Suicidal Ideation in Korean Children at-Risk for Autistic Spectrum Disorder (ASD)

Background: High rates of suicidal behaviors have been reported in individuals with ASD (e.g., Cassidy et al., 2014). Reported risk factors for suicidal behavior in ASD are male gender, comorbid psychiatric diagnosis and peer victimization (see Segers & Rawana, 2014). Studies have been in clinic- or community-based samples of ASD participants; no studies of prevalence or risk factors have drawn from representative, population-based samples. 

Objectives: 1) Examine prevalence of current suicidal ideation (SI) in population-based samples of 7-12 year olds diagnosed with ASD or classified as high- or low-risk for ASD. 2) Determine whether having ASD is a risk factor for SI in addition to previously reported risk factors for SI. 3) Within the ASD group, explore differences between children with and without SI.

Methods: A retrospective cohort design was used in three population-based samples (Korean Prevalence=13,561, Sooncheon=4,837, CHEER=3,702). In the Korean Prevalence sample, 94 children had ASD diagnoses confirmed by comprehensive diagnostic assessment and were compared to 13,467 children with ASSQ scores <14 (low-risk). In the Sooncheon and CHEER cohorts, children were classified as high-risk for ASD based upon a score of >14 on the Autism Spectrum Symptom Questionnaire (ASSQ). Current SI was determined by items from the parent-rated Behavior Assessment for Children, 2^nd Edition (BASC).

Differences in proportions (prevalence) were tested with Chi Square. Logistic regressions were used to determine whether ASD diagnosis/ASD risk status was a significant predictor of SI in addition to previously reported risk factors (gender, BASC scales reflecting emotional/behavior problems and items suggesting negative social experiences). Analyses were repeated for each of the samples. Independent Sample T-tests were used to compare ASD+SI vs. ASD+NoSI on age and BASC scales to inform understanding of factors that may relate to increased risk for SI within individuals with ASD.

Results: For the Korean prevalence study, SI was significantly more common in children with confirmed ASD (14.9%) than children at low-risk for ASD (5.4%; OR=3.09, CI95% 1.75-5.49). Logistic regression models showed that ASD diagnosis was a significant predictor of SI (OR=2.13, CI95% 1.15-3.92). Being teased (OR=1.96, CI95% 1.66-2.33) and chosen last for games (OR=1.84, CI95%=1.56-2.17), and scores from BASC scales of depression, hyperactivity, aggression, conduct problems, anxiety, and attention problems were also significant predictors of SI. Gender was not associated with SI. Children with ASD+SI had more anxiety symptoms than children with ASD+NoSI (p<.01); groups did not differ on age or other BASC scales.

Children at high-risk for ASD had greater prevalence of SI than children with low ASD risk in the Sooncheon (16.6% vs. 3.4%; OR=5.59) and CHEER cohorts (27.4% vs. 6.9%; OR=5.08). Patterns of predictors of SI were similar to the Korean Prevalence sample. 

Conclusions: Children with ASD or suspected high-risk for ASD had increased prevalence of current SI in all three Korean samples. Clinical implications for assessment and monitoring of SI in children with ASD will be discussed.  Interventions in ASD aimed at reducing exposure to modifiable risk factors should focus on emotional and behavior problems, as well as negative social experiences.