Jouni Tuomisto, THL Risk-benefit assessment for plant food supplements (PFS)
New context: shared understanding • A large group of people who try to understand • What issues are on the table • Where are the agreements and disagreements • What are the reasons for disagreements • The aim to understand everyone’s point of view leads to inclusiveness. • ”Scientific” in this context means that a hypothesis can be openly criticised based on observations and relevance.
Guidance based on inclusiveness • Benefits and risks separately or combined? • Do both if you can. • Should DALYs or QALYs be used? • Do both if you can.
What impacts to include? • Health risks of compounds in products. • Health risks of behavioural changes due to consumption. • Health benefits of compounds in products. • Health benefits of placebo effect. • Health benefits of the habit of consumption (tradition of use).
Burden of proof? • Wide use of a product in a society implies health or other benefits in that society. • Requirement of substantial evidence of risk before banning a product. • If there is no use, burden of proof is rather not on risk side.
What do consumers expect? • Large variation between individuals for sure. • Probably those who consume products expect the products to be net beneficial. • Consumer values and expectations should be studied and used as a basis for assessments and regulation. • The motto of THL / Environmental health: • ”People must be able to drink, eat, breathe, and use consumer products trusting that their health is not in danger.” • This should be applied to PFS as a default.
Sharing is learning • People mostly think about one assessment at a time. • We should systematically collect and share data about all important and interesting compounds. • We should try to learn about groups of assessments and groups of chemicals. • Big data: e.g. Google’s influenza predictions..
Sharing is learning (2) • We will never have enough data about many single chemicals, but we do have good data about eg. cancer impacts of all chemical exposures combined. • We should aim at a systematic assessment of all exposures. That is a huge task, though.
Is it technically feasible? • The null hypothesis is usually: ”No impact” • But it could also be: P(X > x) • Probability that the true impact X is actually larger than value x. • It would be possible to assess the highest plausible impacts to see whether conclusions can be made. • Bayesian approaches do this naturally. • Quantitative assessment is feasible more often than thought.