In their article, Waldman et al1 present a novel hypothesis that examines the association between autism and environmental factors using ecologic data from California, Oregon, and Washington. Their findings suggest a positive association between mean annual precipitation and autism prevalence at the county level. Although the authors address confounding, we believe they should have better addressed the potential effect of confounding by county-level urbanicity and the potential for cross-level bias (ecologic fallacy) when interpreting their results.
Braun JM, Kalkbrenner A. Autism Prevalence and Precipitation: The Potential for Cross-Level Bias. Arch Pediatr Adolesc Med. 2009;163(5):492-493. doi:10.1001/archpediatrics.2009.83