Exposure characterization of soil lead and blood lead in an urban border environment: The spatial analysis of data from multiple sampling strata

Blood lead and soil lead data collected under the USEPA/UCI Tijuana Lead Project (1996–1999) was made available for this study to evaluate soil lead sampling from multiple strata (background, schools and parks, point sources, and residential). Geographic information systems were utilized to determine whether the use of multiple strata in sampling improves upon exposure characterization focused upon either residential or point source sampling (traditionally investigated at one level using one stratum). The Bocco and Sanchez model of relative risk by proximity to fixed point sources was tested and validated with USEPA/UCI study data. Associations between the independent variables and to the dependent variable were assessed at the individual, census tract, and regional levels.; Comparison of nonceramic users' and ceramic users' data revealed an increase in the percentage of subjects with blood lead levels ≥10 ug/dl (pe 10) from 7.3% to 18.8%, or an average 1.4 ug/dl increase. Among nonceramic-using children aged 1.5 to 6.9 years, 20,440 are predicted to have blood lead greater than or equal to 10 ug/dl in Tijuana. Individual level data revealed a significant association between dust lead and soil lead (r = 0.70; p < 0.01). At the regional level, significant relationships were found between mean blood lead level and fixed point source soil lead (R2 = 0.72 using a quadratic model) and between residential soil lead and fixed point source soil lead (r = 0.90; p < 0.05). Significant differences were observed between weighted residential soil lead means across regions (p = 0.02). Results of the Bocco and Sanchez model were tested by Chi-Square analysis with no significant difference at the p = 0.02 level, showing that the model was strikingly accurate in predicting the distribution of subjects with blood lead levels ≥10 ug/dl.; Results show that at the regional level, point source soil lead is associated with mean blood lead levels and that concurrent high background and point source soil lead levels may be predictive of a high pe 10. This study revealed that industrial emissions have been a contributing factor to soil lead contamination in Tijuana. Results do not support the expectation that multiple strata sampling will reduce variability in predicting the outcome variable at the individual level. Nevertheless, these findings will be useful in defining regions for targeted or universal soil lead sampling by site type.; It is concluded that when apparently undisturbed sites reveal high soil lead levels in addition to local point sources, dispersion of lead is widespread and will be associated with a high prevalence of elevated blood lead in children. Study findings suggest the utility of multiple strata sampling was not functional in apportioning the relative contribution of each of the strata in a multivariate model. However, multiple strata sampling was shown to be useful in differentiating among sources by site-specific association to mean blood lead and the prevalence of elevated blood lead at the regional level.