812 - Trends in primary care provider autism diagnosis among U.S. children from 2013 to 2023: Effects of child age, race and ethnicity, and poverty level

Publication Number: 812.6827

Vincent Sutherland, Northern Arizona University, Flagstaff, AZ, United States; Veronica I. Underwood Carrasco, Oregon Health & Science University, Portland, OR, United States; David C. Folch, Northern Arizona University, Flagstaff, AZ, United States; Katharine Zuckerman, Oregon Health & Science University School of Medicine, Portland, OR, United States; Olivia Lindly, Northern Arizona University, Flagstaff, AZ, United States

Background: Timely autism diagnosis by a trained specialist is typically difficult and lengthy, disproportionately affecting under-resourced families in the US. Historically, primary care providers (PCPs) have not received training to diagnose autism, though new training initiatives are increasing.

Objective: Identify interactions of age, race-ethnicity, and income over time with the likelihood of PCP autism diagnosis among US children ages 3-17 from 2013-2023.

Design/Methods: This was a secondary analysis of 2016-2023 National Survey of Children’s Health data. Diagnosis year was constructed from child age difference between time of survey and autism diagnosis. We only considered diagnosis after 2012 (2013-2023) to account for small cell sizes and changes in policy and practice (e.g., DSM-V diagnostic criteria), and to reduce recall bias. Diagnosing provider types were collapsed into a binary response variable of PCP vs non-PCP (Figure 1). Diagnosis year, diagnosis age, sex, race-ethnicity, language, income, confounding conditions, and functional impairment were covariates. General Estimating Equations considered interaction terms between diagnosis year and the other explanatory variables. A linear spline with knot=2018 was used to better fit the interaction terms. Nonsignificant explanatory variables were removed.

Results: Overall, lower household income was associated with a higher probability of autism diagnosis by a PCP vs non-PCP. The final model results (Table 2) show several sign reversals for interaction terms around 2018. The interaction term for lowest household income shows the likelihood of PCP diagnosis decreased pre-2018 but then increased post-2018. Similarly, the interaction term for Hispanic vs white non-Hispanic children showed a decreased likelihood of PCP diagnosis pre-2018 but then increased post-2018. The probability of PCP diagnosis decreased overall for older children.

Conclusion(s): PCP autism diagnosis increased among US children in the past decade. Children with lower household income, who are younger, or who are Hispanic were most likely to receive PCP autism diagnosis, with the greatest increases occurring after 2018. PCP autism diagnosis training initiatives may help to reduce persistent autism diagnosis inequities.

Figure 1. Weighted Proportions of Autism Diagnosis by Primary Care Provider vs Other Provider
Note. Other Providers: specialist, school psychologist/counselor, non-school psychologist, psychiatrist, and other providers as labeled in the National Survey of Children’s Health.

Table 1. Primary Care Provider General Estimating Equations Model Summary