Children with ASD exhibited heterogeneous CSHQ total sleep disturbances scores (Fig. 1A) with 70% (166 of 237) exceeding a clinical cutoff of 41 as originally proposed for 4–10 year old children (Owens et al., 2000) and for children with ASD (Katz et al. 2018). The children also exhibited heterogeneous SP scores (Fig. 1B) with > 40% of the children exhibiting scores that were ≥ 2 standard deviations above the population norm in at least one SP subscale (Fig. 1B). Note that the most prevalent sensory problem was in the sensory sensitivity domain, indicating hypersensitivity to stimuli, reported for > 25% of the children. Heterogeneous scores were apparent across children in all 5 ABC subscales (Fig. 1C). There are no clinical cutoffs for the subscales of this questionnaire.
CSHQ scores were not significantly correlated with either the ADOS total CSS (r(239) = 0.04, p = 0.52), the ADOS social affect (SA) CSS (r(235) = 0.050, p = 0.44), or the ADOS restricted and repetitive behaviors (RRB) CSS (r(239) = 0.004, p = 0.95). However, there was a significant negative correlation between the CSHQ score and the cognitive scores of the children (r(195)=-0.14, p = 0.049), indicating that parent reported sleep disturbances explained almost 2% of the variance in cognitive scores across children.
CSHQ scores were significantly positively correlated with all four SP subscale scores (Fig. 2): Sensation Seeking (r(235) = 0.43, p < 0.001), Sensation Avoiding (r(235) = 0.4, p < 0.001), Sensory Sensitivity (r(235) = 0.5, p < 0.001) and Low Registration (r(235) = 0.37, p < 0.001). Similarly, CSHQ scores were significantly positively correlated with all five ABC subscales scores (Fig. 3): Irritability (r(235) = 0.52, p < 0.001), Social Withdrawal (r(235) = 0.42, p < 0.001), Stereotype Behavior (r(235) = 0.41, p < 0.001), Hyperactivity (r(235) = 0.46, p < 0.001), and Inappropriate Speech (r(235) = 0.34, p < 0.001).
There were moderate to strong positive correlations between ABC and SP subscale scores (Table 2). Moreover, strong correlations were also apparent across sub-scale scores of each questionnaire separately (Table 2).
Table 2
Correlations across ABC and SP subscales
| Irritability | Social Withdrawal | Stereotypical Behavior | Hyperactivity | Inappropriate Speech | Sensation Seeking | Sensation Avoiding | Sensory sensitivity |
ABC subscales |
Social Withdrawal | 0.67*** | | | | | | | |
Stereotype Behavior | 0.67*** | 0.68*** | | | | | | |
Hyperactivity | 0.82*** | 0.64*** | 0.69*** | | | | | |
Inappropriate Speech | 0.58*** | 0.50*** | 0.54*** | 0.59*** | | | | |
SP domains | | | | | | | | |
Sensation seeking | 0.51*** | 0.36*** | 0.43*** | 0.60*** | 0.45*** | | | |
Sensation avoiding | 0.60*** | 0.46*** | 0.43*** | 0.52*** | 0.40*** | 0.53*** | | |
Sensory sensitivity | 0.60*** | 0.50*** | 0.53*** | 0.55*** | 0.38*** | 0.62*** | 0.70*** | |
Low registration | 0.49*** | 0.50*** | 0.47*** | 0.47*** | 0.40*** | 0.56*** | 0.68*** | 0.67*** |
*** p < .001 ABC- Aberrant Behavior Checklist, SP – Sensory Profile |
Predicting The Severity Of Sleep Disturbances From Abc And Sp Scores
Regression analyses were used to examine the ability of individual ABC and SP subscale scores to explain the variance in CSHQ scores. We first included all ABC and SP subscale scores and control variables (age, ADOS score and cognitive level) as predictors in a single multiple regression model. This model explained 31% of the variance in CSHQ scores (Table 3).
Of the predictors included in the initial regression model only the sensory sensitivity score from the SP and the irritability score from the ABC reached statistical significance (Table 3). We, therefore, examined another regression model using only irritability and sensitivity as predictors of the CSHQ score while also examining their interaction (Table 3 model 2). The new regression analysis explained 35% of the variance in CSHQ scores. Note that each of these two variables (irritability and sensory sensitivity) single handedly explained 25% of the variance in CSHQ scores in the correlation analyses above (Figs. 3&4).
Table 3
Multiple regression analyses predicting the CSHQ total score from the ABC and SP sub-scales. Model 1: When using all ABC and SP sub-scales as well as age, ADOS, and cognitive scores. Model 2: When using only the ABC irritability sub-scale and the SP sensitivity sub-scale as well as their interaction term.
Model 1: adjusted R2 = 0.29 | β | p |
ABC | Irritability | 2.30 | < 0.05 |
social withdrawal | 1.42 | 0.16 |
stereotypical behavior | 0.28 | 0.78 |
Hyperactivity | -0.56 | 0.57 |
| Inappropriate speech | 0.66 | 0.51 |
SP | Seeking | 1.45 | 0.15 |
| Avoiding | -0.37 | 0.70 |
| Sensitivity | 2.62 | < 0.01 |
| Low registration | -0.77 | 0.43 |
Age | | -0.56 | 0.57 |
ADOS Score | 0.98 | 0.32 |
Cognitive Score | 1.06 | 0.29 |
Model 2: adjusted R2 = 0.35 |
ABC | Irritability | 3.60 | < 0.001 |
SP | Sensitivity | 5.11 | < 0.001 |
Interaction | Irritability*Sensitivity | 3.53 | < 0.001 |
Note: ABC- Aberrant Behavior Checklist, SP – Sensory Profile
Given the significant interaction between irritability and sensitivity scores in the regression model above, we performed a simple slope analysis. We found that sensory sensitivity moderated the association between irritability and CSHQ scores such that the strength of this association increased with the severity of sensory sensitivities (Fig. 4). The association between irritability and CSHQ scores was strongest for children with sensory sensitivity scores of 2 (B = 0.48, t237 = 6.50, p < .0001) weaker for children with sensory sensitivity scores of 1 (B = 0.29, t237 = 2.18, p = p < .0001), and not significant for children with sensory sensitivity scores of 0 or less (B = 0.11, t237 = 1.29, p = 0.19).