International Office for Water
Capacity building for better water management /
Implementation of requirements on Priority substances within the Context of the Water Framework Directive
Contract N° 07010401/2008/508122/ADA/D2
Issue paper on prioritisation process: comments, possible answers and choices made for final stage ranking
(April 2009)
Contact persons:
A. James
T+33 (0) 3 44 55 69 44
F+33 (0) 3 44 55 67 67
Issue paper on prioritisation process: comments, possible answers and choices made for final stage ranking/ Prioritisation methodology based on monitoring data
April 2009
CONTENTS
I. Introduction......
II. General issues......
II.1. A risk based approach......
III. Exposure: PEC issues......
III.1. Identification and treatment of outliers......
III.1.1 Data discard related to quality......
III.1.2 Data discard related to determination limits......
a. Threshold values above which determination limits are deemed unreliable......
b. Method for determining “outliers”......
III.2. Calculation of PEC values......
III.2.1 Geographical heterogeneity of monitoring data and calculation of PECs
III.2.2 Heterogeneity of monitoring data between matrices and calculation of PECs......
III.2.3 90th percentile versus 50th percentile......
III.3. Metals issues : the use of dissolved versus total concentrations......
IV. Effects: PNEC issues......
IV.1. General PNEC issues......
IV.1.1 Indicative tolerable risk levels for cancer
IV.1.2 Metabolites issue
IV.1.3 General remarks on quality and relevancy of data
a. Remark on the sources of data and their reliability......
b. Remark on the methodology adopted for PNECsediment derivation......
c. Remark on the use of mesocosm data for pesticides......
IV.1. Substance specific PNEC issues......
IV.1.1 PNEC for Pesticides
a. Alpha-Cypermethrin......
b. Ametryn......
c. Atrazine/simazine......
d. Azoxystrobin......
e. Bentazone......
f. Chloridazon......
g. Chlorothalonil......
h. Cyprodinil......
i. Desethyl terbuthylazine......
j. Dicamba......
k. Dimethenamid-P......
l. Dimethoate......
m. Fenpropimorph......
n. Iprodione......
o. Lambda-cyhalothrin......
p. Metalaxyl......
q. Metazachlor......
r. Methidathion......
s. Metolachlor......
t. Pendimethalin......
u. Pirimicarb......
v. Pirimiphos-methyl......
w. Prochloraz......
x. Prometryn......
y. Propiconazole......
z. Terbuthylazine......
aa. Terbutryn......
IV.1.2 PNEC for Metals
a. Aluminum......
b. Antimony......
c. Boron......
d. Copper......
e. Iron......
f. Molybdenum......
g. Vanadium......
h. Zinc......
V. Priority algorithms......
V.1.1 “Low” priority rank versus “Further investigation needed” rank
/ – International Office for Water and INERIS –Under contract of the DGENV N° 07010401/2008/508122/ADA/D2 /
Issue paper on prioritisation process: comments, possible answers and choices made for final stage ranking
/ Prioritisation methodology based on monitoring data
April 2009
I. Introduction
This issue paper aims at addressing the different issues commented by WG E members since the document on the first stage ranking process led by INERIS (July 2008) was disseminated. Choices made by INERIS on PEC calculations and PNEC values are justified in this report and will be applied for final monitoring-based ranking exercise of substances in the context of the Water Framework Directive. The report is not summarising the final methodology that will be exhaustively presented in a final report in the coming months.
The comments that have not been taken on board in this report were not considered relevant or appropriate. Also, there is a third category of comments such as comparison with COMMPS that is not supposed to apply many choices and will be addressed in the final monitoring-based prioritisation exercise.
II. General issues
II.1.A risk based approach
As a general remark, some stakeholders commented that prioritisation methodology based on monitoring data (INERIS) does not follow the required risk based approach as it only considers as possible candidates those substances for which monitoring data are available instead of selecting substances based on the risk that these substances might pose.
As reply to this statement it could be recognised that the monitoring data are used as a first stone to establish the manageable list of candidate substances. However the process did not stop at that stage and further investigations have been made. In fact the risk ratio for each candidate substance was determined whenever was possible, for e.g. when a PEC and a PNEC were available, the risk ratio was used as determinant parameter for the priority setting exercise.
Therefore, it has to be stressed that monitoring based priority setting led by INERIS is mainly driven by a risk-based approach.
It has to be recognised that the monitoring based prioritisation methodology has advantages and disadvantages. One of the gaps of this methodology is that the substances that are not usually monitored (as such emerging substances) and/or substances considered toxics even at very low concentrations are not taken on board.
On the other hand, it has to be underlined that the modelling based approach (JRC-UK approach) will to be used as a complementary methodology of the one proposed by INERIS. Consequently, it is envisaged that the issues not covered by INERIS methodology as those mentioned above will be addressed by the JRC-UK approach.
III. Exposure: PEC issues
III.1.Identification and treatment of outliers
III.1.1Data discard related to quality
It was suggested that some data could be considered as outliers when we are reporting to the whole monitoring database (e.g. very high or small values). However, it should be stressed that the monitoring data sent by each Member State to INERIS/IOW consortium and contributing to the entirely monitoring database is the responsibility of Member States. Consequently, the monitoring data considered as relevant for a certain Member States (hot spot site/local problem) cannot be removed from the monitoring database just because this is not applicable also for other Member States.
III.1.2Data discard related to determination limits
a.Threshold values above which determination limits are deemed unreliable
At the last WG E meeting (the 5th meeting), 2 scenarios were presented showing the impact of using 95th and 90th percentile as a thresholds value for determination limit (DL) to discard data instead of the 99th percentile. It was decided, at WG E level, to maintain the choice of the 99th percentile because the percentage of data discarded with this thresholds value (around 3%) is deemed reasonable. The percentage of data discarded per countries depends on the distribution of DL. Discarded data have been re-calculated and are presented below in Table 1.
Table 1: Different scenarios envisaged for discarding process with various thresholds values applied to determination limits for the definition of outliers: number of analyses considered as “suspicious” or “outliers” by country
Member States / 99 th centile / 95 th centile / 90th centileAT / 0.0 / 0.0 / 0.0
BE / 0.1 / 0.1 / 0.1
BG / 0.0 / 2.6 / 2.6
CH / 0.0 / 0.0 / 0.0
CY / 0.4 / 0.9 / 0.9
CZ / 0.1 / 0.2 / 0.7
DE / 0.3 / 0.5 / 0.6
DK / 0.0 / 0.2 / 0.2
EE / 0.0 / 0.0 / 0.0
EL / 0.0 / 0.0 / 1.6
ES / 0.0 / 0.9 / 0.9
FI / 0.0 / 0.0 / 0.0
FR / 1.1 / 1.5 / 1.7
HU / 0.0 / 1.1 / 1.1
IE / 0.0 / 2.1 / 2.1
IT / 0.9 / 1.3 / 1.5
LT / 0.0 / 0.0 / 0.0
LU / 0.0 / 0.3 / 0.5
LV / 0.0 / 0.0 / 0.0
NL / 0.0 / 0.0 / 0.1
NO / 0.0 / 0.0 / 0.0
PL / 0.0 / 3.7 / 3.7
PT / 0.5 / 0.8 / 0.8
RO / 0.8 / 3.0 / 4.3
SE / 0.0 / 0.2 / 0.2
SI / 0.0 / 0.0 / 0.2
SK / 0.0 / 0.4 / 0.7
UK / 0.0 / 0.0 / 0.0
Total / 0.2 / 0.7 / 0.9
b.Method for determining “outliers”
It was commented that “The QA/QC Directive states that for the compliance checking the values <DL should be divided by 2; but the QA/QC Directive states also that the DL (LOQ) should be lowest than the EQS or PNEC. The division by 2 of the DL is useful for calculation if the DL is lower than the EQS.
For the prioritisation process, it has been decided to divide by 2 all the values <DL for the calculation of the PEC even in the case where DL > PNEC.
It is clear that in many cases the final PEC/PNEC ratios can depend of the values <DL above the PNEC and this approach can lead to estimate high concentrations in the environment when analytical methods are not reliable (sensitive). This is not a realistic scenario and can lead to attribute a higher risk to the substances detected with methods less advanced.” It was also added that “the values <DL are useful data that should be taken into account for the PEC/PNEC final results, but when these values are too high respect to the PNEC they have to be discarded.”
The following conditions were proposed, as final decision data should be discarded when:
-DL > 2 PNEC
-90th percentile of PECstation (or from PEClocal) <DL and DL>PNEC
Unfortunately the proposal was sent at a very late stage. The time constraints do not allow applying this proposal in the ranking. However, it was decided to take it into consideration “a posteriori” when double quality checking will be made for substances very highly and highly ranked.
III.2.Calculation of PEC values
III.2.1Geographical heterogeneity of monitoring data and calculation of PECs
Some scenarios showing the great heterogeneity of monitoring data amongst stations, river basins and countries were presented at the 5th WG E meeting. As a consequence of these demonstrations, it was identified that PEC calculation could be highly influenced by the monitoring data supplied by a limited number of countries as well as a limited number of stations within a country and a very limited number of analysis within a station.
Before the 5th WG E meeting, it was commented that prioritisation of substances would be much relevant to be carried out at the river basin level. Therefore, during the 5th WG E meeting, the possibility of an additional level of calculation (river basin) within PEC calculation was discussed. However, some Member States commented that calculating PEC for a certain river basin the problem of heterogeneity would be solved, but this will not address the issue of representativity as long as river basins are very different in terms of environmental characteristics and sizes. Consequently, this proposal will imply that each river basin must have the same weigh in the prioritisation process (e.g. Danube river basin will weigh as much as other river basins much smaller).
Therefore, it was decided to stick to the methodology proposed during the first stage ranking and therefore to calculate the 90th percentile of all arithmetic means of all measures for a given substance.
III.2.2Heterogeneity of monitoring data between matrices and calculation of PECs
Since the first stage ranking report was disseminated, many Member States and stakeholders commented that over 95% of the monitoring data are for water compartment while 3% of concentrations reported are measured in sediment and 1% in biota. Currently the contribution of different matrices to the new monitoring database is the following: 93.2% for water, 6.3% for sediment and 0.5% for biota. It can be added that as all matrices are considered at the same level in final priority ranking, PEC values can be highly influenced by the monitoring data supplied by a limited number of analysis within a matrix.
It was suggested that substances should be ranked on the basis of water monitoring data and that “data on sediment and biota can be used as supporting evidence in the final step, namely the discussion with experts for a final proposal of Priority Substances.”
INERIS proposes to consider the number of analysis within a matrix as an “additional requirement” and to treat it “a posteriori”. In fact, INERIS proposes to calculate risk ratios for all fractions and matrices as it was done for first stage ranking and to come back to the source of data only for the substances very highly or highly ranked. The final prioritisation report will include the outcomes of this “a posteriori” checking.
III.2.390th percentile versus 50th percentile
Since first stage ranking report was distributed, a stakeholder commented that PEC should be calculated with 50th percentile of all arithmetic means rather than 90th percentile of these values.
INERIS was tested this possibility and the results of using 50th percentile instead of 90th percentile were presented at the 5th WG E meeting. These tests led to significant differences in the priority ranking as such decreasing priority ranking for a relative big number of substances. As regards the substances very highly and highly ranked, a great decreasing appears for the substances highly ranked.
It has to be recalled that this is a relative ranking exercise and not an absolute risk assessment. Therefore, it was decided to stick to the first proposal of 90th percentile as being, in principle, more conservative and consistent with the COMMPS procedure.
III.3.Metals issues : the use of dissolved versus total concentrations
Metal industry suggested, after the 4th WG E meeting, to test for metals the possibility to calculate risk ratios with PECtotal expressed as dissolved concentrations via the equilibrium partitioning approach (using Kp values provided by Eurometaux). The results of this test were presented at the 5th WG E meeting. No comment has been made on this issue, during or after the 5th WG E meeting, except the one of metals industry who supports the use of this method.
Given the important uncertainties brought in by the use of equilibrium partitioning approach and given that the new collection data has led to quite a substantial amount of data in dissolved phase (medians for all quantified measures of metals in dissolved phase is know 1142 analysis for 120 stations), INERIS proposes to use the PECdissolved as a basis in the final monitoring-based ranking and to use the PECtotal expressed as dissolved via the use of the equilibrium partitioning approach only if “a posteriori” checking reveal an evidence for using it.
IV. Effects: PNEC issues
IV.1.General PNEC issues
IV.1.1Indicative tolerable risk levels for cancer
Comments made on INERIS proposal (2008) and response to comments
Comments have been received, before and after the 5th WG E meeting, on the Acceptable Daily Intakes (ADIs) used by INERIS in its first stage ranking report (Bonnomet, 2008[1]) and corresponding to a cancer risk of 10-5. It was suggested that a lifetime risk of 10-6 should be used instead of 10-5, and therefore data may have to be re-calculated.
INERIS proposal for final monitoring-based prioritisation ranking (2009)
Two thresholds in cancer risk are being used in the context of REACH for the evaluation of carcinogenicity risk levels for the general population (10-5 and 10-6). REACH Guidance on information requirements and chemical safety assessment (for more details, please see REACH Guidance[2]) states that “[...] cancer risk levels of 10-5 and 10-6 could be seen as indicative tolerable risk levels when setting DMELs for [...] general population”.
Discussions on the acceptance level or cancer risk are controversial and have been following a long period of time without reaching any conclusion.
On the other hand, World Health Organisation (WHO) considers a lifetime cancer risk for consumers of less than 10-5 while EU Drinking Water Directive recommends a value of 10-6 as a limit value for the general population. Moreover, United States and EU risk assessments reports carried out under Regulation 793/93/EC present different values and different reasons for their choice.
However, Appendix R.8-14 of REACH Guidance on assessment of human health risk2 concludes that the cancer risk decision points used for lifetime exposure of the general population are in the range of 10-5 to 10-6 and not recommends one or the other value.
Whether one considers 10-6 as a more reasonable tolerable risk level or not is a tricky question which seems more a political issue than a really scientific one. This issue was raised before the 5th WG E meeting, discussed during the meeting, but no further comments were made or any conclusion was drawn on this point.
INERIS proposal for final monitoring-based prioritisation ranking (2009)
Considering that choosing 10-6 as tolerable cancer risk is more protective for the environment and the human health, it was decided to change the initial proposal made by INERIS and to use 10-6 as a “reasonable tolerable cancer risk level”.
IV.1.2Metabolites issue
Comments made on INERIS proposal (2008) and response to comments
Before the 5th WG E meeting, comments were received on parent compounds and metabolites that should be grouped, but the EQS should be derived for the each separate compound (no group-EQS).
After the 5th WG E meeting, one Member State made the following comment: “agrees that PNECs or EQS should also be derived for metabolites”. However, it was added that “the PNEC for the parent compound should also be derived and compared with monitoring data, because both parent and metabolites may be of concern. For the parent, comparing peak measurements with MAC-EQS may be most appropriate.”
INERIS proposal for final monitoring-based prioritisation ranking (2009)
Given that PNECs are not always available for metabolites and parent compounds, it is proposed that when no PNECmetabolites are available, would be applied PNECparent compound and when no PNECparent compounds are available and the metabolites are known as being more toxic than parent substance, then PNECmetabolites would be applied.
Finally, it is proposed to use a sum of PEC for parent compounds and metabolites in order to compare them to a unique PNEC.
INERIS proposal for final monitoring-based prioritisation ranking (2009)
INERIS proposal for the final monitoring-based ranking exercise remains unchanged for PNECmetabolites and PNECparent compounds which will be compared to the sum of PEC. Finally, if any of those substances are very highly and highly ranked, an “a posteriori” checking will reveal this evidence.
IV.1.3General remarks on quality and relevancy of data
a.Remark on the sources of data and their reliability
INERIS was asked to clarify how the ecotoxicological data used for derivation of PNEC were validated in terms of quality and relevancy.
In the report for the first stage ranking was stated, “the scientific value of the compiled information has not comprehensively been assessed by the INERIS/IOW consortium and the original published data may not have undergone any systematic peer-review process”.
Therefore, it should be specified that quality and relevancy of the data was not assessed when data were extracted directly from the databases and the reporting data have been “peer-reviewed”, “reviewed by an Agency scientist as well as a second supervisory biologist” or when data were directly retrieved from sources know as valid at the international or European level (e.g. OECD SIDS and SIARs, finalised European Union Risk Assessment Reports). On the other hand, it should be underlined that quality and relevancy of ecotoxicological data used for the derivation of PNEC was assessed when values were directly retrieved from databases where no peer-review had been done for their use in the context of risk assessment (e.g. US-EPA ECOTOX database, public literature including articles and reports).
In conclusion, it was decided that the ecotoxicological data extracted from the websites indicated in the table below could be considered valid without restrictions and any further review process.