Risk Assessment of Manufactured Nanomaterials - Environmental and Human Health Impacts of Nanotechnology - page 400

• specifi city
• temporality (effect succeeds action or factor)
• biological gradient (dose response)
• plausibility (biological explanations)
• coherence (agrees with current knowledge)
• randomised experiments (good study design)
• analogy (effect has already been shown)

The use of Hill's criteria to determine whether observations of effect are causally
linked to specifi c experimental conditions is supported by good study design. For
example, reproducibility of results is required to ensure that artefacts are not con-
sidered indicative of a true result; this is also true for the analysis of data collected
which would be covered by the criteria of biological gradient, plausibility, temporal-
ity, specifi city and analogy. It can be argued that a similar version of Bradford Hill' s
criteria should be used to assess all scientifi c studies for use in risk assessments.
Klimisch and colleagues (1997) developed a systematic approach for the evaluation
of the quality of data, establishing the following scoring system for reliability:

1 = reliable without restrictions. Where the data were generated according to inter-
nationally accepted (or validated) testing guidelines (e.g. OECD) and prefer-
ably performed according to good laboratory practice (GLP) or where the test
parameters are closely related to a guideline method;
2 = reliable with restrictions. Where the data were mostly not performed according
to GLP and where the test parameters do not totally comply with the testing
guideline, but are considered suffi cient to accept the data;
3 = not reliable. Where there were interferences between the measuring system and
the test substance, or in which the test systems were used which are not relevant
in relation to the exposure, or were generated using an unacceptable method;
4 = not assignable. The experimental details provided were not suffi cient and were
only listed in short abstracts or secondary literature.

These evaluation criteria have been applied to many risk assessment methods
previously and have proved to be an acceptable method of determining the quality
of the data to be assessed. As long as the initial requirements for information (and
the quality of that information) are established in defi ned technical guidance docu-
ments, and adopted by bodies such as OECD, then the application of the Klimisch
score to collected data is acceptable. However, the information requirements and
the analysis of the collected data must be robust and transparent to ensure that the
Klimisch score is applied correctly and robustly.
The hierarchy of relevance for toxicological data (stated as individual
study in vitro ) is generally accepted. The Klimsch score may

also be taken into consideration. In doing so, experts need experience in evaluating
toxicological endpoints as well as materials characterisation to ensure that the
knowledge presented is reliable from both toxicology and materials view points.
Individual expertise is also required to determine the relevance and association of
cause and effect. For nanomaterials, there exists the possibility of benchmarking
certain lines of evidence by reference to well characterised materials (as analogues)

> human > animal >