Descriptive statistics for the 54 participants across the three groups are presented in . Chi square analysis demonstrated that the three groups did not differ relative to gender χ2 (2, N=54) = 14.52, P<0.05. Univariate ANOVA demonstrated that the three groups did not differ relative to age F(2,51)=0.04 P>0.10. There was a significant difference in Full Scale IQ F(2,51)=6.38, P<0.005. Post hoc planned comparisons revealed that this difference was due to the ASD group being significantly lower than both the ADHD t=(1,34)=2.13, P<0.05 and TYP t=(1,34)=3.31, P<0.01 groups. Subsequently, IQ was used as a covariate in the analyses.
The means and standard deviations and posthoc multiple pairwise comparisons (Tukey HSD) for the Conners Parent Rating Scale are reported in . The average range on this measure is defined by T scores from 40 to 60. ADHD was defined by a score greater than 1.5 standard deviations above the mean (Conners, 2001
). Based on this criteria, all of the ADHD children met criteria, 8 of the ASD children met criteria, and none of the typical children met criteria.
Means and Standard Deviations for the Conners’ Diagnostic Measure.
The results of the MANCOVAs for each domain are presented below. The means and standard deviations and the results of the ANCOVAS of the EF measures grouped by domains are presented in . Posthoc analyses for each domain are presented below. Due to the concern regarding the possible of effects of medication, analyses were conducted comparing the clinical groups between those with and without medication and there were no significant differences (all F’s < 1.0 and P>0.05).
Means and Standard Deviations and Analysis of Variance for the Executive Functioning and Attention Measures.
3.1. Response inhibition
Based on MANCOVA, there was a significant difference between the groups regarding Inhibition F(6,56)=3.99, P<0.005; Wilks’ lambda = 0.548. Post hoc multiple pairwise comparisons (Tukey HSD) for the VRCQ showed significant differences between the ASD and TYP group and ASD and ADHD F(2,51) = 5.33, P<0.01. The ARCQ revealed significant differences between the ADHD and TYP group F(2,51) = 4.59, P<0.01 and between the ASD and TYP groups F(2,51)=5.36, P<0.01. The ASD group demonstrated the lowest performance, followed by the ADHD group and then the TYP group. There were no differences between the ADHD and TYP groups on the DK-INH, but the ASD group performed significantly lower than the TYP group F(2,34)=6.20, P<0.01.
3.2. Working memory
The MANCOVA was significant between the groups for Working Memory F(6,92) = 2.67, P< 0.05, Wilks’ lambda = 0.726. There were significant differences between the ASD and TYP groups for SSP F(2,48)=4.72, P<0.05, and for SWM Btwn Errors F(2,48)=3.95, P<0.05, and SWM Strategy F(2,48)=3.97, P<0.05. There were significant differences between the ADHD and ASD children for SWM Btwn Errors F(2,48)=3.95, P<0.01 and SWM Strategy F(2,48)=3.97, P<0.05 with the ASD group performing more poorly. Regarding the SSP, there was a significant difference between the ADHD and TYP group F(2,51)=4.72, P<0.05.
There were significant differences across the groups for Switching F(6,54) = 3.18, P<0.01; Wilks’ lambda = 0.546. For DK T-Switch, significant differences were found between the ADHD and ASD F(2,33)=7.56, P<0.01 and between the ASD and TYP group, P<0.001. There were no significant differences for the CCTT2 or the ID/ED tasks across the groups.
There were no significant differences between the groups based on MANCOVA for Planning F(6,56) = 1.73, P>0.05, Wilks’ lambda = 0.79, which included SOC Min Moves, SOC Initial Thinking and SOC Subsequent Thinking.
There were no significant differences between the ASD and TYP groups for Fluency F(4,58) = 2.38, P>0.05; Wilks’ lambda 0.738, which included the DK-Letter and DK-Category measures.
Based on MANCOVA, there were significant differences in Vigilance between the groups F(4,98) = 4.63, P<0.01; Wilks’ lambda = 0.707. Subsequently, significant differences were found between ADHD and TYP groups for AAQ F(2,50)=8.28, P<0.001, and VAQ F(2,50)=5.58, P<0.01 with the ADHD group performing more poorly. There were significant differences between the ASD and TYP groups on the IVA for AAQ F(2,50)=8.28, P<0.001, and VAQ F(2,50)=5.58, P<0.01.
3.7. Prediction and Exploratory analysis
Next, we used stepwise multiple regression to examine the relationship between the dependent variables and behavioral indices of ADHD symptoms. Stepwise multiple regression showed that the ADHD index (C-ADHD) was the first to enter the equation for AAQ explaining 28.5% of the variance (t=10.09, P<0.001), when IQ was also entered into the model it predicted 40.7% of the variance (t=3.24, P<0.01). For VAQ, the C-ADHD explained 27.5% of the variance individually (t=9.41, P<0.05) and when IQ (t=4.57) and age (t=3.96) were entered into the equation, 56.3% of the variance was explained (both P<0.05). For D-KEFS Inhibition, diagnosis (t=13.30) along with the ADHD index (t=2.47) explained 35.4% of the variance (P<0.05). Using linear regression analysis, the ADHD index predicted 16.1% and 11.1% of the variance for ARCQ (t=9.27, P<0.001) and VRCQ (t=7.96, P<0.05), respectively. The ADHD index was not predictive for the remaining variables.
Using an exploratory approach we investigated the influence of ADHD within the ASD group by removing these subjects. The ASD group was divided into those without ADHD (N=10) and with ADHD (N=8) based on the ADHD index (≥ 65), and MANCOVAs were conducted excluding the ASD/ADHD participants on domains which more consistently discriminate ADHD group; namely Inhibition (ARCQ, VRCQ) and Vigilance (AAQ, VAQ). The MANCOVA with all subjects included was highly significant for Inhibition (F(4,98) = 3.81, P = 0.006, Wilks’ lambda = 0.75. However, the exploratory MANCOVA without the ASD/ADHD group fell to trend level F(4, 82)=2.24, P=0.07; Wilks’ lambda = 0.81 (See ARCQ ) suggesting that the ASD/ADHD comorbid group significantly contributed to differences in these domains. The original Vigilance MANCOVA F(4,98) = 4.63, P<0.0001; Wilks’ lambda = 0.71 conducted without the ASD/ADHD group was reduced but remained statistically significant F(4,82) = 4.241, P=0.004; Wilks’ lambda = 0.687 suggesting that the results could not entirely be explained by ADHD symptoms (See AAQ ). It is important to note, however, that this approach resulted in unequal groups and a smaller sample size, which reduced the power to detect differences.
Figure 1 Figures 1a and 1b. Scatterplots of the relationship between the C-ADHD (Conners ADHD Index) and the 1a. ARCQ (Auditory Response Control Quotient) and 1b. AAQ (Auditory Attention Quotient) across the groups: TYP = Typical, ADHD = ADHD, ASD = Autism Spectrum (more ...)