Evaluation of selection methods for toxicological impacts in LCA. Recommendations for OMNIITOX.

Henrik Fred Larsen, Morten Birkved, Michael Zwicky Hauschild, David W. Pennington, Jeroen B. Guinée

    Research output: Contribution to journalJournal articleResearchpeer-review

    Abstract

    Goal, Scope and Background. The aim of this study has been to come up with recommendations on how to develop a selection method (SM) within the method development research of the OMNIITOX project. An SM is a method for prioritization of chemical emissions to be included in a Life Cycle Impact Assessment (LCIA) characterisation, in particular for (eco)toxicological impacts. It is therefore designed for pre-screening to support a characterisation method. The main reason why SMs are needed in the context of LCIA is the high number of chemical emissions that potentially contribute to the impacts on ecosystems and human health. It will often not be feasible to cover all emissions with characterisation factors and therefore there exists a real need to focus the effort on the most significant chemical emissions in the characterisation step. Until now not all LCA studies include toxicity related impact categories, and when they do there are typically many gaps. This study covers the only existing methods explicitly designed as SMs (EDIP-selection, Priofactor and CPM-selection), the dominating Chemical Ranking and Scoring (CRS) method in Europe (EURAM) and in USA (WMPT) that can be adapted for this purpose, as well as methods presenting novel approaches which could be valuable in the development of improved SMs (CART analysis and Hasse diagramme). Methods. The included methods are described. General guidance principles established for CRS systems are applied to SMs and a set of criteria for good performance of SMs is developed. The included methods are finally evaluated against these criteria. Results and Discussion. Two of the most important performance criteria include providing consistent results relative to the more detailed, associated characterisation methods and the degree of data availability to ensure broader chemical coverage. Applicability to different chemical groups, user friendliness, and transparency are also listed amongst the important criteria. None of the evaluated methods currently fulfil all of the proposed criteria to a degree that excludes the need for development of improved selection methods. Conclusion and Recommendations. For the development of SMs it is recommended that the general principles for CRS systems as applied to SMs are taken into account. Furthermore, special attention should be paid to some specific issues, i.e. the emitted amount should be included, data availability should enable broad chemical coverage, and when identifying priority chemicals for the characterisation, the developed SM should generate few false positives (chemical emissions classified wrongly as being of high concern) and no (significant) false negatives (classified wrongly as being of low concern) as compared to the associated characterisation method. These recommendations are not only relevant for a stand alone SM, but also valuable when dealing with simple characterisation methods associated with a higher tier characterisation method. Outlook. There are several questions that need to be answered before an optimal SM can be developed, inter alia: Is it optimal to use simple measured data with high availability or are QSAR estimates of more complex and relevant data better? Which key parameters to include and how? Is a statistical approach, like linear regression of characterisation factors or CART analysis, the best solution?
    Original languageEnglish
    JournalInternational Journal of Life Cycle Assessment
    Volume9
    Issue number5
    Pages (from-to)307-319
    ISSN0948-3349
    DOIs
    Publication statusPublished - 2004

    Keywords

    • Simple characterisations methods
    • Toxicity related impact categories
    • Life cycle impact assessment
    • Evaluation criteria
    • Chemical Ranking and Scoring (CRS)
    • Selection methods

    Cite this