ILSI Uncertainty Workshop – Addressing the Known Knowns, the Known Unknowns and the Unknown Unknowns
Written by Robert Coyle
Uncertainty analysis refers to techniques used to account for the aspects of decision making problems that are imprecise or not completely known. The aim of the workshop was to gather stakeholders interested in dietary exposure assessments to discuss the identification, description and classification of uncertainties and their effect in scientific evaluations.
The workshop was organised by an expert group brought together by ILSI Europe and chaired by Susanne Kettler, Group Director for Scientific and Regulatory Affairs at the Coca-Cola Company. Susanne kicked off proceedings with the following delightful nugget from Donald Rumsfeld (at the time US Secretary for Defense):
…as we know, there are known knowns; there are things we know that we know. There are known unknowns; that is to say, there are things that we now know we don't know. But there are also unknown unknowns – there are things we do not know we don't know.
The quote set the scene nicely. In the risk assessment of chemicals in foods or in assessing nutrient intakes in a population, dietary exposure plays a key role. However, an exposure assessment typically involves thousands of inputs (e.g. food consumption surveys, food composition data, chemical occurrence data) and often uses sophisticated statistical and probabilistic modelling techniques. At the end of this process we get an answer (usually summarised by statistics) that is used to describe risks or benefits in a population of consumers. Considering the huge volumes of data that went into the assessment and the inescapable fact that all models are imperfect reflections of reality, we are obliged to ask the simple question: how certain are we about our answer?
The workshop proceeded with a set of presentations on how uncertainties have been evaluated in different areas, in particular flavourings, food additives and total diet studies. Cronan presented a talk on how we used probabilistic modelling to evaluate additive exposure in the FACET project and how the uncertainties were captured in the model and software. After lunch, there followed a break-out session where I ran two one-hour long workshops on some worked examples of uncertainty assessments from the FACET project. I broke the participants into groups and we worked through some actual examples of uncertainties from the project – just to show how challenging it is to carry out this type of analysis!
With participants from EFSA, the WHO, the US FDA, the German Cancer Research Centre, the European Commission, academia, as well major multinationals such as PepsiCo, P&G, Danone and DSM, lively debate ensued throughout the day.
Considering that uncertainty has the potential to impact on any scientific assessment, particularly risk assessment, how we evaluate it will be an ongoing issue. Beyond this, successfully communicating uncertainties to risk managers and those tasked with making decisions will be key – poor communication leads to poor understanding. We can therefore look forward to (or for some, anticipate with trepidation) interacting with uncertainties more and more in the fields of risk, aggregate exposure and predictive intake modelling.
If you have any questions about any of the above or if you are interested in talking to us about accessing our software and services, please contact email@example.com.