Early identification of autism spectrum disorder (ASD) and other developmental disabilities is a significant national priority. As the national prevalence of ASD has steadily increased, the U.S. has seen a concurrent increase in the number of children and families from culturally diverse communities (Baio et al. 2018; U.S. Census Bureau 2017; Van Naarden Braun et al. 2015). Within this context, there are significant disparities in screening rates, age of initial ASD diagnosis and other developmental disabilities, and access to early intervention services, such that children from diverse communities often receive less comprehensive screening and are diagnosed later than White children (Bishop-Fitzpatrick and Kind 2017; Mandell et al. 2009, 2005; Shattuck et al. 2009; Hewitt et al. 2016; Zuckerman et al. 2013). The changing cultural landscape of families with young children in the U.S. highlights a significant need for ongoing developmental screening, early diagnosis, and timely early intervention services in culturally and linguistically diverse communities.

The current study is a part of a larger study entitled Minneapolis Somali Autism Spectrum Disorder Prevalence Project (MSASDPP), which examined ASD prevalence in Minneapolis among children in multiple diverse communities, with a focus on ASD prevalence within the East African Somali community (Hewitt et al. 2016). Minnesota currently has the largest number of Somali immigrants in the U.S. (U.S. Census Bureau 2017), with a large percentage of Somali families living in Minneapolis (U.S. Census Bureau 2017; Mannix 2015; Ronningen 2000). Further, a significant percentage of MN’s Somali population was born outside the U.S. (MN State Demographic Center 2016). The focus of the MSASDPP study came about due to community concerns about the rates of ASD in the Somali community.

Overall findings from the MSASDPP study indicated that ASD prevalence did not differ statistically between White and Somali children, with rates of 1 in 36 and 1 in 32, respectively. White and Somali children had significantly higher prevalence estimates than Hispanic (1 in 80) and non-Somali Black children (1 in 62). One of the most compelling findings of the previous MSASDPP study was the high rate of co-occurring intellectual disability (ID) within the group of Somali children with ASD, with all (100%) of the Somali children with ASD also identified as having a co-occurring ID (Hewitt et al. 2016). The current study explores possible disparities in the age at which ASD is initially identified in Somali children and presents data to evaluate various explanations for discrepancies in the age at which ASD is first identified for those children.

Methods

This study consisted of a single-site, multi-source, records-based public health surveillance methodology based on the CDC’s ADDM Network methodology. To ensure a population large enough to obtain stable prevalence estimates, records were reviewed for children ages seven to nine. University of MN and Minneapolis Public Schools Institutional Review Boards (IRBs) approved record review, and all procedures met privacy and confidentiality requirements under 45 CFR 46 (U.S. Department of Health and Human Services 2013). Full information on the data collection process, including data site selection, is delineated in Hewitt et al. (2016). The project included two data collection phases: (1) screening and abstraction of records and (2) clinician review.

Screening and Abstraction

Children’s educational and health records were reviewed for triggers for inclusion in a clinical review process. There were two types of triggers: (1) ASD diagnosis or description, ASD special education classification, or record of an ASD test or assessment and (2) ASD behavioral triggers. ASD triggers included social, communication, and restricted/repetitive behaviors associated with a diagnosis of ASD (Rice et al. 2007). If a child’s record contained either trigger, abstractors prepared it for clinician review.

Clinician Review

After individually identifiable information was removed, CDC-trained clinicians reviewed the abstracted file to determine the child’s ASD status based on DSM-IV-TR criteria (APA 2000). If records included a comprehensive evaluation by a qualified professional documenting child behaviors at any point from birth through age nine that were consistent with the DSM-IV-TR diagnostic criteria for Autistic Disorder, Pervasive Developmental Disorder-Not Otherwise Specified (PDD-NOS), or Asperger Disorder, the child was considered to have met criteria for an ASD classification.

Race and Ethnicity Information

Race/ethnicity of individual children who met the ASD case criteria was determined from data from school and health/clinic records. Missing race/ethnicity values were replaced by linking files to birth certificates. Children were identified as Somali if either school records or clinic records indicated Somali was spoken as the primary language in the home as analyses identified this as the most reliable indicator.

Results

In the current study sample, 31.5% of the children were White, 35.0% were Black (including Somali children), 20.0% were Hispanic, 7.3% were Asian, and 3.0% were Native American. Somali children represented slightly less than 25% of Black children in the sample and 8.2% of the overall sample.

Age of Earliest Known ASD Identification

Table 1 presents data on the age of earliest known ASD clinical diagnosis and ASD educational eligibility. The average age of initial ASD clinical diagnosis was 4.8 years (range 1.4–9.7 years). The average age of initial ASD educational eligibility was 4.6 years (range 1.6–9.3 years). This difference was not statistically significant.

Table 1 Child’s age at earliest known age of ASD identification including DSM-IV ASD clinical diagnosis or ASD educational eligibility by race/ethnicity, stratified by ASD subtype
Full size table

Diagnostic Subtype

There was a significant disparity in the initial age of diagnosis of Autistic Disorder between Somali and White children, with Somali children (5.9 years) diagnosed later than White children (3.7 years; F(3, 33) = 3.08, p = .041). The current study lacked sufficient power to compare differences between Somali children and other non-White children in diagnostic subcategories. There were no differences in age of first educational eligibility by race/ethnicity (F(3, 95) = 1.78, p = .157).

ASD Developmental Concerns

The age and category of first developmental concerns were also identified through reviews of comprehensive evaluations. More than 80% of children who met ASD case criteria had general developmental concerns documented prior to age three. As seen in Table 2, the pattern of developmental concerns was roughly parallel across race/ethnicity with no statistically significant differences among groups.

Table 2 Number and percentage of children aged 7–9 years who met ASD surveillance case definition with developmental concerns noted in evaluations prior to age 3 years, by race/ethnicity
Full size table

Earliest Comprehensive Evaluation

Given that a large proportion of the Somali population in Minneapolis was born outside of MN (most of whom were likely born outside of the US), this factor was examined as a possible explanation for why Autistic Disorder is first diagnosed later in Somali versus White children (Table 3). Overall, children born outside of MN (54.9 months) were evaluated significantly later than children born in MN (45.7 months; t(253) = 2.44, p = .015, d = 0.40). This extended to a subsample of Somali children born outside of MN, although the sample was very small and not adequately powered for statistical analysis (N = 5). For Somali children not born in MN, the average age for first comprehensive evaluation was 69.6 months (5.8 years old), while for those born in MN, the average age for first comprehensive evaluation was over 2 years earlier at 43.6 months (3.6 years old). Again, sample sizes for this subgroup were extremely small, but these preliminary observations do offer an avenue for future exploration.

Table 3 Child’s age (months) at earliest comprehensive evaluation abstracted by birth state, race/ethnicity and somali status
Full size table

Discussion

Age of Initial Identification of ASD

The average age of initial ASD identification across the entire sample including both clinical and educational eligibility was approximately 4.7 years old. Consistent with previous ASD research (Levy et al. 2003; Shattuck et al. 2009; Mandell et al. 2005; Yeargin-Allsopp et al. 2003), the current study findings suggest, on average, professionals identify children with ASD at a point that corresponds relatively closely with the age of entry into the K-12 public school system.

Somali children were significantly older (5.9 years old) than White children (3.7 years old) when they received an initial clinical diagnosis of Autistic Disorder. These results are consistent with previous research revealing that children from diverse communities often receive an ASD diagnosis later (Durkin et al. 2017; Magaña et al. 2017; Mandell et al. 2007, 2009; Shattuck et al. 2009). On the other hand, given that Somali children with Autistic Disorder in this study also presented with co-occurring intellectual disabilities (Hewitt et al. 2016; Esler et al. 2017), the later age of diagnosis did not follow the expected trend of children with more severe symptoms being identified at earlier ages (Herlihy et al. 2015; Lopez 2014).

No differences in age of identification for ASD special education eligibility were found across groups, suggesting that in school settings Somali children were identified earlier ages, similar to children from other groups, when compared to clinics. This highlights the potential role disparities in access to medical care, developmental screening, and specialty care may contribute to this delay in identification for children from linguistically diverse communities (Bishop-Fitzpatrick and Kind 2017). Sample sizes were small, so caution in interpretation is warranted.

To better understand this potential influence, we examined where the children were born and found children born outside of MN were identified later. While we do not know that the children born outside of MN were born outside of the US, given patterns in the Somali population in Minneapolis as a whole, it is likely that they were. As such, this finding is consistent with existing ASD research that reveals children born outside of the U.S. and born to immigrant mothers are likely to have a later age of initial ASD diagnosis (Valicenti-McDermott et al. 2012). The implications are that increased efforts should be made to conduct developmental screenings and connect families new to the state to resources if there are concerns about their child’s development.

It is notable that more than 80% of children in this study with confirmed ASD had documented developmental concerns prior to age three despite not being diagnosed until they were almost 5 years old. These results mirror the broader ASD literature on presentation of early developmental concerns (Chawarska et al. 2013; Hess and Landa 2012; Wetherby et al. 2004). There were no statistically significant differences in the pattern of developmental concerns across racial and ethnic groups. This suggests that the disparity in the age of identification of Autistic Disorder in Somali versus White children is not due to differences in when or how developmental concerns arise for those children. Factors such as clinician knowledge of ASD and comfort discussing developmental concerns with families and a lack of properly translated screening tools may influence the timing of initial diagnosis, particularly for families from culturally and linguistically diverse backgrounds. These findings may further highlight both the general caution of providers to identify ASD in young children and the need for further education around the early signs of ASD in children under 5 years old.

Limitations of the Current Study

The study surveillance area was Minneapolis. Thus, the findings represent only this community. Further, there was not 100% case ascertainment due to the inability to review all records in public/charter schools and clinics. Our overall sample size was small, with some cell sizes too small to allow for specific subgroup analyses. Additionally, Somali children were identified based on primary home language, as this was determined to be the most reliable indicator; however, there may have been additional unidentified Somali children. A final limitation is the secondary review of records to determine ASD case status. Thus, results are limited to the information in those records rather than a direct assessment, which introduces effects of other variables, such as trends in evaluation, referrals, and clinician knowledge (Pantelis and Kennedy 2016).

Future Directions/Practical Implications

The current study contributes to a limited number of studies on culturally and linguistically diverse populations of children with ASD. The Somali population in this study is one of the largest studied. Future directions include expansion of surveillance area to increase the sample size to identify subgroup trends and increase generalizability.

These findings highlight the practical importance of increasing access to early developmental screening resources for children in diverse communities. Public health outreach campaigns such as CDC’s national “Learn the Signs. Act Early.” (http://www.cdc.gov/actearly) provide evidenced-based and accessible developmental resources for families. These early developmental resources promote increased access to resources outside of conventional screening settings. The related MN Act Early Project has done considerable outreach directly in diverse communities in many nonconventional settings (e.g. culturally focused family organizations, faith communities, community cultural events) to increase reach to families who might not receive these materials or messages in more traditional settings.

Conclusion

These results highlight that the initial age identification of ASD is a complex, multidimensional construct. The initial age of ASD identification continues to be later than desired, and delayed ASD identification may be more pronounced in some diverse communities. As disparities in age of initial ASD identification persist, the field needs to continue to promote early developmental monitoring, develop culturally sensitive screening and assessment tools, and build greater clinician awareness of different cultural views and perceptions of child development (Harris et al. 2014; Magaña et al. 2017).