Background
Autism spectrum disorders (ASD) are a group of disorders characterized by impairments in maintaining reciprocal interaction and communication with others and the presence of narrow interests and stereotyped patterns of behavior and activities [
1]. More recently, a quantitative, dimensional reconceptualization of ASD in the general population has been proposed [
2]. In other words, the concept of the autistic spectrum, originally conceived as a gradient of severity within the clinical range, has been extended to a continuum of autistic traits in the general population [
3‐
5]. This change suggests that typically developing individuals may display autistic traits that vary in both the degree of severity and number [
6]. These traits, known as the Broader Autism Phenotype (BAP), have been examined in the relatives of individuals with ASD and in the general population [
7‐
9]. In a continuum of severity of autistic traits in the general population, the BAP is generally considered to be a subclinical set of characteristics that is milder but qualitatively similar to the diagnosed autism phenotype [
10,
11].
One of the most widely used quantitative measures of BAP is the Autism-Spectrum Quotient (AQ). The AQ is a self-report screening instrument for measuring the severity of autistic traits across five subscales (social skills, communication, attention to detail, attention switching and imagination) in both the general population and the autism spectrum community [
3]. The 50-item questionnaire has good cross-cultural stability [
11‐
14] and demonstrates consistent results across different age groups [
15‐
17]. The AQ exhibits its own advantages compared with other self-administered measures of BAP; for example, the AQ has been shown to distinguish between individuals with high-functioning ASD and individuals with other psychiatric disorders [
18]. These characteristics have demonstrated that the AQ is a reliable instrument for quantifying the BAP, has been widely used with relatives of individuals with ASD, and has the ability of screening for autistic traits in the general population [
5,
7,
11,
19]. The AQ score is related to individuals’ performance on gaze-oriented attention to happy faces [
20] and on the global integration of closed contours [
21], has been found to be linked with white matter fiber tract [
22] and has been associated with white matter volume in the posterior superior temporal sulcus [
23].
It is particularly worth noting that Baron-Cohen et al. have found sex differences in the general population. The mean total AQ score was higher in males than in females [
3]. Similar sex differences in the AQ were found in studies conducted in the Netherlands [
8], Scotland [
9], Italy [
5] and Poland [
13]. The incidence of autism was much higher in males than in females [
24]. The exploration of sex differences in the severity of autistic traits in different groups of people and cultural contexts may contribute to a deeper understanding of the association between autism and gender [
13].
Existing evidence suggests that ASD and psychosis spectrum disorders share the clinical symptoms and manifestations [
25]. Longitudinal studies have found that children having a greater severity of early autistic traits are more likely to have psychotic experiences in early adolescence [
25,
26]. In addition, individuals with ASD were more likely to report schizophrenic symptoms, and schizophrenia (SCH) patients were more likely to report autistic symptoms [
27‐
30]. Similarly, both ASD and obsessive-compulsive disorder (OCD) have common features, such as obsessional interests and repetitive or stereotypic behaviors. Furthermore, certain vulnerability genes may prove to be generalist genes, influencing the phenotypic expression of both ASD and OCD [
31]. Individuals with ASD were more likely to show obsessive-compulsive traits, and patients with OCD were more likely to show autistic symptoms [
32,
33].
Regarding the studies noted above, determining whether the symptoms of individuals with ASD resemble those with SCH or OCD is necessary, and identifying the similarities and differences between ASD and the two other disorders is vitally important [
34]. Therefore, there is a growing demand for research on investigating how ASD can be effectively and sufficiently differentiated from other psychiatric disorders. The AQ is a widely used screening instrument that can be used to distinguish validly between high functioning adults with autism and individuals with other psychiatric disorders [
18]. Previous research has found that, in the autistic traits continuum, subjects with autism lie on one end and neurotypical subjects lie on the other end; the individuals with SCH or OCD can be placed approximately in the middle of this continuum [
28,
35].
To date, no standardized and brief measures that could be practical in ASD screening purposes have been made available in mainland China. The adaptation and validation of the AQ for the Mandarin Chinese-speaking population is important and necessary for both research and ASD screening purposes [
13]. Prior studies from Western cultures have shown that the parents of ASD children score significantly higher than do the parents of TDC children on the AQ, and men tend to have significantly higher scores than women on the AQ. These general patterns were replicated in Japan, India and Malaysia [
12,
36], which suggests that these patterns are stable and are independent of cultural influences, thus providing strong support for the validity of the AQ as an instrument for use in Eastern cultures. However, there are cultural nuances for certain autism-related behaviors, as described by the AQ. For instance, certain behaviors that are considered to be related to attention in Western cultures may actually have a social significance in Eastern cultures. In this study, we wanted to investigate the psychometric properties of the AQ and to examine whether the patterns noted above could be replicated in mainland Chinese populations.
In addition to a group with parents of ASD and a group with parents of typically developing children (TDC), this study included a sample of three different patient groups (a group with ASD, a group of patients with SCH and a group with OCD) and a healthy control group. In accordance with prior researches, we predicted that the AQ scores in these patient groups will be higher than the general population mean. Extremely high AQ scores were expected to be specific to individuals with ASD.
Primarily, the characteristics of the Mandarin Chinese AQ, including the internal consistency, the test-retest reliability and the discriminating power of items, were examined in a large sample of parents of ASD children and TDC parents. In accordance with the findings from previous studies, we hypothesized: 1) AQ scores to be continuously distributed in these two groups of parents; 2) higher AQ scores in parents of ASD children than in TDC parents; and 3) significantly higher mean AQ scores in males compared with females.
Discussion
First, this study set out to determine the basic psychometric properties of the Mandarin Chinese version of the AQ. Next, we studied the distribution of AQ scores in the parents of ASD children and TDC parents and performed an analysis on sex differences in the AQ. Furthermore, this study examined the usefulness of the AQ in differentiating between individuals with ASD, SCH and OCD.
To obtain a better understanding of the differences in AQ scores within the general population, the distribution of total AQ scores in the parents of ASD children and the TDC parents are graphed (refer to Fig.
1 and Fig.
2, respectively). These two figures illustrated that AQ scores followed a continuous distribution in the parents groups. As found in previous studies [
3,
6,
8], AQ scores in the ASD and TDC parents groups had an approximately normal distribution. This study showed that symptom features of the autistic profile in parents of ASD children were not obviously different from that in TDC parents. These findings suggest that the Mandarin Chinese AQ model is fitting for a continuum of autistic expression in agreement with the perspective that autistic traits are part of a broader phenotype, on which features lie along a continuum [
6]. Therefore, the quantitative, dimensional reconceptualization of ASD is reflected not only in the general population but also in parents of individuals with ASD.
The analysis of group differences showed that the parents of ASD children scored significantly higher than the TDC parents in the total AQ scores and the subscales. These results are in accordance with most of the previous related studies [
5,
11] and extend the earlier finding by Bishop et al. [
41], who found that AQ scores differentiate the parents of children with ASD from healthy control parents on two subscales (communication and social skills) in a small sample. To the best of our knowledge, there is one sole study that did not find any significant difference between the parents of children with ASD and control parents [
42]. However, this study included 25 parents with autistic children and 25 control parents. Our study provided more valuable data regarding the discriminant power of the AQ obtained from a comparison of the parents of ASD children with the controls matched for sex, age and educational level.
Although previous studies found sex differences in healthy adults [
3,
8,
43], it is crucial to determine whether the sex effects play a part in the expression of the BAP phenotype in parents. In our study, the analysis of sex differences revealed that, in the total AQ as well as in four of its subdomains (social skills, communication, attention switching and imagination), fathers scored higher than mothers. The exception to this was the subdomain of attention to detail; in this, there was no sex difference. Similar differences have previously been reported in samples of parents [
10] and samples of students [
13,
44]. Dawson et al. using the Broader Phenotype Autism Symptom Scale, which is another quantitative assessment of autism symptom-related traits, also found the sex differences [
7].
With regard to internal reliability, the findings of our study are similar to those obtained by other researchers in other cultures [
13,
18,
45]. Internal consistency was highest for the total AQ, social skills and communication, whereas other subscales demonstrated moderate Cronbach’s α coefficients. It is also worth noting that the lowest Cronbach’s α was found for imagination. These results were in accordance with previous studies in other language samples [
5,
12,
13,
18,
43], regardless of the version of the instrument. In at least four existing factor analytic studies, the subscale imagination was never established as a separate subscale [
8,
19,
45,
46]. Obviously, the items that develop the subscale imagination load on different subscales of the AQ. Therefore, it is possible that the lowest Cronbach’s α for imagination could be due to the questionable validity of this particular subscale [
28].
The results of the item analysis indicated that the discriminating power of most of the items was acceptable. On 2 items of 50 (items 18 and 30), the group of TDC parents scored higher than the group of parents of ASD children, which strongly confirms the value of these items for discriminating between parents of ASD children and TDC parents. However, it should be stressed that, as in this study, the British and French studies [
3,
14] have also found that healthy controls scored higher than ASD participants on item 30. This finding may show that item 30 should be improved or revised.
In this study, participants in SCH and OCD groups scored significantly higher on the total AQ than healthy controls but lower than participants with ASD. It appears that the total AQ score follows a continuous distribution in the healthy controls; participants with ASD fall in the upper end of this distribution [
28], and SCH and OCD patients obtain scores between the healthy controls’ mean and scores typical for participants with ASD. These results demonstrate that the AQ appears useful for distinguishing the SCH and OCD groups from participants with ASD and healthy controls.
More significantly, there were disorder-specific manifestations on the AQ between the SCH and OCD groups. The results showed that the SCH group reported more problems in social skills and communication compared to the individuals with OCD, as was found in the ASD group compared to the OCD group. This finding means that there is potential symptoms overlap between the ASD and SCH groups, and the symptoms here mainly refer to impairments in social interaction. These findings were in accordance with previous studies that found a high degree of similarity in the social function deficits experienced by individuals with ASD and SCH [
47,
48]. Furthermore, our findings revealed that the OCD group reported more problems in attention switching and attention to detail compared to the individuals with SCH, as was found in the ASD group compared to the SCH group. These results imply that ASD and OCD may share attention problems, as was found in another study that suggested attention problems may reflect both symptoms overlap and a common etiological factor underlying ASD and OCD [
49].
Several shortcomings of this study should be noted. First, we did not assess the predictive validity of the AQ. Second, certain clinical features such as the severity of symptoms and the symptom dimensions reported by the patients were not addressed in this study. The presence of such clinical features could influence the AQ score as reported by the patients. Ideally, the validity of the AQ should also have been studied by comparing the AQ score with another self-administered questionnaire designed to measure autistic traits. However, there are no other self-reported questionnaires in mainland China; therefore, we could not administer more questionnaires to the participants.
Abbreviations
AD, autism disorder; AQ, Autism-Spectrum Quotient; AS, Asperger syndrome; ASD, autism spectrum disorders; BAP, Broader Autism Phenotype; HC, healthy controls; OCD, obsessive-compulsive disorder; PDD-NOS, pervasive developmental disorder not otherwise specified; SCH, schizophrenia; TDC, typically developing children
Acknowledgements
We thank all the participants in this study. The authors also want to acknowledge the valuable support from Huacai Zhang, Liping Zheng, Zhitu Liang, Hongqiong Guo, Wei Chen, Chen Hong, Rong Chen.