Request trial now
Have us call you backMaking the Most of Your Food Consumption Databases
Comment on this article in CRN (Post Reply)
This article is taken from a talk I gave at the IFSTI annual meeting in Kilkenny recently. The PowerPoint presentation from that talk can be downloaded from the link below. In that talk, I asked the question:
What more can we do to really understand the exposure levels to food chemicals for different demographics groups of the population?
To answer this question, I discussed the following points:
- The available food consumption databases
- The use of the databases in assessing chemical intakes
- Handling variability and uncertainty.
- The role that Creme 2.0 Food can play in enabling you to get the results you need
Food consumption Databases
There are a range of food consumption databases available including the US NHANES food consumption database, the UK NDNS database, the Chinese food consumption database and the Irish (IUNA) food consumption databases. There are references and links to these data sets available in CRN (click here).These food consumption databases contain a detailed set of demographic information as well as detailed food consumption information for the population. Data sets typically include hundreds or thousands of subjects and a large number of eating events for each subject, including
- Habitual Food and Beverage consumption of each subject
- Height, weight, body fat content of each subject
- Habitual physical activity levels
- Lifestyle characteristics including smoking
- Attitudes to food and health
- Socio-demographic characteristics
The diary table contains every single eating event recorded for each subject. For each eating event, the subject code, the food code, the amount of food consumed (in grams), day, time and meal number is recorded. Further information such as brand code and packaging type can also be recorded.
See slides 8 and 9 of the PowerPoint file for screen shots of sample subject and diary tables.
Food group tables can be used to group food codes of similar chemical characteristics in order to allow chemical concentrations to be linked to groups of foods.
From chemical concentrations to population exposure levels
There is a large amount of data available about chemical or contaminant levels in foods. These come from specific studies in literature, industry information and routine government monitoring programmes.A chemical concentration table is used to store information about chemicals. For each chemical the presence probability and concentration of the chemical in each food or food group is stored.
See slides 16 and 17 of the PowerPoint file for screen shots of sample food group and chemical concentration tables.
Once these data sets are set up, you are now ready to run a probabilistic exposure assessment.
What happens in a probabilistic exposure assessment?
A probabilistic assessment method is required if some of the input values are uncertain or variable. Examples include a presence probability value that is less than 100% or a probabilistic chemical concentration value such as LogNormal(0.14,0.04) or Data(0.15,0.14,0.05,0.06,0.07,0.11). See slide 17 for more examples.For a probabilistic exposure assessment, the software tool runs through every eating event in the diary table. For each eating event the software identifies the food group of the food (if any) using the food groups table. The software will then look up the chemical concentration table for each eating event to assign a chemical concentration to the eating event. A random choice is made based on the presence probability as to whether the chemical was present or not and if present the concentration field is used to assign the chemical concentration value for that eating event. If the concentration is probabilistic a random value from the concentration probability function is generated.
The software will continue through every eating event in the diary file in this way, calculating the chemical intake for each event. Once complete, the total intake from all eating events is then calculated for each subject. We now have a chemical result for each subject in mg of chemical per day or mg per kg bodyweight per day. The mean intake and the high percentile intakes (e.g. P95) can now be calculated for the population.
If there were probabilistic inputs to the calculation, this entire process is repeated a number of times. This results in a range of mean intakes and a range of P95s. The variance in these results is due to the variability and uncertainty represented by the probabilistic input expressions. See slide 22 of the presentation for a typical set of results. You will see that in this case, the expected value of the P90 (the 90th highest person in the population) was around 0.087 mg/day within a range of 0.083 to 0.089 mg/day. These are confidence intervals in your exposure results based on your input data ranges and assumptions.
Creme Food Safety facilitates generating these results efficiently and easily for any selection of demographic groups in the population. This allows you to handle the variability and uncertainty in the assessment with confidence. For more information how we can help your organisation, please submit your questions or comments below.