A new statewide estimate finds that 2.65 percent of Pennsylvania youth ages 3 to 21 — about 1 in 38, or 78,207 individuals — were identified with autism spectrum disorder in linked state administrative records for 2022. The figure comes from a study published in Public Health in Practice by researchers affiliated with a new Autism and Developmental Disabilities Monitoring (ADDM) Network site in Pennsylvania, who integrated the state’s special education records with Medicaid claims to produce prevalence estimates that span childhood, adolescence, and the transition to adulthood.
The ADDM Network has tracked autism prevalence in the United States since 2000, traditionally relying on intensive manual record review at specific geographic sites and focusing on children at key ages of 4 and 8 years. The latest national ADDM estimate, based on 8-year-olds in surveillance areas, is 3.22 percent, or 1 in 31. The Pennsylvania study extends that approach in a different direction: instead of abstracting records by hand, it used exclusively secondary administrative data that states already collect, allowing prevalence to be calculated statewide and at the county level across a much wider age range with far fewer resources.
To build the estimate, the research team linked administrative data from Pennsylvania’s state education system and Medicaid records for all youth ages 3 to 21 enrolled in public schools and the state Medicaid program as of December 2022. That date represents a point-in-time enrollment record captured through Child Find, a federal mandate under the Individuals with Disabilities Education Act requiring the identification of children age 21 and younger with disabilities who receive special education services. Records were matched across the two systems on first name, last name, and date of birth; a sensitivity analysis relaxing the criteria to any two of the three elements showed that more than 98 percent of records matching on either name and date of birth also matched on all three, supporting the validity of the matching strategy.
A child was counted as an autism or intellectual disability case if they met the case definition in either data source. In the education data, classification required a primary or secondary special education designation of autism or intellectual disability. In the Medicaid data, an autism case required at least one inpatient claim or two outpatient claims with an ICD-10 code in the F84 range during 2021 and 2022, with corresponding codes used for intellectual disability. The final analytic sample totaled 78,207 youth, of whom 20.16 percent of those in the education data were matched to Medicaid records and 59.03 percent of those in Medicaid data were matched to education records. Population denominators came from 2022 US Census estimates, supplemented by Surveillance, Epidemiology, and End Results (SEER) Program data where breakdowns by single year of age and race were needed, and 95 percent confidence intervals were calculated with the Wilson score method.
The statewide pattern varied sharply by race and ethnicity. The highest prevalence was observed among youth identified as Multiracial (4.17 percent; 95 percent CI 4.06 to 4.28), Hispanic (3.34 percent; 95 percent CI 3.28 to 3.39), and Black (3.16 percent; 95 percent CI 3.10 to 3.22), compared with American Indian/Alaska Native (2.44 percent), Asian/Pacific Islander (2.26 percent), and White (2.26 percent) youth. Boys accounted for 78 percent of youth with autism statewide, with a male prevalence of 4.04 percent versus 1.19 percent among girls. Across the 65 of Pennsylvania’s 67 counties with non-zero prevalence, estimates ranged from 1.75 percent to 6.00 percent, with a median county-level prevalence of 3.23 percent.
Age produced the most striking gradient. Statewide prevalence peaked among 6- and 7-year-olds, reaching 3.66 percent (95 percent CI 3.57 to 3.76) among 7-year-olds, and fell to its lowest level, 0.77 percent (95 percent CI 0.73 to 0.82), among 21-year-olds. The authors note that the reasons for lower prevalence among transition-age youth ages 19 to 21 could be multifaceted: educational supports end by age 22 in Pennsylvania, cohort effects may play a role because broader diagnostic criteria were not in place during the childhood years of people now in that age group, and differences in access to services such as Early Intervention may have affected identification rates. The study cannot distinguish among these explanations, and the authors frame them as possibilities rather than established causes.
Where cases were identified also differed by age, reflecting how children enter different service systems. Statewide, across ages 3 to 21, autism cases were identified in special education records for 54,075 cases, or 69.1 percent, and in Medicaid claims for 50,388 cases, or 64.4 percent. Among 4-year-olds, most cases appeared in Medicaid data (3,254, or 80.8 percent) compared with special education records (2,299, or 57.1 percent), while among 8-year-olds the pattern reversed, with most cases identified in special education records (3,800, or 73.7 percent) compared with Medicaid (3,310, or 64.2 percent). The authors connect this shift to Pennsylvania’s funding structure, in which Medicaid funds Early Intervention services until age 3, after which funding transitions to a specialized office that shares operations across the state’s Departments of Human Services and Education.
The study also reports estimates for the suburban county containing the new ADDM Network surveillance site, where administrative prevalence for ages 3 to 21 was 2.55 percent, or 1 in 39, with 79.5 percent of identified youth male. In that county, prevalence was highest among Black (2.95 percent) and Hispanic (2.44 percent) youth compared with Asian/Pacific Islander (2.33 percent) and White (1.99 percent) youth. The surveillance county’s administrative prevalence of 3.63 percent for its tracked ages fell in the mid-range of Pennsylvania counties, with 12 counties showing prevalence of 2 percent or lower and 15 counties at 4 percent or higher; 14 counties were suppressed due to low counts. Statewide, 12.00 percent of youth with autism identified in administrative records had co-occurring intellectual disability, as did 10.3 percent in the surveillance county.
Comparisons with ADDM site surveillance reveal systematic differences between the two methods. The CDC ADDM Network reported 2022 Pennsylvania site estimates of 4.74 percent among 8-year-olds and 4.27 percent among 4-year-olds, whereas the administrative method yielded lower figures of 3.49 percent and 2.86 percent statewide, and 3.61 percent and 3.15 percent for the surveillance site county. The authors suggest these differences in case identification across systems are substantive and important for interpreting administrative prevalence, since states differ in how their special education, Medicaid, and other public programs are funded and structured, and in the age of entry and coverage for autism assessments and services.
The study carries important limitations that the authors describe candidly. Administrative claims are coded by providers for billing purposes, so condition misclassification and variation in diagnoses may be present, and underreporting in claims data creates potential selection bias. About half of Pennsylvania youth have private insurance, and diagnoses and services covered by private insurance were not included, which may produce an underestimate of community prevalence; conversely, because the linked dataset may capture a sample more likely to seek services, the estimates could also overestimate prevalence. Pennsylvania’s policies allowing youth to enroll in Medicaid regardless of parent income if they meet disability requirements also make its Medicaid enrollment higher than in other states. The authors therefore caution that the estimates may not be generalizable to Pennsylvania’s population as a whole, and low county populations and suppressed counts limit some comparisons.
Despite those caveats, the authors argue that linked, cross-system administrative data can supplement more resource-intensive surveillance processes to estimate autism prevalence statewide and at the county level, and can highlight resource allocation priorities and reveal disparities by race, age, and geography. The increase to 78,207 identified youth ages 3 to 21 indicates that this total has grown, they write, and warrants implementing a lifespan approach to estimating administrative prevalence to examine differences over time, including the impact of the COVID-19 pandemic on service access and use. Pennsylvania is among the first states with adult-focused Medicaid programs for people age 21 and older with autism that do not require co-occurring mental health or intellectual disability diagnoses, but waiting lists for such programs have existed for decades, and the authors suggest that Medicaid programs supporting adults with mental health or intellectual disability diagnoses may need to be expanded in tandem with autism-specific programs as the growing number of autistic individuals ages into adulthood.
Subject of Research: Statewide administrative estimation of autism spectrum disorder prevalence in Pennsylvania using linked education and Medicaid data
Article Title: Statewide administrative autism spectrum disorder prevalence estimates: Evidence from a new Autism and Developmental Disabilities Monitoring (ADDM) Network site
Article References: Sadowsky, M., Wright, J., Ventimiglia, J., Rast, J., Mandell, D. S., Alford, A., Salett, B., & Shea, L. (2026). Statewide administrative autism spectrum disorder prevalence estimates: Evidence from a new Autism and Developmental Disabilities Monitoring (ADDM) Network site. Public Health in Practice, 12, Article 100875. https://doi.org/10.1016/j.puhip.2026.100875
Image Credits: AI Generated
DOI: 10.1016/j.puhip.2026.100875
Keywords: autism spectrum disorder, ADDM Network, administrative prevalence, Pennsylvania, Medicaid, special education, surveillance, health disparities, county-level estimates, transition-age youth, intellectual disability, public health planning
News Source: Ophelia Keating. (October 28, 2026). Linked Administrative Data Put Autism Prevalence in Pennsylvania at 1 in 38. Scienmag.
