Contrasting norms on the use of evidence in risk assessment: the controversy surrounding the carcinogenicity of glyphosate

Abstract

Glyphosate is the most widely used herbicide in the world, used in farming as well as home gardening since the mid ‘70 s. For decades it was considered by regulators an ‘ideal’ pesticide: deadly to pests, respectful of humans. In 2015, it unexpectedly became highly controversial because of conflicting scientific assessments of its carcinogenic effects. On the one hand the UN International Agency for Research on Cancer (IARC) classified glyphosate as ‘probably a carcinogen’; on the other the European Food Safety Authority (EFSA) concluded that glyphosate should be classified as ‘not a carcinogen’. In this I article I develop an explanation for differences in the assessments made by these two regulatory agencies. Whereas most sociological explanations for regulatory disagreements focus on cultural and organisational factors, in this article I advance a normative institutionalist explanation for differences in the outcome of risk assessments. Accordingly, I posit that formal norms regulating the procedures for appraisal directly affect the final outcome by establishing the criteria for the type, quantity and quality of evidence assessed. On the basis of the glyphosate case study, the article discusses advantages and shortcomings of different procedural norms for the selection and evaluation of scientific evidence, and their implications for the overall quality of risk assessments.

Keywords:

• risk assessment
• regulatory divergences
• pesticides

Acknowledgements

The authors would like to thank the anonymous reviewers and the editor for valuable and constructive suggestions and Professor Daniela Sicurelli, Professor Alessandra Arcuri and Dr Vesko Paskalev for discussion and comments.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. The analysis presented here will not attempt at shedding light on the reasons for the procedural differences between the two agencies. Rather, formal norms will be analysed for their influence on the outcomes of regulatory assessments.

2. For example, Lowndes (2010, p. 65) lists 9 different approaches: historical, normative, rational choice, empirical, international, sociological, network, constructivist, and feminist.

3. See: EFSA Register of Questions at http://registerofquestions.efsa.europa.eu/roqFrontend/wicket/page?3.

4. See: https://www.iarc.fr/iarc-monograph-on-glyphosate-other-related-information/; https://www.efsa.europa.eu/en/topics/topic/glyphosate.

5. ERC funded project ‘Law, science and interests in European policy-making’ (LASI) (https://cordis.europa.eu/project/rcn/105651/factsheet/en); evaluation of the implementation of EU Pesticide Regulation carried out on behalf of the European Parliament Research Service (http://www.europarl.europa.eu/thinktank/en/document.html?reference=EPRS_STU(2018)615668).

6. It is relevant to note that EU agencies differ fundamentally from US ones like FDA and EPA, since the former are given purely advisory competences, while the latter perform scientific evaluations and issue legally binding regulations (Majone, 1999).

7. A complete list of tests to be submitted to EU authorities can be found in the Annexe to Commission Regulation 283/2013.

8. Cohort studies recruit individuals living in a certain area and monitor the exposure to chemicals and health outcomes over a long period of time. Case-control studies recruit a group of individuals who suffer a specific disease – like cancer – and a control group of individuals who do not. Their pattern of exposure to risk factors – like pesticides – is then retrospectively assessed on the basis of self-reported information. The goal is to assess whether there is a statistically significant difference in the rates of exposure between the two groups.

9. See footnote 8.

10. The IARC classification comprises 5 main categories: substances included in the Group 1 are categorised ‘carcinogenic to humans’; Group 2A ‘probably carcinogenic to humans’, meaning that there is limited evidence of carcinogenicity in humans and sufficient evidence in experimental animals; Group 2B ‘possibly carcinogenic to humans’, meaning that there is limited or inadequate evidence in humans and less than sufficient evidence in animals; Group 3 comprises substances that are not classifiable because of inadequate evidence in both humans and animals; Group 4 includes substances that are ‘probably not carcinogenic to humans’.

11. A broader look reveals that over the years IARC produced over 1000 classifications. Of them, 120 have been classified ‘carcinogenic to humans’, 83 ‘probably carcinogenic’, 314 ‘possibily carcinogenic’ and 500 ‘not classifiable’. See https://monographs.iarc.fr/agents-classified-by-the-iarc/.

12. EU agencies eventually made the raw data public, after intense pressure from MEPs and environmental groups.

13. For a summary see IARC (2015) Table 3.1 and 3.2.

Additional information

Funding

This work was supported by theEuropean Research Council - Starting Grant [Grant ID: 313642].