Mortality of Children with Autism Spectrum Disorder Using Data from a Large-Scale Korean National Cohort

Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by deficits in social communication and interaction, restricted interests, and repetitive behaviors.1 Additionally, ASD is considered to be one of the leading causes of disability in children.2, 3 The prevalence of ASD is estimated to be 1.9% in the United States4 and 2.6% in the Republic of Korea.5

Previous studies have reported that adult ASD patients have increased mortality, possibly due to comorbidities associated with the condition, such as epilepsy, as well as intentional and unintentional self-inflicted injuries.6, 7, 8 However, there are limited studies on population-level analyses regarding pre-adults, especially children. Furthermore, it remains unknown whether children with ASD are at a higher predisposition of increased mortality. Meanwhile, recent studies have identified that children with ASD are likely to have hearing impairment from infancy.1, 9 These findings on the associated hearing impairments, in turn, have promoted the early detection of ASD in children, due to the detection of impaired sensory perception from the test conducted during infancy. However, although hearing impairment during infancy has been associated with various adverse health outcomes in the general populations,10, 11 it remains unclear and uninvestigated whether disease prognosis may differ between ASD children with hearing impairment and those without hearing impairment.

Therefore, given the aforementioned gaps in knowledge and limitations of prior studies, we performed longitudinal analyses investigating the association between ASD diagnosis and mortality among children using the data from a large-scale national cohort of Korean infants and children. We also explored the association between hearing impairment detected during infancy and mortality, as a proxy for prognosis, among children diagnosed with ASD.

Table 1 describes the study population, which was followed up until the end of 2020. Of the 3598904 children born between 2007 and 2014, 32878 children (0.9%) were diagnosed with ASD as of 2020. Of those subjects, 24459 (74.4%) were boys and 8419 (25.6%) were girls. Children with ASD were more likely to have hearing impairment (0.9%) than those without ASD (0.4%) (Table 1). Children born earlier during the study period or with lower household income were less likely to participate in the first health screening (Supplementary Table 1, only online).

Table 1
Sociodemographic Characteristics of Children Included in the Analyses, Stratified by ASD Diagnosis

• Click for larger image
• Click for full table
• Download as Excel file

A total of 35843066 person-years were observed for children without ASD and 328036 person-years for those with ASD. The crude hazard ratio (HR) of mortality for children diagnosed with ASD compared to those without ASD was 2.6 [95% confidence interval (CI): 2.3–3.0]. After controlling for the abovementioned potential confounders (i.e., sex, household income, and birth year), the HR was 2.5 (95% CI: 2.2–2.9), slightly lower compared to the crude model (Supplementary Fig. 1, only online).

In the sex-stratified analyses, the HR was higher in girls (HR=4.8, 95% CI: 3.9–5.8) than in boys (HR=1.9, 95% CI: 1.6–2.2) (Table 2), showing a significant difference by sex (p<0.0001). The HRs for the association between ASD diagnosis and mortality were not different between children with hearing impairment (HR=4.3, 95% CI: 3.4–5.5) and those without (HR=4.3, 95% CI: 1.5–12.2).

Table 2
Associations between ASD Diagnosis and Mortality among Children

• Click for larger image
• Click for full table
• Download as Excel file

Among children diagnosed with ASD (n=32878), hearing impairment detected during infancy was associated with a substantially higher risk of mortality (HR=5.8, 95% CI: 2.1–16.0) (Table 3 and Supplementary Fig. 2, only online).

Table 3
Association between Hearing Impairment Detected in Infancy (4–6 Months of Age) and Mortality among Children Diagnosed with Autism Spectrum Disorder^*

• Click for larger image
• Click for full table
• Download as Excel file

In this study using a large-scale national cohort of Korean infants and children, we were able to identify an association between ASD and higher mortality among children. This association was stronger in girls than in boys. We also found that hearing impairment detected during infancy was associated with higher mortality among children diagnosed with ASD.

Given the current literature, the mean age of death for ASD patients was reported to be 36.2 years,8 and previous studies investigating the association between ASD diagnosis and mortality were performed on adult populations only.6, 7, 8 In these studies, ASD patients had a mortality rate that was 2 to 5 times higher than the general population due to comorbidities, such as epilepsy, as well as intentional and unintentional self-inflicted injuries.6, 7, 8 To address the gaps in knowledge and limitations of prior studies, we observed an increased risk of mortality related to ASD among children, who generally have lower mortality risk than adults or the elderly. This interesting finding might be explained by the increased risk of injury (e.g., suffocation, asphyxiation, and drowning) or comorbid conditions8, 15 associated with ASD among children. However, due to insufficient data on the cause of death, this study was not able to pinpoint the reasons for the increased mortality observed in children. Further studies should aim to address such limitations and identify the causes of death to draw more specific public health implications (by investigating, for example, the association between ASD, hearing impairment, and traffic accident mortality).

The point estimates for the association between ASD diagnosis and mortality were larger among girls than among boys in the present study, which was consistent with previous findings among adults in Demark, Sweden, Australia, and the United States.6, 16, 17, 18, 19 Although increased ASD mortality due to certain causes (e.g., injury) is more likely to occur in boys than in girls, the findings in the present study might be attributed to the fact that the general mortality risk is higher in boys than in girls, at least in part.6 In addition, it might also be explained by the fact that more severe brain problems, such as comorbid epilepsy, are required for the typical ASD phenotype to be manifested in girls than in boys.17 Since the reason for this sex-related difference is still unclear, further studies are needed to confirm the observed sex difference in the present study, especially with the information on the cause of death.

Sensory symptoms in the areas of hearing, vision, smell, and touch are prevalent in ASD patients, and can even appear during infancy.20, 21 Atypical sensory features may have detrimental cascading effects on social communication and reactivity,21 and it was reported that sensory symptoms, such as reactivity, at 1 and 2 years of age are associated with ASD severity among preschool children.20 Although we used simple hearing screening questions instead of a comprehensive evaluation of hearing, we found consistently increased risks of mortality for negative response to the question related to hearing impairment. These findings have important implications in public health regarding the significance of detecting sensory symptoms at the early stages of childhood, which would be clinically relevant in identifying priority groups for health intervention.

This study provides some new insights and evidence on the heightened risk of death in children with ASD. Moreover, the present study provides unique evidence on the association between hearing impairment during infancy and mortality among children with ASD. These results produced from a large-scale national cohort of Korean infants and children have substantial public health implications for planning future interventions and targeting priority groups.

However, this study has several limitations. First, the data sourced from the administrative claims-based cohort has inherent limitations. Although national health screening programs for all children began in 2008, the participation rate of children born between 2007 and 2014 in the first health screening was 56.2%, making it difficult for the results of this study to be generalizable (Supplementary Table 1, only online). Moreover, our results still undercount the true number of ASD cases, particularly due to missed diagnoses or untreated variables. Since the early diagnosis of ASD is difficult, the diagnostic codes are often inserted as other codes, such as ICD-10 codes R62.0 (delayed milestone), R62.8 (other lack of expected normal physiological development), and R62.9 (lack of expected normal physiological development, unspecified), instead of ICD-10 codes used in the present study to determine the outcome of interests (i.e., F84.0, F84.1, F84.5, F88.8, and F88.9). This possibility of outcome misclassification should be considered to interpret the results properly. Second, despite utilizing the national claims data, children in lower socioeconomic status households are less likely to visit a physician and seek treatment for ASD-related symptoms, which often leads to poor prognosis.22, 23 However, effect modification of the association between ASD diagnosis and mortality by income levels cannot be evaluated due to the limited power. Therefore, further studies should focus on expanding the sample size and number of variables. Third, although we controlled for sex, household income, and birth year in the present study, there remains a concern of residual confounding by factors, such as severe illnesses (e.g., cancer, congenital defects, and heart diseases). Therefore, studies considering potential confounders, especially severe illnesses, are warranted to confirm the observed association.

In conclusion, ASD diagnosis was associated with higher mortality among Korean children. Although the point estimate for the association were found to be larger among girls than among boys, this result should be interpreted with caution as the reason for this finding, in terms of the cause of death, is still unclear. Hearing impairment detected during infancy was also associated with higher mortality among children diagnosed with ASD. Ultimately, these results stress the importance of targeted public health interventions for children, namely the need to develop more robust programs for those diagnosed with ASD. This study also pinpoints the need to improve interventions for children detected with hearing impairments as early as infancy, which can be screened through simple and cost-effective clinical assessments.

Table 1

Sociodemographic Characteristics of Children Who Participated in the First Health Screening and Those Who Did not Participate

Fig. 1

Kaplan-Meier survival curve displaying the association between autism spectrum disorder (ASD) diagnosis and mortality among children.

Fig. 2

Kaplan-Meier survival curve displaying the association between hearing impairment detected in infancy (4–6 months of age) and mortality among children diagnosed with autism spectrum disorder.