Part A Background
The purpose of the Australian Total Diet Survey (ATDS) is to estimate the level of dietary exposure of Australian consumers to a range of pesticide residues and contaminants that can be found in the food supply. Pesticides are used for agricultural and veterinary purposes for the control of unwanted insects, mites, fungi, rodents, weeds, nematodes and other pests, and for the control of diseases in farm animals and crops.
Pesticides have been used in world agriculture for many years and provide important benefits in agriculture, resulting in a number of benefits to society. Their use provides the community with year-round availability of, and improved quality and variety in, our food supply, and leads to the production of food at a cost to the consumer that would otherwise not be possible.
Although pesticides present the community with significant benefits, there are risks associated with their use. In order to ensure safe pesticide use, a number of Australian government agencies assess the safety of food which contains residues of pesticides before the pesticide is approved for use in Australia. These agencies must be satisfied that the use of the pesticide will result in no appreciable risk of adverse health effects.
Origin of the survey
In Australia, the National Health and Medical Research Council (NHMRC), at its 68th session held in 1969, recommended that a ' market basket ' survey be carried out to examine the levels of pesticide residues and contaminants in foods that constitute a significant part of the normal Australian diet.
The NHMRC conducted the first total diet survey in 1970. Another 15 surveys were conducted by the NHMRC before responsibility passed to the predecessor of the Australia New Zealand Food Authority (ANZFA), the National Food Authority. The 19th ATDS is the fourth survey to be conducted by ANZFA or its predecessor.
The ATDS is conducted approximately every two years. The sampling and analysis of foods usually take place in one year, and the report writing and planning for the next survey take place in the following year. Publication of the report follows peer review of the survey.
Pesticide and contaminant surveillance in Australia
The Commonwealth Government, through the Department of Agriculture, Fisheries and Forestry' Australia, conducts two further programs that collect information on the levels of pesticide residues and contaminants in foods:
- the National Residue Survey; and
- the Imported Food Program, conducted by the Australian Quarantine and Inspection Service, which undertakes the surveillance of imported foods to ensure that they comply with the Food Standards Code.
The main aim of these programs is to monitor pesticide residues and contaminants in food commodities in export and import trade respectively. In contrast, the ATDS aims to estimate the level of dietary exposure to pesticide residues and contaminants in the overall Australian diet.
In addition to these programs, State and Territory health and agriculture authorities carry out surveys of specific contaminants or pesticide residues. These surveys usually investigate specific concerns and determine whether primary producers are complying with the law. They are a valuable source of supplementary information on the contaminant and pesticide status of foods.
Comparison with other surveys
The ATDS differs from other surveys of pesticide residues and contaminant levels in the following ways:
- The ATDS monitors the level of certain substances in the total diet to determine whether they pose an unacceptable risk to human health. Other surveys examine the level of residues and contaminants in individual raw agricultural commodities or foods to determine compliance with the law but do not carry out a comprehensive examination of their significance in the diet.
- The ATDS contrasts with other national surveys in that all ATDS food samples are prepared to a ' table-ready ' state before they are analysed, that is, they are subjected to prescribed preparation or processing steps. Food preparation varies with the type of food. For example, fruits may be peeled if they are usually eaten without their skins, while beef is dry fried because this food is nearly always consumed after cooking. As food preparation is known to affect the concentration of pesticide residues or contaminants in the food, an analysis of prepared foods more accurately reflects the levels of residues or contaminants that are likely to be consumed.
Using information from the survey
Data from the ATDS provides background information for developing food regulatory measures. The ATDS data on the dietary exposure to agricultural and veterinary chemicals is used as a check on exposure assessments undertaken during the registration process at the National Registration Authority for Agricultural and Veterinary Chemicals (NRA).
In addition, the results of the survey are a source of information for Australia' s contribution to the World Health Organization/Food and Agriculture Organization (WHO/FAO) Global Environmental Monitoring System, which monitors food contamination internationally, the Codex Committee on Pesticide Residues, the Codex Committee on Food Additives and Contaminants, and independent researchers both inside and outside government agencies.
Conducting the survey
As usual, this survey was coordinated by ANZFA in cooperation with each of the States' and the Northern Territory' s departments of health or equivalent. A working group, including liaison officers nominated by each State and the Northern Territory, was formed to advise ANZFA on the food and contaminants to be examined in the survey. Other participants in the working group were representatives of the Australian Government Analytical Laboratories (AGAL) and the National Residue Survey as well as ANZFA staff.
State and Territory officers were responsible for arranging the purchase and preparation of food samples. Food was sampled in each State capital city and Darwin - seven jurisdictions in all. Food was sampled over an entire year in four batches in February, April, August and October 1998. This accommodates seasonal variation in foods and allows for the sampling of foods that are available only in certain seasons.
The food was prepared according to strict instructions, frozen and dispatched to AGAL. Analytical chemists in the laboratories undertook the chemical analyses of the foods in accordance with the quality assurance procedures in Part 5 of the supplementary information available on ANZFA' s website. Following analysis, the results were sent to ANZFA where the total dietary exposures were estimated and a report prepared. States and Territories were informed of any results that may indicate a breach of the Australian Food Standards Code.
Foods included in the survey
Foods were sampled according to a schedule that categorises them into core, national or regional foods. This allows a good overview of the Australian diet.
Core foods were defined as foods central to the Australian diet, such as bread, beef, eggs, milk, lettuce, orange juice and potatoes. Lamb' s liver is also sampled as a core food although it is not consumed to a great extent in Australia. Liver, as a major organ of detoxification, may contain high levels of contaminants or pesticide residues and therefore it is appropriate to examine it at the same level as core foods.
Composite samples of core foods, consisting of four purchases each, were collected in each of Australia' s six States and the Northern Territory in each of the four seasons. This would ordinarily result in 28 composite samples of each core food. However, floods in the Northern Territory during one season meant that core foods were sampled only three times from Darwin. This meant that there were only 27 core food samples for the 19th ATDS.
Regional foods were defined as those foods that might be expected to show regional variation of residue and contaminant levels. Regional foods include fruits, vegetables and meats. Three composite samples of these foods, consisting of three purchases each, were collected in each of Australia s six State capital cities and Darwin, making 21 composite samples for each regional food.
National foods were defined as those foods that are available nationwide and are not expected to show regional variation. They are foods, such as cornflakes, canned salmon and infant cereal, that are distributed nationwide from a small number of outlets. Three composite samples, of three purchases each, were collected in three capital cities, making nine composite samples for each national food.
The ATDS Working Group chose foods according to the following criteria:
- The samples in each survey must consist of representative foods from each major food group and therefore the total foods surveyed must be consistent with a nutritionally acceptable diet.
- The most commonly consumed food in each food group, as shown by the National Dietary Surveys (NDS), should be analysed. If the food was examined in a recent survey and caused no concern, another food from the group may be chosen.
- Foods that may be of particular interest from a pesticide or contaminant viewpoint may be included in the survey, although their intake may be low.
- Foods may be included if they form a significant part of the diet of a subpopulation of Australians. For example, wheat is by far the most popular cereal and is the basis for many foods. However, some individuals cannot consume wheat without ill effect and must substitute other cereals. For this reason, rice, tahina and red kidney beans were also examined.
The foods surveyed in the 19th ATDS are shown in Table 1 of the supplementary information on ANZFA' s website. All the foods examined in the survey were prepared to a ' table ready ' state before analysis. For example, meats and eggs were cooked, while fruits that are normally consumed without peel were peeled. In preparing food as ' table ready ' , local tap water is used rather than distilled water to ensure that pesticide residues and contaminants that may be present in tap water are taken into account in the overall estimate of dietary exposure.
Pesticide residues and contaminants examined
All foods were tested for a range of pesticide residues including residues of chlorinated organic pesticides, organophosphorus pesticides, carbamates, synthetic pyrethroids and fungicides (see Table 6 in the supplementary information for a complete list). Fruits and vegetables were also analysed for dithiocarbamate fungicides. All foods were tested for the contaminants antimony, arsenic, cadmium, copper, lead, mercury, selenium, tin and zinc. Walnuts, tahina and roasted salted peanuts were tested for aflatoxins and milk samples were examined for the presence of Aflatoxin M1. All foods were also tested for polychlorinated biphenyls (PCBs).
A glossary of terms used in determining safe exposures and regulatory limits for pesticide residues and contaminants is included in Part 4 of the supplementary information.
What is dietary modelling?
Dietary modelling is a scientific method for estimating the levels of pesticide residues or metal contaminants a person or population may be eating. Dietary modelling techniques have been used by food regulators internationally for a number of years to check that dietary exposure to pesticide residues and metal contaminants is not likely to represent an unacceptable risk to public health and safety.
Dietary modelling is an important part of the ATDS as it translates analytical results for individual foods into exposure data on the total diet that can be compared to health standards. It is generally the exposure to the chemical from the total diet that is of interest when looking at health outcomes rather than the consumption of specific foods.
Dietary modelling at ANZFA
ANZFA uses dietary modelling in areas of work other than the total diet survey. On receiving an application to vary the Food Standards Code, ANZFA must undertake a comprehensive risk assessment. Dietary modelling is an important step in the risk assessment of applications relating to food chemicals, for example pesticide residues, contaminants, food additives or nutrients. Further details on dietary modelling can be found on ANZFA' s website and are available upon request from ANZFA.
How is dietary modelling conducted?
DIAMOND (Dietary Modelling on Nutritional Data) is a computer program developed by ANZFA to computerise dietary modelling calculations. The amount of chemical in each food is multiplied by the amount of that food we consume and summed over all foods to determine the amount of chemical in the whole diet. Once dietary exposure to the chemical from the total diet has been estimated, this is compared to reference health standards to assess the potential risk to human health.
Reference health standards are Acceptable Daily Intakes (ADIs) for pesticide residues and tolerable limits for metal contaminants. These are the amounts of chemical or contaminant that can be consumed without appreciable risk on a daily or weekly basis.
The chemical levels used in dietary modelling for the ATDS are representative levels taken from the tests on each surveyed food conducted by the AGAL. The data on amount of foods consumed are taken from the recent Australian National Nutrition Survey (NNS) that was conducted in 1995 and released in 1998.
A major step in dietary modelling is matching (or mapping) the ATDS foods to the foods reported as consumed in the food consumption data (the NNS foods). This process assigns the pesticide residue and contaminant levels detected in the ATDS survey foods to the appropriate food consumption data to estimate dietary exposure to the chemical. Given that the ATDS cannot survey all foods in the food supply, a single ATDS food (for example milk) may be assumed to represent a whole group of foods (for example milk, ice cream and dairy fats) with appropriate adjustment factors for concentration. Recipes are used for mixed foods to assign ingredients to the appropriate ATDS food. Food mapping is based on traditional nutritional groupings as well as potential or possible pesticide use.
It is recognised that registered pesticide uses may apply only to specific crops (often major crops) in the crop group rather than to the whole group. Therefore, the assumption of a certain residue level in the whole group is conservative in those cases.
Changes from the previous Australian total diet survey
The methods of dietary modelling have been improved from previous surveys, mainly through the use of the DIAMOND technology and the use of more recent food consumption data from the 1995 NNS instead of the 1983 and 1985 NDS used in previous Australian Market Basket Surveys (AMBS).
Use of DIAMOND for dietary modelling brings many benefits. DIAMOND enables the dietary exposure assessments to be conducted more efficiently and accurately. Records from the NNS of actual diets for approximately 13,500 people of all ages are used in place of ' average ' diets that were used in previous surveys. This means that dietary exposure is calculated for each individual in the survey before deriving mean dietary exposure results. Use of this up-to-date food consumption data greatly improves the reliability and accuracy of the dietary exposure estimates, and takes account of the different eating patterns of consumers.
Use of median levels for contaminant concentration levels
In choosing a metal contaminant concentration level for use in dietary modelling, ANZFA used the statistical middle value (median), rather than the mean level as in previous surveys, to represent the most likely level of contaminant in any given commodity. The median level is a more stable central statistic and is not sensitive to skewing by chemical detections above the normally expected range. The median simplifies calculations for surveys containing analytical results below the limit of reporting (LOR) because the position of the median, unlike the mean, is not dependent on the treatment of results below the LOR (WHO 1997). The use of medians is consistent with international practice (WHO 1997) and was used in ANZFA' s review of metal contaminants in food (ANZFA 1999).
In choosing a pesticide residue concentration level for use in dietary modelling, ANZFA chose the mean level consistent with previous surveys as it was recognised that given the high number of results below the LOR the mean level better accounted for detected levels of pesticide residues.
Means and medians are generally well correlated where there are few results reported below the LOR. This is demonstrated by the results for copper and zinc, where the means and medians are very similar.
Limitations and assumptions
Although improvements have been made to the methods of estimating dietary exposure, limitations do exist in the methods as well as in the data itself. For example, we draw conclusions about lifetime eating patterns from food consumption data derived from a single 24-hour diet. More comprehensive data on multiple-day intakes may provide better estimates of long-term dietary exposure and food consumption.
Assumptions were also made about the value of analytical results below the LOR. In the case of pesticide residues, the results that were lower than the LOR were assumed to be zero in dietary modelling. Given that pesticides are selectively applied to food crops, it is reasonable to assume that pesticide residues are not present when pesticide residues are less than the LOR. However, in the case of metal contaminants that occur naturally in the environment, it may not be reasonable to assume that the contaminant is not present at all in the food. For this reason, results below the LOR could be anywhere between zero and the LOR. Results for dietary exposure to metals were presented as a range, based on the two concentration levels of zero and the LOR assigned to results below the LOR.
Construction of the infant diet
As there were no data available from the NNS on children under two years, a diet was constructed to estimate dietary exposure for infants. Recommended energy intake for a nine-month-old boy at the 50th percentile weight was used as the basis for the model diet (WHO 1983). Boys - weights were used because boys tend to be heavier than girls at the same age and therefore have higher energy and food requirements. It was assumed that 50 per cent of the energy intake was derived from milk and 50 per cent from solids (Hitchcock et al. 1986). The patterns of consumption of a two-year-old child from the NNS were scaled down and used to determine the solid portion of the nine-month-old' s diet. Certain foods such as seafood and nuts were removed from the infant diet as it was assumed that infants do not generally consume these products. Consumption of breakfast cereals was assumed to be in the form of infant cereal. All milk consumption was assumed to be in the form of infant formula.
In past surveys, ANZFA has calculated dietary exposures for ' high consumers ' in the population (those eating more than the average person). These exposures were prepared by multiplying the mean dietary exposures by a factor. This factor was calculated by dividing the 95th percentile energy consumption by the mean energy consumption. This approach does not provide an ideal estimate of dietary exposure for high consumers of chemicals because it is based on the energy content of foods and not on the amount of foods consumed. It assumes that the dietary patterns for a mean and a high consumer are the same. This is unlikely.
In assessing the data for the 19th survey, ANZFA investigated the approach of using DIAMOND and the 1995 NNS to calculate the dietary exposure for ' high consumers ' . However, the 1995 NNS is based on 24-hour food consumption data, and research suggests that such surveys underestimate the food consumption for ' low consumers ' and overestimate consumption for ' high consumers ' (Institute of European Food Studies 1998). This is because no one eats the same food in the same amount every day. Surveys conducted over longer periods account for variability in the diet and therefore provide more accurate food consumption data for low and high consumers. However, 24-hour food recall surveys are still appropriate for providing food consumption data for the average person because mean population levels are used. These are generally not affected by the day-to-day variability in our diets.
The lack of reliable food consumption data on high consumers means that it is not possible to calculate realistic dietary exposures for high consumers. ANZFA will be investigating these problems in future surveys to determine if there are other techniques or data that can be used to estimate dietary exposures for habitual high consumers.
Dietary exposure estimates based on the 1983 and 1985 National Dietary Surveys and the 1995 National Nutrition Survey
In addition to the dietary exposures calculated for the 19th ATDS, dietary exposures were also estimated using analytical results from the 1996 AMBS (18th survey) for pesticide residues and contaminants, and food consumption data from the 1995 NNS. The results of the 18th survey previously published were based on food consumption data from the 1983 and 1985 NDS. This enabled comparisons between the 1996 AMBS (18th survey) with the 19th ATDS based on the same food consumption database and using the same dietary modelling techniques (DIAMOND computer program). It also enabled comparisons between the 1996 AMBS dietary exposure estimates based on the 1983 and 1985 NDS and those based on the 1995 NNS data.
A study currently being undertaken by the Australian Food and Nutrition Monitoring Unit, comprising a consortium from The University of Queensland, The University of Sydney and Deakin University, is investigating comparability between the 1983 and 1985 NDS data and the 1995 NNS data. This work aims to determine whether differences in estimates of average food intake between the surveys were due to actual changes in food consumption patterns during the period or due to differences in the sample design or the food intake methodology used to collect the survey data. The Commonwealth Department of Health and Aged Care is funding the work, which is known as the Bridging Study.
Preliminary results reveal that sample design variations, especially changes in the age-range of adult respondents, help to explain observed differences in average food intake estimates between the surveys. Other sample-design-related factors, such as differences in under-coverage and non-response rates between the surveys, are also likely to explain changes in the food intake estimates.
However, these sample design differences were not expected to greatly influence the comparison of food chemical exposure estimates between the 1996 AMBS and the 19th ATDS. In particular, the age-range differences are not relevant to ATDS exposure estimates, which are limited to specific age and sex subgroups. Therefore, it was possible to make direct comparisons between these data sets. Comparisons of the recalculated 1996 AMBS data with the 19th ATDS are detailed in Part B Results.
On comparing the exposures based on the two food consumption databases and 1996 analytical data, exposures for all age groups for both metals and pesticides were very similar. This suggests that general eating patterns and food intake for the identified age groups from 1983 and 1985 to 1995 did not differ significantly enough to influence the estimates of exposure from the total diet. These results also confirm the introduction of DIAMOND technology to estimate dietary exposures for the ATDS.
The 20th survey will, however, review the age groups used in calculating dietary exposures to ensure that they are in line with changing demographics.