Chemical priority settings: Automated data selection routine
Environ. Toxicol. Chem. 19, 2000
2373
cording to the selection criteria of EURAM are compared with
selected data in a limited number of risk assessments of priority
substances already performed within the EU and within the
Organization for Economic Cooperation and Development
(OECD), Paris, France. The comparison is performed for the
environmental endpoints (both exposure and effects). The
comparison did not focus on human health endpoints, because
human health exposure endpoints are similar to environmental
endpoints and, because the data used for the human health
effects scoring in the EURAM are to a large extent the so-
called R-phrases (following Annex I to Directive 67/548 [4]
or the provisional classification and labeling following Annex
VI thereof). A general assumption is that the producer or im-
porter has submitted the R-phrases in the HEDSET, which is
a legal responsibility for companies. Hence, due consideration
of the classification and labeling guide and, therefore, a con-
siderable amount of expert judgment should be associated with
the filling in of these phrases.
Fig. 1. Schematic overview of the division into eight preference clas-
ses (I–VIII) after applying the technical guidance document criteria
within the data sets preferred test results (PTR) and acceptable test
results (ATR). The not acceptable test results (NATR) are not divided
in a preference class. GLP
ϭ good laboratory practice.
Within each of these two categories, the data are divided into
two subcategories, depending on the application of good lab-
oratory practice. The division of preferred test results and not
acceptable test results according to the fulfillment of the re-
quirements of the two categories results in eight preference
classes. Figure 1 illustrates in schematic form the division into
the eight preference classes (class I–VIII). Hence, for each
endpoint the highest preference class is selected by the au-
tomated data selection procedure.
SELECTION CRITERIA
The selection criteria in EURAM are captured by two main
factors: first, the selection of data as laid down in the technical
guidance documents (TGDs) [5], and second, the handling of
nonhomogeneity of the quality and quantity of the submitted
data between substances.
Nonhomogeneous data
The second factor taken into account by the data selection
criteria is the handling of nonhomogeneous data. Because the
resources vary considerably from data submitter to data sub-
mitter, both the quality and the quantity of the data in IUCLID
vary considerably from substance to substance. To handle these
nonhomogeneous data the most conservative value among the
data in the highest preference class is selected. The most con-
servative value means the value that results in a worst-case
calculation of the environmental exposure and effect scores
within EURAM, as is described in detail in a previous paper
[3]. For the calculation of the environmental exposure score
(which includes the use of a Mackay level I model) the most
conservative value of the endpoints boiling point, vapor pres-
sure, and Kow are the lowest values, whereas for the aqueous
solubility the most conservative value is the highest value.
Only data generated under environmental conditions are used.
For boiling point this means measurements at a pressure be-
tween 950 and 1,050 hPa; for vapor pressure, aqueous solu-
bility, and Kow this involves measurements performed at tem-
Technical Guidance Documents
In the TGDs, guidance is given to data selection for risk
assessment of existing chemicals. First, data that are invalid
because of invalid test conditions or poorly reported test results
should be rejected. Second, studies carried out according to
currently accepted methods (e.g., EU, OECD, or other methods
of standardized bodies and or those performed to good labo-
ratory practice [6]) should have greater weight than those not
carried out according to acceptable methods. Finally, prefer-
ence should be given to consolidated data sets, that is, to
HEDSET data generated by two or more cooperating com-
panies.
Applying these data selection rules, guided by the TGD,
the EURAM sorts the data according to the following criteria;
first, the test method used to generate the data; second, whether
the test result was generated using good laboratory practice;
and finally, whether the data originate from a consolidated
data set. The data for a specific substance and endpoint are
divided into three sets: the set of not acceptable test results,
the set of acceptable test results, and the set of preferred test
results. The not acceptable test results are not further taken
into account in the data selection procedure of EURAM. In
general, test results generated using EU or OECD guidelines
or certain other methods of standardized bodies will fall in the
preferred test result category. Because no glossary code exists
in HEDSET and hence in IUCLID, for some methods of stan-
dardized bodies, results from these methods will not be in-
cluded in preferred test results. However, using these data will
be possible through the expert judgment step of the ranking
procedure. Estimated data for logKow and vapor pressure can
be entered directly into the HEDSET/IUCLID; quantitative
structure–activity relationship (QSAR) estimates can be used
if no measured value is present in IUCLID. The use of QSAR
estimates is flagged in the ranking results.
peratures between 15 and 25ЊC. Any tests indicating that they
are not to be measured under environmental conditions will
be included in the not acceptable test results.
The most conservative biodegradation test result is the glos-
sary value (readily biodegradable, inherently biodegradable,
or nonbiodegradable), which gives rise to the lowest fraction
degraded, using the table for fraction degraded in Hansen et
al. [3]. However, if only readily biodegradable and inherently
biodegradable test results are in the highest preference class,
then the substance is seen as being readily biodegradable. If
no glossary entry exists in IUCLID for biotic degradation in
soil, water, sediment, and sewage treatment plant, then the
entries must be extrapolated from available standardized lab-
oratory tests in IUCLID, namely biochemical oxygen demand
(BOD) and chemical oxygen demand (COD) using BOD/COD.
The test results in IUCLID will be converted into readily bio-
degradable, inherently biodegradable, and nonbiodegradable
according Annex VI of Directive 67/548 [4], that is, if the
BOD/COD is greater than 50%, a substance is classified as
readily biodegradable; if the BOD/COD is in between 20 and
The preferred test results and not acceptable test results are
divided into two categories, one with the data originating from
a consolidated data set and one with the nonconsolidated data.