Geographic influences on autism diagnosis: Accessibility of health services and exposure to hazardous air pollutants

We evaluated two geographic determinants of autism: early-life geographic access to health services on the timing of autism diagnosis and perinatal exposure to hazardous air pollutants (HAPs) on the prevalence of autism. We included children identified by the Autism and Developmental Disabilities Monitoring Network records-based surveillance who resided in the surveillance regions at age 8 and at birth (in 1992, 1994, 1996).;Geographic access to health services was measured as individual-level density of and proximity to providers and facilities. We included 206 children from central North Carolina meeting a standardized case definition for autism in a Cox proportional hazards model adjusted for family social class. We found positive associations as shown by adjusted hazard ratios and 95% confidence limits (CL) between earlier age at diagnosis and better accessibility to some health services: higher density of psychiatrists and neurologists: 1.2 (0.8, 1.7), residing < 20 miles to a teaching hospital: 1.3 (0.9, 1.9) and the Infant-Toddler Early Intervention Program: 1.4 (0.8, 2.3), and residence in a non-Health Professional Shortage Area: 1.4 (0.8, 2.5). Greater provider density may support earlier autism recognition.;Exposures to ambient concentrations of 35 metal, particulate, and volatile organic HAPs in the census tract of the child's birth residence were estimated from the 1996 National Air Toxics Assessment annual-average model. We compared HAP exposure among 383 children with autism and 2829 children with speech and language impairment from central North Carolina and West Virginia. Semi-Bayes logistic models included surveillance year, state, a priori confounders from the birth certificate and census (maternal education, age, smoking in pregnancy, race, marital status, census median household income, urbanicity), and covarying HAPs. We estimated mostly near-null associations shown by odds ratios (CLs), positive associations for butadiene: 1.5 (0.7, 3.3) for 79 vs. 18 ng/m3, dichloropropene: 1.9 (0.8, 4.8) for 71 vs. 14 ng/m3, proprionaldehyde: 1.5 (0.6, 3.6) for 208 vs. 56 ng/m3, quinoline: 1.4 (1.0, 2.2) for 0.0051 vs. 0.0001 ng/m3, and styrene: 1.8 (1.0, 3.1) for 41 vs. 5 ng/m3, and negative associations for lead and toluene. Our results support a role for some air pollutants in autism etiology.