Autism spectrum symptoms in children with neurological disorders

Participants

Data stem from the second wave of the BCS, an ongoing, longitudinal population based study of Norwegian children born in the years 1993–1995 (http://www.uib.no/bib). As the protocol and population of the BCS have been detailed elsewhere [40, 41], the current paper will provide only a brief description. The first wave of the BCS was conducted in 2002. Parents and teachers of all children (9430 in total) attending second through fourth grade at elementary school (7–9 years of age) in any public or private school in the city of Bergen were asked to fill in a questionnaire, including the ASSQ. Informed consent to participate was provided by the parents of 7007 children. Based on the teacher ratings of 2544 children for whom parent ratings were not obtained, the impact of non-responses on estimates of mental health problems was examined in wave 1. Negligible non-response bias was found for mean scores and correlations, but logistic regressions indicated that children rated to have mental health problems by their teachers were less likely to participate. This mainly concerned children with moderate levels of symptoms [42]. The second wave was conducted in 2006. Parents and teachers of 5781 children, now in the fifth to seventh grades (11–13 years), participated alongside the children themselves by filling in a questionnaire, which again included the ASSQ. Compared to the wave one sample, the participants in wave two had somewhat lower mean ASSQ total scores as reported by parents and teachers, but differences were very small (ES =.05-.06). Ethnic diversity in this sample is expected to be minimal [43]. For more details about the second wave, see Hysing and collaborators [44]. The BCS was approved by the Regional Committee for Medical and Health Research Ethics Western Norway, and by the Data Inspectorate.

Instruments

Autism spectrum symptoms

The ASSQ is designed to identify school-aged children who may need a more comprehensive evaluation due to suspected ASD. It was designed for completion by lay informants, and identical versions exist for parents and teachers. The ASSQ covers a wide range of symptoms predictive of a diagnosis within the autism spectrum, including difficulties with social interaction, verbal and non-verbal communication, restricted and repetitive behavior, motor clumsiness, and tics. It consists of 27 items scored on a three-point scale: not true (0), somewhat true (1), and certainly true (2). Possible scores range from 0–54, with higher scores indicating a greater symptom load [3]. Children exceeding the 98^th percentile on the parent and/or teacher reported ASSQ total score in the present sample were defined as high-scoring, indicating severe impairment according to a validation study by Posserud et al. [28].

The factor analysis from the first BCS wave (28), including parent and teacher ASSQ reports, was rerun on the full sample from the second BCS wave. The factor structure was confirmed with the following differences: two items, “Different voice/speech” and “Idiosyncratic attachment”, were not included in any factor in the analyses from the first wave, but now loaded clearly onto the second factor (motor/tics/OCD) for both parents and teachers, and were thus incorporated into this factor in the present study. The item “involuntary sounds” was accidentally omitted from the parent questionnaire in the second wave. It was therefore excluded from the teacher questionnaire in the present study. For more details regarding this factor analysis, see Posserud et al. [30].

Missing data

When calculating the ASSQ total score, mean individual scores were inserted for missing items when 4/26 (15%) or fewer items were missing. ASSQ forms with more than four items missing were discarded from analyses of the ASSQ total score. Forms from 5129 parents (88.7%) and 5540 teachers (95.8%) were completed with fewer than four ASSQ items missing. When calculating the ASSQ factor scores, mean individual scores were inserted for missing items when only one item was missing. ASSQ forms with more than one item missing from any factor were discarded from factor score analyses. No missing items were substituted in the single item analyses. Data about intellectual disability and sex was present for 5116 (88.5%) and 5195 (89.9%) children, respectively. Due to the discrepancy between response rates for parents and teachers, and missing data regarding intellectual disability and sex, the number of participants varies across the present analyses.

Statistical analyses

All analyses were performed using PASW Statistics 17. Descriptive statistics, Chi-squares and t-tests were used to explore sample characteristics. One-way between-groups analyses of covariance (One-way ANCOVA) were conducted to explore ASSQ scores reported by parents and teachers in the sample, with intellectual disability and sex as covariates. Due to uneven group sizes and violation of the homogeneity of variance, the Hochberg’s GT2 post hoc test was used to evaluate the results of the one–way ANCOVAs, and the significance level was set to p<.01. Effect sizes (Cohen’s d) were calculated and interpreted according to the guidelines of Cohen [46], in which a d value of 0.20 is small, 0.50 is medium, and 0.80 is large. Cross-tabs were used to calculate the frequency of children for whom the parent and/or teacher reported an ASSQ total score above the 98^th percentile. First, we calculated the frequency of children for whom either the parent or the teacher reported a score above the 98^th percentile; secondly, we calculated the frequency of children for whom both informant groups reported such a score. The same procedure was used to calculate the frequency of children scoring above the 80^th, 90^th, and 95^th percentiles. Bivariate correlations (Pearson’s r) were conducted to explore the agreement between the parent and teacher reported ASSQ scores in children with neurological disorders. These were interpreted according to the guidelines of Cohen (1988), in which an r value of .10 to .29 is low, .30 to .49 is moderate, and .50 to 1.0 is high. For analyses at the item level, scores of 0 (not true) and 1 (somewhat true) were collapsed, and the dichotomy absence of behavior (0, 1 = 0) and presence of behavior (certainly true; 2 = 1) was used. McNemar Chi-square tests were used to explore the relationship between item scores reported by parent and teachers for children with neurological disorders, using Cohen’s kappa [47] to quantify the level of inter-rater agreement. According to the guidelines presented by Fleiss [48], kappa values below .4 indicate poor agreement, values between .40 to .75 indicate fair to good agreement, and values above .75 indicate excellent agreement.

Sample characteristics

The ASSQ total score

The ASSQ high scorers

The ASSQ factor scores

The groups differed significantly in terms of their ASSQ social difficulties factor score as reported by both parents (F(2, 5053) = 77.65, p = .0005, d = 0.35) and teachers (F(2, 5009) = 43.57, p = .0005, d = 0.26). This difference remained significant after controlling for intellectual disability (F(1, 5053) = 30.65, p = .0005, d = 0.16) and sex (F(1, 5053) = 57.09, p = .0005, d = 0.21) for the parent reported social difficulties factor score, and after controlling for intellectual disability (F(1, 5009) = 24.23, p = .0005, d = 0.14) and sex (F(1, 5009) = 180.18, p = .0005, d = 0.38) for the teacher reported social difficulties factor score.

Further, the groups differed to a statistically significant level in their ASSQ motor/tics/OCD factor scores as reported by parents (F(2, 5060) = 74.56, p = .0005, d = 0.35) and teachers (F(2, 5015) = 54.19, p = .0005, d = 0.29). This difference remained significant after controlling for intellectual disability (F(1, 5060) = 14.27, p = .0005, d = 0.11) and sex (F(1, 5060) = 22.89, p = .0005, d = 0.14) for the parent reported motor/tics/OCD factor score, and after controlling for intellectual disability (F(1, 5015) = 44.76, p = .0005, d = 0.19) and sex (F(1, 5015) = 70.09, p = .0005, d = 0.24) for the teacher reported motor/tics/OCD factor score.

Finally, there was a statistically significant difference between the groups in the ASSQ autistic style factor score as reported by parents (F(2, 5061) = 29.51, p = .0005, d = 0.22) and teachers (F(2, 5024) = 17.21, p = .0005, d = 0.17). This difference remained significant after controlling for intellectual disability (F(1, 5061) = 6.99, p = .01, d = 0.06) and sex (F(1, 5061) = 3.64, p = .06) for the parent reported score, and after controlling for intellectual disability (F(1, 5024) = 0.12, p = .73) and sex (F(1, 5024) = 45.64, p = .0005, d = 0.19) for the teacher score. Post-hoc tests showed that both parents and teachers reported children with neurological disorders to have significantly higher factor scores than did children with other or no chronic illnesses (p < .01), with medium to large effect sizes (d = .50-1.00). The mean differences between the groups with other chronic illnesses and no chronic illness were all non-significant (Table 2).

Informant agreement in children with neurological disorders

Summary of findings

In this population-based study of 11–13 year olds, children with neurological disorders obtained significantly higher parent and teacher reported ASSQ scores than did children with other chronic illnesses or no chronic illness. 14.1% of the children in the neurological disorder group obtained a score above the cutoff for positive ASD screening, as derived from combined parent and teacher ASSQ reports. The frequency of children identified with a score greater or equal to the 80^th, 90^th, and 95^th percentile cutoff points was higher than for children belonging to the two other groups. Overall, the agreement between parents and teachers of children with neurological disorders was moderate to high for the ASSQ total and factor scores, and poor for most of the single items.

The ASSQ total score and high-scorers

The elevated ASSQ total score in children with neurological disorders is in accordance with the high rate of ASD symptoms found in studies of children with hydrocephalus [7, 9] and epilepsy [11]. Although our sample represented a heterogeneous group including a range of neurological disorders, the combined parent and teacher ASSQ of 14.1% scoring above the 98^th percentile is in agreement with studies indicating that certain diagnoses included in our sample, for instance cerebral palsy [15], entail a heightened risk of ASD. Furthermore, our results showed that a large number of children with neurological disorders were rated as presenting with several ASD symptoms. A high frequency of children with neurological disorders was found above the 80^th percentile cutoff point on the ASD problem dimension in the total sample. These findings indicate that children with neurological disorders probably affect the percentile cutoffs for the total population by being overrepresented in the highest quartile.

The ASSQ factor scores

The ASSQ scores of the neurological disorder group were elevated across all three factors as compared to the two other groups, i.e. in terms of social difficulties, motor/tics/OCD issues, and autistic style. The fact that the children with neurological disorders obtained the highest mean score and the largest effect size for the social difficulties factor score according to both informants, suggests that social difficulties are observed to be the most frequent symptoms in this group, and that the elevated total scores are only to a lesser extent affected by symptoms which may be more directly linked to the neurological disorders, such as motor difficulties. This result is in accordance with the study by Ekstrom and collaborators [14], in which impairment in social interaction and communication was the main problem in children with myotonic dystrophy and ASD.

Informant agreement in children with neurological disorders

The study yielded high to moderate parent-teacher correlations on the ASSQ scores in children with neurological disorders, as in the study by Huberty and collaborators [39] which investigated informant agreement on behavior ratings in children with epilepsy. The correlation for the ASSQ total score in children with neurological disorders is comparable to the parent-teacher correlation of 0.66 found in the clinical study by Ehlers and collaborators [3]. In contrast, the correlations for children with other chronic illnesses and no chronic illness were mainly low or moderate, which is a general finding in reports on child psychiatric symptoms [34–36]. Furthermore, parents and teachers co-identified 45.2% of the ASSQ high-scorers in the neurological disorder group, which is higher than the equivalent rates for the two other groups, and higher than the 24% of ASD cases which were co-identified by parents and teachers in the study by Mattila and collaborators [38]. Altogether, these findings may indicate that parent-teacher agreement is higher in children with neurological disorders. On the other hand, the level of agreement between parents and teachers regarding the single items was generally poor, as in a previous BCS-study on items assessing ODD symptoms [36].

The relatively high correlation between the parent and teacher reported ASSQ total score may indicate that parents and teachers of children with neurological disorders have a shared understanding of the overall severity of the problem, but that, as indicated by the low level of agreement on the single items, they report different problems. This may be related to the different contexts in which they tend to observe the child. Parents more frequently see the child in a one-to-one setting, in which tics and compulsory behavior may be more easily observable, whereas teachers are better positioned to see the child interact with peers, and may thus be better able to identify social difficulties. Low parent-teacher agreement may also reflect differences in the understanding and interpretation of symptoms. For instance, the item “Bullied by other children” was reported as “certainly true” significantly more often by parents than by teachers. This might suggest that parents, more than teachers, perceive lack of friends or peer interaction as a type of bullying. Low agreement may also be partly explained by de facto differences in individual child behavior across settings. This underscores the general importance of obtaining information from both family and school settings when screening for problems associated with ASD. However, the high parent-teacher correlation on ASSQ scores in children with neurological disorders suggests that information from one setting may still be sufficient when assessing these children in a pediatric clinic.

Strengths and limitations

The main strength of the present study is its population-based design. The total sample size and the size of the neurological disorder group increased the power and generalizability of the results. Another strength is the use of a validated instrument which included both parent and teacher reports.

This study has several limitations. First, chronic illness was assessed by parent reports, without medical verification of the diagnosis or its severity. Secondly, parent reports of intellectual disability were not validated by a formal test of intellectual function. Third, sensitivity to a prior ASD diagnosis may have biased the way in which parents and teachers responded to the ASSQ. Fourth, the overall participation rate was 70% in wave one and well above 50% in wave two. This attrition rate may have affected the representativity of the wave two participants, although the differences in mean ASSQ total scores between the wave one and wave two samples were very small. The lack of diagnostic assessment of ASD was another major limitation of the study. Parent and teacher reports on a screening questionnaire cannot replace a clinical validation of diagnosis. However, although no clinical validation study was performed in this second wave of the BCS, the rate of “clinically diagnosable” ASD in the children with a parent and/or teacher ASSQ total score above or equal to the 98^th percentile is likely to be high [28]. Nonetheless, caution should be applied in interpreting these results as indicating ASD rates, as firm conclusions require clinical confirmation.

Clinical implications

The present findings of a high frequency of ASD related symptoms in children with neurological disorders emphasize the importance of clinicians having access to, and making skilful use of, screening instruments enabling them to recognize and identify problems which may pose a great challenge to the affected children and their families. The elevated scores on the social difficulties factor highlight the need for interventions targeting such difficulties in this group as a whole, and not only in children who meet criteria for a clinical diagnosis of ASD. As ASD share symptoms with other developmental/psychiatric disorders, such as ADHD, differential diagnostic evaluations are advised. The present study highlights for the first time the utility of the ASSQ to identify problems in children with neurological disorders. Its results call for future clinical studies of the children defined as “screen positives” for ASD in the present study.