Benefit-risk assessment of food supplements
This page is a method.
The page identifier is Op_en4210
|Moderator:Jouni (see all)|
Give your opinion to the peer rating of the content of this page.
- Name: DWP5-1 Risk-benefit methodologies
- Responsible partner: THL, Jouni
- Deadline: June 2011
- Status: Draft submitted, comments received and included
<section end=plantlibra />
- 1 Question
- 2 Answer
- 3 Rationale
- 4 See also
- 5 Keywords
- 6 References
- 7 Related files
What is a good method for making benefit-risk assessments for plant-based food supplements such that it has the following properties?
- The valuations and expectations are in line with what a typical user of PFS would think (i.e., if a PFS BRA and its conclusions are explained to a typical user, the user is not negatively surprised about what was done).
- It offers a clear guidance for a relevant authority to perform a PFS BRA in practice.
- It offers a clear guidance for a relevant authority to evaluate nutrition and health claims of a PFS.
- The method is coherent with the methods used for the neighbouring areas, namely foods and medicine, especially with borderline products.
How is the method validated?
In this context, the methodology is designed for making benefit-risk assessments (BRA) for plant-based food supplements (PFS). However, it should be noted that there is a clear and explicit attempt to develop an assessment methodology that is as much context-independent as possible and practical. Therefore, in the text we avoid talking about plant-based food supplements specifically and rather talk about e.g. exposures. Also, we avoid talking about benefit-risk assessments specifically and rather talk about assessments in general without excluding other types of assessments, such as those about risk only, about cost-benefit, or about compliance to standards.
The methodology for benefit-risk assessment (BRA) for plant-based food supplements (PFS)
Most assessment methods are built on the idea that the main or only purpose of an assessment is to produce scientifically valid estimates about the measure of interest, such as acceptable daily intake or net health impact of consuming a plant-based food supplement (PFS). The proposed method has a different starting point. It is based on the idea of a society (e.g. the EU) that needs to take actions related to a particular product, compound, or situation. An assessment is performed to inform the society (or its representatives such as EFSA) to make wise decisions about the issue. The decision-making process is inevitably bound to information available, information produced by an assessment, and value judgements in the society. Therefore, the assessment should try to capture also the practical needs of decision making and the values of the society. Hence, a useful assessment is much wider a thing than a net impact estimate. The assessment is essentially intertwined with the decision-making (i.e., the use process of the information produced), and this connection has a profound impact on the assessment methodology. Due to open attitude towards participation in assessments, the methodology to be presented here is called open assessment.R↻
The suggested open assessment methodology has three tiers or phases. The main flow of work proceeds from I to III but in practice, there is a lot of feedback from subsequent tiers, which results in revisits and rethinking based on new information occurring during the assessment.
The three tiers described here have a close resemblance to the four phases in the IEHIA approach (integrated environmental health impact assessment, developed in INTARESE project). The four phases are Issue framing (similar to Tier I), Design (similar to Tier II), Execution (similar to Tier III), and Appraisal (which is incorporated in all tiers in this approach), or IDEA in short. It is also useful to compare this to the BRAFO (benefit-risk assessment for foods) approach. BRAFO also has tiers, but BRAFO assumes that the issue framing is more or less clear in the beginning of the assessment. The BRAFO tiers mostly focus on deciding the magnitude of effort needed in Tier III (or the Execution phase of INTARESE).
What is clearly different in IEHIA and open assessment is the timing and role of information produced in an assessment. IN IEHIA, appraisal is a separate phase after the (scientific) execution of information collection and analysis. Also stakeholders may participate in appraisal by commenting assessment results and their implications on decision-making. However, in open assessment, appraisal, participation, and information use happen all the time during an assessment and therefore they are not a separate tier. In contrast, it is specifically the use of information that guides the work within and between tiers. Whether it is time to collect new information (Tier III), to adjust decision criteria (Tier II), or even to rethink the main questions (Tier I), or stop the whole assessment, it depends on the needs of the intended use of the assessment.
To clarify the crucial role of information use as the guide for making an assessment, we will try to give some practical examples what this could be in practice.
Tier I: Definition of the use purpose and scope of an assessment
In Tier I, the use purpose of an assessment is defined by answering an overall question: what are the practical needs of a particular decision-making situation at hand, and what information should be produced in an assessment to serve this practical need? To answer this question, several more specific sub-questions are asked:
- What are the research questions whose answers will bring the information needed by decision makers in this situation? Common types of questions include but are not limited to following:
- Is the health or content claim X (e.g. improves immunity, rich source of antioxidants) of a product Y acceptable for marketing, i.e. does it hold against observations and scientific criticism?
- What are the magnitudes of beneficial and harmful health effects of a compound due to a consumption of or an exposure to relevant levels?
- What are the best decision options that can lead to improved health or other outcomes in a particular situation?
- What are the spatial, temporal, population, and other boundaries within which the assessment should be performed?
- Who are the intended users of the assessment and what is the intended use? It is especially important that the users identified actually participate in the making of the assessment, not just read a ready-made report. This is because user's needs should drive the purpose, content, and making of an assessment, and this is best ensured by active participation.
- Who is not allowed to participate in the assessment? The default is that anyone interested is allowed. Exceptions should be explicitly mentioned and reasons for exclusion should be given. In traditional risk assessments, it was typical that the actual user (e.g. a policy-maker) was not allowed to participate in an assessment due to a potential attempt to manipulate the assessment outcomes from a scientific ideal. In the proposed methodology, it is essential that the users participate, as was mentioned before. In an open assessment manipulation of results is prevented by other means than restricting participation, notably by open and science-based criticism of statements in an assessment. Actually, it is essential that several other parties (with conflicting interests) participate to identify and prevent any attempts to manipulate outcomes by users or any other group.
- What are the particular scenarios that are considered in the assessment (e.g. business as usual vs. targeted restrictions of use of a particular product)?
An assessment can stop at Tier I if e.g. the answers to the questions above show that there is a good scientific consensus about things but rather a conflict of values. In such a situation, an assessment may not bring any added value, and the matter must be solved by voting or other tools of decision making.
Tier II: Definition of the decision criteria
In Tier II, the criteria that the decision maker wants to use in the decision making process are defined. There are several ways to approach this, and again the definition is based on the practical needs of the particular situation. It should be noted that there are numerous approaches even at a crude level, and each can have different nuances in specific criteria. Only a few examples are listed here.
- Margin of exposure (MOE): Based on toxicological information, a safe exposure level (NOAEL) is estimated. This level is divided by the current exposure levels. The higher the ratio (MOE), the more sure we can be that there is no hazard.
- Acceptable daily intake (ADI): Maximum amount of an agent, expressed on a body mass basis, to which an individual in a (sub)population may be exposed daily over its lifetime without appreciable health risk. If the actual exposure is smaller than ADI, there is no concern.
- Quantitative benefit-risk assessment: both beneficial and harmful health effects of a PFS product are quantitatively estimated. Typically, different effects are summed up using e.g. disability weights, and decisions are made based on the net health effects. Also monetary units may be used. A quantitative risk assessment is a sub-category of this: risks are not compared to benefits but to some previously defined acceptable level of risk.
- Estimation of a safe level of exposure: A product may have a safe level of exposure. This means that there is a level where the exposure is high enough to cause a low probability of no benefit but low enough to cause a low probability of harm.
- Avoidance of harm: This is an approach where the emphasis is on harms: if exposure to a product may cause harm, its use should be restricted and possible benefits are irrelevant. Only if potential harm is low, benefits are evaluated to find out whether the product can be recommended.
- If the question is about acceptability of a health or content claim of a product, there is a need for clear criteria about what evidence is enough to accept or reject the claim.
- Multi-criteria decision analysis is an approach where there are several conflicting objectives or several decision-makers with differing opinions. Criteria are needed on how to weigh different objectives or opinions.
An assessment can stop on Tier II, if the main dispute is about selection of decision criteria instead of scientific facts. An example of this is a classical case of persistent pollutants in farmed salmon, which caused an outrage in 2004 (Hites et al, Science 2004).  However, there was any reason for outrage only if risk-only approach was used to assess the situation. Instead, if net-health-effect approach was used, it was clear that health benefits of farmed salmon outweigh the risks of pollutants, and there was no need for further benefit-risk assessment or concern (Tuomisto et al, Science 2004). 
Tier III: Information production
Tier III is about producing information needed in the decision making as described in Tiers I and II. It is useful to note that e.g. the BRAFO approach tiers mostly belong to Tier III. Therefore, the proposed methodology has a wider scope within the assessment-management sphere.
This tier can be described by listing several tasks related to information production.
- Collecting data
- Building (quantitative) models
- Making conclusions and writing descriptions based on tasks above.
- Making value judgements about e.g. the estimated outcomes of interest.
- Evaluating the content within the assessment: peer review, discussions, hypothesis testing.
Because this is the last tier, it is obvious that an assessment can stop here. However, it should be emphasised again that instead of stopping, an assessment may revisit previous tiers and adjust research questions and decision criteria. It is common to learn fundamental things during an assessment, so that only after an extensive period of work, the participants suddenly realise what they actually should have been assessing.
General comments for all tiers
The purpose of an assessment is to make recommendations and conclusions based on information collected and value judgements used to interpret the information, and then use this increased understanding in real decision making. Dissemination of information has a crucial role in this. However, it should be noted that as most interested parties are expected to participate in the whole assessment process, dissemination happens during the whole work and not just after final conclusions have been made. It is not a separate phase in an assessment unlike in many traditional assessment methods.
An important aspect is the evaluation of the status of the assessment. Is the current information good enough to answer the practical needs identified in the beginning? Are the current conclusions scientifically acceptable and useful? Should more work be performed in some parts of the assessment to improve them? The evaluation of the status typically leads back to one of the previous tiers, until the practical needs have been addressed and the assessment ends.
Based on findings reported in the review (DWP5-1) and a preliminary conceptual workshop held at the kick-off meeting (MWP5-1), a risk-benefit assessment methodology will be proposed and tested in selected case-studies. The methodology, to be developed in Opasnet, should be organized in tiers, allowing to identify risk-benefit scenarios and to address qualitatively scenarios (e.g., through MOE, ADI) where risks or benefits are not overlapping. After preliminary testing of the methodology on at least 5 case-studies drawn from the PlantLIBRA list of case-studies, the report will be submitted for review to partners and to the three advisory boards; their feedback will be discussed, aligned with outcomes of WP2 (benefits, specifically definition of benefits and integration of various types of evidence), WP3 (risks and new approaches for risk assessment) and WP4 (acute risks, specifically epidemiological evidence), and integrated at the second project meeting. Uncertainty will be described and estimated when possible. The final report will be published alongside with a generic Analytica software tool to explore risk and benefits (DWP5-1).
Task leader: THL
PM: THL 5, Hylo 3, UMIL 0.5
Which compounds and plants should be used for five case studies in the WP5.2. Five suggestions and short comments why we selected them are given below. The purpose of selecting the plants for the case studies is not necessarily the significance of health effects, but to test benefit-risk methodology. For this purpose, we should include plants requiring both with 1) simple approaches and 2) complex requirements for methodology.
Suggested case-studies (updated list):
- Green tea
Bilberry Vaccinium myrtillus L.
Health effects: Astringent, enhance vascular tone, antioxidant, antiphlogistic
Bilberry, a close relative to blueberry is used as common food item in several countries. Additionally, it is a seasonal product, therefore intakes might occasionally be high.
Cinnamon Cinnamomun zeylanicum
Health effects: Genotoxicity, antioxidant, several others to be considered
Cinnamon was selected from Annex 2 listing. It's used commonly as spice in several dishes, particularly in the Asia.
Maidenhair tree Ginkgo biloba L.
Health effects: Memory and concentration enhancer
We have expertise on gingko inside the PlantLIBRA project so this one is a natural choice.
Consultation of interested parties
Special expertise is needed to make sure that other PlantLIBRA WPs related to benefits/risks provide the necessary data and check that all the relevant fields are available in the database (e.g.. effective dose, intake information, etc.). Below, other experts related to project are proposed:
- Prof. Galli - University Of Milan
- Bernard Bottex – safety assessment guidelines of EFSA
- Prof. Silano
- Prof. Vartiainen - EVIRA, Finland
A review of the concepts of validation and fit-for-purpose in the field of risk, benefit and risk- benefit assessment will be conducted by Hylo in peer-reviewed literature and reported (DWP5- 1). Following review by partners and advisory boards, a protocol with requirements to assess fit-for-purpose and to validate (potentially against unused data-sets) the different assessment methodologies will be prepared using the Opasnet platform and reported by THL (MWP5-2).
Task leader: THL
PM: THL 6, Hylo 3
Safety factor thinking would lead to bizarre outcomes with benefits. For risks, the question is: "Taken into account human variability, what is a dose where we can be sure that there is no risk?" Then we apply a safety factor, e.g. divide a NOAEL with 10. But with benefits, using the same logic, we should ask: "Taken into account human variability, what is a dose where we can be sure that there is no benefit?" Although the question is logical, we have no use whatsoever for the answer. Instead, if we ask: "...what is a dose where we can be sure that there IS benefit?", then we must confess that there is no such number that we could use to multiply a dose; many compounds simply do not have any benefit, whatever the dose.
The working conclusion is that for a coherent BRA method, safety factors should not be used at all.
- Margin of exposure for selected food supplements
- State of the art of environmental health assessments
Assessment method, plant-based food supplements, benefit-risk assessment.
- The research leading to these results has received funding from the European Union Seventh Framework Programme under grant agreement number 245199.
- R. A. Hites et al. Global assessment of organic contaminants in farmed salmon. Science, 9 Jan. 2004, p. 226
- Tuomisto JT, Tuomisto J, Tainio M, Niittynen M, Verkasalo P, Vartiainen T, Kiviranta H, Pekkanen J. Risk-benefit analysis of eating farmed salmon. Science. 2004 Jul 23;305(5683):476-7 Read the article