Spatial Surveillance of Childhood Lead Exposure in a Targeted Screening State: An Application of Generalized Additive Models in Denver, Colorado.

Context: The targeted nature of Colorado's childhood lead screening program presents several analytical issues that complicate routine epidemiologic surveillance.

Objectives: To analyze spatial patterns of childhood lead exposure among children younger than 6 years, identifying areas of increased risk along with associated covariates.

Methods: We analyzed a spatial case-control data set of childhood lead poisoning using generalized additive models. Incident cases were represented by the residential locations of children younger than 6 years with confirmed elevated blood lead levels (EBLL) of 5 μg/dL or more recorded between calendar years 2010 and 2014, and controls were sampled from the population at risk. We modeled the effect of spatial location, adjusting for potential spatial confounders. We also adjusted for a number of covariates previously identified in the childhood lead poisoning literature to understand the ecologic-level drivers of spatial variation in risk.

Main Outcome Measure(s): Crude and adjusted spatial odds ratios describing the relative frequency of lead poisoning among different locations in Denver, Colorado.

Results: We found evidence of statistically significant spatial clustering in incident cases of lead poisoning even after adjustment for age, sex, year, season, and spatially smoothed screening rate. Spatial confounder-adjusted odds ratios in the Denver study area ranged from 0.22 to 2.7. Adjusting for additional ecologic-level covariates effectively accounted for the observed spatial variation. We found that ecologic-level indicators of low socioeconomic status, Hispanic ethnicity, Asian race, and older housing age were all positively and significantly associated with an increased EBLL risk.

Conclusion: Housing and socioeconomic factors continue to be the primary ecologic risk factors associated with childhood lead exposure and can be used to predict risk at a fine spatial resolution in the Denver study area. Our analysis demonstrates how other targeted screening states can be proactive about childhood lead surveillance within their major population centers and enhance the spatial specificity of lead mitigation efforts.