The following questions / concerns were discussed:
- Would the methods reviewed in this workshop be accepted for use in regulatory assessments under current guidance? If not, what steps would be needed to facilitate their acceptance in the future? What are the opportunities to update technical guidance?
The syndicate group discussed and concluded that some uses of SSDs are already accepted in some jurisdictions, e.g. ETx in Europe, Canada, Australia and WebICE in USA. The group identified that the type of tools that are most appropriately applicable is strongly depending on the regulatory setting:
- Evaluation of water quality.
- Site-specific risk assessment.
- Retrospective assessment and assignment of causality to effects in ecosystems.
- The protection of endangered species.
- Non-regulatory settings, e.g. internal business assessments.
It was concluded that SSDs could in principle be applied in appropriate ways to all settings. However, the group mentioned that there was more confidence in applying SSDs in prospective applications, e.g. ETx where the statistical methodology is well developed (compared to retrospective/diagnostic applications). A remark was that the discussion could be better structured by addressing precision, accuracy and domain for different policy uses (hereunder retro- and prospective risk assessments) in the use of SSDs, and the specific research needs to increase the applicability and acceptance of SSDs.
Regulatory acceptance is considered to require:
- Retrospective analyses showing the reliability of the proposed approaches.
- The availability of guidance and decision-trees on how to use SSDs in each setting and how to overcome limitations.
- Confidence needs to be built about the extrapolation to untested species.
- The concept of the dependency of the slope of the SSD on toxic mode of action urgently needs to be validated before it can be used for purposes of extrapolation.
- New approaches should be developed in collaboration with regulators taking account of their needs, including preference for rules on what to use and how in each context.
- Benefits of the use of SSDs need to be demonstrated – this should include the benefits to the regulators themselves.
- Clear communication of approaches, results and limitations is needed.
- Formalised but appropriately conservative criteria are to be set in Tier-1, more complex use of weight of evidence and expert judgement is needed in higher tiers?
The group does not expect a major contribution from SSD to the reduction of animal testing. Coping with the trend for less testing in some jurisdictions will be a challenge for the SSD approach.
- Should current guidance on the use of SSDs be revised in the light of the issues and approaches discussed in this workshop, e.g. number of species?
The group concluded that more guidance is needed, mainly with respect to:
- The number and nature of tested species required for the construction of a valid SSD.
- The same holds for the number and nature of taxonomical groups to be tested.
Lengthy discussions lead to the conclusion that data quality is a major concern:
- Firm and consistent criteria for data-quality should be formulated and standardised for all uses of SSDs – lack of this now is considered a problem.
- The group is concerned about the possibilities for manipulating the SSD output by the selection of input data (cherry picking).
- The group was uncertain about the benefit of adding more data with weighting methods for reliability – more research is needed before this can be recommended.
- Lack of established testing guidelines for non-standard species is a potential problem.
- What implications are there for the interpretation of SSDs and HC5s in risk assessment and risk management?
The final topic addressed by the syndicate group on the interpretation value of SSDs leads to a very short conclusion: Garbage input will automatically lead to garbage output.
- What are the research needs?
The syndicate group identified the following research needs:
- Guidelines need to be developed on how to deal with data quality.
- Guidance needs to be formulated for the use of non-standard test species.
- Guidance should be developed on which methods and tools can be used to generate SSDs – this requires sensitivity analysis, identification of causes of differences, etc.
- Methods need to be developed to include censored input data, such as greater-than values for toxicity endpoints. These are addressed in SSD-Master.
- Methods may be developed to expand on data availability by adding less strictly selected input data and putting less weight on their inclusion, based on reliability of data.
- Main overall conclusions
- SSDs can, in principle, be applied to all regulatory settings if appropriately done as suggested in the following points.
- Regulatory acceptance may require the formulation of a challenging set of arguments and proof of concept.
- A strong focus on data quality and desire for strict standardisation of approaches is needed.
- There is a need to demonstrate when and where criteria of acceptance of data and/or requirements can be relaxed.