The Children's Sleep Habits Questionnaire: Evaluating Subscales for Sleep Problems in Children with Autism Spectrum Disorder

Background:  Children with Autism Spectrum Disorder (ASD) face many developmental obstacles, including elevated sleep problems. To accurately identify these problems, psychometrically sound measures are important. The Children’s Sleep Habits Questionnaire (CSHQ; Owens et al., 2000) is the most common measure of sleep problems in early childhood—including children with ASD. However, despite its common use, few studies have assessed the accuracy of the CSHQ in screening sleep problems in this population.

Objectives:  We aimed to explore the basic psychometric properties of CSHQ subscales in a sample of children with ASD. We further aimed to determine whether these subscales were useful in screening pediatric sleep problems.

Methods: This study included 41 children with ASD between the ages of 2-10 (M= 5.5). Caregivers completed the CSHQ as part of a larger study of sleep and behavior in the context of early intervention. The CSHQ includes 33 parent-report items with eight empirically derived subscales (Owens et al. 2000). This measure was recently validated in children ages 2-5, resulting in five subscales for use in toddlers and pre-school aged children (CSHQ-T; Sneddon et al., 2013). Children were first grouped into ‘sleep problem’ and ‘non-problem’ categories via parent reports at time of enrollment (Does your child have a sleep problem?). Using data from published validation studies (Owens et al., 2000; Sneddon et al., 2013), mean subscale scores for the ‘sleep problem’ group were compared to a clinical sample, and mean subscale scores from the ‘non-problem’ group were compared to a community sample. Differences between samples were evaluated using Welch’s t tests.

Results:  Cronbach’s alpha ranged from .64 to 1.00 for Owens’ subscales, and .78 to 1.00 for the CSHQ-T. When comparing Owens’ community sample with the ‘non-problem’ group, two subscales were significantly different (p < .05): Parasomnias and Daytime Sleepiness (Table 1). In this comparison, mean subscale scores were higher in the ‘non-problem’ group than the community sample. When comparing Owens’ clinical sample with the ‘sleep problem’ group, only sleep duration was significantly different (p < .01). Duration difficulties (mean scores) were higher in our ‘sleep problem’ group than the clinical sample (Table 1). Finally, two subscales differed significantly when comparing the CSHQ-T community sample and the ‘non-problem’ group, with higher mean scores in the ‘non-problem’ group. However, there were no significant differences between our ‘sleep problem’ group and the clinical sample (Table 2).

Conclusions: This work contributes to a growing body of literature on sleep problems in children with ASD. Specifically, this is the first known study to explore basic psychometric properties of the CSHQ-T in a sample of children with ASD, including statistical comparisons with published clinical and community samples. Overall, both sets of subscales had acceptable psychometric properties and accurately distinguished between problem and non-problem sleepers in our sample. Significantly higher scores for children in the ‘non-problem’ group compared to community samples may reflect under-reporting of ASD sleep problems at time of enrollment. Future directions include validating the CSHQ-T with a larger, representative sample of children with ASD.