The role of epistemic policies in regulatory science: scientific substantiation of health claims in the European Union

Abstract

This paper presents an analysis of the concept of scientific substantiation in European health claims regulation. It focuses on the controversies about the demand for the establishment of cause-and-effect relationships between food consumption and health outcomes in claim substantiation. Our analysis, on the basis of regulatory and scientific documents, identifies two opposing views about the aims of health claims regulation. Each of these two stances links certain regulatory objectives with specific epistemic policies, that is particular sets of scientific methodology, criteria, and procedure. The regulators, in selecting a demanding evidentiary approach based on a hierarchy of methodologies that requires causal data for substantiation of claims, give priority to preventing the authorization of false claims. The opposing view, espoused by the critics of this approach, opts for less demanding requirements for substantiation, implying the market availability of a wider range of products with health claims that may provide individual as well as public health benefits. We argue that one of the objectives that underlie the European regulators’ demand for causal data is to protect their own credibility, by trying to isolate them from value-laden debates about the limitations of scientific methodologies, as well as the societal and policy implications of regulatory decision-making.

Keywords:

• health claims
• causality
• regulatory science
• scientific evidence
• randomized controlled trials
• benefit assessment

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. The emphasis that EFSA places on RCTs shows that the interpretation of causality that underlies the assessments is a probabilistic one (Worrall 2007; Cartwright 2010).

2. There are, however, important differences between regulatory practices. In the US, for instance, less than a third of all authorized claims are fully authorized SSA claims (Rowlands and Hoadley 2006). In the Japanese regulatory framework, the difference between standard, fully authorized (ordinary) and qualified claims depends on the availability and statistical significance of the RCT data. Without RCT data or RCT data of low statistical significance (p < 0.1), claims are automatically categorized as qualified claims (Ohama, Ikeda, and Moriyama 2006).

3. The language used in many of the European regulatory documents is not particularly explicit. For example, in accordance with the NHCR, as well as EU and EFSA guidelines, applicants must state ‘the extent to which’ a causal relationship has been established. But, as we will show, the actual evaluation requirements specify (and EFSA’s NDA Panel makes explicit) that in order to receive an authorization, the Panel has to consider that a cause-and-effect relationship has in fact been established (EFSA 2009, 2011b).

4. These criteria figure among those proposed by Hill (1965) for being able to establish causal relationships. In fact, they constitute the key elements of Hill’s proposal, and – with the exception of plausibility– are in scientific practice best operationalized by way of the RCT methodology.

5. Pertinent studies are defined as ‘studies from which scientific conclusions can be drawn for the substantiation of the claim’ (EFSA 2011b, 9). EFSA decides on the pertinence of studies (being human studies the most relevant ones) by checking the following items (EFSA 2011b, 11): (i) use of appropriate outcome measures; (ii) sufficient definition and characterization of the ingredient used in the studies; (iii) appropriate design and quality of the studies; (iv) appropriate simulation of real use conditions of the ingredient; and (v) representativeness of the study group vis-à-vis the foreseen user population.

6. As different from observational studies that are understood to present fewer operational limitations if well planned and executed.

7. Both the Passclaim and Fufose projects, along with several other specific research programs, conducted EC-funded research in preparation for the future health claims Regulation.

8. The other two main criteria that have to be fulfilled – as we have already seen – are that the ingredient is fully characterized and that the claimed effect is well defined and considered a beneficial physiological effect (apart from other, minor criteria related, e.g. to the proposed intake of the food or wording of the claim).

9. Related to the question of the possible overreliance on RCTs is the issue if food constituents can be characterized by reducing them to singular (or very few) active components that then imply a single measurement with respect to a single, well-defined effect. As we have seen, the NHCR and EFSA currently focus approvals on single components that are linked to single, measurable endpoints, obviating the multitude of components present in foods, as well as their multifactorial physiological (pleiotropic) effects. This leads, for instance, to EFSA’s negative scientific opinions on honey (because of insufficient characterization due to natural variations in honey) but positive opinions on copper (a well-characterized and clearly measurable ingredient) (de Boer, Vos, and Bast 2014).

Additional information

Funding

This work was supported by the Spanish Ministry for the Economy and Competitiveness and European Commission FEDER funds (research project ‘La evaluación de beneficios como ciencia reguladora: las declaraciones de salud de los alimentos funcionales’ [grant number FFI2013-42154-P]).