J.H. Cunningham
Dr Judy Cunningham is an officer of the Commonwealth Department of Industry, Technology & Commerce, 51 Allara Street, Canberra ACT 2601.
Sampling of foods for analysis, and their subsequent preparation prior to analysis, are critical steps in the production of good quality and useful food composition data. If sampling and sample preparation are done incorrectly then all subsequent analyses are a waste of time and money, as a mistake in sampling can only be corrected by repurchasing and repreparing the sample. This review of the process of sampling and sample preparation for nutrient composition studies is written from a practical viewpoint pointing out important considerations and likely pitfalls. Sampling considerations are relevant not only to those who will actually be involved in the generation of data or the preparation of tables; users of food composition tables should also be aware of the factors that influence sampling and hence the analytical data ultimately reported.
In the present context, sampling is the selection and collection of items of food defined in number, size and nature to represent the food under consideration; and the consolidation and reduction of the collected items to form the portion for analysis (Horwitz & others 1978).
The aim of sampling for food composition purposes is to obtain an analytical sample that is representative of the foods available to, or consumed by, the population concerned (Greenfield & Southgate 1985). This aim implies that the participants in the program have a detailed knowledge of the population in question and its food habits. In general, food samples obtained from a single source or produced under special conditions are not suitable for generation of data for food composition tables as they are rarely representative of the foods available to the majority of the public (Greenfield & Southgate 1985). Before any samples of food are purchased for a food composition program, a sampling plan should be developed and assessed critically to ensure that the foods purchased will be representative of those available. The analyst and program supervisor should be closely involved in the development of the sampling plan. In Australian studies the usual practice for obtaining representative samples has been to sample foods available in the major urban areas due to constraints of cost and distance. In contrast, however, the bush foods program at the University of Sydney analysed samples primarily collected in the Northern Territory by Aborigines, Aboriginal health workers and dietitians (Brand & others 1983).
Type and source of food to be analysed
Within the general category of foods to be analysed a survey should be carried out of the range of products available and their sources. The same general type of food may be available from a range of outlets (restaurants, supermarkets, delicatessens, take-away food shops, fast food chains, health food shops), in a range of forms (fresh, frozen, canned, dried) and from a number of manufacturing sources (factory-processed, home-made etc). For example, in a 1981 study of Greek foods samples were purchased from delicatessens, restaurants and takeaway food bars (Greenfield & others 1983).
If a food is factory-produced it is likely to be relatively consistent in nutrient composition regardless of where it is purchased. Fresh produce, restaurant foods or home-made foods are much more likely to show wide variations in composition according to their source and the chosen sampling plan must reflect this variation. The more variable in nutrient composition a food is likely to be, the greater the number of samples that should be purchased to achieve a representative result. For example, in the University of NSW nutrient analysis programs the approach used has been to purchase four units of standardised food, such as a manufactured breakfast cereal, and ten units of non-standardised foods, such as meat pies produced in different small cake shops (Wills & others 1988).
One important variable that may affect non-factory produced foods is the socio-economic status of the area in which a sample was purchased. In a large program of analysis of the composition of Australian fresh meats, a sampling plan was devised, based on published demographic and social studies, dividing Sydney into ten ranked socio-economic areas and then selecting one suburb out of each of these ten areas from which to purchase meats (Greenfield & others 1987). Some foods of the type being examined may only be available in certain suburbs or towns; for example some fresh delicatessen-style products may be readily available in a yuppie suburb but may be unheard of in a middle class mortgage belt suburb. Similarly foods consumed by particular ethnic groups may only be readily available in certain cities or certain welldefined suburbs.
Suppliers of foods for analysis should not be made aware of the purpose for purchasing, to reduce bias in samples and the subsequent analytical results.
Type of representative sample
Ideally if one were to purchase ten units of a food sample (eg Valencia oranges) then each unit would be analysed separately and the final result calculated as the mean of these units. Such samples are referred to as single replicate samples. As this is obviously an expensive approach it is infrequently used and an approach described as single composite sampling is more common. With single composite samples, the ten units of Valencia oranges would be mixed together to produce a composite sample for analysis. A third class of sample is the multiple composite sample where foods from different regions, seasons, cultivars or brands are combined together to form one composite sample. Continuing the analogy of oranges, Washington Navel and Leng Navel oranges would be combined with Valencia oranges to form a composite sample called ‘oranges’. The multiple composite approach is most suitable for foods that are not staples of the diet, such as specialty fruit or frozen pizza (Greenfield & Southgate, 1985). If the food in question is a significant component of the diet separate samples are usually drawn for different cultivars, different growing areas, different seasons and different processing and preparation methods.
Trial runs
A trial run of the entire procedure should always be done to identify and resolve problems prior to the actual purchase of samples for analysis. Typical problems and questions include: what type of transport is needed? (is a small car satisfactory or will a delivery-size van be required?); what type of containers are needed for sample storage during transportation? (perishable products need to be placed directly into insulated containers); how long will it take to visit every outlet required? (in large cities it can take one or more working days just to collect the samples); if samples are being flown in from overseas, customs and quarantine clearances will be necessary; is there enough storage space in the refrigerator and freezer at the laboratory to store all the samples before and after preparation?; how long does it take to prepare one laboratory sample?; what equipment is needed to prepare and homogenise the sample?; is there sufficient time remaining on the day of preparation to perform the immediate analyses of moisture and vitamin C? (reagents and equipment should be prepared in advance for these analyses); is there a washing-up area readily available?; if samples are to be cooked are there hot plates, ovens and utensils available?
Sample preparation
Sample preparation often generates many problems. When dealing with perishable foods it is preferable to purchase only a few samples at a time so that the preparation and immediate analyses can be completed rapidly and without mistakes. A range of standard kitchen equipment is needed to prepare samples prior to analysis. This equipment includes cutting boards, knives, spoons, spatulas, strainers, bowls, blenders or homogenisers, plastic storage bags, aluminium foil and indelible marking pens. Plastic utensils are desirable where possible to reduce the risk of metal contamination. Ideally, a food composition laboratory will have a dedicated set of utensils as well as a dedicated sample preparation area. The sample preparation area should be free from environmental contamination such as dust and insects, particularly when trace metals are to be analysed.
Registering the sample
All relevant details about the sample should be entered immediately into a bound register. Loose pages are not acceptable as they can become lost. Registration details can be transferred at a later time to a computer system but it is often impractical to site a computer close to a messy food preparation area. Details that should be recorded about the sample include: common names(s), sometimes the same food will have different common names in different areas or among different ethnic groups and as many common names as possible should be reported; systematic name for plant foods or seafoods (in particular); serving or purchase size in grams or millilitres; edible portion weight together with a description of the material constituting both the edible and inedible portions; description of the state of the food as purchased (whether raw, cooked, frozen etc.); description of the components of the foods (eg the proportion of pastry in a pie) with photographs if possible; date marking or product code if available; place and date of purchase; name and address of manufacturer, if available; label claims and ingredient lists, if provided; details of the cooking method, if the sample is subsequently cooked (Greenfield & Southgate 1985).
Cooking of samples
Care must be taken in the selection of cooking method when samples are to be cooked prior to analysis. Popular cookbooks should be consulted to obtain recipes. In the case of foods from another culture, advice may have to be sought from members of this culture. For packaged foods it is acceptable to follow the manufacturer's instructions. If oil, fat or water are added during the preparation this should be recorded and, in the case of fats, the type of fat noted. Other details such as the amount of sample cooked, time and temperature of cooking and the type of cooking method and utensils must be recorded. Samples should be weighed before and after cooking (Greenfield & Southgate 1985).
Homogenising the sample
If reproducible, reliable results are to be obtained the laboratory sample must be reduced to a homogeneous mass. A poorly homogenised sample will invariably give nonreproducible results. A study by Wills & others (1980) found that domestic food processors were as effective as special laboratory homogenisers in obtaining homogeneous samples for a range of foods. Many foods are extremely difficult to homogenise and food processors may not be entirely satisfactory. In this case additional equipment and trials are required. For hard, dry foods such as grains and legumes it is generally necessary to use a laboratory mill or a coffee grinder. However the extensive grinding required can result in metal contamination of the food leading to high iron levels. A solution to the problem of metal contamination may be to use samples coarsely ground, by a mortar and pestle for mineral analysis, and a finely ground sample for the remaining analyses. Although sieving will remove large pieces of food some component of the food, such as the bran, may be selectively removed also. Extensive blending or grinding may produce sufficient heat to damage some vitamins. Sticky or sugary foods may require coarse grating and mixing rather than homogenisation. This is, however, likely to result in a greater degree of variation between analytical duplicates in subsequent analyses. Excessive blending of high oil content foods such as nuts should be avoided as the oil may separate. Similarly some fruits such as bananas or papaya should be mashed rather than blended at high speed.
During the process of sample preparation it is important to minimise sample exposure to light, air and heat as these will lead to destruction of vitamins. At least 500 g to 1000 g of sample should be obtained if possible. This should provide sufficient material for the inevitable repeat analyses.
Storing the sample
Samples should be stored in such a manner as to minimise deterioration or change in nutrient composition. Usually samples are frozen although some researchers have used freeze-drying (Brand & others 1983). Frozen samples must be stored in containers that are impervious to water. Strong polythene bags are satisfactory and cost-effective; the sample should be placed within three bags to minimise the risk of leakage, each bag being carefully sealed (eg with a heat sealer) and clearly labelled with the product identity. Plastic bags may not be suitable for high water content foods such as fruits, as the sample may swell when frozen and burst the bag. Rigid plastic dishes with strong seals may be used as an alternative. Headspace within solid containers should be minimised to reduce the likelihood of sample oxidation. Glass containers should be avoided where possible due to the risk of breakage. Code names on sample labels should be avoided unless an efficient and permanent record for deciphering the code is available. The homogenised samples should be divided into at least two separate sub-samples which should, preferably, be stored in different freezers to avoid the loss of all sample material in the event of a thaw-out. During frozen storage some of the water or fat in samples may separate out; in this case it is necessary to reblend the sample prior to taking an aliquot for analysis.
Conclusion
During the process of sampling and sample preparation compromises will have to be reached. Indeed the entire process of sampling can be regarded as a compromise because it is clearly impossible to sample every piece of food throughout the country. Provided the sampling process is approached with knowledge, a clear mind and careful planning this most vital of steps in a food composition program will, however, be successfully negotiated.
References
Brand, JC, Rae, C, McDonnell, J, Lee, A, Cherikoff, V & Truswell, AS. 1983. The nutritional composition of Australian Aboriginal bushfoods. I. Food Technol. Aust. 35: 293–8.
Greenfield, H, Lerogiannis, V, Makinson, J & Wills, RBH. 1983. Composition of Australian foods. 19. Greek foods. Food Technol. Aust. 35: 84–6.
Greenfield, H, Kuo, YL, Hutchison, GI & Wills, RBH. 1987. Composition of Australian foods. 33. Lamb. Food Technol. Aust. 39: 202–7.
Greenfield, H & Southgate, DAT. 1985. A pragmatic approach to the production of good quality food composition data. ASEAN Food J. 1: 47–54.
Horwitz, W, Cohen, S, Hankin, L, Krett, J, Perrin, CH & Thornburg, W. 1978. Analytical food chemistry. Inhorn, SL (ed). Quality assurance practices for health laboratories. Washington DC: American Public Health Association. 545–646.
Wills, RBH, Balmer, N & Greenfield, H. 1980. Composition of Australian foods. 2. Methods of analysis. Food Technol. Aust. 32: 198–204.
Wills, RBH & Greenfield, H. 1988. Laboratory instruction manual for food composition studies. Kensington NSW: University of NSW, Dept. of Food Science & Technology.
