Creme Global

What is Probabilistic Food Safety Exposure

- By:

Courtesy of Creme Global

What is Probabilistic Food Safety Exposure Assessment?

There are a myriad of chemicals in food products consumed by populations. Each chemical in its own right may not be harmful to the population, but in combination with other chemicals, there may be risk of high levels of exposure to potentially harmful substances.

An effective probabilistic exposure assessment needs to handle the variability and uncertainty of this real world situation. Variability characterises the natural differences between consumers' characteristics and habits. These are natural variations such as people's bodyweight, different people eating different things or people changing their own eating habits from day to day. Note that the variability in these values would not be reduced by further measurement.

Uncertainty on the other hand is an actual lack of information about the situation. The exact concentration of a pesticide on a food item being consumed is not known exactly (this data is not measured for every single food item), hence the analyst is genuinely uncertain as to the value of the chemical concentration of each food item. The uncertainty of a situation could be reduced by further measurement, but this is not always practical or feasible.

Uncertainty and variability have to be treated using a probabilistic method such as the Monte Carlo simulation. Probabilistic methods are well established in other areas of risk analysis such as financial risk. By analysing large samples of the population's eating habits, including quantifying the concentration of the contaminants, hazards and chemicals in their diet, we get a better picture of how people's health is affected by what they eat.

In order to capture a full picture of the variability of the situation, an effective food safety assessment system should facilitate the use of the full consumption data set in your analysis. This will include all of the natural variability contained within that data set, such as individual bodyweight and consumption patterns. Considering variability facilitates an accurate analysis of key demographics of the population that may be at risk (for example children or consumers with a high intake of a certain chemical).

It can happen that concentration data include 20 different concentration measurements for a chemical within a certain food. All this data can be added to the exposure assessment by inputting the raw data itself or by fitting a distribution to the data and inputting the parametric distribution (such as LogNormal(2.3, 0.45)).

In the case where the raw data was inputted, a Monte Carlo approach will sample randomly from the raw data in order to assign concentrations to the relevant food consumption eating events. For the parametric approach, the system will sample from the parametric distribution assigning a concentration value from the distribution at each eating event.

Because of the stochastic nature of this process, the calculation has to be repeated a number of times each time running through all of the food consumption data sets and assigning a concentration value from the range of plausible values for each event. Repeating this process allows the analyst to calculate the 'expected value' of the statistics of interest (such as the 97.5th percentile population exposure), it also allows the analyst to report the confidence intervals for the expected value.

A common issue in the area of food safety assessment is the use of vastly conservative deterministic estimates of exposure. These estimates are arrived at by multiplying two ultra-conservative figures together (such as an assumed extremely high food consumption multiplied by the maximum legal limit of chemical concentration). While these conservative estimates are designed to ensure that the population is protected from chemical exposure, they do not provide any understanding of the actual exposure levels of the population to chemicals.

It is important not only to assess the exposure based on a conservative approach, but also to assess how likely that exposure level is to be reached by the population. This is only possible through analysis of all eating event data available.

CREMe specialises in handling these issues for our clients, who include government regulators, food and packaging manufacturers, consultants and researchers. With our state-of-the-art software service which utilises techniques that are used by major financial institutions to manage financial risk, we will provide the rapid and reliable results that you need.

Customer comments

No comments were found for What is Probabilistic Food Safety Exposure. Be the first to comment!