Prevalence protocol study design

The present ASD prevalence study, conducted between February 24th, 2016, and February 23rd, 2018, was promoted by the Italian Ministry of Health in collaboration with the MoE and coordinated by the National Observatory for ASD. The study utilized an adapted version of the ASDEU project protocol. The protocol for the study was approved by the Ethic Committee of the ISS on March 15th, 2016.

Geographical areas selection and rural/urban schools’ invitation

Three geographical areas were selected according to the following criteria established by the ASDEU project: a. well-defined and delimited geographical and administrative area (s); b. stable population; c. compulsory education system at the ages of the study subjects; d. existence of a Public Health Care System covering nearly 100% of the population; e. accessibility of data from educational and special educational sources; f. no potential selection bias due to the existence of reference services of ASD diagnosis, treatment, or special education facilities, which are located outside the area but close enough for children living within the area to access—this could result in missing children within the study area.; g. accessibility to cases’ clinical records; h. data accessibility from clinics and institutional private services; i. rural and urban residence of schools should be considered. Finally, ASD representative organizations and other regional stakeholders must be involved in the study. By adhering to these criteria, the study aimed to ensure a comprehensive and representative assessment of ASD prevalence within the chosen geographical areas.

The selected three geographical areas were: the city of Lecco and the area of Monza-Brianza (North area), the city of Rome and its province (Center area), and the city of Palermo and its province (South area). Each of these areas had a national clinical referral center specializing in the diagnosis of ASD in children (Lecco-Monza-Brianza: IRCCS E. Medea, La Nostra Famiglia, Bosisio Parini; Rome and its province: ASL Roma 1- 'La Scarpetta'; Palermo and its province: ASP Palermo). In adherence to the ASDEU protocol, both rural and urban schools were invited within each of the three areas. This ensured a representation of different educational settings and demographics within the study population.

PopulationSample size calculation

The sample size was determined by considering two key factors: (1) estimated target population: children between the ages of 7 and 9 years who were residents of the areas during the study period (children born between January 1st, 2007, and December 31st, 2009); (2) expected number of ASD cases based on existing data or previous studies on ASD prevalence rates. By combining information on the estimated target population and the expected number of ASD cases, the researchers were able to determine an appropriate sample size that would provide statistically reliable results for estimating the prevalence rate of ASD in the areas and age group.

According to the National Institute of Statistics (ISTAT) [20] data, in the northern area, the estimated target population of 7–9-year-olds as of January 1st, 2017, was approximately 26,066 children in Monza Brianza and 9971 children in Lecco. This totals to an estimated population of 36,037 children in the northern area. In the center area, which includes Rome and its province, the estimated target population of 7–9-year-olds was 124,346 children. In the southern area, which includes Palermo and its province, the estimated target population of 7–9-year-olds was 37,632 children.

At the time of the study, the expected number of ASD cases was approximately 1% based on previous research [13]. The precision of this estimation was around 0.2 with a 95% confidence level. To obtain reasonably accurate estimations, the ASDEU protocol recommended selecting a sample size of 8,000–10,000 children within the specified age range. Considering an estimated dropout rate of 20%, a sample size of 12,500 children per area was calculated. This size would help ensure that enough data were collected to accurately estimate the prevalence of ASD in the three areas.

Procedure

The present study was divided into two phases: phase (1) identification of children already certified with ASD from the MoE. These records provided information on children who had previously received a formal diagnosis of ASD and were already recognized by the educational system as having the condition; phase (2) screening and identification, by clinical assessment, of ASD cases not registered with the MoE. Parents of children in the target age range (7–9 years old) attending mainstream schools have been invited to fill in the SCQ-L. Children who scored equal or above the cut-off point of 15 (≥ 15) on the SCQ-L were invited to undergo an expert clinical consultation for a comprehensive assessment of ASD symptoms, as well as cognitive and daily life skills. This phase aimed to identify children with ASD who may not have been previously diagnosed or certified.

Phase 1: Certified ASD diagnoses obtained from the Ministry of Education's records

In Italy, students with ASD or other disabilities requiring special educational support are assisted in their learning and daily living skills, as well as in developing relationships with their typically developing peers, by support teachers and/or educators, in accordance with Italian Law 104/1992 (Gazzetta Ufficiale della Repubblica Italiana) [18]. The MoE maintains the unique regional register of children with disabilities, including those with ASD per se or with other comorbidities, requiring health and educational support. To be included in the registry, parents of ASD children present at the MoE the clinical certification of diagnosis produced by a child psychiatry unit of the Nation health system following a multidisciplinary evaluation with standardized tools. Unfortunately, diagnoses are included in the MoE registry either with the ICD-9, ICD-10, DSM-IV-TR and DSM 5 diagnostic codes.

In the present study, access to the number of ASD diagnoses in the MoE's Office for Disability registry was made possible through an agreement with the ISS. A designated contact person was assigned by the MoE’s regional/local offices for Disability in each of the three areas. This contact person assisted the ISS researchers in establishing initial contact with the school deans. The MoE’s regional/local officers for Disability provided to ISS records of children born between January 1st, 2007 and December 31st, 2009 (aged 7–9 years), who were attending primary mainstream schools in the three areas and had received a diagnosis and certification of disability.

Subsequently, the ISS's clinical team reviewed all provided records to identify children diagnosed with ASD: (1) Diagnostic criteria from the DSM-5 [1]; (2) International Classification of Diseases [41] codes 299.0, 299.80, and ICD-9 nosographic labels F84.0, F84.5, F84.8, F84.9; (3) DSM-IV-TR [2]. In cases where conflicting information was found in the records, the local Child Psychiatry Unit responsible for the diagnosis and disability certification was contacted for clarification and resolution.

Phase 2: Screening of the general population

The second phase of the present study involved the screening for ASD among the general population of 7–9-year-old children born between January 1st, 2007, and December 31st, 2009, in invited schools. Children who scored equal or above the cut-off 15 on the SCQ-L were further assessed for ASD diagnosis.

The ISS and the MoE compiled a list of schools in the three selected areas. As per the ASDEU project requirements, local and national stakeholders were contacted to promote the study. The MoE regional/local officers sent official letters to the deans of the schools, inviting their participation, and explaining the study's aims, as well as its public health and scientific significance.

During school visits, the study protocol was presented by the ISS researchers and MoE regional/local officers in the presence of the school's dean, teachers, and parents. Parents of the 7- to 9-year-old children attending the invited schools received an envelope containing a letter explaining the purpose of the study, an informed consent form, and the SCQ-L.

The SCQ-L, a validated tool for ASD screening, consists of 40 yes/no response items completed by parents or primary carer. It assesses language abilities, presence of siblings with ASD, social-relational development at 4–5 years, and atypical behaviors at the time of assessment. A cut-off score of 15 was selected, as it is commonly considered suggestive of ASD presence [8, 33].

Two weeks after the delivery of the envelopes, the ISS team collected the sealed envelopes containing the completed SCQ-L questionnaires and signed informed consent forms. The signed informed consent was required for the analysis of the SCQ-L data.

Clinical assessment procedure

In cases where the child’s SCQ-L score was equal to or greater than 15, the ISS team invited the child and their parent to the selected reference center for ASD. The clinical team consisted of a child psychiatrist from each area's center and a psychologist from the ISS. The assessment of the children was conducted at the clinical center using the Autism Diagnostic Interview-Revised (ADI-R) [34, 35], Autism Diagnostic Observation Schedule-2 (ADOS-2) [23], and clinical judgment, following the criteria outlined in the DSM-IV-TR and DSM-5.

The ADI-R is a comprehensive 93-item interview that covers the individual's developmental history and focuses on three functional domains: language and communication, reciprocal social interactions, and restricted, repetitive, and stereotyped behaviors and interests. ADOS-2 is a semi-structured and standardized assessment with four modules based on the individual's level of expressive language ability. It includes play-based activities to gather information on communication, reciprocal social interactions, and restricted and repetitive behaviors associated with an ASD diagnosis. The ADOS-2 allows for comparison across modules through an algorithmic scoring system, ensuring consistent scores regardless of the module used. For this study, module three, designed for children with fluent language skills, was administered, while module two was used for children with limited language abilities. Algorithmic scores were calculated for all participants.

The intellectual abilities and cognitive functions of verbal children were assessed using the Wechsler Intelligence Scale for Children-IV (WISC-IV) [30, 39], which evaluates verbal comprehension, perceptual reasoning, working memory, and processing speed. Nonverbal children's cognitive abilities were assessed using the Leiter international performance scale-revised (Leiter-R) [33], which targets nonverbal intelligence in fluid reasoning, visualization, visuospatial memory, and attention. The brief form of the Leiter-R, including subdomains such as Figure Ground, Form Completion, Sequential Order, and Repetitive Pattern, was administered to obtain a standardized nonverbal Brief Intelligence Quotient (IQ). Each subtest and brief IQ score demonstrated excellent validity and reliability [32].

Furthermore, parents were interviewed using the Vineland Adaptive Behavioral Scales (VABS) [5, 37], a standardized parent interview consisting of 297 items that assess daily skills in four domains of functioning: Communication Skills, Daily Living Skills, Social Skills, and Motor Skills. Equivalent ages based on published Italian norms were derived from the adaptive behavior composite score for each domain.

Upon receiving an ASD diagnosis or any other diagnosis, the clinical team developed a specific and personalized recommendations/interventions plan for the child and their family.

To minimize the risk of false negatives, the ISS team randomly selected the 20% of children who scored between 11 and 14 on the SCQ-L and invited their primary carer to participate in a semi-structured interview over the phone using the ADI-R, following the approach described by Carpenter and colleagues [7]. It has been suggested in previous studies to lower the conventional SCQ-L cut-off point of 15 when screening the general school population [27]. Some studies propose using a lower cut-off point of 11 [3, 17, 29, 40] or 12 [11] to improve sensitivity and reduce the likelihood of false negatives.

Statistical analysis

The prevalence of ASD (reported as number of ASD cases per 1000 children) and its 95% confidence interval (CI 95%) were estimated in the present study, both overall and stratified by geographical areas (North, Center, South), SCQ-L score classes (≥ 15, 11–14, < 11), gender (male vs female), and residence of the school (rural vs urban). Participants who did not complete the SCQ-L or did not provide a signed informed consent were excluded from the estimation.

ASD cases were counted independently from the source of data for diagnosis (MoE records –phase 1– or clinical screening –phase 2). The survey data commands (svyset) in Stata 15 were used to specify the survey design by setting weights for participants and SCQ-L classes as strata. For children with a certified diagnosis in the MoE database (phase 1), a weight of 1 was assigned. For children who underwent clinical screening (phase 2), weights were calculated based on the reciprocal of the probability of being clinically screened. For children with SCQ-L scores in the range of 11 to 14, weights accounted for 20% sample. These weights were calculated separately for each geographical area and SCQ-L score class, by accounting for the certified ASD diagnoses in the MoE database and the number of children who underwent clinical screening.

Prevalence estimates were obtained using the "proportion" command, considering the specified survey settings. The statistical analyses were conducted using Data Analysis and Statistical Software STATA (version 15.1; Stata Corp., College Station, TX, USA).