Behavior and Sleep - Associations Across Childhood and Adolescence In Autism Spectrum Disorder

Background: Autism Spectrum Disorder (ASD) is characterized by deficits in three major domains: social interaction, communication, and sensory issues, restricted interests and/or stereotyped and repetitive behaviors. Sleep concerns are common, with prevalence rates estimated to range from about 50-80%. Sleep problems persist into adolescence with adolescents experiencing problems with sleep onset, shorter sleep duration and daytime sleepiness. The magnitude of the association of sleep problems and problematic daytime behaviors across the age span remains unclear.

Objectives: To determine the magnitude of sleep and behavior problems across childhood and adolescence in a large cohort. To estimate the relative risk and 95% CI of poor sleepers having a behavioral problem, and  to identify variations in the sleep-behavior paradigm across this age-span. 

Methods: We assessed the relationship between behavior and sleep in 1784 children, ages 3–18, mean (standard deviation) 6.7 (3.5) years, with confirmed diagnosis of ASD participating in the Autism Treatment Network. The Parental Concerns Questionnaire (PCQ), a validated questionnaire specific to ASD, was used to evaluate behavioral concerns and to define good or poor sleepers. Sleep was evaluated using the Children’s Sleep Habits Questionnaire (CSHQ). Log-binomial models were used to evaluate relative risk (95% confidence intervals) and stepwise logistic regression analysis to determine which CSHQ subscales were the most predictive of being a poor sleeper.

Results: Parents of 1200 (67%) children reported their child to be a good sleeper, and 584 (33%) reported their child as a poor sleeper. Over 60% of the children had problems with language use and understanding, attention span, and social interactions. Poor sleepers had a higher percentage of problems on all PCQ scales than good sleepers. Over 75% had problems with attention span and social interactions. In age adjusted log-binomial models, good sleepers (compared to poor sleepers) had a lower likelihood of being reported as having problems with a specific behavior on the PCQ  for all behaviors except language use and understanding. A poor sleeper had a 10% higher likelihood of having problems with self-injurious behavior, and a 6 % higher likelihood of mood swings, aggression and compulsive behavior problems. The effect of age in the models was significant with language, anxiety, aggression, hyperactivity, and eating problems. In all cases except anxiety problems, older children had fewer problems.

In age adjusted multivariate logistic regression models, the CSHQ parasomnia scale was the one scale that most consistently remained as a sleep problem. For a one-unit increase on the parasomnia scale, there was 20% increase in the odds of the parent reporting a problem on the anxiety, sensory issues, aggression, hyperactivity, attention span, mood swings, and self injurious behavior questions when all other variables in the model were held constant.

Conclusions: Across childhood and adolescence poor sleepers are at higher risk for behavioral problems than good sleepers. Future work should focus on refining the sleep-behavior phenotypes to provide the foundation for focused studies targeting treatment of sleep and behavior in ASD.