Autistic traits in children with ADHD: preliminary evidence for a subgroup with elevated rule-breaking and aggressive behaviors Authors: Yu-Wei Fan, Yuan Wang, Te Fan, Liang-Wen Chen, Xiao-Lan Cao, Chao Dong, Jun Zhu, Lin-Lin Zhang, Qian Sun, Jing Chen, Li Zhang, Geng-Tian Zheng, Xiao-Shan Qiu, Yu Lan, Ying Li, Lu Liu, Bin-Rang Yang, Juan Liu Published: 2026 ABSTRACT AbstractBackgroundAutistic traits (ATs) are prevalent and clinically significant among children with Attention Deficit/Hyperactivity Disorder (ADHD); however, their specific behavioral correlates have not been fully characterized. This cross-sectional study investigated the presence of ATs and their relationship with behavioral profiles within this population.MethodsA total of 154 children were enrolled, including 48 healthy controls (HCs), 69 children with ADHD, and 37 children with comorbid ADHD and Autism spectrum disorder (ASD). The Autism Spectrum Screening Questionnaire (ASSQ) and the Child Behavior Checklist (CBCL-ASD), composed of Withdrawn, Thought Problems, and Social Problems subscales, were used to assess ASD symptoms and explore the structure of ATs. ATs were defined as a total score of ASSQ ≥ 12, leading to the subdivision of the ADHD group into ADHD without ATs (ADHD-ATs, n = 40) and ADHD with ATs (ADHD + ATs, n = 29). To characterize the non-ASD behavioral features associated with ATs, behavioral differences across HCs, ADHD-ATs, ADHD + ATs, and ADHD + ASD groups were compared using CBCL non-ASD scales.Results1. ATs in ADHD: Compared to HCs, the ADHD group showed significantly higher scores on all ASSQ and CBCL-ASD measures except the ASSQ communication problems subscale (all P < 0.05). However, their scores were lower than those in the ADHD + ASD group (all P < 0.05). When further divided by ATs, the ADHD + ATs subgroup exhibited widespread ASD-related impairments. Their ASSQ scores and CBCL-ASD social problems scores exceeded those of both ADHD-ATs and HCs (all P < 0.05), and were comparable to the ADHD + ASD group. 2. Non-ASD Behaviors in ADHD + ATs subgroup: Apart from anxiety/depression, the ADHD + ATs group scored higher than both ADHD-ATs and HCs on all CBCL-non-ASD scales, and exceeded ADHD + ASD in aggressive behavior and rule-breaking (all P < 0.05). Multiple linear regression showed that aggressive behavior was predicted by social-interaction, withdrawn and thought problems scores; rule-breaking was predicted by communication problems scores, withdrawn and thought problems; attention problems was predicted by all CBCL-ASD scales; somatic complaints was predicted by social-interaction and thought problems.ConclusionsATs may represent a clinically characterized subgroup within ADHD, characterized by global impairments across all core ASD symptom domains, alongside elevated non-ASD behavioral problems, particularly in externalizing behaviors. Notably, specific ASD symptom profiles differentially predict distinct patterns of non-ASD behaviors. Exploring the ATs subgroup can help elucidate the heterogeneity and complexity of ADHD, thereby contributing to precision treatment for the disorder.Clinical trial numberNot applicable.Keywords: Attention-deficit/hyperactivity disorder (ADHD), Autistic traits (ATs), Behavior, Comorbidity Background Attention-Deficit/Hyperactivity Disorder (ADHD) is a prevalent neurodevelopmental behavioral disorder in children and adolescents, characterized by developmentally inappropriate levels of inattention, hyperactivity, and impulsivity [1]. The global prevalence of ADHD among children and adolescents is 7.2%, with a reported rate of 6.26% in China [2]. Autism spectrum disorder (ASD) is characterized by impairments in social interaction and communication, as well as restricted and repetitive interests and behaviors (RRBs) [1]. Globally, ASD affects approximately 1 in 31 individuals [3], with a prevalence of 0.7% in China [4]. Both ADHD and ASD are neurodevelopmental disorders contributing substantially to the childhood disease burden. According to the Global Burden of Disease Study 2019 [5], ADHD and ASD rank 6th and 8th, respectively, among all mental disorders in terms of disability-adjusted life years (DALYs). The age-standardized DALY rates for ADHD and ASD were 32.5 and 55.5 per 100,000 population, respectively, highlighting their importance as major public health concerns [5]. Moreover, ASD and ADHD frequently co-occur [6]. This high comorbidity is supported by evidence of shared genetic risk factors, as demonstrated by twin and family studies [7]. Neuroimaging research has also revealed overlapping neurobiological substrates [8], and common impairments in executive functions are observed in both disorders [9]. The co-occurrence of ADHD and ASD exacerbates the core behavioral symptoms of both disorders [6, 10–12]. In addition to the comorbidity between ADHD and ASD, individuals with ADHD may show autistic traits (ATs) even without a formal ASD diagnosis. These traits do not meet the diagnostic criteria for ASD, but share several clinical features with ASD [13, 14]. Previous research has reported a significantly higher prevalence of ATs among children with ADHD, with rates ranging from 15% to 64.3% [15]. Compared to children without ATs or Healthy controls (HCs), those with ADHD who exhibit ATs have an increased risk of multiple psychiatric comorbidities, including oppositional defiant disorder, conduct/antisocial disorders, anxiety disorders, major depressive disorder, bipolar disorder [16], sleep disorders [17], substance use disorder [18], and language disorders [19]. It still remains unclear whether ATs in ADHD are conceptually consistent with ASD or represent a subgroup shaped by ADHD. Although both children with ADHD and ASD lead to impaired social function, their underlying neurocognitive mechanisms and etiologies differ greatly. Social difficulties in children with ADHD may result from poor behavioral control, such as interrupting others or making impulsive social approaches, with mild social cognitive impairments [6]. In contrast, social impairments in ASD often arise from core deficits in social motivation and cognition, including a lack of active social initiation (e.g., initiating contact, maintaining eye contact), deficits in understanding social cues, and poor theory of mind [6]. Interestingly, recent studies have further pointed out that typical ADHD symptoms like talkativeness and hyperactive and impulsive behaviors (sometimes perceived as eccentricity) are not simply behavioral deficiencies; they may also serve as potential advantages in social interaction [20]. Given the complexity, it remains unclear whether ATs in ADHD represent a phenotype closer to ASD, a subgroup intrinsic to ADHD itself, or a combination of both. Further studies are needed to clarify the specific social characteristics and clinical influences of this special group. Numerous studies have focused on the comorbidity between ADHD and ASD, while relevant research concerning the clinical features of ATs in ADHD children and their correlations with functional impairment remains relatively insufficient. Thus, this study intends to explore the presence of ATs in children with ADHD, clarify their core symptom composition, and analyze the behavioral features associated with these co-existing traits. The results are expected to provide a basis for formulating targeted intervention measures and further deepen the understanding of clinical heterogeneity among children with ADHD. Methods Participants and procedures A cross-sectional study was conducted involving children who visited the Children’s Healthcare and Mental Health Center, affiliated with Shenzhen Children’s Hospital of Shantou University Medical College (SCH, Shenzhen, China) from June 2017 to April 2020. Written informed consent was obtained from the parents of all participants, and the study was approved by the Medical Ethics Committee of SCH. Patients came to the clinic on their first visit. They underwent a semi-structured clinical interview that used the Kiddie Schedule for Affective Disorders and Schizophrenia Present and Lifetime Version (K-SADS-PL) to confirm the diagnosis of ADHD and exclude other psychiatric disorders [21]. Additionally, they were also evaluated by an experienced psychiatric specialist for ASD, sleep disorders, and learning disabilities. Inclusion criteria for children with ADHD were as follows: (a) met the diagnostic criteria for ADHD according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) and the K-SADS-PL; (b) age 6–16 years; (c) Full-Scale Intelligence Quotient (FSIQ) ≥ 70; and (d) without any drug, behavioral, or psychological intervention for ADHD. Exclusion criteria for children with ADHD were as follows: (a) any other psychiatric diagnosis according to the K-SADS-PL(such as tic disorders, intellectual disability, conduct disorder, schizophrenia, or affective disorder), as well as a diagnosis of ASD, sleep disorder diagnosis, or learning disability; and (b) physical illnesses and neurological disorders such as epilepsy, short stature, congenital heart disease, enuresis, or immune encephalitis. The inclusion and exclusion criteria for ADHD + ASD were identical to those for ADHD, with the additional requirement that participants meet the DSM-5 diagnostic criteria for ASD, confirmed by positive results on either the Childhood Autism Rating Scale (CARS) or the Autism Diagnostic Observation Schedule (ADOS). The inclusion criteria for the HCs group: (a) did not meet criteria for any disorder listed in the K-SADS-PL or for ASD, sleep disorders, or learning disabilities according to DSM-5. Exclusion criteria included any past history of psychiatric illness (even if currently resolved); the remaining criteria were the same as those for the ADHD group. Each child completed an intelligence assessment administered by a qualified examiner, and parents of all participants filled out the CBCL and ASSQ to evaluate ADHD and ASD symptoms. An ASSQ total score ≥ 12 was used to identify children with ATs [22]. The ADHD sample was therefore split into ADHD + ATs and ADHD–ATs subgroups. In total, 48 HCs, 69 children with ADHD, and 37 children with ADHD + ASD were included in the behavioral analyses. Measurements Wechsler intelligence scale for children – fourth edition (WISC-IV) The WISC-IV is used to evaluate the intellectual developmental level of children over 6 years of age. The scale was administered by trained professionals, and the FSIQ was recorded for the study [23]; a score of FSIQ below 70 was considered abnormal. Autism spectrum screening questionnaire (ASSQ) The ASSQ was developed by Ehlers et al. in 1999 [24]. The scale comprises 27 items rated on a three-point scale (0 = “normal”, 1 = “somewhat abnormal”, 2 = “definitely abnormal”): nine items assess social interaction, seven assess communication problems, and eleven assess RRBs [22]. The scale has demonstrated good internal consistency (Cronbach’s α = 0.86) [25]. To define the ADHD + ATs group in our study, we used the Mandarin Chinese version, for which a cut-off score of ≥ 12 distinguishes individuals with ASD from unaffected controls [22]. Achenbach child behavior checklist (CBCL) The parent-report version of the Achenbach Child Behavior Checklist was used to assess children’s psychological and behavioral problems [26]. The scale comprises eight syndrome subscales: anxious/depressed, withdrawn, somatic complaints, thought problems, aggressive behavior, rule-breaking behavior, attention problems, and social problems. It contains 113 items, each rated by parents on a 3-point scale (0 = “not true”, 1 = “somewhat/sometimes true”, 2 = “very/often true”). The score for each syndrome scale is the sum of its item scores, with higher scores indicating greater severity. The CBCL has high test-retest reliability and good internal consistency [7, 27]. Some studies have shown that the combined scores of the withdrawn, social problems, and thought problems subscales effectively screened for ASD [16, 28, 29]. For instance, Biederman et al. (2010) [29] found that these three subscales best discriminate ASD status (AUC = 0.86) and proposed a “CBCL‑ASD profile” as a rapid screening tool. Joshi et al. (2020) [16] later applied this CBCL profile (Withdrawn + Social + Thought Problems aggregate score ≥ 195) to operationalize autistic traits in youth with ADHD, demonstrating its stability over a ten‑year follow‑up and its prognostic utility for adverse functional outcomes. Based on this evidence, we therefore divided the CBCL scores into two parts: CBCL-ASD, reflecting ASD symptoms, and CBCL-non-ASD, reflecting other behavioral problems. Statistical analysis Categorical variables are presented as frequencies and percentages (n, %), with group comparisons performed using the chi‑square test or Fisher’s exact test, as appropriate. For continuous variables, normality of distribution was assessed prior to analysis. Normally distributed data are expressed as mean ± standard deviation (SD) and were compared using one‑way analysis of variance (ANOVA). Non‑normally distributed data are presented as median (interquartile range, IQR) and were compared using the Kruskal–Wallis H test. The overall effect size for the Kruskal–Wallis test is reported as epsilon squared (ε²), and the effect size for each Dunn’s post‑hoc comparison is reported as r (rank‑biserial correlation). To adjust for potential covariates (age, male sex, and FSIQ), analysis of covariance (ANCOVA) was employed for other continuous variables. Parametric ANCOVA was used for variables with normally distributed residuals. Effect sizes are reported as partial eta‑squared (ηₚ²) for the overall group effect and Cohen’s d for the standardized mean difference between groups after covariate adjustment. In cases where the assumptions of homogeneity of variance or normality of residuals were violated, Quade rank‑transformed analysis of covariance (Quade ANCOVA) was applied. For the Quade ANCOVA, the overall effect size is reported as partial eta‑squared (ηₚ²) based on the rank‑transformed data, and the effect size for each pairwise comparison is reported as r (rank‑biserial correlation). Regarding the choice of post-hoc pairwise comparison methods for multi-group comparisons above (ANOVA/ANCOVA, Kruskal–Wallis H and Quade ANCOVA tests), given the exploratory aim of our study, we applied the judicious Sidak correction to core indicators (ASSQ and CBCL total and subscale scores) and the conservative Bonferroni correction to the non-core indicator (FSIQ), balancing power and stringency according to indicator importance. Stepwise multiple linear regression analyses were performed to quantitatively explore the relationship between ASD symptoms and externalizing behaviors within the pure ADHD group. ASD symptoms were assessed using the ASSQ and CBCL‑ASD scale scores, and externalizing behaviors were assessed using the CBCL‑non‑ASD subscales. Predictors were retained only if the variance inflation factor (VIF) was less than 5. Statistical analysis was performed using SPSS 26.0 software. P < 0.05 was considered significant. Results The demographic characteristics of the three groups (HCs, ADHD, and ADHD + ASD) are presented in Table 1. Regarding sex distribution, the ADHD group had a significantly higher proportion of males than the HCs group and a significantly lower proportion than the ADHD + ASD group (all P < 0.017, χ² tests with Bonferroni-corrected α = 0.017). For FSIQ scores, the Kruskal–Wallis H test revealed a significant overall difference across the three groups (H = 26.09, df = 2, P < 0.001, ε² = 0.17). Post-hoc Dunn’s tests with Bonferroni correction showed that the HCs group had significantly higher FSIQ scores than both the ADHD group (median [IQR]: 103.00[99.25-110.75] vs. 91.00[86.00-102.00]; adjusted P < 0.001, r = 0.46) and the ADHD + ASD group (median [IQR]: 103.00[99.25-110.75] vs. 93.00[80.00-105.50]; adjusted P = 0.001, r = 0.40), whereas no significant difference was found between the ADHD group and the ADHD + ASD group (P > 0.05). In terms of ADHD clinical subtypes, both the ADHD and ADHD + ASD groups were predominantly composed of the combined (62.32% vs. 43.24%) and inattentive subtype (36.23% vs. 51.35%), with very few cases of the hyperactive-impulsive subtype (1.45% vs. 5.41%). The overall distribution of subtypes did not differ significantly between the two groups (P > 0.05). No significant difference was found in age distribution among the three groups. Table 1Comparisons of demographic characteristics of the three groupsHCs 1(N = 48)ADHD 2(N = 69)ADHD + ASD 3(N = 37)χ2/F/H aP aPairwisecomparison bAge in years(mean ± SD)9.46 ± 1.249.05 ± 1.738.67 ± 1.682.710.070-Male [n, (%)]35(72.92)58(84.06)37(100.00)11.660.0011 < 2 < 3FSIQ[Median (IQR)]103.00(99.25-110.75)91.00(86.00-102.00)93.00(80.00-105.50)26.09< 0.0011 > 2 = 3ADHD subtypes [n, (%)] ADHD-IA-25(36.23)19(51.35)-0.097- ADHD-HI-1(1.45)2(5.41) ADHD-C-43(62.32)16(43.24)SD standard deviation. IQR Interquartile Range. FSIQ Full-Scale Intelligence Quotient. ADHD-IA ADHD with inattention; ADHD-HI ADHD with hyperactivity and impulsivity; ADHD-C a combination of botha P values were calculated using: one-way analysis of variance (ANOVA) for age; the Kruskal–Wallis H test for FSIQ; and the χ² test or Fisher’s exact test for categorical variables (male and ADHD subtypes) among the three groupsb Pairwise comparisons for FSIQ were performed using Dunn’s test with Bonferroni correction Comparisons of demographic characteristics of the three groups Age in years (mean ± SD) FSIQ [Median (IQR)] SD standard deviation. IQR Interquartile Range. FSIQ Full-Scale Intelligence Quotient. ADHD-IA ADHD with inattention; ADHD-HI ADHD with hyperactivity and impulsivity; ADHD-C a combination of both a P values were calculated using: one-way analysis of variance (ANOVA) for age; the Kruskal–Wallis H test for FSIQ; and the χ² test or Fisher’s exact test for categorical variables (male and ADHD subtypes) among the three groups b Pairwise comparisons for FSIQ were performed using Dunn’s test with Bonferroni correction ATs in ADHD The presence of ATs As shown in Table 2, comparisons of ASD symptom scores on the ASSQ and CBCL-ASD among the HCs, ADHD, and ADHD + ASD groups indicated that, except for the communication-problems subscale of ASSQ, the ADHD group scored significantly higher than the HCs on all other ASSQ and CBCL-ASD scores (all P < 0.05), but significantly lower than the ADHD + ASD group (all P < 0.05). Table 2Comparison of core ASD symptoms among the three or the four groupsHCs I①(N = 48)ADHD II(N = 69)ADHD + ASD III④(N = 37)F #aP #aηp2#aPairwisecomparison #bADHD-ATs ②(N = 40)ADHD + ATs ③(N = 29)F &aP &aηp2&aPairwiseComparison &bASSQ [Median (IQR)]Total score3.00(0.00–5.00)11.00(4.00-16.50)20.00(12.00–31.00)38.123< 0.0010.340Ⅰ < Ⅱ < Ⅲ4.00(2.00–8.00)17.00(15.00–23.00) *79.756< 0.0010.619① = ② < ③ = ④Social interaction1.00(0.00–2.00)3.00(1.00–7.00)9.00(6.00–11.00)50.625< 0.0010.406Ⅰ < Ⅱ < Ⅲ1.00(0.25–2.75)7.00(5.50–8.50) *86.730< 0.0010.639① = ② < ③ = ④Communication problems1.00(0.00–2.00)3.00(1.00–5.00)5.00(3.00–9.00)12.003< 0.0010.140Ⅰ = Ⅱ < Ⅲ1.00(0.00-2.75)5.00(3.00–6.00) *20.589< 0.0010.296① = ② < ③ = ④RRBs0.00(0.00–2.00)3.00(1.00–6.00)7.00(3.00-11.50)27.949< 0.0010.274Ⅰ < Ⅱ < Ⅲ1.00(0.00–3.00)6.00(5.00-8.50) *52.506< 0.0010.517① = ② < ③ = ④CBCL-ASD [Median (IQR)]Withdrawn0.00(0.00–1.00)2.00(1.00–4.00)4.00(2.00–6.00)26.907< 0.0010.267Ⅰ < Ⅱ < Ⅲ2.00(1.00–3.00)3.00(1.00–4.00)19.327< 0.0010.283① < ② = ③ = ④Social problems1.00(0.00–2.00)4.00(2.00–6.00)6.00(4.00–9.00)42.131< 0.0010.363Ⅰ < Ⅱ < Ⅲ3.00(1.00-4.75)6.00(4.00–7.00) *39.224< 0.0010.445① < ② < ③ = ④Thought problems0.00(0.00–0.00)1.00(0.00–2.00)2.00(0.00–3.00)16.427< 0.0010.182Ⅰ < Ⅱ < Ⅲ1.00(0.00-1.75)1.00(0.00–2.00)10.981< 0.0010.183① < ② = ③ = ④ASSQ: Autism Spectrum Screening Questionnaire. RRBs: restrictive and repetitive behaviors. CBCL: Child Behavior Checklist. CBCL-ASD: including scores of withdrawn, social problems, thought problems in CBCL profilea Rank-transformed analysis of covariance (ANCOVA) was performed adjusting for FSIQ, age, and male sex, among these groups. Partial eta squared (ηₚ²) represented overall effect sizeb Pairwise comparisons with Sidak test among groups# among HCs, ADHD and ADHD + ASD groups& among HCs, ADHD − ATs, ADHD + ATs and ADHD + ASD groups* Effect sizes (r) for the comparison between ADHD-ATs and ADHD + ATs (② vs. ③): For ASSQ total score, r = 0.79; for social interaction, r = 0.77; for Communication problems, r = 0.57; for RRBs, r = 0.72; for social problems, r = 0.49. All comparisons were significant at P < 0.001 Comparison of core ASD symptoms among the three or the four groups ASSQ: Autism Spectrum Screening Questionnaire. RRBs: restrictive and repetitive behaviors. CBCL: Child Behavior Checklist. CBCL-ASD: including scores of withdrawn, social problems, thought problems in CBCL profile a Rank-transformed analysis of covariance (ANCOVA) was performed adjusting for FSIQ, age, and male sex, among these groups. Partial eta squared (ηₚ²) represented overall effect size b Pairwise comparisons with Sidak test among groups # among HCs, ADHD and ADHD + ASD groups & among HCs, ADHD − ATs, ADHD + ATs and ADHD + ASD groups * Effect sizes (r) for the comparison between ADHD-ATs and ADHD + ATs (② vs. ③): For ASSQ total score, r = 0.79; for social interaction, r = 0.77; for Communication problems, r = 0.57; for RRBs, r = 0.72; for social problems, r = 0.49. All comparisons were significant at P < 0.001 The structure of ATs The comparisons of ASD symptom scores on the ASSQ and CBCL-ASD across the HC, ADHD-ATs, ADHD + ATs, and ADHD + ASD groups revealed an interesting finding that after stratifying ADHD participants by ATs status, the ADHD + ATs subgroup exhibited significantly higher ASSQ total and all subscale scores (including the communication problems subscale) than both the HC group and ADHD-ATs group. Moreover, these scores reached severity levels comparable to the ADHD + ASD group. No significant differences were observed between the ADHD-ATs group and the HC group on any ASSQ measures (Table 2). A similar between-group difference was also observed on the social problems subscale of the CBCL-ASD when comparing the ADHD + ATs and ADHD-ATs groups. In contrast, comparisons on the withdrawn or thought problems subscales revealed no significant differences between these two groups (all P > 0.05, Table 2). Non-core ASD behaviors in ADHD + ATs subgroup Categorical comparisons As shown in Table 3, after controlling for FSIQ, age, and male sex, the ADHD + ATs group scored significantly higher than both the ADHD-ATs and HC groups on most CBCL-non-ASD subscales (all P < 0.05), with the exception of Anxious/Depressed symptoms. Furthermore, compared to the ADHD + ASD group, the ADHD + ATs group exhibited significantly higher scores on Rule‑breaking behavior (post‑hoc r = 0.27, P = 0.007) and Aggressive Behavior syndrome (post‑hoc Cohen’s d = 0.47, P = 0.012). Detailed pairwise comparisons and overall effect sizes (ηₚ²) are presented in Table 3. Table 3Comparison of the CBCL-non-ASD behavior symptoms among the four groupsHCs①(N = 48)ADHD-ATs ②(N = 40)ADHD + ATs ③(N = 29)ADHD + ASD④(N = 37)F aP aηp2aPairwisecomparison bCBCL-non-ASDSomatic complaints0.00(0.00–0.00)0.00(0.00–2.00)1.00(0.00-2.50) *0.00(0.00–1.00)6.559< 0.0010.118① < ② < ③ = ④Anxious/ depressed0.00(0.00–1.00)3.00(1.00-5.75)4.00(1.50-8.00)4.00(2.00–6.00)21.298< 0.0010.303① < ② = ③ = ④Attention problems1.00(0.00–2.00)8.00(5.00–10.00)10.00(8.50–12.00) *9.00(7.00–12.00)48.142< 0.0010.496① < ② < ③ = ④Rule-breaking behavior1.00(0.00–2.00)2.00(1.00-4.75)4.00(3.00–6.00) *3.00(1.50-4.00)18.967< 0.0010.279① < ② = ④ < ③Aggressive behaviorsyndrome2.75 ± 2.7310.43 ± 5.6913.97 ± 7.53*10.84 ± 5.9928.378< 0.0010.367① < ② = ④ < ③ASSQ: Autism Spectrum Screening Questionnaire. RRBs: restrictive and repetitive behaviors. CBCL: Child Behavior Checklist. CBCL-non-ASD: including scores of Somatic complaints, Anxious/depressed, Attention problems, Rule-breaking behavior and Aggressive behavior syndrome in CBCL profilea Rank‑transformed analysis of covariance (Quade‑ANCOVA) or conventional parametric ANCOVA; presented as median (IQR) or as mean ± SD; all controlling for FSIQ, age and male sex; Partial eta‑squared (ηₚ²) for overall effect sizeb Pairwise comparison with Sidak test among groups* Effect sizes for the comparison between ADHD-ATs and ADHD + ATs (② vs. ③): For somatic complaints, r = 0.25; for attention problems, r = 0.35; Effect sizes for the comparison between ADHD + ATs and ADHD + ASD (③vs④): for rule-breaking behavior, r = 0.27; for Aggressive behavior syndrome, Cohen’s d = 0.47; All comparisons were significant at P < 0.05 Comparison of the CBCL-non-ASD behavior symptoms among the four groups Aggressive behavior syndrome ASSQ: Autism Spectrum Screening Questionnaire. RRBs: restrictive and repetitive behaviors. CBCL: Child Behavior Checklist. CBCL-non-ASD: including scores of Somatic complaints, Anxious/depressed, Attention problems, Rule-breaking behavior and Aggressive behavior syndrome in CBCL profile a Rank‑transformed analysis of covariance (Quade‑ANCOVA) or conventional parametric ANCOVA; presented as median (IQR) or as mean ± SD; all controlling for FSIQ, age and male sex; Partial eta‑squared (ηₚ²) for overall effect size b Pairwise comparison with Sidak test among groups * Effect sizes for the comparison between ADHD-ATs and ADHD + ATs (② vs. ③): For somatic complaints, r = 0.25; for attention problems, r = 0.35; Effect sizes for the comparison between ADHD + ATs and ADHD + ASD (③vs④): for rule-breaking behavior, r = 0.27; for Aggressive behavior syndrome, Cohen’s d = 0.47; All comparisons were significant at P < 0.05 Quantitative relationship between ASD symptoms and CBCL-non-ASD behaviors Given the qualitative finding that children with ATs in the pure ADHD group showed more externalizing behaviors than those without ATs and those with comorbid ASD, we further conducted stepwise multiple linear regression analyses to quantitatively explore the relationship between ASD symptoms and externalizing behaviors within the pure ADHD population (n = 69). ASD symptoms were assessed using the ASSQ and CBCL-ASD scale scores, and externalizing behaviors were assessed using the CBCL-non-ASD. We verified no severe multicollinearity existed, with all variance inflation factors below 5. All models were adjusted for age, sex, and FSIQ. The results indicated that: Aggressive behavior: The overall regression model was significant (F = 13.590, P < 0.001, adjusted R² = 0.357). Aggressive behavior was predicted by social interaction (Std. β = 0.293, P = 0.004), withdrawn (Std. β = 0.324, P = 0.005) and thought problems scores (Std. β = 0.277, P = 0.015; Table 4). Table 4Independent variables of ASD core symptom scores with CBCL-measured non-ASD behavioral symptom scores in children with pure ADHD group (n = 69)Dependent variableIndependent variableStandardizedcoefficientt-valueP1r2 /r2adj P Aggressive behaviorSocial interaction0.2932.9910.0040.385/0.357< 0.001Withdrawn0.3242.9350.005Thought problems0.2772.5010.015Rule-breaking behaviorCommunication problems0.2322.2720.0260.348/0.318< 0.001Withdrawn0.2382.0900.041Thought problems0.3452.9960.004Attention problemsWithdrawn0.3053.474< 0.0010.661/0.640< 0.001Social problems0.3904.711< 0.001Thought problems0.2723.2330.002FSIQ-0.224-3.0540.003Somatic complainSocial interaction0.2342.0270.0470.130/0.1040.010Thought problems0.2492.1520.035ASSQ: Autism Spectrum Screening Questionnaire. RRBs: restrictive and repetitive behaviors. CBCL-ASD: including scores of withdrawn, social problems, thought problems in CBCL profile. r2: determination coefficient; r2adj: adjusted determination coefficient, P: overall significance of multiple stepwise linear regression, P1: significance of standardized partial regression coefficient Independent variables of ASD core symptom scores with CBCL-measured non-ASD behavioral symptom scores in children with pure ADHD group (n = 69) ASSQ: Autism Spectrum Screening Questionnaire. RRBs: restrictive and repetitive behaviors. CBCL-ASD: including scores of withdrawn, social problems, thought problems in CBCL profile. r2: determination coefficient; r2adj: adjusted determination coefficient, P: overall significance of multiple stepwise linear regression, P1: significance of standardized partial regression coefficient Rule-breaking behavior: The overall regression model was significant (F = 11.567, P < 0.001, adjusted R² = 0.318). Rule-breaking behavior was predicted by thought problems (Std. β = 0.345, P = 0.004), withdrawn (Std. β = 0.238, P = 0.041) and communication problems scores (Std. β = 0.232, P = 0.026; Table 4). Attention Problems: The overall regression model was significant (F = 31.256, P < 0.001, adjusted R² = 0.640). Attention problems were predicted by all CBCL-ASD scales (Std. β = 0.305, P < 0.001; Std. β = 0.390, P < 0.001 and Std. β = 0.272, P = 0.002, for withdrawn, social problems and thought problems scores respectively; Table 4). Somatic complaints: The overall regression model was significant (F = 4.939, P = 0.010, adjusted R² = 0.104). Somatic complaints were predicted by social interaction (Std. β = 0.234, P = 0.047) and thought problems scores (Std. β = 0.249, P = 0.035; Table 4). Discussion This cross-sectional study explored the symptom structure and behavioral correlates of the ATs subgroup among children with ADHD. The results suggest the presence of a clinically characterized subgroup with prominent ATs in children with ADHD and show its structural similarity to ASD. Notably, children with ADHD + ATs exhibited more severe overall non-ASD behavioral impairment. In particular, externalizing behaviors, such as rule-breaking and aggression, were even more severe than those in children with ADHD + ASD. Other difficulties, including attention problems and somatic complaints, were also more prevalent in the ADHD + ATs group than in the ADHD-ATs group. Further analyses also found quantitative associations between specific ASD symptom domains and distinct patterns of non-ASD behaviors. Structural alignment of ATs with ASD core domains In our study, the prevalence of ATs measured by the ASSQ was approximately 42%, which is consistent with the widely reported range of 15%-64.3% [15]. The clinical existence of this ATs subgroup was further supported by categorical group comparisons. Overall, except for the communication problems subscale, the ADHD group scored between the HCs and the ADHD + ASD group in all other ASSQ dimensions and CBCL-ASD measures. This result is partly consistent with the findings of Holtmann, Bölte, Poustka (2007) and Hoffmann, Weber, König (2016) [30, 31]. However, subgroup classification of children with ADHD based on ATs status revealed preliminary clinical characteristics. Specifically, the ADHD + ATs subgroup had elevated scores across all ASSQ symptom domains. Their scores were significantly higher than those of the ADHD-ATs group and HCs. Notably, their scores in social interaction, communication problems and RRBs reached a severity level similar to that of children with ADHD + ASD. This finding supports the symptom continuum between ADHD and ASD. It suggests that ADHD with ATs can produce clinical impacts comparable to a formal ASD diagnosis [15]. This point has long been insufficiently discussed, since most existing studies mainly focus on direct diagnostic comorbidity of the two disorders [6]. Therefore, the symptom composition of ATs is consistent with core clinical manifestations of ASD, which further supports the neurodevelopmental continuum theory. As an intermediate phenotype on this spectrum, the ATs subgroup may exert a specific relationship with the development and clinical performance of non-ASD symptoms within ADHD. Exploring these subgroup-specific correlates can help refine the clinical recognition of ADHD combined with ATs. Furthermore, defining ATs according to the standard ASSQ total score could effectively screen for this specific subgroup within the ADHD population. There were no significant differences in all ASSQ subscale scores between HCs and the ADHD-ATs group. This finding indirectly supports the reliability of the ASSQ for identifying ATs in children with ADHD, which is also verified by published evidence [32, 33]. Results derived from the CBCL-ASD scale further supported most of the above conclusions. The ADHD + ATs subgroup displayed more severe social problems than the ADHD-ATs subgroup, and their scores were comparable to those of the ADHD + ASD group. However, no significant differences were observed in withdrawn or thought problems between the ADHD groups with and without ATs. A possible reason is that the CBCL is a comprehensive behavioral assessment instrument. Although some researchers recommend the CBCL-ASD subscale as a screening tool for ASD [28, 30], its ability to specifically identify autistic symptoms may be lower than that of the ASSQ. The reasons why ATs are present in children with ADHD may be illustrated by neuroimaging. Although ASD and ADHD overall possess distinct brain network characteristics, both conditions share patterns of overconnectivity between the Default Mode Network (associated with internal thought and social information processing) and the Dorsal Attention Network (involved in guiding external attention). This shared abnormality in neural circuitry may underlie the similar functional impairments observed in both groups regarding social interaction difficulties, restricted interests, and challenges with attentional shifting [34]. Di Martino et al. [35] also demonstrated it. Besides, Liu’s research identified brain regions specifically associated with ATs: the right middle temporal gyrus [33]. Diffusion Tensor Imaging (DTI) studies also indicate that within ADHD patients, white matter alterations unique to ATs are primarily located in the right posterior limb of the internal capsule/corticospinal tract, the right cerebellar peduncle, and the midbrain [36]. Concurrently, ADHD severity is associated with white matter microstructure in the left subgenual cingulate gyrus [37]. Shared functional brain alterations in both ADHD and ASD, and the special ATs-related brain activity alterations in ADHD, suggest the presence of a clinically preliminary subgroup. Future clinical studies focused on neuroimaging mechanisms of ATs in children with ADHD may help to explain the clinical heterogeneity and complexity of ADHD. The presence of ATs exacerbates externalizing behaviors in children with ADHD This study found that children with ADHD + ATs presented the most severe externalizing behaviors, including aggressive and rule-breaking behaviors, which exceeded not only those in ADHD-ATs and HCs groups, but also those in the ADHD + ASD group. This result is consistent with previous studies by Grzadzinski et al. [13] and Mulligan et al. [38], and further supports that ATs are a significant risk predictor of externalizing behaviors among children with ADHD. Further stepwise multiple linear regression analyses showed that aggressive behavior was predicted by scores of social interaction, withdrawn, and thought problems. Meanwhile, rule breaking behavior was mainly predicted by scores of communication problems, withdrawn, and thought problems. In this study, the use of the withdrawn and thought problems subscales of the CBCL to identify ASD-related traits is grounded in empirical evidence [16]. Although the CBCL is not a dedicated diagnostic scale for ASD, several of its subscales have been consistently linked to core ASD symptoms [39, 40]. Specifically, the withdrawn subscale, which assesses social avoidance and lack of interaction, can well reflect social communication deficits typical of ASD [40]. Furthermore, the thought problems subscale is capable of capturing rigid thinking, stereotyped behaviors and atypical sensory abnormalities, all of which belong to the restricted/repetitive behaviors and interests (RRBs) domain of ASD [41–43]. Therefore, combined with the above theoretical basis and our regression results, future research should further clarify how the two core ASD symptom dimensions, namely social communication deficits and RRBs, specifically correlate with externalizing manifestations such as aggressive and rule-breaking behaviors among children with ADHD and ATs. Our regression analysis indicated that aggressive behavior was predicted by subscales reflecting social deficits (social interaction, withdrawn) and RRBs (thought problems). This suggests that within this population, social frustration and rigid thinking patterns may interact to exacerbate emotional impulsivity and aggressive reactions [44]. Rule-breaking behavior was primarily associated with dimensions reflecting RRBs (thought problems) and specific communication difficulties, implying that such behaviors may arise from rigid comprehension of social norms, disrupted repetitive routines, or confrontational reactions caused by poor social communication [45]. Collectively, these findings support the hypothesis that the externalizing behaviors among children with ADHD + ATs are not simply a result of inattention or impulsivity but may be partly driven by the interactive effects of underlying core ASD-like symptoms [46]. Building on the hypothesis that externalizing behaviors among children with ADHD + ATs may arise from interactions of ASD-like symptoms, a critical and seemingly counterintuitive question arises: why are such behavioral symptoms more severe than those seen in children diagnosed with both ADHD and ASD? Two plausible explanations are proposed as follows. First, social motivation is different between children with ADHD + ATs and those with ADHD + ASD. Children with ADHD + ATs typically retain adequate social motivation [47] but display heightened sensitivity to frustration and others’ comments. In contrast, children with ADHD + ASD have diminished social motivation, tend to actively avoid social contact, and are less responsive to social and environmental cues [12, 47]. Second, children with ADHD + ATs frequently present with emotional dysregulation and severe impulsive symptoms [47]. The combination of ADHD-related impulsivity, emotional lability, and social cognitive deficits linked to ATs greatly increases their risk of aggressive, destructive and other externalizing behaviors [47]. Therefore, for children with ADHD who exhibit significant externalizing behaviors such as rule-breaking and aggression, clinicians should conduct a comprehensive assessment for potential ATs. Beyond conventional treatment for ADHD, interventions targeting social interaction and communication may provide clinical benefit for children with this ATs subgroup within ADHD. The presence of ATs exacerbates attention problems and somatic complaints in children with ADHD, compared to those without ATs This study found that children with ADHD + ATs had more severe attention problems than those without ATs, and such differences could not be explained by different ADHD clinical subtypes. Further quantitative analysis revealed that attention problems were independently predicted by the overall CBCL-ASD scales. Multiple studies have confirmed that social problems in children with ADHD are linked to deficits in executive functions, particularly deficits in working memory and inhibitory control, which are core cognitive features of poor attentional regulation [48–50]. Notably, children with ASD also present similar executive function impairments, including deficits in working memory, processing speed, and response inhibition [51]. Furthermore, children exhibiting withdrawn social motivation frequently demonstrate diminished attentional control. Concurrently, the Thought Problems subscale of the CBCL-ASD scales mainly reflects the RRBs. It may hinder flexible attention shifting in tasks. Such excessive focus on specific stimuli easily leads to task-irrelevant inattention when individuals need to switch their attention flexibly [52]. Thus, the exacerbation of attention problems among children with ADHD + ATs may be explained by executive dysfunctions shared between ADHD and ASD, such as impaired working memory and inhibitory ability, together with ASD-specific impairments in cognitive flexibility, and poor social competence. Future research focusing on executive function in this ATs subgroup will help clarify its underlying cognitive mechanisms. This study also found that ATs exacerbate somatic symptoms, one dimension of internalizing behaviors, in children with ADHD more severely than in those without ATs. Quantitative analysis revealed that somatic symptoms were independently predicted by social interaction and thought problems. This finding is consistent with previous longitudinal observations. Joshi et al. (2020) [16] demonstrated that the children with ADHD + ATs experienced greater long-term mental health burdens and more severe somatic complaints than those with pure ADHD. Theoretically, prominent ATs are recognized risk factors for internalizing problems [53]. Mechanistically, social difficulties may hinder emotional expression and regulation, while rigid thinking heightens anxiety levels and physical discomfort sensitivity. These factors may transform psychological distress into actual somatic symptoms [16, 54]. Consequently, clinical management of ADHD with ATs should include systematic evaluation of somatic complaints, alongside targeted interventions addressing their core social-cognitive features. The ATs subgroup did not exacerbate levels of depression and anxiety No statistically significant difference in anxiety and depression levels was found based on the presence of ATs. This may indicate that anxiety and depression are not core features of the ADHD + ATs phenotype. Alternatively, the absence of group differences could be due to measurement limitations. Our study relied on parent‑reported CBCL scales, which may not fully capture the internalizing distress of children with ATs, as these children may exhibit less overt emotional expression [53, 55, 56]. Given that our study did not directly assess children’s self‑reported emotional states, it remains unclear whether the lack of group differences represents a true phenotypic characteristic or a measurement artifact. Future studies should incorporate self‑report anxiety and depression scales to better capture the subjective internalizing experience of children with ADHD + ATs. This study has some limitations. First, this study adopted a cross-sectional design, which cannot support causal inference and only allows exploration of correlational relationships. Second, the relatively limited sample size may affect the stability and generalizability of the results. In addition, all participants were recruited from a single clinical center, which further restricts the external validity of our results. Multi-center and longitudinal follow-up investigations are needed to verify and extend our conclusions. Third, currently, ATs are often defined based on positive results on screening scales, and no superior method exists for defining ATs. Future research is warranted to more precisely define ATs. Fourth, this study primarily relied on screening scales to examine relationships among behavioral measures, without including more precise and mechanistic indicators such as cognitive tests or neuroimaging biomarkers. Future studies incorporating these measures are needed to further refine and extend our preliminary findings. Additionally, all behavioral assessments depended purely on parental reports, which may introduce inevitable informant bias. Future research should incorporate multiple informants (e.g., teachers, clinicians) and collect more comprehensive demographic information to better control for residual confounding. Fifth, although age, sex and FSIQ were adjusted in statistical analyses, several potential confounders were not controlled for in this study, such as socioeconomic status and detailed family environment factors. Sixth, our stepwise regression analyses may be prone to overfitting and model instability, particularly given the modest sample size (n = 69). Therefore, the findings should be interpreted as hypothesis‑generating, and future studies should employ more robust methods (e.g., LASSO, hierarchical regression, or independent validation samples) to confirm them. Despite these limitations, this study still possesses distinct analytical perspectives and practical clinical implications. Compared with previous studies, the innovation of this study is that it not only subdivided children with ADHD into those with and without ATs, but also included children with comorbid ADHD + ASD as a reference group. This approach enables systematic exploration of the gradual shifts in social symptoms and behavioral characteristics along the clinical spectrum ranging from pure ADHD to ADHD + ATs to ADHD + ASD. Conclusions Our study suggested the presence of a potential ATs subgroup among children with ADHD. This subgroup is characterized by widespread deficits in social communication and interaction, as well as obviously elevated non-ASD externalizing behaviors, particularly aggressive behavior and rule-breaking problems. Furthermore, we found that core ASD symptoms predict non-ASD externalizing behavioral symptoms in children with ADHD, which offers preliminary evidence for relevant underlying mechanisms. Clinically, our results emphasize the importance of systematic screening for ATs in children with ADHD. Recognizing this subgroup is essential for targeted evaluation and individualized intervention, and further research highlights the urgent need to develop targeted treatment strategies. Acknowledgements The authors are grateful to all participants of the study. Abbreviations Attention Deficit/Hyperactivity Disorder Autistic traits Autism spectrum disorder ATs Children with ADHD and ATs ATs Children with ADHD but without ATs ASD Children with ADHD and ASD Healthy controls Disability-adjusted life years Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition Autism Spectrum Screening Questionnaire Child Behavior Checklist Full-Scale Intelligence Quotient Kiddie-Schedule for Affective Disorders and Schizophrenia Present and Lifetime Version Childhood Autism Rating Scale Autism Diagnostic Observation Schedule Analysis of variance Analysis of covariance Standard deviation Least significant difference Variance Inflation Factor Restricted and repetitive behaviors Diffusion Tensor Imaging Working Memory Author contributions YWF, JL and BRY conceived of and designed the study. JC, LZ, GTZ, XSQ, YL, YW, LWC, CD, JZ, LLZ, XLC, YL, QS, LL and TF organized the database. JL and YWF designed the statistical analysis. YWF and JL performed the statistical analysis. YWF, JL and BRY interpreted the results. YWF finished the draft. All authors contributed to manuscript revision and read and approved the submitted version. Funding This work was supported by Shenzhen Science and Technology Program (JCYJ20220530155416036, JCYJ20220818102800001), the Guangdong Medical Science and Technology Research Fund Project (A2025078). Data availability The datasets presented in this article are not readily available due to regulations governing ethical approval. Declarations Ethics approval and consent to participate Approval was obtained from the Medical Ethics Committee of SCH. The procedures used in this study adhered to the tenets of the Declaration of Helsinki. Informed consent was obtained from all participants and their parents. Consent for publication We obtained consent for publication from the children or their parent or legal guardian. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The datasets presented in this article are not readily available due to regulations governing ethical approval.