Application of health-based screening levels to ground-water quality data in a state-scale pilot effort

Abstract

A state-scale pilot effort was conducted to evaluate a Health-Based Screening Level (HBSL) approach developed for communicating findings from the U.S. Geological Survey (USGS) National Water-Quality Assessment Program in a human-health context. Many aquifers sampled by USGS are used as drinking-water sources, and water-quality conditions historically have been assessed by comparing measured contaminant concentrations to established drinking-water standards and guidelines. Because drinking-water standards and guidelines do not exist for many analyzed contaminants, HBSL values were developed collaboratively by the USGS, U.S. Environmental Protection Agency (USEPA), New Jersey Department of Environmental Protection, and Oregon Health & Science University, using USEPA toxicity values and USEPA Office of Water methodologies. The main objective of this report is to demonstrate the use of HBSL approach as a tool for communicating water-quality data in a human-health context by conducting a retrospective analysis of ground-water quality data from New Jersey. Another important objective is to provide guidance on the use and interpretation of HBSL values and other human-health benchmarks in the analyses of water-quality data in a human-health context. Ground-water samples collected during 1996-98 from 30 public-supply, 82 domestic, and 108 monitoring wells were analyzed for 97 pesticides and 85 volatile organic compounds (VOCs). The occurrence of individual pesticides and VOCs was evaluated in a human-health context by calculating Benchmark Quotients (BQs), defined as ratios of measured concentrations of regulated compounds (that is, compounds with Federal or state drinking-water standards) to Maximum Contaminant Level (MCL) values and ratios of measured concentrations of unregulated compounds to HBSL values. Contaminants were identified as being of potential human-health concern if maximum detected concentrations were within a factor of 10 of the associated MCL or HBSL (that is, maximum BQ value (BQmax) greater than or equal to 0.1) in any well type (public supply, domestic, monitoring). Most (57 of 77) pesticides and VOCs with human-health benchmarks were detected at concentrations well below these levels (BQmax less than 0.1) for all three well types; however, BQmax values ranged from 0.1 to 3,000 for 6 pesticides and 14 VOCs. Of these 20 contaminants, one pesticide (dieldrin) and three VOCs (1,2-dibromoethane, tetrachloroethylene, and trichloroethylene) both (1) were measured at concentrations that met or exceeded MCL or HBSL values, and (2) were detected in more than 10 percent of samples collected from raw ground water used as sources of drinking water (public-supply and (or) domestic wells) and, therefore, are particularly relevant to human health. The occurrence of multiple pesticides and VOCs in individual wells also was evaluated in a human-health context because at least 53 different contaminants were detected in each of the three well types. To assess the relative human-health importance of the occurrence of multiple contaminants in different wells, the BQ values for all contaminants in a given well were summed. The median ratio of the maximum BQ to the sum of all BQ values for each well ranged from 0.83 to 0.93 for all well types, indicating that the maximum BQ makes up the majority of the sum for most wells. Maximum and summed BQ values were statistically greater for individual public-supply wells than for individual domestic and monitoring wells. The HBSL approach is an effective tool for placing water-quality data in a human-health context. For 79 of the 182 compounds analyzed in this study, no USEPA drinking-water standards or guidelines exist, but new HBSL values were calculated for 39 of these 79 compounds. The new HBSL values increased the number of detected pesticides and VOCs with human-health benchmarks from 65 to 77 (of 97 detected compounds), thereby expanding the basis for interpreting contaminant-occu