Epidemiologic Designs for Studying Environmental Factors in Autism

Objective: This presentation introduces the various types of epidemiologic study designs that can and have been used in autism research, what we have learned from them, and where the field is heading.  The focus is on investigation of environmental factors and the underlying theme is the relevance of neuro-immunomodulatory mechanisms.

Methods: Several designs have been used for epidemiologic research aimed at identification of the environmental determinants of autism.  These determinants are either causes or proxies for causes, i.e., factors that contribute to the incidence or severity of autism.  The principal designs, their implementation, and their advantages and disadvantages will be discussed, including the ecological study, cohort study and case-control study.  Pitfalls in using ecological correlation analyses will be described. 

Results: Some retrospective cohort studies have examined existing databases such as medical records or national registries to document both the cases and the exposures of interest.  Case-control studies using exposure databases for pesticides or air pollutants have also been employed, as well as population-based specimen banks.  Finally, comprehensive case-control studies, such as the CHARGE and SEED studies in which cases are confirmed using standardized instruments, exposure is collected systematically for research purposes, and specimens are obtained for biologic measurements have begun to produce intriguing findings with regard to immune dysregulation and/or pesticides in autism in several studies. 

Conclusion: Major weaknesses in much of the existing epidemiologic literature on environmental causes of autism include: reliance on ecological correlation analysis, lack of standardized confirmatory testing of cases, and error-prone exposure assessment methods.  Large-scale epidemiologic studies hold the greatest promise for identification of environmental determinants as well as gene-environment interactions.  These prototypic studies are characterized by individual exposure assessment and outcome ascertainment, linked to molecular and other mechanism-oriented methods for analysis of biologic specimens from population-based participants.