Variability of household-water lead levels in American cities. Final draft report

The authors estimated the most significant sources of variability in repeat samples of water lead concentrations from household taps and service lines in six U.S. cities: Bennington, VT; Boston, MA; Chicago, IL; New Bedford, MA; Newport News, VA; and Seattle, WA. The Boston, Bennington, and Seattle samples include both baseline samples and samples after pH and alkalinity adjustment (treatment) has been carried out. The most significant sources of variability included: differences in household plumbing, differences in hardness or corrosivity of household water as indexed by pH and alkalinity, and seasonal effects (possibly related to water temperature). Repeat sampling variability was often large (lognormal coefficient of variation greater than 0.58) or very large (lognormal coefficient of variation greater than one). In Boston, estimated 90th percentiles decreased by a factor of 1.5 within four months after treatment began, and by a factor of 3.4 after 3 years.