Classifying Drinking Water Contaminants for Regulatory Consideration (2001)

This appendix provides an overview of two European rule-based schemes for identifying and prioritizing substances (primarily chemicals) that may pose risks to freshwater and marine environments and human health through these aquatic environments. Although neither scheme exclusively addresses drinking water contaminants, they are provided to illustrate how the complex and often contentious task of identifying, ranking, and culling multitudes of substances to much smaller numbers that will receive regulatory and research consideration has recently been approached in Western Europe. For example, they serve as two very clear and relevant examples of how expert judgment is vital and integral to the design, implementation, and validation of these types of prioritization schemes. In a broader capacity, several facets of their design can be compared and contrasted with the chemical prioritization schemes reviewed in the committee’s first report (NRC, 1999a) and the approach recommend in this report for the development of future Drinking Water Contaminant Candidate Lists (CCLs).

Currently 15 countries belong to the European Union, which is becoming increasingly important in its role of environmental protection through the European Commission (EC). Among other requirements, Article 16 of the (1999) European Parliament and Council Directive re-

quires the commission to establish a List of Substances based on their risk to the aquatic environment and to human health through the aquatic environment. To create such a list expediently, the EC decided that a simplified risk-based assessment procedure was needed that would account for the intrinsic health and environmental hazards of substances of concern based on monitoring and modeling data. More specifically, the approach should consider the aquatic ecotoxicity and human toxicity of a substance through various aquatic exposure routes and other related factors that may indicate the possibility of widespread environmental contamination, such as chemical production volume and use patterns.

On this basis, the COMMPS procedure was developed in collaboration with the (German) Fraunhofer Institute for Environmental Chemistry and Ecotoxicology and subsequently accepted by the European Commission to establish the priority list. The current Version 2 of COMMPS is based on an approach that combines automated risk-based ranking with subsequent expert judgment for the final selection of priority substances. That is, the procedure is essentially a series of simplified substance-by-substance risk assessments. (Notably, only chemicals were assessed and included on the first priority list.) The report, Proposal for a European Parliament and Council Decision Establishing the List of Priority Substances in the Field of Water Policy (EC, 2000) summarizes the background, design, development, intended uses, and other related information concerning COMMPS. Further information about the European Commission and COMMPS is available on the Web at http://www.europa.eu.int.

In brief, the automated risk-based assessment results in two different types of ranking lists—one type based on monitored exposure levels and the other on modeled exposure estimates—which are in turn based on production volumes, use patterns, environmental distribution, and biodegradation as input parameters. More specifically, the first use of the COMMPS procedure comprised the following five steps:

1. Selection of candidate substances for the ranking procedure. For this step, a “list-based” approach was used in which the original candidate substances were selected from eight existing official lists and monitoring programs. (The committee notes that this approach is very similar to the approach used by the EPA to develop the draft 1998 CCL; EPA, 1997;NRC, 1999b.)

2. Calculation of exposure scores. In this step, two ranked lists were established for organic chemicals in the aquatic compartment, one

based on surface water monitoring data and the other on modeling data. Further lists were established for pollutants adsorbed by sediments and for metals based exclusively on monitoring data.

3. Calculation of health effect scores. One or more such lists were established for organic pollutants in the aquatic compartment, for sediments, and for metals based on test data.

4. Computation of the risk-based priority index. Ranked lists were calculated by multiplying the exposure and the corresponding health effects index for each substance. Two lists were ultimately developed for organic chemicals based on aquatic monitoring and modeling data, respectively. One list was obtained based on sediment monitoring data, and several lists were obtained for metals.

5. Recommendation of priority substances. For this purpose, a two-step procedure was applied. In the first step, the ranked lists were screened to cull a subset of candidate priority substances from each of the lists. These highly ranked substances were further screened and reorganized based on two criteria: (1) the grouping of substances that normally occur as mixtures and (2) the elimination of candidate substances if their marketing and use are already severely restricted or prohibited in Western Europe (i.e., “historic” pollution). In the second step, expert judgment was used on a substance-by-substance basis to make a final decision on whether a particular candidate priority substance would be included or excluded from the list.

In general, candidate substances selected based on monitoring exposure data were accepted for inclusion in the List of Priority Substances unless there was strong evidence against their high relative rank. In contrast, substances taken from the modeling list were accepted only if additional information was available (e.g., additional monitoring data) that supported the high relative rank of the substance. In all, 658 substances were compiled and evaluated using this approach. In June 2000, the European Parliament and Council adopted a total of 32 chemicals (e.g., pesticides, solvents, metals) that were selected and recommended through use of the COMMPS procedure and expert judgment.

The OSPAR Commission was founded as a result of the 1992 Oslo and Paris (OSPAR) Convention for the protection of the Northeast Atlantic marine environment. It includes 16 Western European countries together with the European Community (represented by the European Commission). In addition, participants and observers from more than two dozen nongovernment organizations representing various environmental groups and industry also contribute to OSPAR’s activities. One of the major goals of OSPAR is to develop programs and measures to identify, prioritize, monitor, and control the emissions, discharges, and losses of hazardous substances that may reach the Northeast Atlantic Ocean. In this regard, in 1998 the OSPAR Commission established an Ad Hoc Working Group on the development of a dynamic mechanism for selecting and prioritizing hazardous substances (hereafter referred to as DYNAMEC) to update the existing 1998 OSPAR List of (15) Chemicals (and groups of related chemicals) for Priority Action.

In brief, the purpose of DYNAMEC is to serve as a tool to enable the OSPAR Commission—in a transparent manner and using sound information—to identify and select those hazardous substances that have to be addressed by the commission as a whole. The tool is then used to determine those hazardous substances that should be given priority in OSPAR’s activities. In broader terms, DYNAMEC should help the OSPAR Commission as a first step in the implementation of its long-term strategy on the elimination of anthropogenic inputs of hazardous and radioactive substances to the Northeast Atlantic Ocean “within one generation,” that is, by 2020. The DYNAMEC mechanism consists of several interrelated steps and procedures that are summarized below and illustrated in Figure A-1.

The OSPAR report Briefing Document on the Work of the DYNAMEC and the DYNAMEC Mechanism for the Selection and Prioritisation of Hazardous Substances (OC, 2000) provides an introduction and description of the DYNAMEC mechanism and other related information. Further information about the OSPAR Commission, its policies on hazardous substances, and DYNAMEC is available on the Internet at www.ospar.org.

DYNAMEC considered that there are approximately 250,000 manmade chemicals in the so-called chemical universe. Thus, it would clearly not be possible to assess and rank all of these chemicals in a substantive manner. Moreover, the vast majority would invariably not be of concern in the marine environment. Therefore, as a first step, DYNAMEC incorporated the chemicals included in three large and well-established European environmental databases: (1) the Nordic Substance Database (approximately 18,000 substances); (2) the Danish Environmental Protection Agency quantitative structure-activity relationship (QSAR) database (more than 166,000 substances); and (3) the database of the Netherlands’ BKH/Haskoning report (approximately 180,000 substances) for initial assessment. Thus, DYNAMEC also relied on a list-based approach for the initial identification of chemicals for subsequent consideration.

DYNAMEC identified three intrinsic criteria to assess all the substances compiled in the initial selection step. The working group then established five sets of cutoff values (ranging from the most to the least restrictive) to be applied to these criteria. In brief, the criteria assess whether a substance is persistent (P), toxic (T), or liable to bioaccumulate (B). However, after taking into account the overall structure and purpose of DYNAMEC, the least stringent selection criteria and corresponding cutoff values were ultimately applied to the hazardous substances under consideration. After establishing and applying the PTB criteria, the criterion for persistency was developed further to render it more specific to the marine environment. In a separate validation exercise, the cutoff criteria were also applied to the 246 substances (or groups of related substances) included on the OSPAR 1998 List of Candidate Substances. The outcome of this exercise indicated that only 61 of the substances were identified as being of possible concern, while the remaining 185 were not—due mainly to a lack of data and a very low potential for bioaccumulation.

Under DYNAMEC, “hazardous substances” refers not only to substances or groups of related substances that are toxic, persistent, and liable to bioaccumulate, but also to those that are deemed by OSPAR to require a similar assessment approach—even if they do not meet the criteria for toxicity, persistence, and bioaccumulation. To help select substances with an “equivalent level” of concern, DYNAMEC agreed to supplement the initial selections by a “safety net” procedure. Specifically, DYNAMEC experts reviewed proposals from interested parties to include substances on the preliminary List of Substances of Possible Concern that they felt achieved such an equivalent level of concern. Thus, several substances were ultimately included on the preliminary list using this mechanism. The safety net procedure is also intended to address those substances (e.g., metals, inorganic compounds, endocrine disruptors) for which the criteria of persistency and bioaccumulation are generally not applicable.

The results of the initial selection of substances were examined by a group of experts established by DYNAMEC in order to check the plausibility and consistency of the substance-specific data and exclude those substances that had been incorrectly selected.

The ultimate outcome of the initial selection procedure was a List of Substances of Possible Concern for the marine environment. However, DYNAMEC noted that the status of this list is not definite and could change as further information becomes available and in light of improved knowledge.

DYNAMEC decided that fact sheets should be prepared to aid further assessment of all listed substances of possible concern. These fact

sheets would provide comprehensive but concise background information, such as physical-chemical properties and production/use volume information (where available). After producing and distributing the first set of fact sheets, subsequent work focused on expanding the fact sheets for 80 chemicals and groups of related chemicals that were later determined to require priority action (i.e., so-called “selection box” substances described later in this appendix). DYNAMEC noted that additional related work would be necessary to complete fact sheets for all remaining substances of possible concern and to help locate and ascertain relevant data to fill gaps on the existing fact sheets.

For a variety of reasons, the substances and groups of related substances identified by the initial selection, process will give rise to differing levels of concern. In particular, a given substance may (1) have intrinsic properties similar to persistent organic pollutants (POPs) and fulfill the most restrictive set of cutoff points for PTB criteria; (2) have suspected endocrine disrupting properties; and (3) already be adequately addressed in other forums. Regarding the latter, OSPAR could then evaluate whether to await the outcome of any relevant action or to initiate specific OSPAR action. Since DYNAMEC sought to produce a comprehensive and feasible list of substances that are a threat to the marine environment, OSPAR agreed that any substances falling into one or more of these three categories should be “flagged” to ensure consideration in the revision of the existing List of Chemicals for Priority Action.

In order to rank all substances or groups of related substances on the Preliminary List of Substances of Possible Concern, each was characterized with respect to its production volumes, use patterns, and/or measured occurrence in the environment. The level of potential concern for each substance was assessed through use of an effect score (relative toxicity and liability to bioaccumulate) and an exposure score (relative level of predicted or measured occurrence in the environment). The mathematical product of these two scores was used to help determine the relative risk for each listed substance. This process included automated

data processing and was followed by expert judgment (e.g., on the basis of chemical fact sheets). In addition, DYNAMEC decided that calculated exposure estimations and monitored freshwater concentrations, both for the aquatic phase and in sediment, should be accounted for in the ranking process.

It is important to note that these ranking algorithms were based on those that had already been established for use in the previously reviewed COMMPS procedure. However, some algorithms or weighting factors were modified to render them more suitable for the marine environment. In some cases, conservative default values were used when certain substance-specific data were not known or available. In addition, a significant obstacle that DYNAMEC had to overcome concerned restricted access to some data on production/use volumes for certain substances for reasons of confidentiality. This meant that the application of the ranking algorithms, assessment of the outcome of the ranking, and the data used could be undertaken and validated only by a limited number of experts with unrestricted access to the data.

For substances without sufficient information available to carry out the ranking, further action could not be undertaken until either adequate information became available or some other approach for determining the status of such substances was developed. The ranking of the List of Substances of Possible Concern resulted in four lists:

1. substances associated with marine waters based on measured environmental concentration and the properties of the substances;

2. substances associated with marine waters based on modeled exposure scores (in turn based on calculation from production volume and use patterns);

3. substances associated with marine sediments based on measured environmental concentration and the properties of the substances; and

4. substances associated with marine sediments based on modeled exposure scores (in turn based on calculation from production volume and use pattern).

Although final selection of substances for priority action is ultimately a policy decision by the OSPAR Commission itself, it was agreed that DYNAMEC should continue to provide information and expert advice to support revision of the existing OSPAR List of Chemicals for Priority Action.

To facilitate these discussions, a selection box of 80 substances (all chemicals) was extracted by combining the 48 top-ranked substances from the four ranked lists (excluding certain substances already included on the 1998 OSPAR List of Chemicals for Priority Action) with all initially selected substances that could fulfill the most stringent cutoffs for the PTB criteria or those that were previously flagged as endocrine disruptors.

DYNAMEC experts examined the 80 selection box substances on the basis of their expanded chemical fact sheets and established a basis for grouping these substances that is described in Table A-1. A complete listing of selection box substances by group is provided in Appendix 4 of the briefing document for DYNAMEC (OC, 2000).

Based on these groupings, DYNAMEC recommended that the OSPAR Commission consider adding the 12 substances included in Groups A and B when it revises the OSPAR List of Chemicals for Priority Action. Regarding the 20 total Group A and B substances that might be in doubt, DYNAMEC recommended that they should not presently be considered priority substances. However, interested parties were invited to provide more reliable data for these substances in 2000–2001 so that they might be considered with the rest of the Group A and B substances. DYNAMEC further recommended that the 15 substances in Groups C and D should not be considered as priority substances unless new data could be provided expeditiously to support their consideration.

Lastly, DYNAMEC recommended that OSPAR consider initiating monitoring activities with respect to some of the heavily regulated substances in Group E to help determine whether concentrations observed in the environment result from historic uses, unintended or by-product emissions and discharges, or long-range (atmospheric) transport. No recommendations were made in the DYNAMEC report (EC, 2000) concerning Group F endocrine disruptors; however, OSPAR has established a separate List of Priority Research and Development Actions for Endocrine Disruptors within its overall strategy for hazardous substances.

EC (European Commission). 2000. Proposal for a European Parliament and Council Decision Establishing the List of Priority Substances in the Field of Water Policy Study on the Prioritisation of Substances Dangerous to the Aquatic Environment. Document 500PC0047 online at http://europa.eu.int/eur-lex/en/com/dat/2000/en_500PC0047.html.

EPA (U.S. Environmental Protection Agency). 1997. Announcement of the Draft Drinking Water Contaminant Candidate List; Notice. Federal Register 62(193):52194–52219.

NRC (National Research Council). 1999a. Setting Priorities for Drinking Water Contaminants. Washington, D.C.: National Academy Press.

NRC. 1999b. Identifying Future Drinking Water Contaminants. Washington, D.C.: National Academy Press.

OC (OSPAR Commission for the Protection of the Marine Environment of the North-East Atlantic). 2000. Briefing Document on the Work of the DYNAMEC and the DYNAMEC Mechanism for the Selection and Prioritization of Hazardous Substances. Reference Number:1998–16.