Participants

Participants included individuals who received diagnostic neurodevelopmental evaluations at an autism specialty clinic, referred by neontologists, pediatricians, geneticists, pediatric neuropsychologists, psychiatrists, as well as other professionals who work closely with the families (i.e., speech pathologists, occupational therapists). Clinical diagnoses were made by licensed psychologists with extensive experience in the assessment and treatment of ASD and common co-occurring conditions (5–15 years of clinical experience). Licensed psychologists integrated multiple sources of information, including medical history, educational records, caregiver reports, neuropsychological assessments, and comprehensive interviews regarding autistic characteristics to make diagnostic decisions based on DSM-5 criteria. The SCQ was collected as part of the diagnostic evaluation but was not the sole factor considered when making a diagnosis. The methods were approved by the Institutional Review Board at the University of Minnesota.

From 2016 to 2019, data from 793 consecutive evaluations were entered into a de-identified clinical database. To examine the current study hypotheses, only initial evaluation data (N = 519) were included in the analysis. Data of individuals who were younger than 4 years old at their initial evaluations (N = 142) were excluded from analyses. Individuals who did not complete or only partially completed the SCQ (N = 162) were also excluded. Finally, only participants who had completed all the measures used in the present study (described below) were included. Ultimately, initial diagnostic evaluations from 187 children and adolescents were included in the present study.

Among all individuals, 133 (94 males and 39 females) were clinically diagnosed with ASD, and 54 (34 males and 20 females) were identified as non-spectrum with other neurodevelopmental challenges. Within the non-spectrum sample, 32 received a primary diagnosis of ADHD (59.3%), 10 were diagnosed with a language disorder (18.5%), 5 had behavioral disorders (9.3%; e.g., oppositional defiant disorder, conduct disorder, disruptive behavioral disorder), 27 had anxiety disorders (50.0%), 7 had mood disorders (13.0%), and 8 had other genetic and/or physical disabilities (14.8%; e.g., fragile X syndrome, Williams syndrome, or mild cerebral palsy).

Measures

Demographic characteristics were collected through a clinical intake questionnaire and clinical interviews with the primary caregivers. All participants were given the Autism Diagnostic Observation Schedule, Second Edition (ADOS-2), and caregivers were administered the Autism Diagnostic Interview-Revised (ADI-R). Cognitive and language skills were assessed using clinically appropriate measures depending on the child's age and function level. Scores on individual sub-tests are standardized against age-specific norms and then grouped to produce separate measures of verbal and non-verbal IQ, with the former encompassing those tests most related to verbal skills and the latter being more independent of verbal skills. Full Scale IQ (FSIQ) is the composite of these verbal and non-verbal skills. All individuals also completed the following measures.

Social Communication Questionnaire, Lifetime Version (SCQ)

The SCQ [1] is a 40-item questionnaire that measures the symptomatology associated with ASD (e.g., certain communitive behaviors, language uses, and stereotyped behaviors) focusing on the behaviors that are rare in non-affected individuals, based on an established diagnostic interview, the Autism Diagnostic Interview-Revised (ADI-R) [30]. The Lifetime SCQ version asks respondents to focus on characteristics of the individual at age 4 to 5 years for developmentally influenced behaviors, or at any point in their lifetime for behaviors that are atypical at any age (e.g., repetitive motor movements), while the Current SCQ version focuses on characteristics present within the previous 3 months [1]. Total scores range from 0 to 39, with higher scores reflecting the presence of more symptoms.

Behavioral Assessment System for Children, Parent Rating Scale (BASC-PRS)

The BASC-PRS [31] is a parent-report questionnaire using a multi-dimensional approach to evaluate behaviors and adaptive skills for children ages 2 years, 6 months to 21 years. This sample completed the BASC-PRS, Third Edition, which generates four composite scales (Externalizing Problems, Internalizing Problems, Behavioral Symptoms Index, and Adaptive Skills) and 14 Primary Scales (Hyperactivity, Aggression, Conduct Problems, Anxiety, Depression, Somatization, Atypicality, Withdrawal, Attention Problems, Adaptability, Social Skills, Leadership, Activities of Daily Living, and Functional Communication). For behavioral scales, higher T scores indicate greater difficulties, whereas for adaptive scales, lower T scores represent greater challenges [31].

Vineland Adaptive Behavior Scale (VABS)

The VABS [32, 33] is a standardized clinical assessment tool that utilizes a semi-structured interview to measure adaptive behaviors and skills for individuals with developmental challenges. For this sample, the third edition of the VABS, Comprehensive Interview Form, was used to assess participants’ adaptive skills. The VABS consists of three subscales, including Communication (receptive, expressive, written), Socialization (interpersonal relationships, play and leisure, coping skills), and Daily Living skills (person, domestic, community), yielding an overall composite score of adaptive skills (Adaptive Behavior Composite). The VABS also provides an indirect measure of gross and fine motor skills, yielding a Motor skills domain [32, 33].

Autism Diagnostic Observation Schedule, Second Edition (ADOS-2)

The ADOS-2 [34] is a standardized, semi-structured observational assessment used to assess language and communication, reciprocal social interaction, imagination/creativity, as well as stereotyped behaviors and restricted interests to inform diagnosis of ASD. The ADOS-2 is organized into five modules based on the individual’s expressive language level (and, in some cases, chronological age), ranging from preverbal to verbally fluent. The diagnostic algorithm provides separate total scores for the Social Affect (SA) and Restricted and Repetitive Behavior (RRB) domains, as well as a cut-off for the sum of the two domains to provide instrument classifications of autism, autism spectrum, or non-spectrum. The Calibrated Severity Score (CSS) is a standardized version of ADOS-2 raw total scores aimed to minimize the impact of factors such as age, language, and cognitive ability. The ADOS-2 CSS has been suggested as a measure of symptom severity independent of these developmental factors [35].

Analytic approach

Data collected through the comprehensive evaluations were entered, coded, and checked for errors and logic using a standardized procedure. Subsequently, the entered data were transferred to SAS 9.4 (SAS Institute Inc., Cary, NC, USA) to perform range checking and internal consistency checking. Based on the clinical diagnosis of ASD and suggested cut-off score of the SCQ (summed score of 15 or above) [1], individuals were separated into four groups to test our hypotheses: true positives (TP; have a clinical diagnosis of ASD and the SCQ score is above 15; N = 82), false negatives (FN; have a clinical diagnosis of ASD but the SCQ score is below 15; N = 51), false positives (FP; does not have a clinical diagnosis of ASD but the SCQ score is above 15; N = 22), and true negatives (TN; does not have a clinical diagnosis of ASD and the SCQ score is below 15; N = 32). Demographic homogeneity of the groups was assessed using chi-square tests and post-hoc analyses (Fisher’s exact test) for discrete variables. Continuous variables, such as sum or mean score of cognitive measures, symptomatologic items, as well as emotional, behavioral, and adaptive measures, were compared between groups by using independent t tests and post-hoc analyses (Bonferroni correction). To determine the performance of the SCQ as a screener for ASD in this clinical-referred sample, the sensitivity (i.e., number of true positive/[true positive + false negative]) and specificity (i.e., number of true negative/[true negative + false positive]) were calculated at multiple cut-off points. In addition, the area under the curve (AUC; the area under the receiver operating characteristic (ROC) curve) is considered to be a metric of fit between the true positive rate (sensitivity) and the false positive rate (1—specificity). AUC values can range from 0.5 to 1.0, with estimates closer to 1 indicating greater accuracy. Hanley and McNeil [36] suggested that a screener is a failure when an AUC is between 0.5 and 0.6, poor between 0.6 and 0.7, fair between 0.7 and 0.8, acceptable between 0.8 and 0.9, and perfect between 0.9 and 1.0.