Water Lead Levels of Private Domestic Well-Supplied Homes in Three Rural Illinois Counties

Background: In the wake of Flint, MI, there has been increased focus on impact of the corrosion of old service lines, premise piping, and solder on water lead levels (WLLs), and their contributions to population and community-wide lead intake. However, there has been less focus on the WLLs resulting from groundwater corrosion of private domestic wells (PDWs). Previous studies in Virginia, Pennsylvania, Wisconsin, and New York have uncovered elevated water lead levels, but have failed to characterize corrosivity. Research Question: To evaluate corrosivity and housing age as predictors of WLLs in PDW-supplied homes in rural Illinois, and develop a decision rule for local health department (LHD) testing prioritization. Design: Cross-sectional Setting: Jackson, Kane, and Peoria County, Illinois. Methods: We teamed up with LHDs to collect stagnant and flushed samples from households using PDWs. Year of home construction was determined by residents, and samples were analyzed for lead and corrosivity-related parameters (chloride, sulfate, and alkalinity). We conducted stratified bivariate analysis and logistic regression modelling to characterize relationships between WLLs, corrosivity (Larson Skold Index (LSI), Chloride to Sulfate Mass Ratio (CSMR)), and age of housing. Results: About 47% (n=97) of our first draw (stagnant) samples had detectable levels of lead, with only 3.1% exceeding EPA regulatory levels of 15 µg/L. LSI values did not vary significantly by county, while the CSMR was significantly higher in Kane than Jackson County (p<0.0001). We found significant differences in median WLLs of homes built before and after 1986 (before: 1.11, after: 0.54, p=0.01). There were statistically significant associations between LSI empirical cutoffs and lead detection and exceedance of 1.57 µg/L (67th percentile). Logistic regression modelling revealed stratum specific significant effects of lead detection (OR: 4.13, 95% CI: 1.02, 16.68) in the presence of pre-1986 homes and LSI values exceeding 0.59 (the 75th percentile). The modelled decision rule had low sensitivity (31.0%). Discussion and Conclusion: Limitations to this study include low generalizability, reliance on participant sampling and recall, corrosivity metrics limitations, and lack of direct plumbing/well inspections. Our decision rule was of limited utility, and LHDs should develop well and plumbing inspection expertise to compliment it.