Autism and Distribution of Hazardous Air Pollutants at Birth in Southern California

Background: The dramatic increase in the number of children reported with autistic spectrum disorders (ASD) has re-focused the search for causes and preventive measures.  Non-genetic factors likely explain some of the increase, leading to an interest in environmental exposures.  In a previous study, we linked our autism surveillance system for 1994 births in the San Francisco Bay Area to estimated hazardous air pollutant (HAP) concentrations compiled by the U.S. EPA. We found elevated adjusted odds ratios between ASD and higher concentrations at the birth address of several HAPs, including metals and chlorinated solvents.
Objectives: Investigate environmental risk factors for ASD by substantiating prior results in another geographic area of California and another time period.
Methods: Our target population was the nearly 700,000 singleton births occurring during 1996-98 in the South Coast Air Basin, in which we ascertained 3,386 ASD cases from the statewide Developmental Disabilities Services (DDS) agency. A 5% sample of non-case births that survived to age one were selected as controls (n=30,906), frequency-matched to cases by date of last menstrual period (LMP). The residence at birth was geo-coded and the census tract assigned for linkage to the U.S. EPA 1996 HAPs database. We identified 24 chemicals as potential neurotoxicants, developmental toxicants, and/or endocrine disruptors, as well as focusing on the six prior associated HAPs. Because concentrations of many HAPs were highly correlated, we also combined the chemicals into mechanistic and structural groups, calculating summary index scores. These group scores or individual chemical concentrations were categorized as quartiles and odds ratios (OR) calculated for ASD, adjusting for demographic factors from the birth certificate.
Results: Comparing mean concentrations, cases had slightly higher means than controls for mercury, most aromatic solvents, vinyl chloride, diesel PM, hexane, PAHs and a few other compounds, but lower mean cadmium and manganese means. Focusing on the six compounds of primary interest, most quartile ORs were significantly elevated (1.4-2.1) compared to the lowest quartile, except for cadmium. However, there was little dose-response pattern by quartile. Adjustment for demographics had little effect on the ORs, but including the regional center of case ascertainment reduced them (1.1-1.3), although still significant. Examining the chemical groups, metals showed little association with ASD, whereas the quartile ORs were slightly elevated for aromatic solvents.
Conclusions: Examining HAPs in another region of California with higher concentrations and larger numbers did not replicate the original results. Metals as a group were not associated, but instead aromatic solvents yielded the highest risks. Specific chemicals, including mercury and nickel, were associated with increased risk of ASD in a threshold type pattern. The magnitude of the associations were reduced when controlling for regional center, a proxy for likelihood of inclusion in DDS, but also geographically based. Examination of HAPs and autism in other areas of the country and with more recent births may shed further light on the potential associations. Studies with biomarkers of environmental exposure would be ideal, but are difficult to conduct unless peri-natal specimens are archived.