Food Standards Australia New Zealand

• Learning Centre
• Media centre
• Translated material
• Publications
• Monitoring and Surveillance
• Science in FSANZ
  • Risk Analysis
  • Expertise
  • Dietary exposure assessments at FSANZ
    • Dietary exposure and intake assessments at FSANZ
    • FSANZ Dietary Exposure Assessment Computer Program
    • Food consumption data used in dietary exposure assessments
    • Protecting ‘high consumers’
    • FSANZ Dietary Modelling Review
  • Food Composition Program
  • Nutrient reference values in the Australia New Zealand Food Standards Code
  • FSANZ Fellows
  • Student projects
  • From the desk of the Chief Scientist
• APEC 2011
• About FSANZ

• Home
• Science and Education
• Science in FSANZ
• Dietary exposure assessments at FSANZ

Protecting ‘high consumers’

Background

FSANZ uses dietary modelling techniques to estimate the potential dietary exposure of different population groups to a range of food chemicals. These chemicals include food additives, contaminants, novel foods and food ingredients, pesticide and veterinary drug residues and nutrients. In simple terms, dietary exposure is estimated as:

 

Dietary exposure = food chemical concentration x food consumption

One population sub-group of potential concern is one we call ‘high consumers’. High consumers can be:

• people who consume a lot of one food that contains a chemical of interest
• people who consume smaller amounts of a number of different foods that all contain the same chemical, or
• people who consume small amounts of a food which contains a high concentration of a chemical.

Where relevant, FSANZ reports dietary exposure for high consumers when assessing dietary exposure to food chemicals, as well as population average exposure, so that we know we are considering the potential risks across the population.

How do we identify the high consumers?

We use Australian and New Zealand national nutrition survey (NNS) data to model exposure to food chemicals. These surveys only captured a single 24 hour time period for all participants, although a subset were studied for two separate 24 hour periods. Across the thousands of people who participated in these surveys, we will typically find a wide range of eating patterns and therefore a wide distribution of dietary exposure to food chemicals.

Part of the survey population may have no exposure to the chemical in question, because they didn’t report eating food/s containing that chemical on the day of the survey. These are the ‘non-consumers’. The sub-group who are exposed, the ‘consumers’ are often the focus of a dietary exposure assessment. Each consumer’s exposure will be estimated from their individual consumption records and then ranked from highest to lowest. Average consumer exposure will be reported, as will that of the high consumer. Until recently, we have used the 95^th percentile of exposure as the point in a distribution of exposures that represents a high consumer. This means that 95% of the population will be exposed to less than this amount and 5% will be exposed to more than this amount on any one day.

There are international conventions regarding dietary exposure assessment reports in regulatory risk assessment. Where possible, FSANZ adopts principles and procedures from international best practice. However there is no agreed international position for reporting cut-offs for chronic (long term) dietary exposure estimates for high consumers of a food chemical. This is because each country or region has different types of food consumption surveys to draw on. When assessing dietary exposure for high consumers, countries with one or two days of food consumption data tend to report at a lower percentile than those with surveys of a longer duration. This is because use of only one or two days of data tends to exaggerate high consumption, and therefore dietary exposure, as discussed below.

Following a recent international peer review of our dietary modelling program and procedures, we have changed the way we report chronic dietary exposure for high consumers to align with the reviewer’s recommendations.

For more information on the peer review see the final report . (Hyperlink to report to be inserted ) .

What has changed?

In summary, FSANZ has changed the ‘definition’ of a high consumer from being represented by the 95^th percentile of dietary exposure to the 90^th percentile of exposure. We did this because we believe it is a more realistic measure of what might be a population’s long term exposure to a food chemical.

This change will only apply to chronic dietary exposure assessments for food additives, contaminants and novel foods and ingredients, where estimates of dietary exposure are based on food consumption data from a single 24-hour recall from NNSs. FSANZ will use this approach for standards development work, other ad-hoc risk assessment and dietary exposure assessment for surveillance activities (e.g. Australian Total Diet Study).

We will use other percentiles to represent the high consumer where two days of consumption data are available, for nutrient intake assessments and for exposure assessments of agricultural and veterinary chemical residues (see below for further information). In some circumstances risk assessors and/or risk managers at FSANZ may choose to use other reporting cut-off points, depending on the purpose of the risk assessment. In these cases, the reasons for doing so would be fully explained in the relevant FSANZ report, along with any accompanying evidence supporting the decision

Supporting evidence for the change

FSANZ sought scientific evidence to support this change. We reviewed and evaluated published literature written by experts in the area of food consumption data, risk assessment and dietary exposure assessments. What follows is a summary of the findings of this research.

Using one 24-hour food consumption record, which is generally what is available for Australia [1] and New Zealand, may capture an unusual eating event for an individual that does not describe how they normally eat. This could potentially over or under-estimate their typical food consumption. It could also exaggerate the reported extremes of food consumption across the survey group – on the day of the survey they may have eaten much more or much less of a food than their usual eating pattern.

We found that mean daily food consumption amounts estimated from survey data for consumers of those foods may decline as the length of a survey increases (see Figure 1), depending on the type of food consumed. European research [2] indicates that for frequently consumed foods, the mean amount of food consumed per day, calculated from 1, 2, 7 or 14 day data may not change significantly. For example, we continue to eat two slices of bread or drink a 200 ml glass of milk every day. However, the daily mean consumption of occasionally consumed foods will decrease if more than one day of data is considered. Figure 1 clearly shows this for pizza.

The distribution of food consumption amounts for a survey of one 24-hour period is much broader than that of two or more days. Therefore, the number of days of food consumption data affects the predicted high food consumption amount, and therefore affects estimated high consumer dietary exposure, particularly for food chemicals in occasionally consumed foods. Figure 2 shows, for sausages, how the use of one day’s consumption data exaggerates high consumption amounts. The data in Figure 2 are from the subset of 1995 Australian NNS participants who completed a second 24-hour recall and ate sausages on both survey days. In this example, the 95^th percentile of sausage consumption based on one day of data was 12% above the 95^th percentile of consumption assessed over two days, and 60% above at the 99^th percentile.

In addition, population food consumption, and hence dietary exposure, generally has a right hand (positively) skewed distribution. This means a few high consuming individuals could artificially increase the mean and high percentile food consumption amounts. Consequently, the estimated dietary exposure based on these data would also be higher. Where there are relatively few consumers of a food containing a specific chemical, a single high food consumption amount is more likely to distort the dietary exposure distribution.

Figure 1: Effect of survey duration on estimate of daily amount of food consumed (adapted from IEFS 1998)

 

Figure 2: Effect of survey duration (one or two days) on estimates of daily amount of sausages consumed (adapted from NNS 1995 using DIAMOND)

In summary, we believe that we should use the 90^th percentile of exposure to represent the high consumer when only single 24-hour records are available for use in assessments examining chronic dietary exposure.  

Assessments for nutrients and agricultural and veterinary chemical residues

This change to the use of the 90^th percentile does not apply to estimated dietary exposure for agricultural and veterinary chemical residues. We estimate chronic dietary exposure for agricultural and veterinary chemical residues using a specific internationally accepted methodology where only the mean (average) dietary exposure for the population is reported. Acute, or short term, dietary exposure estimates for agricultural and veterinary chemical residues are based on food consumption data at the 97.5^th percentile for consumers of the food, which is also an internationally accepted methodology (WHO, 2008) [3] .

Furthermore, this change does not apply to nutrient intake assessments. For nutrients, we have two days of intake data for a subset of NNS respondents. Using an established statistical adjustment method we use this second day of nutrient data to estimate ‘usual’ nutrient intake in all NNS respondents, where the intakes are normally distributed. This adjustment can be used for nutrients, as opposed to other types of food chemicals, because nutrients are widely dispersed in foods and therefore all respondents will have a nutrient intake on both days on which they were surveyed. A 10% (1995 Australian NNS), 15% (1997 and 2002 New Zealand NNSs) or 100% (2007 Australian NNS [4] ) sub-sample for which two 24-hour recalls are available is generally sufficient to allow estimation of the within-person standard deviation for nutrients.

FSANZ considers it reasonable to continue to use the 95^th percentile to report high percentile nutrient intakes after an adjustment has been made for the second day of intake. Figure 3 illustrates the effect of using this statistical adjustment on the predicted distribution of nutrient intakes and the potential this can have to alter the interpretation of a population’s nutritional status, that is, the proportion of a population estimated to be above or below a reference health standard.

Assessments for acute exposure to food chemicals

An acute exposure assessment examines dietary exposure over a short time period, for example a single meal or 24 hour period only. It is not concerned with long term exposure. Therefore it is appropriate to continue to use a high percentile (such as the 97.5^th consumers only) from single day food consumption records to represent the high consumer.

Figure 3: Comparison of the predicted distribution of nutrient intakes estimated using one day of food consumption data or data adjusted to reflect longer term intakes. Points A and B represent lower and upper reference health standards, respectively

Risk management implications

Overestimating chronic dietary exposure assessments (using the 95^th percentile for high consumers) could potentially result in FSANZ adopting a more stringent risk management approach than is required to protect public health and safety. Therefore, we believe it is important to employ international best practice and sound science when estimating and reporting dietary exposure for high consumers. By reporting high consumer dietary exposure at the 90^th percentile, we ensure that the estimated exposure is not greatly overestimated. At the same time, we will continue to ensure that any risk management strategy chosen will protect high consumers.

Further information

You can find more detailed information on FSANZ’s dietary exposure assessments in “Principles and Practices of Dietary Exposure Assessment for Food Regulatory Purposes ”