Note: This information is provided for reference purposes only. Although the information provided here was accurate and current when first created, it is now outdated.

TRAC 5/28-29/98

BACKGROUND:

Exposure assessments for drinking water using statistical methods based on probability (probabilistic) are very valuable in making good risk management decisions. Their value lies in providing accurate predictions of the number of people who are likely to be exposed to different amounts of a pesticide. However, this type of assessment requires reliable data both on how much water people drink (i.e., the distribution of water consumption values) and on the concentrations of a pesticide over time in drinking water in areas where the pesticide is used (i.e., the distribution of pesticide concentrations).

We have appropriate data from USDA on the consumption side; however, we do not have sufficient, appropriate data on pesticide concentrations in drinking water for the vast majority of pesticides. Although we do not have all the data we need for probabilistic assessments, we do know enough to conclude that some pesticides are likely to be present in drinking water in some locations at levels that could be significant from a human health perspective. The overarching public policy issue before us today is what to do now in these cases, in the context of complying with FQPA.

CURRENT APPROACH (Since November 1997)

OPP scientists begin by reviewing substantial amounts of registrant-submitted data. These data tell us whether the pesticide will easily move to groundwater or surface water and whether it will degrade quickly or persist. OPP scientists then use the data to develop screening-level mathematical models to provide rough estimates of pesticide concentrations in highly vulnerable surface water (i.e., a small pond on the edge of a treated field) and groundwater (i.e., shallow aquifers under sandy soils). These rough estimates are used primarily to screen out those pesticides that will not be a drinking water concern. EPA recognizes that it would be better to use a larger body of water (e.g., a small reservoir) rather than the "farm pond" for surface water, and OPP expects to have a reservoir-based model ready for peer review in July.

OPP's next step is to compare the model estimates to human health-based "drinking water levels of concern" (which are arrived at after having first considered all food-related exposures). Some pesticides at this point are "cleared" from a drinking water perspective. The remaining pesticides undergo further analysis (i.e., an evaluation of available monitoring data) to produce more realistic estimates. If there are no monitoring data available (or if the data are not adequate), OPP makes a risk management decision as to the need for groundwater monitoring, surface water monitoring, and interim risk mitigation measures while monitoring is underway.

If the pesticide is not cleared using model estimates, and there are monitoring data, OPP gathers and analyzes water monitoring data and supporting information (e.g., information on how the samples were collected and analyzed, where they were collected, when they were collected, and why they were collected). OPP attempts to fully characterize the range of values reported, the highest values reported, the 95th percentile value, and the mean value. If the data are adequate, a regional-based picture of the distribution of measurements is completed as well. OPP then considers the results of the analysis of monitoring data, appropriate short-term (for acute effects) and/or longer-term average (for chronic effects or cancer) and selects drinking water concentrations to use in the food and residential exposure analyses to complete the aggregate exposure assessment.

CASE STUDIES:

Availability and quality of drinking water data can be very good to limited. For example, for many pesticides, registrants have submitted substantial amounts of data that describe how the pesticide behaves in the environment, along with initial model-based estimates of pesticide concentrations in highly vulnerable groundwater and surface water, but we lack data on actual concentrations in drinking water. Additional data are available for some pesticides, such as monitoring data from studies that were not specifically designed to look for the pesticide in question. Other studies that provide useful information include controlled field trials involving application of a known amount of pesticide, with measurements made of pesticide concentrations in groundwater over several growing seasons, as well as studies of surface and groundwater in areas where a specific pesticide is used, to find out how much is reaching the water. All these data sets contribute to the level of analysis that is possible.

The corn herbicides are the only group of pesticides for which sufficient data currently are available to conduct probabilistic exposure assessments. For other pesticides, rather than waiting for data and tools to be developed that would allow us to conduct state-of-the-art exposure assessments, we are using the techniques described above so we can include drinking water estimates in human health risk assessments.

QUESTIONS THAT TRAC MIGHT ADDRESS:

1. Should EPA consider another approach as an alternative to its current approach of use the best available data along with existing models and scientific judgment in assessing drinking water exposure? Should there be any limitation for pesticides that are clearly persistent and mobile in the environment?

   Description: Some have suggested that, given the current wide variation in data availability and quality, the Agency should generally defer extensive consideration of drinking water exposure. In such cases, the need for retaining some or all of the FQPA safety factor because of uncertainty in the drinking water exposure also needs to be addressed. In specific cases, where sufficient reliable data are available, the Agency could conduct refined exposure assessments as part of its aggregate exposure analysis.

2. Considering the range of data availability for pesticides, how should OPP select a reasonable high end value (or values) from available valid monitoring data? How would one define a "reasonable high end value"?

   Description: Data show that a particular pesticide does not degrade rapidly in soils and water in certain areas of the country and that it does not bind strongly or readily to soils. Available surface water and groundwater monitoring data are of variable quality. The majority of monitoring data are from studies not designed to specifically look for the pesticide, and we do not know whether it was actually used in these areas. Many of the values (thousands of values) reported in these non-specific monitoring studies are "zeroes" (that is, "below the limits of detection"). Maximum measurements in these general surface water studies range between 0.012 ppb and 17 ppb. Maximum measurements in these general groundwater studies range between 0.17 ppb and 36.6 ppb. There are four groundwater studies and two surface water studies that did specifically look for the pesticide, using controlled field trials. Maximum measurements in these specific surface water studies are 23 ppb and 150 ppb. Maximum measurements in the specific groundwater studies are 3.0 ppb, 9.0 ppb, 51 ppb and 71 ppb.

   Given these data, and a strong scientific basis for believing that water contamination at some levels in some locations is likely, how should EPA factor this information on the potential for drinking water exposure into the tolerance decision? Specifically, how should OPP approach the selection of a value or values for use in the human health dietary risk assessment?

3. In the absence of monitoring data, or where time is very short, as with emergency exemption requests, what should EPA do to address the drinking water issue in the context of complying with FQPA?

   Description: Currently, in the absence of monitoring data, OPP uses model estimates to screen out those pesticides that are not likely to pose any significant "threat" to drinking water. For those that are not cleared, OPP's risk managers make decisions as to the need for surface water and/or groundwater monitoring and/or interim risk mitigation measures while monitoring is conducted.

   Section 18s pose a special problem, in that there are surface water and/or groundwater monitoring data available in some cases--but, given the emergency nature of the action (and the tight time constraints), OPP often lacks the time to collect, thoroughly analyze, and fully consider these data. Given the nature of section 18 decisions (to allow or not to allow) and the tight timeframes, what role should model estimates play in making this decision?

   ---

   TRAC Home | OPP Home | EPA Home | Comments
   Site Map | Search OPP | Search EPA
   

   www.epa.gov/pesticide/trac
   updated May 17, 1998