Objectives: The purpose of this study was to model spatio-temporal patterns of risk for ASDs in three birth cohorts in a three county surveillance region in Utah. The following objectives were addressed: 1) identify areas of significantly heightened relative risk, 2) determine the temporal persistence of areas of heightened relative risk across birth cohorts, and 3) describe changes in the size and geometry of relative risk hotspots across years.
Methods: Using an administrative multisource record review methodology, ASDs (n = 590) were identified in eight-year-old children born in 1994, 1998, and 2000 and residing in a three county surveillance region by the Utah Registry of Autism and Developmental Disabilities (URADD). The control population (n = 10,534) was comprised of a gender-matched, random selection of children from these birth cohort’s birth certificates. Maternal residential birth addresses of cases and controls were geocoded by the Utah Department of Health and then linked to URADD. Spatio-temporal relative risk was modeled using adaptive kernel-smoothed relative risk functions (Kelsall and Diggle 1995), and areas of heightened relative risk (or relative risk hotspots) were identified using asymptotic normality approximations (Davies and Hazelton 2010).
Results: Heightened areas of spatial relative risk were identified in all three birth cohorts. Relative risk varied from 0.5 to 2.4 throughout the surveillance region. Three areas of heightened relative risk were observed in the 1994 and 2000 birth cohorts. Four areas of heightened relative risk were identified in the 1998 birth cohort. Two areas of heightened relative risk persisted across the three study years. The temporally stable relative risk hotspots covered a larger area of the surveillance region in 2002 than in 2006 but increased in size again in 2008.
Conclusions: Both temporally stable and ephemeral areas of heightened relative risk for ASD were identified. In these hotspots, children were at 1.4-2.4 times greater risk for ASD than children residing elsewhere in the surveillance region. The existence of temporally stable and single-year relative risk hotspots indicates that both long-term and short-term effects influence the spatial pattern of relative risk in Utah. The next step will involve statistical modeling to elucidate the potential variables responsible for these different spatio-temporal patterns including neighborhood effects, locally distributed environmental or biological risk factors, social or familial effects, ascertainment bias, and/or sociodemographic variables.
See more of: Epidemiology
See more of: Prevalence, Risk factors & Intervention