted patients (53.3%), followed by trauma (27.3% and 17.7%, respectively). In Figure 2 broad etiology group prevalence was assessed by patient outcome. Across the entire cohort, trauma had the lowest rate of ROSC on hospital arrival (9.6%), while the figure for medical etiology was 15%. Drowning (31.1%) and drug overdose (40.0%) had the highest rates of ROSC on hospital arrival.
Supplementary Figure 2 shows, in detail, the prevalence of each etiology, in which sub-etiologies of medical cause are shown. When etiologies were analyzed individually, trauma had the highest prevalence (27.3%) in DOA patients, followed by asphyxia (23.4%). Asphyxia was the dominant condition in resuscitation patients who sustained ROSC to hospital, representing 20.5% of this cohort. Sudden unexplained infant death (SUID) surpassed other etiologies in prevalence within the resuscitation without ROSC cohort at 24.8%.

Age Analysis in Resuscitation Attempt Cohort
Of those who received resuscitation attempts, early adolescents and infants remained the most prevalent groups (31.5% and 35.0%, respectively). Infants had the highest unwitnessed arrest rate, this generally decreased with age, and correspondingly, older age groups had increased rates of ROSC on hospital arrival (Table 2). Figure 3 shows how etiological prevalence changed between age groups. Across the entire resuscitation attempt cohort, SUID accounted for the highest prevalence (20.3%). However, when separated by age categories this condition was, by definition, only highly prevalent in infant (59.4%) and toddler (17.8%) cardiac arrests. As age increased, other medical etiologies decreased, and traumatic causes became more prominent. Drowning superseded all other etiologies in toddlers and early childhood (38.9% and 35.4%), trauma dominates the middle childhood group (29.3%), while asphyxia (hanging) events were the most common cause of cases in the early adolescent cohort (36%). Of the 311 cases of asphyxia or drug overdose of patients 6 years or older, 288 (92.6%) were considered suspected suicide attempts.

Paramedic Treatment Analysis in Resuscitation Attempt Cohort
The median scene interval for patients who received a resuscitation attempt was 26 minutes (interquartile range of 14-36 minutes). Transport interval from the scene to the hospital had a median of 11 minutes (interquartile range of 6-19 minutes). Supplementary Figure 3 describes the percentages of the cohort that received various treatments from paramedics. Most patients received intravenous or intraosseous access (51.8%), bag-valve-mask ventilation (79.7%), epinephrine at the scene (60.1%), and endotracheal tube intubation (56.1%).

Factors Associated with ROSC on Hospital Arrival
Univariate analysis of the level two sample is shown in Supplementary Table 1, to identify individual variables with statistically significant relationships to ROSC on hospital arrival. Both age and age group were found to be significantly associated with ROSC on hospital arrival (p < 0.001); however, for ease of analysis age group was the variable used for the age factor in the subsequent logistic regression. The multivariate logistic regression analysis in Figure 4 shows that drowning (AOR 3.70, 95% CI 2.13-6.41) and asphyxia (AOR 2.35, 95% CI 1.33-4.13) are independently associated with increased odds of ROSC on hospital arrival compared to medical etiology. Among the age groups, arrest in early adolescents was associated with 2.4 times (95% CI 1.44-4.00) increased likelihood of ROSC on hospital arrival by comparison to infants. Other independent factors positively associated with ROSC on hospital arrival include witnessed arrest, initial shockable rhythm, and arrest occurring in metropolitan areas (all p < 0.001).

Discussion
The aim of this study was to conduct an epidemiological analysis of all pediatric OHCA attended to by QAS paramedics, and to identify regional or age-based differences in occurrence, as well as factors associated with improved prehospital outcomes. The purpose of this analysis is to inform service provision and clinical training. We found that  children accounted for 3% of all OHCA patients attended by QAS paramedics. The overall rate of ROSC on hospital arrival for pediatric OHCA was 26.8%. A meta-analysis of international data found the figures for adult OHCA generally slightly higher, with an average 29.7% ROSC rate (24). The results from this study in children compare favorably to international results with rates of ROSC on hospital arrival ranging 11.1-19.3% (Australia, America, Denmark, and Korea) (9,(25)(26)(27).
There was a clear disparity in outcomes within the cohort based on individual causality; for example, rate of ROSC on hospital arrival of trauma (9.6%) or medical (15.0%) vs drug overdose (40.0%). This may be related to the reversibility of the underlying cause of arrest, as drug overdoses can be more readily reversed than major trauma, if found in time, and association between etiologies and likelihood of being witnessed (28). However, there is also potential that such disparities may inform the need for further investigation in whether current paramedic practice or emergency medical services (EMS) system design could be modified to improve patient outcomes, for instance via resuscitation protocols that incorporate etiology, as well as typical guidelines such as age and weight.
With such a high prevalence of OHCA in the infant cohort (28.9% of study population), consideration should be given to the high SUID rate without ROSC (24.8% of the no ROSC cohort) in the resuscitation attempt cohort. This may be attributed to the tendency of paramedics to attempt resuscitation for compassionate reasons when it may be clear that the attempt will be futile (29). It is hypothesized that this may be skewing the data, and further consideration should be given to including an additional data point to capture this for future studies as currently there is no data capturing reason for attempted resuscitation. This issue identifies a need to focus training for paramedics on how to handle these emotionally challenging and confronting situations, support the families, and ensure that there is followup support for the attending paramedics (given the high lethality of SUID).
SUID is also hard to quantify in medical data as an independent etiology due to the difficulty to diagnose by paramedics. Within these data it was often combined within the medical etiology for ease of analysis and understanding. The medical etiology, which equated to 43.7% of DOA patients and 53.3% of resuscitation, may be skewed with SUID responsible (by paramedic diagnosis) for 18.2% of DOA and 20.3% of resuscitation attempted patients. Consideration should also be taken for the common misdiagnosis of nonaccidental trauma in infants and toddlers. A large scale study found that EMS personnel were reporting child maltreatment and abuse of pediatric patients <3 years old at an incidence of 0.10%, where the national incidence was 1.46%, This drastic difference in reporting highlights a need for improving EMS detection and diagnosis of non-accidental trauma in pediatric patients (30).
This study highlights several priorities for further research into community prevention methods. First, 92.6% of all asphyxia and drug overdose cases of or above 6 years old in this study were suspected suicide attempts; that is 17.9% of the entire cohort of pediatric OHCAs. Pediatric suicide is a major concern in this state. Community suicide prevention techniques have proven to be effective. A systematic review of 29 youth suicide prevention programs found that 60% of school-based programs and 43% of communitybased interventions were effective in significantly reducing suicidal ideation or attempts in pediatrics (31). Yet, whilst there is government interest in youth suicide prevention programs in Australia, the literature highlights a major obstacle in the research of its effectiveness in this country due to the insufficient coordination of suicide prevention programs, which also leads to a limited opportunity to access these resources as a result of the convolution of the process (32). Given that prehospital clinicians have access to the home and high levels of community trust, there may be an opportunity here for future research into the use of paramedics as a community resource for just-in-time injury prevention and risk mitigation.
Drowning has been recognized within this study as a preventable high-yield etiology. With a 9.2% rate of occurrence in the pediatric OHCA cohort and a 31.1% ROSC rate, this is a cause of arrest that is highly preventable. Literature suggests the effects preventative steps in the community can have as participants of community education programs regarding water safety anecdotally report improvements to their drowning awareness; however, high-quality quantitative research on the effectiveness of these prevention methods is limited (33). There is a research and community need here that is not currently being met and must be investigated thoroughly in the future.
The study shows that patients in metropolitan areas have an increased likelihood of survival compared to rural patients. Response intervals in rural areas do not account for this disparity, as they are comparable to those in metropolitan Queensland. Whilst bystander-witnessed arrest rates are also comparable, Young et al. (34) suggests that a low population density with large property sizes mean it may take longer to discover an unwitnessed arrest, and extended transport to hospital distance in a rural setting increases the timeline of the chain of survival (34). Rural cases had a much lower rate of CCP attendance than metro (63.1% vs 92.2%, respectively). CCP attendance has been demonstrated previously to improve survival outcomes (19). Further research into increasing availability of CCP presence in these communities may be one strategy to compensate for the lower survival rates experienced in these regions. There is also a significant difference in etiology in these areas, where trauma is 10% more common. This etiology, with the lowest rate of ROSC on hospital arrival, elucidates the significant difference in prehospital outcomes of rural patients. This supports the academic priority to understand how better to prevent and react to traumatic incidents in children; any advancement here could support the pediatric population and improve rural outcomes.
To our knowledge this is the first study to analyze these patient characteristics within the pediatric population by defined age stages. This research showed that in comparison to infants, older age groups had better rates of ROSC on hospital arrival. There are several factors that may potentially explain this finding. For instance, children develop physiological robustness with age as the youngest age groups are anatomically disproportionate and take years to establish stable metabolisms and normative cerebral blood flow (35). Younger pediatric patient conditions can change more rapidly, being able to decline and improve drastically and suddenly (36). Given the sleeping patterns in the first 2 years of life, this younger age group also experience less supervision for more hours of the day, leaving them susceptible to unwitnessed arrest (37). These factors and statistics can also be attributed to the susceptibility to SUID of this younger age group (38). Future investigation may help us to better understand the different survival rates of the conditions more prevalent in the younger demographic and may suggest better practice strategies that could improve this statistic. To strengthen future comparisons, an agreement of pediatric age group definitions is required.
Several variables have been identified in current literature to be predictors of ROSC in the prehospital setting. Banerjee found that every minute delayed in EMS response time decreased the chance of ROSC and hospital admission by 5%. The Banerjee study had a ROSC rate of 20% with an interquartile range for response interval of 9-62.5 minutes (11). While our study reported a response interval interquartile range of 6-12 minutes. Another study found administration of epinephrine to be a predictor of prehospital ROSC, finding that 11.1% of pediatric patients who receive epinephrine by EMS achieved ROSC while 3.75% who don't receive epinephrine achieve ROSC. That study found only 7.7% of patients receive epinephrine from EMS and the cohort had an overall ROSC rate of 4.32% (10). Our study reported 60.1% received epinephrine at the scene and a further 6.5% during transport. The fact that Queensland reports higher statistics in all of these variables may be a factor in explaining the higher ROSC rate of 26.8% in this study compared to others in the literature, and supports the importance of these variables in pediatric OHCAs.
Witnessed arrest and shockable initial rhythm had such a significant positive association to survival outcomes, yet there were insufficient instances of bystander defibrillation to show effect (four patients [0.2%] of the entire study sample where 5.1% of patients had shockable rhythms). This suggests a need for community awareness campaigns into the use of defibrillators on children. Future exploration into the social reflexivity surrounding the causes and effects of the minimal use of automated external defibrillators (AED) on pediatrics is required. Many factors should be considered; the public's apprehension to defibrillate children due to a lack of education, the infrequency or inconvenience of AED positioning in public places, or the societal need for the innovation of a more affordable AED to increase the availability of these devices. The incorporation of new technologies, including bystander smart phone applications, to aid in the bystander administration of critical medical aid to these patients, such as public access defibrillation, should be further explored to increase rates (39). An opportunity is identified to study the prevalence of etiologies and their outcomes for all pediatric OHCA patients treated by paramedics nationally, allowing for a more comprehensive analysis. A national analysis of pediatric OHCA could help to get a better understanding of the factors independently associated with occurrence and survival, identify areas of disparity for addressing, and subsequently review current paramedic practice with the specific aim of improving the mortality rate for this cohort irrespective of causation.
This study is not without limitations. The study dataset was retrieved from the ambulance service and does not include hospital or 30-day survival data; therefore, only prehospital outcome (ROSC on hospital arrival) was within the scope of this study. The patient etiology within the QAS dataset was as decided by the attending paramedic, so it is possible that cause of death may be recorded differently than it would have been if an autopsy had been performed. This was also a retrospective study so any trends discovered may not be entirely current, as several factors influencing outcome may have changed or advanced throughout the dataset dates.

Conclusion
This study found that, as for all OHCA, overall rates of ROSC on hospital arrival in children are lower than adults. Etiology, age group, rurality of arrest location, and witness status have associations with outcomes. Lack of research amongst pediatrics means an inability to extrapolate robust knowledge into best practice. Further research is required to better understand current pediatric OHCA survival rates, how pediatric OHCA data are being recorded and analyzed, how community education and earlier intervention can improve pediatric outcomes, and how the prevalence of different etiologies in different communities' affects survivability. This research has the potential to advance the treatment for this population group and improve pediatric survival in this country.