![[ Visit Client Website ]](images/banner.jpg)
BACKGROUND: The majority of Utah’s population lives in the Wasatch Front, over a quarter of which is less than 15 years of age. As the Wasatch Front population grows, it is important to understand health in relation to the built environment.
METHODS: Using data-linkage techniques, the strengths and weaknesses of three methods in evaluating childhood leukemia (<15 years old on diagnosis) and proximity to roadways and gas stations were assessed. The buffer method analyzed exposures within a specified radius around each case, the distance method analyzed the distances between cases and exposure locations, and the last method was an analysis by census area. All of the models used Poisson regression controlling for potential confounders and weighting factors, including annual average daily traffic, the number of pumps at the gas stations, age, sex, race, and ethnicity. Additional potential confounders as identified in the literature included birth weight, mother’s age, smoking, and Down syndrome prevalence by category for each census area.
RESULTS: The methods were compared based on several factors, including the ease of data-linkage, the number of assumptions required to fit a model, and the goodness of fit to the data. The distance method included areas beyond those usually considered for pollutant exposures, but was straight-forward for modeling. The buffer method captured everything within the exposure radius. The census area analysis was too large and assigned exposures too broadly or narrowly depending on the size of the census area.
CONCLUSIONS: The best method for this linkage study was the distance method with the incorporation of weighting. The technique was the most straightforward and required the fewest assumptions. The buffer method required additional assumptions, such as no exposure beyond the established distance and all exposure occurred at the residence. The polygon method did not work well because some census areas were too big to properly assign exposures based on proximity to gas stations and roadways. Thus, the point-distance method was best at assessing the relationship between roadways and gas stations and risk of childhood leukemia. Strong analysis of methods is crucial for the future use of data-linkage studies.