We explored the overall relationship between the symptoms of autism through a large sample of children with autism and an integrative approach-network analysis in this study. Overall, there were broad relationships between the core and associated symptoms included in this study; and almost all of them were positive, with only one negative relationship existed in “anxious/depressed problems” and “awareness.” The associations between some of these symptoms were stronger than the other associations. In consistency with the clinical characteristics of ASD, the symptom of “communication” was the most central to the network. The strength centrality and network edges were highly stable, indicating that our results are highly reliable.
- Central symptoms in ASD
The strength centrality index demonstrated excellent reliability; thus, we focus our discussion on symptom centrality on strength.
The social communication problem, the core symptom in ASD, was the most central symptom in our study. Previously, there have been few network analyses on the multidimensional relationship between symptoms of autism. Most of them were compared with other disorders on one or two symptoms, such as depression[19] and obsessive-compulsive disorder[20], or evaluated with associated symptoms of autism[21]. What is consistent with our research results is that, in Anderson’s studies about network analysis of autism, “social” and “communication” domains of Autism Diagnostic Observation Schedule (ADOS) ranked top two central items[21]; and the “anxiety” node in both ADOS and Revised Child Anxiety and Depression Scale (RCADS) items were highly peripheral[12, 21], although previous study found a lack of correlation between items scores of CBCL and ADOS[22, 23]. Here, “communication” was the most central node in the network, and “anxious/depressed problems” node in the CBCL items had relatively more peripheral centrality estimates. The lowest centrality node was “somatic complaints”, indicating that somatic problems were a peripheral symptom in ASD. However, there were quite a number of people with autism who had experienced varied degree of chronic gastrointestinal (GI) problems during different stages of the disorder[24, 25], and previous study suggested a interaction between anxiety and GI problems[26]. Additionally, GI problems were associated with externalizing problem behavior and internalizing symptoms, and the relationship varies in different ages of patients[25]. Therefore, further studies are needed to explore the factors of the symptoms’ centrality estimates.
- Connections among the core and associated symptoms of ASD
There were broad correlations between core and noncore symptoms in patients with ASD. In this study, a direct relationship was found between anxiety problems and core symptoms, such as restricted interests and repetitive behavior. A recent study indicated an insufficient evidence to support a two-way relationship between anxiety and core symptoms; however, it was also found that anxiety would gradually increase the risk of social communication dysfunction, and the treatment of anxiety symptoms would help improve social and emotional functions[27]. Furthermore, several studies demonstrated that the presence of anxiety and depression symptoms leads to increased severity of ASD symptoms[28, 29]. Similarly, aggressive behaviors and sleep problems had connections with core symptoms and could link via some nodes within the network to the core symptoms. Previous studies suggested that sleep disturbance (such as short sleep duration) was associated with the severity of core symptoms[30] and related symptoms[8, 31] in ASD. Additionally, our previous work found that aggressive behaviors were significantly associated with sleep problems and ADHD symptoms, which were treatable[14], and brought new insight into the treatment of aggressive behaviors in ASD.
High comorbidity rates of ADHD were observed across ASD[32]. An increasing amount of evidence suggests that ASD and ADHD may have shared genetic risk patterns[33-36]. Family-based studies show that siblings of patients with ADHD have an increased risk of developing ASD symptoms[34], while studies of twins show that common genetic factors may explain 50% to 70% of the covariance between ASD and ADHD symptoms[35, 37]. Here, we found that there were direct or indirect relationships between the characteristic symptoms of ADHD, attention problems, and the core symptoms of ASD—social function and autistic behavior. We can learn that there is also an internal relationship between the clinical phenotypes of ASD and ADHD. Both as neurodevelopmental disorders, ASD may overlap with ADHD in genetic and clinical perspectives that brings new insights into recognizing them.
Relationships within associated symptoms were also found in this study. The edge between sleep problems assessed by CBCL and CSHQ, in line with expectation, was stronger than all other edges in this network that indicated the high consistency of the two scales and the good sleep status of children with autism. Stronger connections also existed between emotional problems and anxious/depressed problems, emotional problems and aggressive behavior, and withdrawal and attention problems. This means that these symptoms could interact with each other, and a stronger symptom leads to a greater degree of other symptoms. As mentioned earlier, there is a high prevalence of these symptoms in patients with ASD. However, previous studies have not clearly explained the relationship between these symptoms. In Duvekot’s study, however, results demonstrated that the treatment of anxiety symptoms would help improve emotional function[27]. Furthermore, the only negative connection existed in anxious/depressed problems and social awareness assessed by SRS. Stephen et al. found that ASD patients with higher SRS scores reported greater degree of anxious/depressed problems[38]. Other studies also suggested that social function reflected by SRS scores could predict the severity of anxiety[39, 40]. Mikle et al. found that anxiety symptoms assessed by different scales had significant correlations with the five subscales of SRS, including the social awareness domain[41]. These findings were contrary to our results.
Previous research and clinical evidence show that the treatment of the core symptoms of autism is not effective; whether drug or nondrug treatment, its effect is very small. Other ASD-related symptoms, including attention problems, hyperactivity, anxiety or depression, and sleep disturbance, can also cause clinically significant problems[4]. Many tools can be used to manage these symptoms, such as drug treatment, cognitive-behavioral therapy, and psychosocial treatment[42]. As shown in this study, these ASD-associated symptoms have a complex positive correlation with core symptoms in children with ASD and can interfere with each other, indicating that we can alleviate the core symptoms by treating them and then improve the prognosis of patients, which requires follow-up data for confirmation.
Limitations
The current study also had some limitations. First, as a cross-sectional study, the network analyses was conducted by using cross-sectional data, so the analysis of some network characteristics is not sufficient. It remains unknown that whether the network structure will change with the progress of the course of the disease or the change of the severity of symptoms. Second, some nodes may actually be measuring overlapping constructs (e.g. “sleep problems” measured by CBCL and CSHQ respectively), which could artificially inflate edge weights and centrality. Third, a group of typically developed children were not enrolled as a controlled sample, and parent questionnairs were used in this study, which may lead to the existence of some confounding factors. In the future, further research can be carried out through a well-designed follow-up study.