Housing-based lead paint dust is the most common source of lead exposure for US-born children. Although year of housing construction is a critical indicator of the lead hazard to US children, not all housing of the same age poses the same risk to children. Additional information about housing condition is required to differentiate the housing-based lead risk at the parcel level. This study aimed to identify and assess a method for gathering and using observations of exterior housing conditions to identify active housing-based lead hazards at the parcel level. We used a dataset of pediatric blood lead observations (sample years 2000-2013, ages 6-72 months, n = 6,589) to assess associations between observations of exterior housing conditions and housing-based lead risk. We used graphical and Lasso regression methods to estimate the likelihood of an elevated blood lead observation (≥3.5 μg/dL). Our methods estimate a monotonic increase in the likelihood of an elevated blood lead observation as housing conditions deteriorate with the largest changes associated with homes in the greatest disrepair. Additionally we estimate that age of home construction works in consort with housing conditions to amplify risks among those houses built before 1952. Our analysis indicates that a survey of external housing conditions can be used in combination with age of housing in the identification process, at the parcel level, of homes that pose a housing-based lead hazard to children.
Pediatric lead poisoning remains a persistent public health problem. Children in the US spend the preponderance of their time at home; thus, housing is an important social determinant of health. Improving health outcomes derived from housing-based sources involves differentiating the risks posed by the existing housing stock. In this paper, we developed a parcel-level lead risk index (LRI) based on external housing conditions and the year of home construction. The purpose of this study was to introduce a housing-based lead risk index (LRI), developed using retrospective data, to estimate parcel-by-parcel variation in housing-based lead risk. We described how the LRI is constructed, relate it to the likelihood of a pediatric occupant’s blood lead level (BLL) > 3.5 µg/dL using Lasso regression (n = 6589), visualized this relationship graphically, and mapped the outcome. We found that mapping the LRI provided more information at a more precise geographic level than was possible using other public health surveillance methods.