Abstract
Autism Spectrum Disorder (ASD) prevalence estimates have varied by region. In this study, ASD prevalence, based on active case finding from multiple sources, was determined at the county and school district levels in the New Jersey metropolitan area. Among children born in 2008, residing in a four-county area and enrolled in public school in 2016, ASD prevalence was estimated to be 36 per 1,000, but was significantly higher in one region -- 54 per 1,000 and greater than 70 per 1,000, in multiple school districts. Significant variation in ASD prevalence by race/ethnicity, socioeconomic status (SES) and school district size was identified. Highest prevalence was in mid-SES communities, contrary to expectation. Prevalence among Hispanic children was lower than expected, indicating a disparity in identification. Comprehensive surveillance should provide estimates at the county and town levels to appreciate ASD trends, identify disparities in detection or treatment and explore factors influencing change in prevalence.
Keywords: ASD, Autism, Epidemiology, Prevalence
Lay Summary
We found autism prevalence to be 3.6% in New Jersey overall, but higher in one region (5.4%) and in multiple areas approaching 7.0%. We identified significant variation in ASD prevalence by race/ethnicity, socioeconomic status (SES) and school district size. Mapping prevalence in smaller, well-specified, regions may be useful to better understand the true scope of ASD, disparities in ASD detection and the factors impacting ASD prevalence estimation.
Background
Autism Spectrum Disorder (ASD) is a complex neurodevelopmental disorder characterized by social communication impairments in combination with restricted interests and/or repetitive behaviors (American Psychiatric Association, 2013). The most recent reports suggest autism prevalence affects approximately 2% of United States (US) children and indicates that ASD prevalence has increased 200% since 2000 (ADDM-CDC, 2007a, 2007b; Fombonne, 2018; Maenner et al., 2020), making ASD one of the most common childhood developmental disorders (Van Naarden Braun et al., 2015).
In the United States (US), ASD prevalence has been estimated by three different systems: the National Survey of Children’s Health (NSCH), the National Health Interview Survey (NHIS) and the Autism and Developmental Disabilities Monitoring (ADDM) Network. NSCH and NHIS provide national estimates based on parent report. The ADDM Network tracks ASD through an active, bi-annual, population-based multi-state US surveillance system (Kogan et al., 2018; Rice et al., 2007; Zablotsky, Black, Maenner, Schieve, & Blumberg, 2015). One of the advantages of ADDM Network surveillance is the ability to identify undiagnosed ASD cases, compared to NSCH and NHIS systems. Since 2000, ADDM Network ASD prevalence estimates have risen three-fold -- from 6.7 per 1,000 (95% CI: 6.3-7.0) in 2000, to 18.5 per 1,000 (95% CI: 18.0-19.1) in 2016 (ADDM-CDC, 2007b; Maenner et al., 2020). Differences in ASD prevalence across US sites have been noted consistently (ADDM-CDC, 2007a, 2007b, 2009, 2012; Baio et al., 2018; Maenner et al., 2020), but not investigated.
The sex ratio estimate is one of the most consistent epidemiologic findings on ASD. Across multiple epidemiological investigations, a 4:1 (male/female) sex ratio is disclosed (ADDM-CDC, 2007a, 2009; Baio, 2012; Baio et al., 2018). Regarding the distribution of ASD by race/ethnicity, the most recent ADDM report indicated that prevalence estimates may be equalizing across race and ethnicity, though disparities remain (ADDM-CDC, 2007a, 2007b; Baio et al., 2018; Maenner et al., 2020). While multiple studies reported a positive association between SES and ASD prevalence, between 2000 and 2010 (Durkin et al., 2017; Durkin et al., 2010; Nevison & Parker, 2020; Thomas et al.,) Recent studies suggest that the association may be shifting (Nevison & Parker, 2020; Winter, Fountain, Cheslack-Postava, & Bearman, 2020).
The distribution of factors contributing to ASD prevalence variation locally has been examined utilizing available (federal and state) administrative data by multiple studies (Maenner & Durkin, 2010; Mandell et al., 2010; Mandell & Palmer, 2005; Palmer, Walker, Mandell, Bayles, & Miller, 2010; Shattuck, 2006). However, ASD prevalence estimates derived from administrative data often underestimate ASD prevalence and/or provide information reflecting only some sectors of the total population. For example, it has been shown that reliance on (Autism) special education classification underestimates ASD prevalence (Baio et al., 2018). For example, across ADDM sites, Autism special education classification ranges from 36.5% to 75% (Baio et al., 2018). More local or granular analyses of ASD prevalence information established by active surveillance may inform the understanding of ASD patterns overall and are likely to inform the allocation of resources to individuals with ASD. Additionally, understanding local variations in autism prevalence allows us to consider and control for multiple factors that are often cited as sources of variations in ASD prevalence estimates, including different methodologies, different regional policies and to some extent level of awareness and access to services (Broder-Fingert, Sheldrick, & Silverstein, 2018; Fombonne, 2018). Furthermore, data from epicenters of ASD may lead to understanding future trends in ASD prevalence and/or lead to innovative strategies for addressing the increase in ASD prevalence estimates.
This study determined ASD prevalence in a diverse and populous US region in New Jersey at the county and school district level and examined ASD prevalence variation in relation to sex, race, socioeconomic status (SES) and school district size. The main objectives were to: 1) describe variation in ASD prevalence in New Jersey using population-based data from an active ASD surveillance system and 2) examine sociodemographic factors related to ASD prevalence.
Methods
Study Design
Cross-sectional data from active, multiple-source, (ADDM Network) ASD monitoring of 8-year-olds, born in 2008 and residing in the New Jersey surveillance region in 2016, was utilized. Data were from population-based surveillance for ASD, developed by the CDC--National Center on Birth Defects and Developmental Disabilities (NCBDDD). Surveillance was conducted in Essex, Hudson, Ocean and Union counties, a region consisting of 76 populous and diverse urban and suburban school districts (Supplemental Figure 1). Two districts did not participate and data from these districts were excluded. The counties differed in their race, SES, and district size profiles (Table 1).
Table 1 –
Distribution of sex, race/ethnicity, socioeconomic status and school district size (overall and by county) in the surveillance region for 8-year old children in 2016
NJAS surveillance Region |
Distribution of demographic and ecological factors by county | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Essex | Hudson | Ocean | Union | p-value | |||||||
Category | Pop. | % | Pop. | % | Pop. | % | Pop. | % | Pop. | % | |
Sex | 0.35 | ||||||||||
Female | 12837 | 49.2 | 4163 | 48.5 | 2959 | 49.9 | 2304 | 48.9 | 3412 | 49.7 | |
Male | 13246 | 50.8 | 4413 | 51.5 | 2974 | 50.1 | 2403 | 51.1 | 3456 | 50.3 | |
Race/Ethnicity | <0.001 | ||||||||||
White, Non-Hispanic | 8835 | 33.9 | 2292 | 26.7 | 1045 | 17.6 | 3209 | 68.2 | 2289 | 33.3 | |
Black, Non-Hispanic | 5317 | 20.4 | 3053 | 35.6 | 697 | 11.7 | 218 | 4.6 | 1350 | 19.7 | |
Hispanic | 9860 | 37.8 | 2520 | 29.4 | 3568 | 60.1 | 1042 | 22.1 | 2731 | 39.8 | |
SES | <0.001 | ||||||||||
Low (<$45,000) | 8482 | 32.5 | 4405 | 51.4 | 933 | 15.7 | 969 | 20.6 | 2175 | 31.7 | |
Mid ($45,000-$100,000) | 12811 | 49.1 | 1718 | 20.0 | 4851 | 81.8 | 3735 | 79.3 | 2507 | 36.5 | |
High (>$100,000) | 4809 | 18.4 | 2459 | 28.7 | 150 | 2.5 | - | - | 2191 | 31.9 | |
School district size | <0.001 | ||||||||||
Small (<250 students) | 4357 | 16.7 | 699 | 8.2 | 691 | 11.6 | 1427 | 30.3 | 1540 | 22.4 | |
Medium (250-999 students) | 13918 | 53.4 | 5153 | 60.1 | 3162 | 53.3 | 2210 | 47.0 | 3393 | 49.4 | |
Large (>999 students) | 7810 | 29.9 | 2724 | 31.8 | 2080 | 35.1 | 1070 | 22.7 | 1936 | 28.2 | |
Total | 26083 | 8576 | 5932 | 4707 | 6868 |
Pop=Population based on NJ Department of Education public school enrollment data for 2016-2017 school year
SES is based on U.S Census 2016 American Community Survey Median Household Income by County Subdivision equivalent to school districts
Abbreviations: NJAS = New Jersey Autism Study; SES = Socioeconomic Status
p-value for differences between counties
The ADDM methodology has been extensively described elsewhere, but briefly it is an active surveillance method, using a two-phase approach to ASD ascertainment, using standard case identification procedures and standard DSM-specified ASD criteria (ADDM-CDC, 2007a, 2007b, 2012; Avchen et al., 2011; Baio, 2012; Baio et al., 2018; Christensen et al., 2016; Maenner et al., 2020). In phase 1, records from hospital-based developmental centers and special education records were reviewed for all children meeting age (born in 2008) and residency criteria, in 2016. Complete records of children showing one or more specific, pre-determined, ASD indicators in professional evaluations conducted for educational services or clinical diagnosis or care, were comprehensively abstracted. Information contained in professional evaluations by community providers (e.g., Psychologists, MD Developmental Specialists, Speech and Language Pathologists, Social Workers, Occupational and Physical therapists) was reviewed by trained researchers. ASD indicators included: ASD diagnosis , Autism special education classification information and/or documented indication of one or more specific ASD-associated behaviors such as “poor, variable or no eye contact,” “inability to form peer relationships” and “preference for solo play,” among others. Information from comprehensive evaluations was compiled into a de-identified, chronological record, from birth to age 8, for each individual child.
In phase 2, clinician reviewers satisfying specialized CDC training and reliability criteria, used standardized scoring and case definition procedures to confirm ASD cases. The ASD case definition was satisfied if 1) behaviors documented in abstracted professional evaluations met the DSM-5 criteria as specified by the surveillance case definition and/or 2) if abstracted information disclosed an ASD diagnosis by age 8-years.
Assessment of ASD Prevalence
Records of 5,453 (8-year-old) children were reviewed. Information from 2,520 children was abstracted, consequent to a documented (ASD) indicator, in one or more professional evaluation. 1,036 children satisfied the surveillance ASD case definition. Ninety-four (94) confirmed ASD cases did not attend a public school in 2016 and were excluded from analysis, yielding 942 ASD children enrolled in public school. The study denominator – 26,083 individuals included all 8-year-olds attending public school in the surveillance counties, in 2016 (NJ Department of Education, 2017). ASD prevalence was estimated at the district and county levels.
Estimates were based on population denominators obtained from New Jersey Department of Education (NJ DOE) public school enrollment data (26,083 8-year-old children) (NJ Department of Education, 2017). Population denominators obtained from the National Center for Health Statistics (NCHS) vintage 2018 postcensal estimates for the surveillance region were obtained at the county level to estimate the number of 8-year-old children not enrolled in public school (33,031 8-year-old children) (National Center for Health Statistics (NCHS), 2018). We estimate that 6,948 (21%) children of the birth cohort were not enrolled in a public school in 2016. Since NCHS does not provide district level population estimates, public school enrollment information, was used to estimate the denominator on behalf of ASD prevalence determination at the district and county level.
Demographic and ecological variables
Data reflecting demographic and ecological factors potentially related to ASD prevalence, including sex and race/ethnicity were abstracted. Race was categorized as White (Non-Hispanic), Black (Non-Hispanic) and Hispanic. Race and ethnicity information was obtained from medical and educational sources and supplemented with information from birth certificates when source data on race and ethnicity was missing. Ecological factors assessed included: socioeconomic status (SES) and school district size. SES by county was determined by averaging median household income across the school districts within each county. Median household income (MHI) was defined at the district level, based on 2016 intercensal 5-year estimates and served as the proxy for SES. MHI estimates ranged from $33,025 to $190,625, with a median of $73,596 (standard deviation: $35,171) among districts. SES was categorized as Low SES (MHI <$45,000), Mid SES (MHI $45,000-$100,000) and High SES (MHI >$100,000) (US Census, 2016).
District size was defined as: small (<250), medium (250 to 999) or large (>999), according to the number of 8-year-olds in the district. District size categorization was based on enrollment data from the National Center for Education (NCES) (Gray, Bitterman, Goldring, & Broughman, 2013).
Data Analysis
Prevalence estimates were based on the number of confirmed ASD cases in public school, identified by surveillance, divided by the total population of public school enrolled 8-year-old children. Wilson score method was used to calculate 95% confidence intervals for prevalence estimates and prevalence ratios. Pearson chi-square and Fisher exact tests were used to compare distributions across counties. Prevalence rate ratios and 95% confidence intervals at the county level and overall surveillance region were assessed by sex (reference: female), race/ethnicity , SES (reference: low SES) and school district size (reference: small district size). Sociodemographic analysis at the school district level was not included due to small sample sizes.
This study was approved by the Institutional Review Board (IRB) of Rutgers University – New Jersey Medical School and ASD surveillance was conducted under waiver of informed consent, consistent with public health investigation standards.
Results
Description and characteristics of the surveillance region
The surveillance region was densely populated, entirely urban & suburban, racially and ethnically diverse, with White (Non-Hispanic), Black (Non-Hispanic) and Hispanic children comprising 33.9%, 20.4% and 37.8% of the population, respectively (Table 1). The distribution of race/ethnicity, socioeconomic status (SES) and school district size varied across the four counties (p<0.001) (Table 1).
Table 2 describes the characteristics of children identified with ASD at the County level. Overall, 767 children (81.4%) were diagnosed with ASD by a community provider such as a Developmental Pediatrician or a Pediatric Neurologist, and 175 children (18.6%) did not have an ASD diagnosis but met the surveillance case definition for ASD. ASD diagnosis by a community provider was highest in Hudson County (90.1%) and lowest in Union County (73.8%) (p =0.0003).
Table 2 –
Characteristics (overall and by county) in the surveillance region for 8-year old children in 2016
NJAS surveillance Region |
Distribution of demographic and ecological factors by county | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Essex | Hudson | Ocean | Union | p-value | |||||||
Category | Pop. | % | Pop. | % | Pop. | % | Pop. | % | Pop. | % | |
ASD Community Classification | |||||||||||
ASD Diagnosis | 767 | 81.4 | 230 | 79.9 | 163 | 90.1 | 211 | 83.7 | 163 | 73.8 | 0.0003 |
Autism Special Education Classification | 485 | 40.7 | 184 | 63.9 | 121 | 66.9 | 90 | 35.7 | 90 | 40.7 | <.0001 |
ASD Diagnosis and Autism Classification | 478 | 40.7 | 177 | 61.5 | 121 | 66.9 | 90 | 35.7 | 90 | 40.7 | <.0001 |
Special Education | |||||||||||
Receiving school services | 886 | 94.1 | 279 | 96.9 | 177 | 97.8 | 226 | 89.7 | 204 | 92.3 | 0.0004 |
Intellectual Ability (n = 736) | <.0001 | ||||||||||
IQ ≤ 70 | 182 | 25.2 | 64 | 30.6 | 53 | 36.3 | 23 | 11.6 | 42 | 24.7 | |
IQ > 70 | 541 | 74.8 | 145 | 69.4 | 93 | 63.7 | 175 | 88.4 | 128 | 75.3 | |
Number with IQ data | 723 | 76.8 | 209 | 72.6 | 146 | 80.7 | 198 | 78.6 | 170 | 77.0 | 0.18 |
p-value for differences between counties
Abbreviations: ASD = Autism Spectrum Disorder; NJAS = New Jersey Autism Study; IQ = Intelligence Quotient
In general, while most children identified with ASD received special education services through their local public school system in 2016 (94.1%), only 40.7% (n=485) of children with ASD were educated under the Autism (special education) classification and special education classification varied by county. Hudson County had the highest Autism classification (66.9%) and Ocean County had the lowest classification (35.7%) (p<.0001). Among children with ASD with documented intelligence quotient (IQ) data, 74.8% (n=541) children did not have IQ in the intellectual disability range. Intellectual ability also differed by County. In Ocean County, 88.4% of children with IQ data had IQ scores above 70, while in Essex County 69.4% had IQ scores above 70 (p<.0001).
ASD prevalence comparisons at the county level
Overall, ASD prevalence was 36 per 1,000 (95% CI: 34-38) in the combined four-county surveillance region (Table 2). ASD prevalence estimates varied by county, sex, race/ethnicity, socioeconomic status and school district size. ASD prevalence estimates for Hudson County were lowest and highest in Ocean County (Table 3). Across the region, ASD prevalence was consistently (4.0-4.8 times) higher among male children, compared to female children (Table 4). Ocean County had higher ASD prevalence across all subgroups, except for Hispanic children (Table 3).
Table 3.
Autism Spectrum Disorder prevalence at the county level in New Jersey among 8-year old children (overall and by sex, race/ethnicity, socioeconomic status, and school district size) in 2016
Region | NJAS surveillance Region |
ASD prevalence per 1,000 children by county | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Essex County | Hudson County | Ocean County | Union County | p-value | |||||||
Category | ASD Cases |
Prevalence (95%CI) |
ASD Cases |
Prevalence (95%CI) |
ASD Cases |
Prevalence (95%CI) |
ASD Cases |
Prevalence (95%CI) |
ASD Cases |
Prevalence (95%CI) |
|
Sex | |||||||||||
Female | 180 | 14 (12-16) | 53 | 13 (10-17) | 37 | 13 (9-17) | 51 | 22 (17-29) | 39 | 11 (8-16) | 0.003 |
Male | 762 | 58 (54-62) | 235 | 53 (47-60) | 144 | 48 (41-57) | 201 | 84 (73-95) | 182 | 53 (46-61) | <0.0001 |
Race/Ethnicity | |||||||||||
White, Non-Hispanic | 402 | 46 (41-50) | 75 | 33 (26-41) | 39 | 37 (27-51) | 198 | 62 (54-71) | 90 | 39 (32-48) | <0.0001 |
Black, Non-Hispanic | 181 | 34 (29-39) | 102 | 33 (28-40) | 30 | 43 (30-61) | 10 | 46 (25-82) | 39 | 29 (21-39) | 0.29 |
Hispanic | 272 | 28 (25-31) | 89 | 35 (29-43) | 79 | 22 (18-28) | 29 | 28 (19-40) | 75 | 27 (22-34) | 0.02 |
SES | |||||||||||
Low SES (<$45,000) | 285 | 34 (30-38) | 157 | 36 (31-42) | 13 | 14 (08-24) | 41 | 42 (31-57) | 74 | 34 (27-43) | 0.003 |
Mid SES ($45,000-$100,000) | 521 | 41 (37-44) | 67 | 39 (31-49) | 162 | 33 (29-39) | 211 | 56 (50-64) | 81 | 32 (26-40) | <0.0001 |
High SES (>$100,000) | 136 | 28 (24-33) | 64 | 26 (20-33) | 6 | 40 (18-85) | - | - | 66 | 30 (24-38) | 0.62 |
School district size | |||||||||||
Small (<250 students) | 147 | 34 (29-40) | 19 | 27 (17-42) | 23 | 33 (22-49) | 65 | 46 (36-58) | 40 | 26 (19-35) | 0.02 |
Medium (250-999 students) | 447 | 32 (29-35) | 150 | 29 (25-34) | 73 | 23 (18-29) | 109 | 49 (41-59) | 115 | 34 (28-41) | <0.0001 |
Large (>999 students) | 348 | 45 (40-49) | 119 | 44 (37-52) | 85 | 41 (33-50) | 78 | 73 (59-90) | 66 | 34 (27-43) | <0.0001 |
Total | 942 | 36 (34-38) | 288 | 34 (30-38) | 181 | 31 (26-35) | 252 | 54 (47-60) | 221 | 32 (28-37) | <0.0001 |
ASD cases are based on DSM 5 ASD case definition using ADDM methodology
Population denominators are based on public school enrollment in 2016-2017 school year
Prevalence per 1,000 8-year old children
95% Confidence Interval based on Wilson score method
SES is based on U.S Census 2016 American Community Survey Median Household Income by County Subdivision equivalent to school districts
No cases were identified in Ocean County in the High SES category
p-value for differences between counties
Abbreviations: ASD = Autism Spectrum Disorder; NJAS = New Jersey Autism Study; SES = Socioeconomic Status; CI = Confidence Interval
Table 4.
Prevalence ratio of ASD prevalence by sex, race/ethnicity, socioeconomic status and school district size
NJAS surveillance Region |
ASD prevalence ratio in each county in the surveillance region | ||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Essex | Hudson | Ocean | Union | ||||||||||||
Category | PR | 95%CI | p-value | PR | 95%CI | p-value | PR | 95%CI | p-value | PR | 95%CI | p-value | PR | 95%CI | p-value |
Sex | |||||||||||||||
Female (reference) | |||||||||||||||
Male | 4.3 | 3.6-5.1 | <0.001 | 4.4 | 3.2-5.9 | <0.001 | 4.0 | 2.8-5.8 | <0.001 | 4.0 | 2.9-5.5 | <0.001 | 4.8 | 3.4-6.8 | <0.001 |
Race/Ethnicity | |||||||||||||||
BNH:WNH | 0.7 | 0.6-0.9 | 0.001 | 1.0 | 0.8-1.4 | 0.90 | 1.2 | 0.7-1.9 | 0.55 | 0.7 | 0.4-1.4 | 0.34 | 0.7 | 0.5-1.1 | 0.10 |
Hisp:WNH | 0.6 | 0.5-0.7 | <0.001 | 1.1 | 0.8-1.5 | 0.62 | 0.6 | 0.4-0.9 | 0.01 | 0.4 | 0.3-0.6 | <0.001 | 0.7 | 0.5-0.9 | 0.02 |
BNH: Hisp | 0.8 | 0.7-1.0 | 0.03 | 1.1 | 0.8-1.4 | 0.70 | 0.5 | 0.3-0.8 | 0.01 | 0.6 | 0.3-1.2 | 0.16 | 0.9 | 0.6-1.4 | 0.79 |
SES | |||||||||||||||
Low (<$45,000) (reference) | |||||||||||||||
Mid ($45,000-$100,000) | 1.2 | 1.1-1.4 | 0.01 | 1.1 | 0.8-1.5 | 0.53 | 2.4 | 1.4-4.3 | 0.01 | 1.4 | 1.0-1.9 | 0.08 | 0.9 | 0.7-1.3 | 0.74 |
High (>$100,000) | 0.8 | 0.7-1.0 | 0.09 | 0.7 | 0.5-1.0 | 0.03 | 2.9 | 1.1-7.9 | 0.02 | - | - | - | 0.9 | 0.6-1.2 | 0.47 |
School district size | |||||||||||||||
Small (<250 students) (reference) | |||||||||||||||
Medium (250-999 students) | 1.0 | 0.8-1.1 | 0.60 | 1.1 | 0.7-1.7 | 0.78 | 0.7 | 0.4-1.1 | 0.12 | 1.1 | 0.8-1.5 | 0.60 | 1.3 | 0.9-1.9 | 0.14 |
Large (>999 students) | 1.3 | 1.1-1.6 | 0.01 | 1.6 | 1.0-2.7 | 0.05 | 1.2 | 0.8-2.0 | 0.37 | 1.6 | 1.2-2.3 | 0.01 | 1.3 | 0.9-2.0 | 0.17 |
Abbreviations: ASD = Autism Spectrum Disorder; NJAS = New Jersey Autism Study; SES = Socioeconomic Status; PR = Prevalence Ration; CI = Confidence Interval; WNH = White, Non-Hispanic; BNH = Black, Non-Hispanic; Hisp = Hispanic
p-value <0.05 represents significant difference in prevalence between groups
Comparisons by race/ethnicity, SES and school district size
Race and ethnicity-based differences were observed, across counties. Hispanic children had significantly lower identified ASD prevalence compared to White (Non-Hispanic) children (prevalence ratio (PR) = 0.6; 95%CI: 0.5-0.7; P<0.001). ASD prevalence estimates were 30-60% lower among Hispanic children in three of four counties (Table 4). Mid SES districts had significantly higher ASD prevalence (PR = 1.2; 95% CI: 1.1-1.4; p = 0.01) compared to Low and High SES districts (PR= 0.8; 95% CI: 0.7-1.0; p=0.09; Table 4). School district size was consistently and positively associated with ASD prevalence estimates (PR = 1.3; 95% CI: 1.1-1.6; p = 0.004) (Table 4). This difference was observed across counties and yielded ASD estimates ranging from 1.2 to 1.6 times higher in large districts, compared to small districts.
ASD prevalence at the school district level
ASD prevalence estimates ranged from 8 to 108 per 1,000, at the district level (p<0.0001). Thirteen of 74 school districts (18%) showed ASD prevalence greater than 50 per 1,000.
Among the four largest school districts (age 8 enrollment >1,000), ASD prevalence was highest in Toms River (Ocean County), 73 per 1,000 (95% CI: 5.9-9.0).
Discussion
Our findings are consistent with prior ADDM reports and other epidemiologic studies showing wide variation in ASD prevalence across geographic regions. This study demonstrates that significant variation in ASD prevalence is also present at the local level. Variation in ASD prevalence across countries is likely the result of using different methods to estimate ASD prevalence (Chiarotti & Venerosi, 2020; Fombonne, 2018). In contrast, the ADDM Network employs a standard and consistent ascertainment method. Observed inter-state variation in ASD prevalence in the ADDM Network may be a function of differences in policy, awareness and/or access to professional services (Broder-Fingert et al., 2018; Fombonne, 2018; Pinborough-Zimmerman, Bilder, Satterfield, Hossain, & McMahon, 2010). Within New Jersey, however, the observed variations are not likely to be due to differences in policy or awareness but may reflect differences in utilization of services or access to care and should be considered, systematically.
Our examination of ASD prevalence at the county and school district levels confirms that ASD is not uniformly distributed, even within a region, and highlights counties and districts with higher-than-expected prevalence. For example, in Ocean County, over 5% of public students had ASD and nearly one in five districts in our surveillance region had ASD prevalence between 5% and 10%. We also found differences between counties in identification of ASD by a community provider. For example, while 90% of children in Hudson County had an ASD diagnosis, in Union County only 73.8% of children had a diagnosis. Furthermore, Ocean County had the highest proportion of children with ASD and IQ >70 suggesting better identification of children with ASD with borderline and average intellectual ability.
Even when utilizing a rigorous and standardized ascertainment method, in a region known for policies aiming to ensure access to high quality educational resources (EducationWeek, 2019), significant variations in ASD identification, diagnosis and educational classification exist. Active surveillance in metro New Jersey indicated that 3.6% of (8-year-old) children in the public education system had ASD. By focusing ‘down’ to the more granular county level, we were able to see that the identified rate of ASD ranged from approximately 3% in Hudson County, to 5% in Ocean County. Focusing still further ‘down’ -- to the district level, we recognized that many communities in our region, approximately one in five, including some of the largest, had ASD rates between 5% and 10%.
This study discloses an important and continuing disparity – Hispanic children with ASD are less likely to be identified than White. Non-Hispanic peers. When we parsed the surveillance data at a more granular level, we detected meaningful county and district level differences in the identification of Hispanic children with ASD. Case identification by the active multiple source method depends on the quality and quantity of information in professional evaluations. If Black and/or Hispanic children received services less frequently, they might be less likely identified. That possibility is supported by recent studies showing that Hispanic children are less likely to receive occupational and physical therapy, compared to White, Non-Hispanic children (Bilaver & Havlicek, 2019) and indicating that case-relevant information was more likely to be missing for Black, Non-Hispanic, and Hispanic children than for White, Non-Hispanic peers (Imm, 2019). Even in regions with high levels of awareness, support for and access to services, disparities in ASD identification may persist. If disparity is identified, local level information may lead to the provision of focused information and resource sharing with districts needing the most help. Accurate specific information at the local level is likely to be most useful for planning and implementation.
Previous epidemiologic studies from New Jersey and the ADDM Network observed a significant and persisting stepwise association of SES and ASD prevalence, between 2000 and 2010(Durkin et al., 2010; Durkin & Yeargin-Allsopp, 2018; Thomas et al., 2012). Surprisingly, in our (2016) population, ASD prevalence rates were highest among children from Mid SES communities. ASD prevalence estimates in High SES communities were lower than in Low-income communities, contrary to expectation. Our findings support the possibility of a shift away from the positive SES gradient for ASD observed from 2000 to 2010 and are consistent with recent ASD trend reports (Nevison & Zahorodny, 2019; Winter et al., 2020). Additional research and ongoing surveillance are necessary to clarify these observations and understand the drivers of the shift in ASD demographics. Previous studies based on administrative data have shown a relation between ASD identification and school district characteristics (Palmer, Blanchard, Jean, & Mandell, 2005). This study also found that higher rates of ASD were identified in large school districts. Multiple school districts in the New Jersey metro region had higher ASD prevalence compared to the average for the entire surveillance region and compared to the ADDM Network average for the period. Zero cases and sporadically higher than expected estimates would be predicted in small districts, as a function of small numbers. However, ASD rates were highest in some of the largest school districts. For example, in Newark, the largest urban district in New Jersey, identified ASD prevalence was 4.4%, while in Toms River, the largest suburban district, ASD prevalence was 7.1%. Larger districts may provide more services from a greater number of professionals or have additional resources for detection or care of ASD. It is possible that parents of children with learning or developmental disorders, including ASD, relocate from small districts to large districts, to maximize their children’s educational attainment. Additional studies are needed to specify the impact of district size and the potential influence of in-migration on ASD prevalence. If large districts are better able to identify and serve children with ASD, it might be useful for small districts to consider consolidating special education services at a county or regional level or to facilitate ASD-specific training and education of staff in small sized districts.
Public schools are the primary source of interventions to students with autism. Overall, 94% of our total population received special education services in the study year, indicating the importance of the public education system to students with ASD. Four in ten (41%) ASD students were served under the Autism classification, suggesting that the actual scope of ASD is dramatically underrepresented by the Autism classification count. Similarly, while about 80% of surveillance identified ASD cases have an ASD diagnosis by age nine, that leaves one in five ASD children undiagnosed and potentially underserved. If ASD diagnosis is associated with more robust services, evidence from this active surveillance system indicates a general area for improvement as well as providing the method for identifying the groups that are most likely to benefit from targeted action.
Significant resources are needed to care for, educate and support children with ASD. Effective planning and action are best served by accurate appreciation of the scope of a challenge. Aggregate estimates and general averages of ASD prevalence can obscure the useful information conveyed at the local level. ASD rates of 5% and higher were already evident in multiple New Jersey communities in 2016 and there is no reason to believe that similar rates of occurrence are not the case in other US metropolitan areas. Moreover, future autism estimates are likely to increase, as detection improves in underserved communities. The findings emphasize the need for increasing resources to and evidence-based planning of services to children with ASD and the enhancement of systems which monitor the expression of ASD at the local level and seek to define the social determinants which influence the identification and distribution of ASD (Krieger, 2011).
An important strength of this study was use of a rigorous, comprehensive, and validated method of active ascertainment in a diverse metropolitan Area. The active ascertainment method has multiple advantages, including detection of undiagnosed ASD cases, use of a standardized and reliable case definition based on DSM criteria and surveillance coverage of the total population, including children from under-represented populations. These ASD prevalence estimates were determined in the context of ongoing ASD monitoring by experienced investigators with access to information from multiple clinical and educational sources in a populous, diverse population.
Several limitations are acknowledged. The study denominators represent children attending public schools, only. Approximately 80% of children in the region attended public schools. ASD prevalence among non-public (private school and home schooled) students was not considered. Overestimation of ASD prevalence is possible among children enrolled in public school, as children in private schools and home-schooled, are less likely to have significant impairment in learning, requiring special education services. However, underestimation is also possible since our findings reveal disparities in the identification of Hispanic children. While ASD prevalence estimates where stratified by sociodemographic factors, this study did not examine whether SES and race-based differences exist across public and non-public school students. SES and school district size are ecological factors and do not reflect individual-level information. As in many epidemiological studies, residual confounding is possible, given that SES and school district size are broad categories. Surveillance was conducted in only four urban-suburban New Jersey counties representing approximately 25% of the total state population. No rural area was included, and findings may not be representative of the entire state. Additional sources of ascertainment bias cannot be ruled out. The ADDM active ascertainment method relies on access to a wide range of information from multiple sources and the completeness of existing information. Incomplete records due to socio-economic and/or race/ethnic disparities may have led to underestimation of ASD, in some districts. However, these ASD estimates may still understate actual ASD prevalence. There are children with ASD who first come to attention after age 8 and, therefore, would not be identified. Additionally, it is likely that SES and/or race & ethnicity-based disparities in ASD identification, in parts of our region, led to underestimation of ASD. Finally, a considerable limitation of this study is the sample size. Small sample sizes may produce higher prevalence estimates (Fombonne, 2002). There were several small sample sizes when the data was further stratified by race/ethnicity and SES.
The findings suggest that the true scope of ASD may be under-represented by extant national and state estimates. If ASD rates of 4 to 7% were recorded in the NJ-NY area, it is possible that similar levels might be detected in other US metro regions. Exclusive reliance on Autism classification or ASD diagnosis data may bias the appreciation of ASD prevalence on individual districts or regions. Additional research is needed to identify specific systemic or local conditions or practices that contribute to the use of health or education resources and, in turn, may affect the estimation of ASD prevalence.
Supplementary Material
Acknowledgement
This study was made possible by support from the Centers for Disease Control and Prevention (CDC) (1U53DD001172) and NIH-NIEHS P30 ES005022. The efforts and expertise of Audrey Mars MD, Mildred Waale LDTC, Arline Fusco PsyD, Tara Gleeson NP, Gail Burack PhD, Paul Zumoff PhD, Kate Sidwell, BA, Rita Baltus, MD, Cindy Cruz, Michael Verile, and Yuriy Levin as well as the cooperative support and participation of the New Jersey Department of Health and Education and the many school districts and health centers in our region is gratefully acknowledged. The authors have no conflict of interest to declare.
Grant Sponsor: CDC
Grant number: NU53DD001172
Grant Sponsor: NIH-NIEHS
Grant number: P30 ES005022
References
- ADDM-CDC. (2007a). Prevalence of autism spectrum disorders--autism and developmental disabilities monitoring network, 14 sites, United States, 2002. MMWR Surveill Summ, 56(1), 12–28. [PubMed] [Google Scholar]
- ADDM-CDC. (2007b). Prevalence of autism spectrum disorders--autism and developmental disabilities monitoring network, six sites, United States, 2000. MMWR Surveill Summ, 56(1), 1–11. [PubMed] [Google Scholar]
- ADDM-CDC. (2009). Prevalence of autism spectrum disorders - Autism and Developmental Disabilities Monitoring Network, United States, 2006. MMWR Surveill Summ, 58(10), 1–20. [PubMed] [Google Scholar]
- ADDM-CDC. (2012). Prevalence of autism spectrum disorders--Autism and Developmental Disabilities Monitoring Network, 14 sites, United States, 2008. MMWR Surveill Summ, 61(3), 1–19. [PubMed] [Google Scholar]
- American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (DSM-5) (5th ed.) (5th ed. ed.). Washington, DC: American Psychiatric Publishing. [Google Scholar]
- Avchen RN, Wiggins LD, Devine O, Van Naarden Braun K, Rice C, Hobson NC, … Yeargin-Allsopp M (2011). Evaluation of a records-review surveillance system used to determine the prevalence of autism spectrum disorders. J Autism Dev Disord, 41(2), 227–236. doi: 10.1007/s10803-010-1050-7 [DOI] [PubMed] [Google Scholar]
- Baio J (2012). Prevalence of Autism Spectrum Disorders — Autism and Developmental Disabilities Monitoring Network, 14 Sites, United States, 2008. MMWR Surveill Summ, 61(03), 1–19. [PubMed] [Google Scholar]
- Baio J, Wiggins L, Christensen DL, Maenner MJ, Daniels J, Warren Z, … Dowling NF (2018). Prevalence of Autism Spectrum Disorder Among Children Aged 8 Years - Autism and Developmental Disabilities Monitoring Network, 11 Sites, United States, 2014 (Table 6). MMWR Surveill Summ, 67(6), 1–23. doi: 10.15585/mmwr.ss6706a1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Broder-Fingert S, Sheldrick CR, & Silverstein M (2018). The Value of State Differences in Autism When Compared to a National Prevalence Estimate. Pediatrics, 142(6). doi: 10.1542/peds.2018-2950 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chiarotti F, & Venerosi A (2020). Epidemiology of Autism Spectrum Disorders: A Review of Worldwide Prevalence Estimates Since 2014. Brain Sci, 10(5). doi: 10.3390/brainsci10050274 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Christensen DL, Baio J, Van Naarden Braun K, Bilder D, Charles J, Constantino JN, … Prevention. (2016). Prevalence and Characteristics of Autism Spectrum Disorder Among Children Aged 8 Years--Autism and Developmental Disabilities Monitoring Network, 11 Sites, United States, 2012. MMWR Surveill Summ, 65(3), 1–23. doi: 10.15585/mmwr.ss6503a1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Durkin MS, Maenner MJ, Meaney FJ, Levy SE, DiGuiseppi C, Nicholas JS, … Schieve LA (2010). Socioeconomic inequality in the prevalence of autism spectrum disorder: evidence from a U.S. cross-sectional study. PLoS One, 5(7), e11551. doi: 10.1371/journal.pone.0011551 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Durkin MS, & Yeargin-Allsopp M (2018). Socioeconomic Status and Pediatric Neurologic Disorders: Current Evidence. Semin Pediatr Neurol, 27, 16–25. doi: 10.1016/j.spen.2018.03.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- EducationWeek. (2019). Quality Counts 2019. Retrieved from https://www.edweek.org/leadership/quality-counts-2019-grading-the-states [Google Scholar]
- Fombonne E (2002). Epidemiological trends in rates of autism. Mol Psychiatry, 7 Suppl 2, S4–6. doi: 10.1038/sj.mp.4001162 [DOI] [PubMed] [Google Scholar]
- Fombonne E (2018). Editorial: The rising prevalence of autism. J Child Psychol Psychiatry, 59(7), 717–720. doi: 10.1111/jcpp.12941 [DOI] [PubMed] [Google Scholar]
- Gray L, Bitterman A, Goldring R, & Broughman S (2013). Characteristics of Public School Districts in the United States: Results From the 2011–12 Schools and Staffing Survey. July 2013: National Center for Education Statistics - Institute of Education Sciences [Google Scholar]
- Kogan MD, Vladutiu CJ, Schieve LA, Ghandour RM, Blumberg SJ, Zablotsky B, … Lu MC (2018). The Prevalence of Parent-Reported Autism Spectrum Disorder Among US Children. Pediatrics, 142(6). doi: 10.1542/peds.2017-4161 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Krieger N (2011). Epidemiology and the people's health : theory and context. New York: Oxford University Press. [Google Scholar]
- Maenner MJ, & Durkin MS (2010). Trends in the prevalence of autism on the basis of special education data. Pediatrics, 126(5), e1018–1025. doi: 10.1542/peds.2010-1023 [DOI] [PubMed] [Google Scholar]
- Maenner MJ, Shaw KA, Baio J, Washington A, Patrick M, … Dietz PM (2020). Prevalence of Autism Spectrum Disorder Among Children Aged 8 Years - Autism and Developmental Disabilities Monitoring Network, 11 Sites, United States, 2016. MMWR Surveill Summ, 69(4), 1–12. doi: 10.15585/mmwr.ss6904a1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mandell DS, Morales KH, Xie M, Polsky D, Stahmer A, & Marcus SC (2010). County-level variation in the prevalence of medicaid-enrolled children with autism spectrum disorders. J Autism Dev Disord, 40(10), 1241–1246. doi: 10.1007/s10803-010-0982-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mandell DS, & Palmer R (2005). Differences among states in the identification of autistic spectrum disorders. Arch Pediatr Adolesc Med, 159(3), 266–269. doi: 10.1001/archpedi.159.3.266 [DOI] [PubMed] [Google Scholar]
- National Center for Health Statistics (NCHS), C. f. D. C. a. P. C. U. S. D. o. H. a. H. S. U. D. (2018). Bridged-Race Population Estimates, United States July 1st resident population by state, county, age, sex, bridged-race, and Hispanic origin. Compiled from 1990-1999 bridged-race intercensal population estimates (released by NCHS on 7/26/2004); revised bridged-race 2000-2009 intercensal population estimates (released by NCHS on 10/26/2012); and bridged-race Vintage 2018 (2010-2018) postcensal population estimates (released by NCHS on 6/25/2019). Retrieved Accessed at http://wonder.cdc.gov/bridged-race-v2018.html on Jul 24, 2021 1:14:32 PM, from CDC WONDER Online Database [Google Scholar]
- Nevison C, & Parker W (2020). California Autism Prevalence by County and Race/Ethnicity: Declining Trends Among Wealthy Whites. J Autism Dev Disord, 50(11), 4011–4021. doi: 10.1007/s10803-020-04460-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Nevison C, & Zahorodny W (2019). Race/Ethnicity-Resolved Time Trends in United States ASD Prevalence Estimates from IDEA and ADDM. J Autism Dev Disord, 49(12), 4721–4730. doi: 10.1007/s10803-019-04188-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- NJ Department of Education. (2017). 2016-2017 Enrollment District Reported Data. Retrieved from https://www.nj.gov/education/data/enr/enr17/
- Palmer RF, Blanchard S, Jean CR, & Mandell DS (2005). School district resources and identification of children with autistic disorder. Am J Public Health, 95(1), 125–130. doi: 10.2105/AJPH.2003.023077 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Palmer RF, Walker T, Mandell D, Bayles B, & Miller CS (2010). Explaining low rates of autism among Hispanic schoolchildren in Texas. Am J Public Health, 100(2), 270–272. doi: 10.2105/AJPH.2008.150565 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Pinborough-Zimmerman J, Bilder D, Satterfield R, Hossain S, & McMahon W (2010). The impact of surveillance method and record source on autism prevalence: collaboration with Utah Maternal and Child Health programs. Matern Child Health J, 14(3), 392–400. doi: 10.1007/s10995-009-0472-3 [DOI] [PubMed] [Google Scholar]
- Rice CE, Baio J, Van Naarden Braun K, Doernberg N, Meaney FJ, Kirby RS, & Network A (2007). A public health collaboration for the surveillance of autism spectrum disorders. Paediatr Perinat Epidemiol, 21(2), 179–190. doi: 10.1111/j.1365-3016.2007.00801.x [DOI] [PubMed] [Google Scholar]
- Shattuck PT (2006). The contribution of diagnostic substitution to the growing administrative prevalence of autism in US special education. Pediatrics, 117(4), 1028–1037. doi: 10.1542/peds.2005-1516 [DOI] [PubMed] [Google Scholar]
- Thomas P, Zahorodny W, Peng B, Kim S, Jani N, Halperin W, & Brimacombe M (2012). The association of autism diagnosis with socioeconomic status. Autism, 16(2), 201–213. doi: 10.1177/1362361311413397 [DOI] [PubMed] [Google Scholar]
- US Census. (2016). 2016: ACS 5-Year Estimates Subject Tables - Median Income in the Past 12 Months (In 2016 Inflation-Adjusted Dollars. In: Census.gov. [Google Scholar]
- Van Naarden Braun K, Christensen D, Doernberg N, Schieve L, Rice C, Wiggins L, … Yeargin-Allsopp M (2015). Trends in the prevalence of autism spectrum disorder, cerebral palsy, hearing loss, intellectual disability, and vision impairment, metropolitan atlanta, 1991-2010. PLoS One, 10(4), e0124120. doi: 10.1371/journal.pone.0124120 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Winter AS, Fountain C, Cheslack-Postava K, & Bearman PS (2020). The social patterning of autism diagnoses reversed in California between 1992 and 2018. Proc Natl Acad Sci U S A, 117(48), 30295–30302. doi: 10.1073/pnas.2015762117 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zablotsky B, Black LI, Maenner MJ, Schieve LA, & Blumberg SJ (2015). Estimated Prevalence of Autism and Other Developmental Disabilities Following Questionnaire Changes in the 2014 National Health Interview Survey. Natl Health Stat Report(87), 1–20. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.