Autism spectrum disorder prevalence in Italy: a nationwide study promoted by the Ministry of Health

Autism Spectrum Disorder (ASD) is a neurodevelopmental condition characterized by impairments in socio-relational domains and both verbal and nonverbal communication skills, as well as the presence of restricted, repetitive, and stereotypical interests and activities [1]. ASD is a diverse condition encompassing individuals with varying support needs and associated psychiatric and medical co-occurring conditions.

In recent years, there has been an observed increase in the prevalence of ASD. This rise can be attributed to several factors, including improved availability of diagnostic services, and increased public awareness among laypeople and professionals [14]. Additionally, the expansion of diagnostic criteria and assessment tools for ASD has contributed to the identification of individuals with the condition [15].

Epidemiological studies on ASD have significantly increased in recent years; however, they often yield varying and conflicting results. This can be attributed to numerous complex factors, including differences in the abilities of clinicians to identify the condition, variations in the screening tools utilized, diverse models of conceptualization and categorization of ASD, discrepancies in methodological protocols, and variations in sample recruitment [36]. Furthermore, the introduction of the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition [1], may have adjusted the threshold for clinical diagnosis, potentially leading to an increased prevalence of individuals diagnosed with a milder phenotype of ASD who would have previously remained undiagnosed [26]. Lastly, the contribution of environmental risk factors cannot be excluded when considering the increase in ASD prevalence [21].

According to a comprehensive review of studies worldwide, the estimated overall prevalence of ASD ranges from 1.1 per 10,000 to 436.0 per 10,000, with a calculated average of 100 per 10,000 [43]. In the United States, the Autism and Developmental Disabilities Monitoring (ADDM) Network reported an overall prevalence estimate of 27.6 per 1000 children aged 8 years in 11 states in 2020, equating to a ratio of 1 in 36 children [24]. In Europe, several studies have been published, including four in Italy [9]. The first one was part of the Autism Spectrum Disorder in the European Union (ASDEU) project and estimated a prevalence in the city of Pisa of 11.5 per 1000 children aged 7 to 9 years, using probabilistic calculations to adjust for non-responses [28]. Lower prevalence rates for children aged 6–10 years were reported by the Child and Adolescent Mental Health Units of two regional healthcare information systems: SMAIL in Piemonte (4.2 per 1000) and ELEA in Emilia-Romagna (4.3 per 1000) regions (as reported in the National Guidelines approved by the National-Regions Conference by Atti n. 53/CU of 10/05/2018) [9, 10]. Lastly, a prevalence study conducted in the Abruzzo region, baministrative data of the Autism Regional Reference Center, estimated a prevalence of 8.0 per 1000 in children aged 6 to 8 years [38]. Due to the lack of an Italian ASD prevalence estimate and of a national health information system or ASD registry, the Italian Ministry of Health funded this comprehensive population-based study.

The present study aimed to estimate the prevalence rate of ASD in Italian children aged 7–9 years. The study was conducted by the National Observatory for ASD, coordinated by the National Institute of Health (Istituto Superiore di Sanità, ISS), using a modified version of the ASDEU project protocol. This protocol was the result of the ‘European Protocol for Autism Spectrum Disorder Prevalence [31] within the European Autism Information System (EAIS) project aimed to facilitate a common format to be tested in the European Member States to determine the ASD prevalence in the EU. The prevalence of ASD was measured in three geographical areas: North, Central, and South, encompassing both urban and rural residence of the school. This study is the first to provide an Italian ASD prevalence estimate based on a large sample size conducted across different geographical areas throughout the country. In addition, for the first time, this study collected data records from the Italian Ministry of Education (MoE) and screened all children within the target age group attending mainstream schools using the Social Communication Questionnaire—Lifetime version (SCQ-L) [33] completed by their parents. A similar ASDEU study conducted in Denmark, Finland, Iceland, and France collected ASD prevalence data using nationwide health registry systems and population-based regional registries [12]. In our study, although on a smaller scale, we identified children diagnosed with ASD in the three geographical areas by accessing the data records from the MoE. To ensure comprehensive case identification, we combined access to the MoE records with population screening, which is less influenced by bias from clinical, educational, or registered records, and is considered a high research standard. The goal was to support the Italian government with reliable data for defining suitable health, social, and educational services and public health policies tailored and targeted to the specific needs of individuals with ASD and their families.

The present study, implemented as part of a national strategy of the Italian Ministry of Health, was carried out by the National Observatory for Autism (ISS) in collaboration with the MoE with the aim of determining the prevalence of ASD in Italy. The findings revealed a prevalence rate of 1 in 77 children aged 7–9 years old between 2016 and 2018. Our study protocol exhibited several notable strengths when compared to previous research conducted in Italy. It encompassed three distinct and sizable geographic areas in the North, Center, and South of Italy, including both rural and urban areas. Furthermore, the study benefited from a substantial sample size, with a total of 16,293 SCQ-L respondents out of the 35,823 individuals invited, corresponding to a response rate of 45.0%.

The present study followed the ASDEU methodological criteria and was conducted in two phases. In the first phase, we accessed the MoE registry to identify children with certified ASD who were already enrolled in mainstream schools within the selected areas. The second phase involved screening all children in the invited schools using the SCQ-L and assessing those who obtained a score equal to or higher than 15 on the SCQ-L for ASD symptomatology, cognitive abilities, and daily skills. This study protocol allowed us to capture data on both children who had already received an ASD diagnosis and had access to educational (e.g., support teacher, educator, individualized learning program, personalized test, and periodical learning reports by teachers) and healthcare services (Phase 1) and children who had not yet been diagnosed (Phase 2).

The present project presented several advantages compared to previous ASDEU studies conducted in Italy [28] and Spain [16]. Firstly, our study was carried out in three extensive geographical areas, including schools from both urban and rural residence. The study was planned by enrolling schools distributed both in rural and urban areas. Overall, analyses found no significant differences in the percentages of certified children, the percentages of children assessed among those who were to be assessed, or the percentages of new diagnoses in assessed children among rural and urban areas, across all SCQ classes. These findings strengthen our estimate, as bias due to geographical localization (urban vs rural) may likely be excluded. This differed from previous studies that focused only on the metropolitan area of Pisa (Italy) and the County of Gipuzkoa (Spain). Secondly, our study considered the unique challenges faced by children in rural areas, as evidence suggests that they are at a higher risk of receiving a late or missed diagnosis. Moreover, accessibility to services has a significant impact on epidemiological rates, and our study accounted for this by including both urban and rural areas, as greater service availability facilitates early screening and diagnosis of ASD. Lastly, in the school population screening stage, our study involved testing all children attending the invited schools using the SCQ-L, whereas Narzisi and colleagues [28] identified children for further evaluation based on the Teacher Nomination form (TN) filled out by teachers and then screened them using the SCQ-L.

Our study’s prevalence rate of 13.4/1000 was higher compared to other Italian studies conducted in the Piemonte region (4.2/1000), Emilia-Romagna region (4.3/1000), and Abruzzo region (7.8/1000) according to the administrative regional reports (reported in the Ministry of Health Guidelines approved by the National-Regions Conference, May 10th 2018) [10] and the study by Valenti and colleagues [39]. These differences in the estimate of prevalence rates could be attributed to the different methods employed. In addition, our estimate may be an overestimate given the low rate of parental participation; parents with children at higher risk or with greater signals picked up by parents may have been more likely to have responded. Generally, prevalence rates obtained from registries tend to be lower compared to rates measured through screening the general population. Registry-based prevalence may miss individuals who do not have access to services or those who are undiagnosed.

In addition, our study reported a higher ASD prevalence compared to the ASDEU study conducted in Pisa (11.5/1000; [28]). This difference could potentially be attributed to the method used for selecting children to be assessed with the SCQ-L. In the Pisa study, children were selected based on the TN completed by teachers, which could introduce a selection bias. This selective approach may lead to an underestimation of ASD prevalence as it could miss children who may not exhibit obvious signs or have not been identified by teachers. In contrast, our study employed a broader screening approach, testing all children attending the invited schools using the SCQ-L. This method allowed for a more comprehensive assessment of the general population, including children who may not have been previously identified or suspected of having ASD. By adopting a population-based screening approach, we were able to capture a higher prevalence of ASD cases in our study. This strength should be considered in light of a low percentage of parents who completed SCQ-L and the low percentage of parents with children with positive SCQs who accepted the clinical assessment.

Our study's male to female prevalence ratio of 4.4:1 was consistent with the ratios reported in other studies conducted in the United States (4.5:1, [4]) and various European countries, such as South-Eastern France (4.0:1), Iceland (4.4:1), and Denmark (3.9:1, [12]). It is worth noting that the Italian ASDEU study conducted by Narzisi and colleagues [28] reported a higher male to female prevalence ratio of 5.2:1 compared to our overall ratio. However, when comparing the specific findings from the city of Pisa, located in the Center of Italy, the prevalence rate reported by Narzisi and colleagues was similar to our local prevalence rate in the Center area (5.3:1). These similarities suggest that there may be regional variations in ASD prevalence within Italy, and the differences in the male to female ratio could be influenced by geographical or environmental factors specific to each region. The integration of phase 2 (screening of the general population) with phase 1 (certified ASD diagnosis records of the MoE) permitted to identify 0.2% more cases with ASD that would not otherwise have been detected. Computing the incidence of new ASD diagnosis for the Italian population (N = 1,718,545) aged 7–9 years old on 1st January 2017 [19], the expected number of undiagnosed children with ASD was 2062.

Notably, the majority (89.5%, n = 17) of the children diagnosed with ASD and not registered with the MoE in our study exhibited preserved language skills, IQ within the average range, and had low support needs. This finding is consistent with the observation that children with milder or less apparent ASD symptoms often receive a diagnosis later in life [36]. Among the 19 children diagnosed with ASD, two of them also presented intellectual disability. In these specific cases, the children had previously received a diagnosis of intellectual disability but had not undergone a specific assessment for ASD. It is crucial to conduct comprehensive evaluations that include the assessment of ASD in individuals with coexisting intellectual disability to ensure accurate diagnoses and provide tailored intervention strategies. To optimize outcomes for children with ASD, it is essential to equip parents and teachers with targeted intervention tools [25]. The cases of ASD identified in this project have been integrated into the support and intervention network for ASD within the public Italian National Health System, ensuring that these children receive appropriate services and support for their ASD needs.

The overall school participation rate in the project was considered satisfactory, with 62.6% of schools participating. However, it was observed that schools in the Central area, specifically in Rome and its province, had lower participation rates (48.3%). This could be attributed to the proximity of these schools to major universities and research centers, as they may have already been involved in other research projects. On the other hand, schools in the South area exhibited the highest participation rate, reaching 80.6%. This could be due to the limited opportunities for schools in the South to participate in a national research screening program, likely because of their geographical distance from the inland and research institutions.

Overall, 45.5% (n = 16,293) of parents of children attending the invited schools participated by filling out the SCQ-L. The participation rate of parents was lower in the Centre area (36.7%) and South area (40.2%) compared to the North area (61.4%). This difference in participation could be attributed to the variation in availability of child and adolescent mental health services across different regions of Italy, with the North having higher availability compared to the Centre and South regions [6]. Parents in the North area may be more informed about the importance of early detection of NDD and more willing to participate in research projects related to this topic. It is concerning that the majority of the parents (60.6%) refused to have their children assessed for ASD despite their SCQ-L scores being above the cut-off. There could be several reasons for this refusal. Some parents may feel that their child does not require clinical evaluation because they do not perceive any social or learning difficulties. Despite explaining the benefits of obtaining a diagnosis at excellent ASD centers without incurring any costs, some parents may not grasp the significance of participating in this ASD prevalence study for the community and their own child. Stigma associated with an ASD diagnosis may also be a concern for some parents. Additionally, parents from rural areas may experience difficulties in accessing clinical centers located in urban areas for the assessment. Future studies should consider providing the option of conducting assessments at home or remotely when feasible. Efforts should also be made to improve the participation rates of schools and parents. This can be achieved through media campaigns that promote the project and emphasize its importance for the general population, including those not directly affected by ASD. Providing incentives such as awards for teachers who actively participate or promote the project and offering benefits to ASD organizations that decide to participate in research projects could also help enhance participation rates.

While this study employed a rigorous methodological protocol to determine the prevalence of ASD in Italy, it is important to acknowledge certain limitations. Firstly, it is possible that the prevalence rate obtained in this study is underestimated due to a high rate of parents (44.9%) who did not provide consent for the evaluation of their children when their SCQ-L scores exceeded the cut-off. Low parental participation rates in projects of this nature are common [13].

Secondly, there is a possibility that our screening process may have missed identifying some children at risk for ASD due to potential misunderstandings of the SCQ-L questions by parents [42]. To minimize the risk of false negatives, we conducted assessments on 20% of the sample. This additional assessment process aimed to identify any potential cases that might have been overlooked during the initial screening phase. Finally, this study encompassed both urban and rural areas near a national clinical referral center specializing in the ASD diagnosis in children, mitigating potential biases associated with service accessibility. Nonetheless, it is essential to acknowledge that various factors could contribute to the diversity in estimated prevalence rates, including social factors (e.g., family socio-professional category, family composition) and environmental factors (e.g., exposure to pollutants), which remained unaccounted for in the current study.

The present study found an ASD prevalence of approximately one in 77 children among 7- to 9-year-olds attending schools in three regions of Italy. The study protocol utilized a combination of registry data and screening of the general population, coordinated by the Italian National Institute of Health. The findings of this study are valuable for informing policies aimed at monitoring service activities, enhancing their quality, and customizing them to better meet the needs of children with ASD and their families. As suggested by Lord and colleagues [22], establishing formal documentation and support systems within healthcare, education, and social care sectors is crucial for recognizing and addressing the unique challenges and inequalities faced by individuals with autism. The epidemiological data obtained from this study provide important insights for guiding such initiatives and improving the overall well-being of individuals with ASD.