Retrospective Analysis of Burn Injuries in Children with Autism Spectrum Disorder

Abstract

Children with neurodevelopmental disorders are at risk for burn injury, but the clinical outcomes, particularly mortality, are unknown in this patient population in the United States (U.S.). The main objectives of this study are to evaluate (1) subject characteristics; (2) burn injury type; (3) clinical care provided; and (4) mortality in children with autism spectrum disorder (ASD), hypothesizing that this patient population has similar mortality and critical care management requirements when compared to children without ASD. This is a retrospective observational cohort study utilizing the TriNetX^® electronic health record database of subjects aged 0 to 18 years with burn injury associated diagnostic codes. Data were analyzed for demographics, diagnostic, medication, procedural codes, and mortality. We analyzed 99,323 subjects (n, %) coded for a burn injury [3083 (3.1%) with ASD and 96,240 (96.9%) without ASD]. Children with ASD had a higher odds of 1-year all-cause mortality [1.9 (1.06, 3.40), p = 0.004], need for critical care services [1.88 (1.40, 2.52), p < 0.001], and mechanical ventilation [2.69 (1.74, 4.17), p < 0.001] compared to those without. Our study found that U.S. children with ASD who had a burn injury had a higher odds of mortality and critical care needs when compared to children without ASD. Future studies are needed to understand the impact of burn injuries and factors associated with mortality in this patient population.

In the United States (U.S.), and worldwide, burn injuries are one of the leading causes of preventable death in children (Children’s Safety Network, 2021; Chong et al., 2020). A majority of burn injuries occur in the home and children less than 5 years are most at risk (Children’s Safety Network, 2021; Jordan et al., 2022). Burns severe enough to require hospitalization are not common in developed countries; however, they are responsible for a disproportionate share of injury-related death, disability, and resource utilization (Bessey et al., 2006). Additionally, burn injuries can have significant psychological, social, and economic impacts on a child’s life and the lives of their caregivers (Dhopte et al., 2017).

One group of children that may be more susceptible to burn injuries are those with neurodevelopmental challenges, particularly autism spectrum disorder (ASD) (DiGuiseppi et al., 2018; Sadeq et al., 2020). ASD is a condition characterized by difficulty with social communication, intense and focused interests, repetitive behaviors, and challenges with sensory processing (National Institute of Mental Health, 2023). These factors may impact their ability to navigate and interact with their external environment, potentially increasing injury risk (DiGuiseppi et al., 2018). Studies worldwide have demonstrated that certain behavioral traits increase risk of burn injuries in children, some of which may also be present in children with ASD. These include impulsiveness, lack of awareness, higher activity levels due to curiosity, and total dependency on caregivers (Dhopte et al., 2017). Additionally, children with ASD can struggle with fine motor skills, which may impact their ability to protect themselves from injury (DiGuiseppi et al., 2018).

Despite this theoretical increased risk, burn injuries in this patient population have not been extensively studied. In general, few studies have reported on the association between pediatric burn injury and neurodevelopmental disorders. In particular, data examining injury risk has been inconsistent and the complications as well as the type of clinical care children with ASD receive to treat burn injuries have not been well described (Chan et al., 2023). To our knowledge, this is the first study examining the impact of burn injuries in children with ASD in the U.S. An understanding of how this patient population is impacted by burns can guide anticipatory guidance for children with ASD and improve their care.

Utilizing a U.S. based multi-center electronic health record database with longitudinal capabilities, the objectives of this present study are to evaluate the (1) subject characteristics; (2) burn injury type; (3) clinical care provided; and (4) mortality in children with ASD. We hypothesized that children with ASD have similar mortality and clinical management requirements when compared to children without this condition.

Methods

Study Design

This is a retrospective observational longitudinal cohort study utilizing the TriNetX^® electronic health record (EHR) database of pediatric subjects aged 0 to 18 years who were reported to have an International Classification of Diseases, 9th (ICD-9) and 10th edition diagnostic codes (ICD-10) associated with burn injuries (“T20-T25”, “T26–28”, and “T30–32”).

Data Source

TriNetX is a global federated research network that delivers aggregated data elements (i.e. diagnoses, procedures, laboratory values) of approximately 68 million patients in 56 large healthcare organizations (HCOs) predominantly in the U.S. These data elements are de-identified, collected, and organized in real-time. TriNetX has multi-center and longitudinal capabilities. Thus, if a subject has multiple encounters (visits) within a TriNetX associated healthcare system, all the EHR data elements (even those within a different health system) are available for analysis allowing for a longitudinal evaluation of a subject before and after the clinical period examined. TriNetX, LLC continues to maintain compliance with the Health Insurance Portability and Accountability Act (HIPAA) and is certified to the ISO 27001:2013 standard. Further details regarding its maintenance as well as the de-identification procedures of patient level data have been previously described (Krawiec et al., 2022; Kridin et al., 2023).

Ethical Considerations

Because this study utilized only de-identified patient records and did not involve the collection or transmission of individually identifiable data, our university’s Institutional Review Board pre-determined this study to be non-human research (Institutional Review Board #: STUDY00020794).

Data Collection

The data used in this study was collected on September 7th, 2023, from the TriNetX Research Network. After receiving all the diagnostic, procedure, medication, and encounter codes in a given subject’s database history, it was pre-processed and organized utilizing Python (primarily using the pandas package) (Python). Due to similar codes for a given variable (i.e. critical care services was associated with multiple possible codes due to the different ages), definitions were developed to ensure all variables of interest were captured in a categorical manner. Thus, if one subject was given a Current Procedural Terminology (CPT) code of 99,291 (“Critical care, evaluation and management of the critically ill or critically injured patient; first 30 to 74 min”) and another subject with a CPT code of 99,475 (“Initial inpatient pediatric critical care, per day, for the evaluation and management of a critically infant or young child, 2 through 5 years of age”), both subjects were classified as receiving critical care services. International Classification of Diseases, 9th edition (ICD-9) codes were mapped to ICD-10 codes as defined by TriNetX database [i.e. (ICD-9 code 948.10, “Burn [any degree] involving 10–19 percent of body surface with third degree burn, less than 10 percent or unspecified” was deemed equivalent to ICD-10 code T31.10, “Burns involving 10–19% of body surface with 0% to 9% third degree burns.” Please see Table Supplementary for the classifications applied to the codes utilized in this study.

Inclusion and Exclusion Criteria

We included subjects aged 0 to 18 years of age with a burn associated ICD-9 and/or ICD-10 diagnostic code. We excluded subjects who did not have any encounters or visits at least one time before the burn injury diagnostic code entry. This increased the likelihood that if a patient had ASD as a pre-existing condition, it was evaluated and documented as a diagnostic code by a provider before the index date.

Variables

The primary outcomes of interest were the need for critical care services, hospitalization, mechanical ventilation, burn categorizations both 7 days before and after the index date, and all-cause 1-year mortality. Patient characteristics (age, sex, race, ethnicity, social determinants of health) were reported.

Because this dataset was de-identified and no date of birth was provided, ages are approximate (within 1 year). For example, a subject born in 2012 reported to have received a Burn of first degree of trunk (ICD-10, “T21.1”)] diagnostic code on January 1st, 2020, was determined to be 8 years of age at the time of this diagnosis.

For patient privacy reasons, exact death dates were not provided by this database, only the month and year. To meet the definition of death within 365 days, we first added 365 days to the first recording of a burn associated diagnostic code. If the month and year of the death date provided matched this calculation, then the patient met the definition of death within 365 days. For example, if the patient had a burn associated diagnostic code entered on January 1, 2020 (i.e., day of burn injury) the patient would not have been classified as a death within 365 days if the death date was reported as February 2021 but would have been if the death date was reported as January 2020.

Using ICD-9, ICD-10, ICD-10-PCS, baseline pediatric complex chronic conditions (PCCC) were identified and categorized using the R PCCC package (DeWitt, 2020; Feinstein, 2018). In addition, ASD and attention deficit hyperactivity disorder (ADHD) were evaluated using any associated diagnostic, procedure, or medication code prior to the first reported date of the burn injury. Presence of a social determinants of health (SDOH) diagnostic code was also collected, denoting any patients with potential health hazards related to socioeconomics and psychosocial circumstances.

Clinical Period Definitions

During data pre-processing, we identified the index date as the first time a burn injury diagnostic code was entered in a subject’s database. Codes for acute interventions (critical care services, mechanical ventilation, hospitalization, tracheostomy, transfusion) and categorization of burn injury (burns classified based on the extent of body surface involved) were evaluated 7 days before and 7 days after the index date. Because of the known clinician variability in coding, this time period was selected to increase the likelihood that all diagnostic and intervention associated codes entered for the subject’s burn injury encounter were captured (Ford et al., 2016). Pre-existing comorbid conditions (including pediatric complex clinical conditions, ASD, ADHD, social determinants of health) were evaluated any time before the index date.

Designation of Cohorts

The study population was divided into two cohorts (ASD and no ASD) and analyzed.

Statistical Analysis

We applied summary statistics using mean and standard deviation or proportions for the reported clinical and demographic characteristics of the subjects included in this study. Analyses were performed using R 4.2.0 17 (R Core Team, 2022). We applied the Fisher’s exact test where a two-sided p-value less than 0.05 was considered statistically significant and employed the epitools R package to compute odds ratios and 95% confidence intervals using the “Wald” method (R Core Team, 2022).

Results

Overview

Our study included 99,323 subjects [3083 (3.1%) with ASD and 96,240 (96.9%) without ASD]. Mean ages, race, and ethnicity appeared similar between the two groups [Table 1]. Of patients with ASD, 76.3% were male, 6.7% had a diagnostic code related to SDOH, 24.8% were diagnosed with a PCCC, and 39.8% had an additional diagnosis of ADHD. Patients with ASD were more likely to have a diagnostic code related to SDOH [2.11 (1.82, 2.44), p < 0.001] and be diagnosed with a PCCC [2.78 (2.55, 3.02), p < 0.001] or ADHD [6.75 (6.26, 7.28), p < 0.001]. Subjects with ASD had significantly higher odds of all-cause mortality at 1 year [1.90 (1.06, 3.40), p = 0.047].

Table 1.

Patient demographics

	ASD n = 3083	No ASD n = 96,240	OR [CI]	p value
Age in years, mean (SD)	5.9 ± 4.5	5.8 ± 4.9
Sex
Female	724 (23.5%)	44,225 (46.0%)
Male	2352 (76.3%)	51,949 (54.0%)
Unknown	7 (0.2%)	66 (0.1%)
Race
American indian or alaska native	14 (0.5%)	490 (0.5%)
Asian	84 (2.7%)	2865 (3.0%)
Black or african american	725 (23.5%)	24,517 (25.5%)
Native hawaiian or other pacific islander	6 (0.2%)	356 (0.4%)
White	1628 (52.8%)	46,419 (48.2%)
Unknown	626 (20.3%)	21,593 (22.4%)
Ethnicity
Hispanic or latino	514 (16.7%)	16,732 (17.4%)
Not Hispanic or latino	2171 (70.4%)	62,833 (65.3%)
Unknown	398 (12.9%)	16,675 (17.3%)
Social determinants of health diagnostic code presence	206 (6.7%)	3166 (3.3%)	2.11 [1.82, 2.44]	< 0.001
Pediatric clinical complex condition	765 (24.8%)	10,229 (10.6%)	2.78 [2.55, 3.02]	< 0.001
Attention deficit hyperactivity disorder	1228 (39.8%)	8596 (8.9%)	6.75 [6.26, 7.28]	< 0.001
1 Year all-cause mortality	12 (0.4%)	198 (0.2%)	1.90 [1.06, 3.40]	0.047

Data is presented as n (%) unless otherwise stated

ASD Autism Spectrum Disorder; SD standard deviation

Burn Categories and Total Body Surface Area Burn Classifications

Categorical data regarding the diagnostic code categories reported for burn categories and burns classified by total body surface area are summarized in Table 2 and 3. Children with ASD were more likely to have burns confined to their eye or adnexa [1.21 (1.01, 1.44), p = 0.047] but less likely to have burns of their head, face, or neck [0.89 (0.79, 1.0), p = 0.049] or lower leg [0.89 (0.8, 0.98), p = 0.02]. Subjects with ASD were more likely to have burns involving 60 to 69 percent of their body surface [5.21 (1.81, 15.0), p = 0.01]. [Table 2] [Table 3].

Table 2.

Burn categories

	ASD n = 3083	No ASD n = 96,240	OR [CI]	p value
Burn and corrosion confined to eye and adnexa	129 (4.2%)	3365 (3.5%)	1.21 [1.01,1.44]	0.047
Burn and corrosion of ankle and foot	276 (9.0%)	7961 (8.3%)	1.09 [0.96,1.24]	0.184
Burn and corrosion of head, face, and neck	303 (9.8%)	10,543 (11.0%)	0.89 [0.79,1.0]	0.049
Burn and corrosion of lower limb, except ankle and foot	439 (14.2%)	15,181 (15.8%)	0.89 [0.8,0.98]	0.0208
Burn and corrosion of other internal organs	41 (1.3%)	1160 (1.2%)	1.1 [0.81,1.51]	0.503
Burn and corrosion of respiratory tract	5 (0.2%)	84 (0.1%)	1.86 [0.75,4.59]	0.204
Burn and corrosion of shoulder and upper limb, except wrist and hand	540 (17.5%)	16,468 (17.1%)	1.03 [0.94,1.13]	0.560
Burn and corrosion of trunk	481 (15.6%)	15,385 (16.0%)	0.97 [0.88,1.07]	0.583
Burn and corrosion of wrist and hand	1007 (32.7%)	30,909 (32.1%)	1.03 [0.95,1.11]	0.531
Burn and corrosion, body region unspecified	1099 (35.6%)	33,125 (34.4%)	1.06 [0.98,1.14]	0.160
Burns classified according to extent of body surface involved	1 (0.0%)	55 (0.1%)	0.57 [0.08,4.1]	1.0
Corrosions classified according to extent of body surface involved	180 (5.8%)	5379 (5.6%)	1.05 [0.9,1.22]	0.550
Burn of unspecified body region, unspecified degree	848 (27.5%)	25,445 (26.4%)	1.06 [0.97,1.14]	0.191
Burn, unspecified site	3 (0.1%)	39 (0.0%)	2.4 [0.74,7.78]	0.141

Data is presented as n (%)

ASD Autism Spectrum Disorder; CI confidence interval; OR odds ratio

Table 3.

Total body surface area burn classifications

	ASD n = 3083	No ASD n = 96,240	OR [CI]	p value
Burns involving less than 10 percent of body surface	723 (23.5%)	23,145 (24.0%)	0.97 [0.89, 1.05]	0.454
Burns involving 10 to 19 percent of body surface	72 (2.3%)	1830 (1.9%)	1.23 [0.97, 1.57]	0.095
Burns involving 20 to 29 percent of body surface	11 (0.4%)	383 (0.4%)	0.9 [0.49, 1.63]	0.884
Burns involving 30 to 39 percent of body surface	7 (0.2%)	155 (0.2%)	1.41 [0.66, 3.01]	0.358
Burns involving 40 to 49 percent of body surface	1 (0.0%)	63 (0.1%)	0.5 [0.07, 3.57]	0.725
Burns involving 50 to 59 percent of body surface	4 (0.1%)	45 (0.0%)	2.78 [1.0, 7.73]	0.065
Burns involving 60 to 69 percent of body surface	4 (0.1%)	24 (0.0%)	5.21 [1.81, 15.02]	0.010
Burns involving 70 to 79 percent of body surface	2 (0.1%)	15 (0.0%)	4.16 [0.95, 18.22]	0.096
Burns involving 80 to 89 percent of body surface	1 (0.0%)	16 (0.0%)	1.95 [0.26, 14.72]	0.415
Burns involving 90 percent or more of body surface	2 (0.1%)	20 (0.0%)	3.12 [0.73, 13.37]	0.148

Data is presented as n (%)

ASD Autism Spectrum Disorder; CI confidence interval; OR odds ratio

Medical Care Provided

While most patients did not require hospitalization (95,071, 95.7%) or critical care services (98,471, 99.1%), subjects with ASD had a higher odds of requiring critical care services [1.88 (1.40, 2.52), p < 0.001], hospitalization [1.41 (1.21, 1.65), p < 0.001], and mechanical ventilation [2.69 (1.74, 4.17), p < 0.001]. There was no difference in need for emergency services, tracheostomy, or transfusion. [Table 4].

Table 4.

Medical care provided

	ASD n = 3083	No ASD n = 96,240	OR [CI]	p value
Critical care services	48 (1.6%)	804 (0.8%)	1.88 [1.4,2.52]	< 0.001
Emergency department services	1032 (33.5%)	32,345 (33.6%)	0.99 [0.92,1.07]	0.892
Hospitalization	181 (5.9%)	4071 (4.2%)	1.41 [1.21,1.65]	< 0.001
Mechanical ventilation	22 (0.7%)	256 (0.3%)	2.69 [1.74,4.17]	< 0.001
Tracheostomy	0 (0.0%)	18 (0.0%)	–	1.0
Transfusion	6 (0.2%)	195 (0.2%)	0.96 [0.43,2.17]	1.0

Data is presented as n (%)

ASD Autism Spectrum Disorder; CI confidence interval; OR odds ratio

Discussion

In this study, we aimed to evaluate subject and clinical characteristics as well as outcomes of burn injuries in children with ASD utilizing a multi-center EHR database centered in the U.S. We found that children with ASD and a burn injury have higher odds of having a SDOH diagnostic code as well as a PCCC or ADHD diagnosis. There were no clear differences in burn patterns or percentage of total body surface area burned between children with or without ASD. Children with ASD did have higher odds of all-cause 1-year mortality after a burn injury and need for hospitalization, critical care services, and mechanical ventilation after a burn injury. These findings may indicate increased injury severity in this population, which could have significant implications regarding anticipatory guidance and prevention strategies for burn injuries in children with ASD.

Children, especially of younger ages, are at risk of suffering burn injuries. Given that a majority of these injuries are preventable, pediatricians work to provide anticipatory guidance to assist families in understanding how to best protect their children from injury (Health, 2023; American Academy of Pediatrics, 2017). Despite these efforts, children continue to be disproportionately affected by burn injuries (Krishnamoorthy et al., 2012). Therefore, it is necessary to develop a better understanding of the risk factors associated with burn injuries. This may help inform clinicians of the best advice to give families regarding reducing burn injury risk for their children.

A majority of pediatric burn injuries are non-fatal, but they represent a source of profound morbidity (Krishnamoorthy et al., 2012). The initial injury, depending on severity, can lead to release of inflammatory and vasoactive mediators, causing a systemic inflammatory release syndrome, significant myocardial depression, and hypotension requiring aggressive fluid and nutritional support (Krishnamoorthy et al., 2012). If an inhalational injury was suffered, the child often requires intubation, which may be complicated by edematous, compromised airways in the setting of impending respiratory failure (Krishnamoorthy et al., 2012). To assist with both respiratory and pain management, pediatric burn patients often require sedation, but the systemic effects of burn injuries alter the clearance of medications used in sedation of these patients (Krishnamoorthy et al., 2012). This often leads to the administration of high-dose opioids, which may result in iatrogenic opioid dependence for patients (Krishnamoorthy et al., 2012). Given these potential complications, burn injuries represent a unique source of morbidity for pediatric patients, requiring intensive monitoring and extensive critical care interventions.

While burn injuries represent a threat to the pediatric population as a whole, those with ASD may be at increased risk given their susceptibility to traumatic injuries as a whole (Guan & Li, 2017). Previous studies where excess mortality has been documented in pediatric patients with ASD, however, focused primarily on asphyxiation and drowning (Guan & Li, 2017). Our study highlights burns as an additional traumatic injury that poses a threat specifically to the pediatric ASD population. Due to difficulties in sensing their surroundings or movements, children with ASD may not be able to detect or avoid danger, and potentially may not perceive the consequences of entering a high temperature area (National Institute of Mental Health, 2023). People with ASD often struggle with social, emotional and communication skills (National Institute of Mental Health, 2023). They might repeat certain behaviors and might not want change in their daily activities. These factors may alter how children with ASD incorporate information about danger avoidance. Some ASD patients are also susceptible to wandering (Anderson et al., 2012). This behavior may allow children to have unrestricted, unsupervised access to a burn source, increasing their injury risk. Finally, behavioral problems may also cause patients with ASD to act impulsively or be hyperactive, especially in those experiencing comorbid ADHD (40% of our study population) (Sadeq et al., 2020).

Despite knowing there potentially is an increased injury risk within the ASD population, little is known about the outcomes of these injuries when they do occur. Our study found that patients with ASD had a higher odds of all-cause mortality within 1 year of their burn injury. At this time, no U.S. based studies have focused on all-cause mortality specifically for children with ASD. However, three non-U.S. based studies report between a 2 and 2.5-fold increased risk of mortality, similar to the risk found within our study population (Kim et al., 2021; Schendel et al., 2016; Yoo et al., 2022). While the underlying cause behind this association warrants additional study, several mechanisms could explain these findings. Children with ASD often have comorbidities that may result in behaviors that have placed the subject at increased risk of more severe injury (Chen et al., 2007). For example, our study demonstrated that 39.8% of subjects with ASD had ADHD, which has previously been shown to increase risk of burns as well as burns covering larger surface areas (Ghanizadeh, 2008; Mangus et al., 2004; Sadeq et al., 2020).

Our study also found that despite overall low need for hospitalization or critical care services for pediatric burn patients, a higher odds of hospitalization, use of critical care services, and need for mechanical ventilation was observed in children with ASD after a burn injury. Due to limitations in our database, we were unable to assess for severity of burn injury itself; however, need for hospitalization, critical care interventions, and mechanical ventilation may imply that the injuries within the ASD population were more severe. However, further study is indicated to determine the nature of this association. Besides burn severity, mechanical ventilation may be necessary for patients with ASD due to higher need for sedation to control pain and agitation in this population. For example, if a child needs regular sedation in order for the medical team to perform dressing changes, this may result in the need for mechanical ventilation. In addition to ADHD, ASD can be comorbid with a wide range of conditions, including neurological, genetic, respiratory, and gastrointestinal conditions (Al-Beltagi, 2021). Depending on the severity of these conditions and resulting disability status of the child, for example, chronic lung disease or seizure disorders, they may be more likely to require sedation and/or intubation during their treatment course. Further study would be indicated to elucidate these associations given the wide range of conditions comorbid with ASD.

This study has several limitations. First, we were limited to the clinical data available within the TriNetX database. Our study was therefore limited to those receiving medical care, and minor burn injuries may have not been identified. Due to the retrospective, multi-institutional nature of the study, it is unknown how race/ethnicity data were obtained. Because the database is limited to certain regions and center, this could have contributed to an under collection of data from smaller and under-represented communities. Mortality is more likely to be reported in this database in hospitalized children. Thus, it is possible that subjects who expired outside of the institution that participates in TriNetX were unable to be identified. Lastly, because convenience sampling was utilized for this study, power analysis was not performed.

Conclusion

Our study found that children with ASD who had a burn injury had a higher odds of mortality and critical care needs when compared to children without ASD. Future studies are needed to understand the impact of burn injuries and factors associated with mortality in this patient population.

Abbreviations

ADHD

Attention Deficit Hyperactive Disorder

ASD

Autism spectrum disorder

CPT

Current Procedural Terminology

EHR

Electronic health record

ICD

International Classification of Diseases

PCCC

Pediatric complex chronic conditions

SDOH

Social Determinants of Health

Data Availability

The data that support the findings of this study are available from TriNetX^®. Restrictions apply to the availability of these data, which were used under license for this study.