The Relationship Between Autism Spectrum Disorders and the Distribution of Hazardous Air Pollutants in Four New Jersey Counties

Background:  A California study reported an association between the diagnosis of autism spectrum disorder (ASD) and residence in regions with elevated levels of air pollutants. Since 1980, the reported prevalence of ASD has risen in the U.S. by a factor of 10, with the highest prevalence in New Jersey (NJ). This increase has been attributed both to increased awareness and diagnosis and a higher incidence due to environmental factors.

Objectives:  Using methods from the California study, this study examined the association of prior exposure to air toxics with an ASD diagnosis by age 8 in four NJ counties: Essex, Hudson, Ocean and Union.  We characterized cases of ASD, delineated confounding variables, and assessed association.

Methods: We obtained mean annual concentrations of 37 air toxicants by census tract in the four counties from the 1996 EPA online NATA database.  Hazardous Air Pollutants (HAPs)were divided into 5 independent groups based on their structure: metals, aromatic and chlorinated solvents, chemicals of interest (e.g. PAHs) and other chemicals.  We also divided the chemicals by their mechanistic effects into 7 groups: reproductive, developmental, neurologic, immunologic and endocrine toxicants, carcinogens and those used for pesticides.  We acquired data on 8-year-old ASD cases (born in 1992) from the NJ Autism and Developmental Disabilities Monitoring Network.  NJ Department of Health and Senior Services provided birth certificate data for the cases and 4 matched controls (birth month and gender) per case for linkage to the exposure data at the census tract level. Using a 4:1 matched case control study design we assessed the strength of association of exposure with case/control status. 

Results: When considering the entire analytical sample, matched regression analyses yielded no association of case/control status with exposure. However, when the case/control sets were stratified by expert estimation of of clinical impairment clinical into 3 levels of severity, the composition of each stratum was strikingly different both demographically e.g.  male/female ratios: mild - 2.5:1, moderate - 4.5:1 and severe - 7:1 ,  and clinically  e.g. almost half of cases in the severe stratum were diagnosed with regression and fewer co-diagnosed with psychological disorders. In the structural groups, the Adjusted Odds Ratios (AORs) for the severe ASD cases were significantly elevated and associated with the top 25% of concentrations of aromatic (4.22) and chlorinated (3.82) solvents, PAHs (5.40) and all other chemicals (9.30). When examined mechanistically, AORs in the severe stratum were elevated for endocrine disruptors (4.08), reproductive (4.29) and neuro (4.63) -toxicants, carcinogens (3.93) and pesticides (5.56) at the same concentration.

Conclusions: Results of this study suggest an increased risk of diagnosis with clinically severe ASD in NJ counties with high ambient levels of HAPs, emphasizing the value of stratification in the ASD population.  These results also may explain why previous studies that have not stratified subjects by severity of impairment report widely varying results.