1.1 Introduction
Persistent organic pollutants (POPs) comprise a wide range of environmental chemicals that can occur in food. The most widely characterised and studied are undoubtedly the polychlorinated dibenzo-p-dioxins (PCDD) and diben-zofurans (PCDDF) and polychlorinated biphenyls (PCB), both of which were included in the first listing under the Stockholm Convention. Also of concern and the subject of significant research, particularly in the last decade, are the brominated flame retardants (BFR), such as polybrominated diphenyl ethers (PBDE), listed under the Stockholm Convention since 2009 (UNEP, 2009), and hexabromocyclododecanes (HBCDD), currently under consideration for listing. A further group of recently-recognised halogenated POPs are the perfluorinated alkyl substances, of which perfluorooctane sulphonate (PFOS), its salts and perfluorooctane sulphonyl fluoride were also added to the Stockholm list in 2009. Other categories of POPs are emerging, many with dioxin-like properties, including brominated and mixed halogenated dioxins, furans and biphenyls, polychlorinated naphthalenes and chlorinated and bro-minated polycyclic aromatic hydrocarbons (PAH).
Most POPs are anthropogenic in origin, although non-anthropogenic dioxins have been found in clay deposits from diverse geographical regions (Ferrario et al., 2000). Due to their persistent nature and ability to undergo long-range transport, POPs are ubiquitous in the environment. They are found in soil, lake and river sediment, benthos, water columns and the atmosphere. They enter and accumulate in the food chain through various pathways, such as deposition onto crops, soil ingestion by grazing animals, and bioaccumulation up through trophic levels. At concentrations typically found in food, the adverse health effects caused by POPs are almost entirely chronic and include cancer, disruption of the endocrine system, neurotoxicity, and damage to the developing foetus. Recent reports also suggest that some POPs may be obesogens (La Merrill and Birnbaum, 2011).
1.2 Dietary exposure and total diet studies (TDSs)
Where a POP is known or suspected to be present in food, there are two general areas in which information is required in order to assess the risk of adverse health impacts through consumption: toxic effects (acute and chronic) and levels of dietary exposure (of the general population, high consumers and sensitive groups). Detailed toxicological characterisation of POPs is rarely available, with most information coming from animal testing, quantitative structure-activity relationship (QSAR) models and epidemiological studies. The development and application of analytical methods to measure the typically low concentrations of POPs in foods is challenging and can be costly, although the cost is likely to be around an order of magnitude lower than that of large toxicological studies. Consequently, to provide evidence of the need for the latter it is generally preferable to begin with an assessment of levels in the diet. In simple terms, this can be achieved in one of two ways - either by measuring levels in individual food samples, targeting those in which the compounds of interest are most likely to be found at significant levels and then estimating exposure through consumption of typical portions, or by carrying out a total diet study (TDS), which attempts to estimate the exposure to the compound(s) of interest through the whole diet. For a TDS, large numbers of samples are required to account as far as possible for the full range of variables that might affect the level of the compound of interest in an individual sample (Peattie et al., 1983). To give an example, variables for a sample of beef may include the age, breed and gender of the source animal, its geographical origin, and the time of year it was slaughtered (due to seasonal differences in feeding). The latter is important as localised variations in levels of contamination are possible, for instance where cattle are grazed on flood plains (Lake et al., 2006), feeding regimen and general health, as well as the possibility of further contamination between slaughter and reaching the eventual retail outlet. The contaminant levels to which the consumer would be exposed would then be further influenced by the methods of preparation, e.g. trimming fat from a portion of meat and cooking (grilling or frying might remove further fat, and therefore lipophilic compounds). Similar variables would exist for other food types. Testing large numbers of samples of every possible food type individually would be highly resource-intensive, especially if different methods of preparation and cooking also had to be taken into account. Generally, therefore, testing is carried out on composites of large numbers of individual samples, representing consumption over a significant period of time.
In the UK, the preferred approach has been to purchase samples from a random selection of different locations (cities, large and small towns, rural centres) in England, Scotland, Wales and Northern Ireland, weighted according to population distribution, over a 12-month period. This is intended to take account of geographical differences and differences in production scale as well as possible climatic and seasonal variation. Contaminants susceptible to such variation include agricultural contaminants such as nitrates, for which levels in leafy greens tend to be higher when there is less sunlight, and myco-toxins, which are likely to be higher when climate conditions favour growth of field and storage fungi. Seasonal variation may also impact sources of imported food, notably between the Northern and Southern hemispheres. Levels of environmental contaminants in food would not normally be affected by climate, although seasonal differences could be associated with changes to feeding and husbandry practices, for example in winter when the main source of exposure changes from pasture to feed. Even then, short term changes would only be expected to be seen in milk and possibly free range eggs. Geochemical sources may account for geographical differences in metal concentrations in food where, for example, lead ores may lie just below or even extrude through topsoil. In the case of POPs, however, whilst they are ubiquitous at low levels, localised hotspots are also possible and these would be difficult to take account of in a randomised, widespread sampling programme. The long half-life of most POPs in animals also means that seasonal fluctuations in exposure are unlikely to be associated with significant variations in levels in the animal. Consequently, a TDS for POPs may not require full 12-month sampling.
Once individual samples have been collected they are prepared and cooked using a typical range of methods, then they are homogenised and grouped into composites. The choice of groups will be influenced by the nature and detail of the consumption information available. For instance, the UK uses 20 food groups (Table 1.1).
Table 1.1
Food groups used for TDSs in the United Kingdom
Bread | Oils and fats | Fresh fruit |
Miscellaneous cereals | Eggs | Fruit products |
Carcass meats | Sugars and preserves | Beverages |
Offal | Green vegetables | Milk |
Meat products | Potatoes | Dairy products |
Poultry | Other vegetables | Nuts |
Fish | Canned vegetables | |
The individual food groups can be tested for the POPs of interest and population exposure may be calculated based on information about the quantities of each food group in the typical diet, gathered as part of a rolling programme of National Diet and Nutrition Surveys (NDNS, Department of Health, 2011). Adjustments can be made for high consumers of specific food groups (for example, those who consume a lot of fish) as well as different eating habits across age and special interest groups (vegetarians, vegans etc.).
The information generated by a TDS has a number of uses. For contaminants newly identified as a risk to the food chain, TDS ...