Consumers fear the food they eat. A survey in Europe showed that people think that the risk from food is nearly twice as high as the risk from traffic accidents.

Fears have grown as food production and processing has become more industrialised, and sources of food have become more distant. The scale and impact of food scares in recent years, from mad cow disease to dioxins and most recently Sudan I and avian flu, have fuelled these fears.

Popular feeling is summed up in a recent article from the Guardian in the UK:
“We thought we were more likely to run risks by eating in a ‘hole in the wall’ take-way or from a fair ground stall, but now we see how the centralised supermarket food chain can be a highly effective machine for compromising the health of millions of people in a matter of hours.”

These fears cannot simply be addressed by scientific argument, or by pointing to statistics that show food poisoning has been significantly reduced over the last thirty years. Consumers don’t think that they are being given the real facts (less than 30% believe what they are told by food companies) and they are likely to react strongly – some say over-react – when they think that there is a problem.

The major Coca Cola recall in France and Belgium in 1999 had a direct cost to the company of $60 million, and an indirect cost to its brand image which was probably much higher. There was never any proven contamination of the product.

This food scare in Belgium occurred shortly after another major alert caused by the release of dioxins into the animal feed supply (a contamination which is believed to have come from just 1g of dioxins entering a tank of vegetable oil). This alert probably cost the Belgian food industry more than €1 billion, and is credited with causing the defeat of the government in the election which followed shortly thereafter. There was never any proven effect on humans, although without effective traceability it can’t be said who ate contaminated products.

The recent Sudan 1 food alert in the UK concerns an ingredient without any proven risk to human health, but which due to the precautionary principle of food safety is illegal -- following the results of certain laboratory experiments with mice. The recall has been estimated at a cost of £100 million for just one company at the centre of the crisis, Premier Foods, and there have been further costs due to the disruption caused. British products are now suspect on the world food market and are subject to special checks as far away as China.

There are new fears developing about food, particularly those relating to bio-terrorism. Last year, Tommy Thompson, then US Secretary of Health and Human Services, said:
“For the life of me, I cannot understand why the terrorists have not attacked our food supply, because it is so easy to do”

Naturally the politicians have reacted to the fears of the general public and, notably in Europe, have embarked on the program of food safety improvement which has been named “from the farm to the fork”. The General Food Law (regulation 178/2002), which was approved in January 2002, has now come fully into effect and this states that “the traceability of food, feed, food-producing animals, and any other substance intended to be, or expected to be, incorporated into a food or feed shall be established at all stages of production, processing and distribution.”

Interestingly it should be noted from this legislation that traceability has been made obligatory not only to help assure food safety, but also to avoid disruption to the market through targeted and precise withdrawals and to give information to consumers (recital 28). This implies that food operators have to go further than assuring food safety and need to implement precise traceability systems and procedures.

The practical implications of these traceability obligations are enormous, and the reality today is a long way from full traceability the length of the food chain. The recently-published guide on traceability from the Spanish Food Safety Agency sums this up well: “The implementation of traceability implies a challenge which will be tackled gradually.”

Since the first laws were introduced which made food traceability obligatory, more and more measures have been introduced – extending the regulations to food packaging and other “contact” materials and introducing specific additional measures for certain foodstuffs. As commented by a representative of the European association which represents 12 million small businesses: “Pandora’s box has now been opened.”

The magnitude of the challenge to implement traceability throughout the food industry is immense. There are hundreds of thousands of food operators in every medium-sized country, ranging from one or two person agricultural holdings to food processors employing tens of thousands, from corner shops and bars to major supermarket chains. Each type of business, in each sector of the food business and in each stage of the food chain, has a different set of issues – not to mention differences in capabilities from one region to another. Many billions of food products are consumed each day.

Food, as we know, is an extremely complex product. Even without processing it undergoes chemical changes, depending on time and storage conditions, and undesirable characteristics can migrate from one food product ingredient to another. No two ingredients are exactly alike, and they can be affected by treatment and environmental conditions in all phases of growing, processing and distribution.

Traceability, as defined by the International Standards Organisation is “the ability to trace the history, application or location of an entity by means of recorded information”. Taken to its logical conclusion, food traceability applies not only to the ability to physically trace a product but also to the ability to recall details of all processes applied to the product and to any of its antecedents which could have an effect on the product.

This is not quite as bad as it sounds. The food industry is inexorably moving towards full record-keeping of all processes applied to food products and their “inputs” – specifically because this is an essential step in maximising food safety. The new European regulations on hygiene (852/2004 and 853/2004), which enter into force in January 2006, extend HACCP obligations to all businesses except primary producers (which are likely to included later) and require detailed documentation, including traceability, of all treatments applied to fields, crops and food-producing animals.

The challenges are how to record and join up all this information in order to enable full traceability of the food chain, and how to link specific products to the right set of information, for example by grouping in lots.

The traditional interpretation of the meaning of traceability in the food industry was the ability to find out the source, and ideally the full history, of a product. Nowadays this is often referred to as “upstream traceability”, in comparison with “downstream traceability” which is the ability to find the current location of a product – or of all products which contain a particular ingredient. Some people use the term “track and trace”, with “track” meaning to follow a product through its life (downstream) and “trace” meaning to find its origins (upstream).

Upstream traceability can have its challenges, but it is relatively easy. If a product is well labelled, then you can easily find your way back to the source of the label and identify the process origins of the product. The concept of lot labelling, which became obligatory ten years ago in Europe but which can be omitted if the best-before date is included, was in response to the demands for upstream traceability.

Downstream traceability is harder to implement. If you want to know the current location of a product you can’t rely on the label, because you don’t have the label to look at. The only way to accomplish this is to keep records at every stage the product moves location. This not only significantly increases the record-keeping requirement but also implies the need to retrieve and quickly organise this information when the need arises – a task which is almost impossible with paper-based records, except when the scale of activity is very low.

The traceability downstream of products which pass through different companies in the food chain has further complications. Unlike upstream traceability, where the information on a label allows you to leapfrog intermediaries back to the source of the product, the records from several companies have to be joined together in order to reconstruct the traceability chain downstream. Current European legislation obliges each company to provide their own records to food safety authorities on request, but this leaves the authorities with an extremely difficult job to piece together these records – since there are no standards on the delivery media or format of the records.

The European General Food Law requires that all food products be identified in order to facilitate traceability, and this is in any case the starting point for any traceability implementation. The identification is used as the reference key to store and retrieve information related to the product.

This identification should at the least be able to categorically indicate the product type and its lot of production or of shipment – often this is done with a single reference number or code, which can be used to consult further details. A lot should describe a group of products which in principle are identical, having been produced under the same conditions and been subject to the same conditions subsequently (including the same storage conditions). Instead of creating a lot number, time stamping can be used, which in effect creates very precise lots. He simplest case of traceability is when products are stored and distributed in discrete, packed units, and do not undergo a later transformation or change of packing. The identification code can be applied via a label on the packing (for example, on a cardboard packing case), and this same code can be used by subsequent companies who handle the product.

It is only slightly more complicated when packing containers are changed, or if packs are broken down into smaller units, a split. The traceability system should record the input identifier(s) and output identifier(s) and establish a link in the database between the two.

Complications increase when processing occurs, such as in the manufacture of a product from several ingredients. The principles remain the same: all ingredients should have their own identification code (traceable back to a lot), and these should be recorded as inputs; the manufactured products are the outputs and are given their own new codes; the traceability system establishes the linkage, and should also record the process applied. It is fundamental to time-stamp all records.

The service TraceCheck (www.tracecheck.com) has defined a little over 20 “tracepoints” for gathering traceability data.

Continuous manufacturing processes can be handled in the same way as discrete lots, even though there is often not a one-to-one relationship between the input ingredients and the output products. Inputs and outputs are recorded, and the traceability system establishes the appropriate level of linkage. A special formula may be required in the traceability system to establish the correct linkage between ingredients and product (for example, to model the time relationship between adding an ingredient to the process and its inclusion in finished products).

In the real world, there is no perfect traceability. Two or three batches of input ingredients may be used to make one batch of products. The tolerance in the manufacturing process may not permit very precise modelling of the appropriate formula.

Ingredients or products stored in bulk create particular challenges for traceability, as do liquids and gases. Firstly it is difficult to apply an appropriate identifier (such as a label) since the material does not have discrete packing, and so the identifier is instead applied to the container or pipe – in effect, the material is given an “indirect identifier”, through the identification of the container combined with the time of filling or emptying.

Bulk containers often contain material from more than one delivery batch. For example, a silo may be continuously filled from the top and emptied from the bottom. Apart from applying procedures where possible to avoid batch mixing, such as operating silos in rotation and cleaning between each batch, study can be made of the likely mixing within the silo and this formula used within the traceability system.

Several types of coding can be used for identifying packs or containers, but there are two key rules to be applied. Firstly the codes should be unique (they should never be repeated for another object as this could lead to ambiguity in the traceability records) and secondly they should be recognisable at later stages in the traceability chain by any system which needs to record or read information about the object.

Good industry practices are that the codes are globally unique (i.e. that they cannot be used by another company) and that they are globally recognisable (i.e. that they can be recognised by other companies, for the sharing of information and to facilitate chain traceability). One example of such codes is the GS1 (formerly EAN) Serial Shipping Container Code (SSCC), which is normally applied to shipping pallets.Another example is the RG Codes RGID, which is a general-purpose unique identifier set up to facilitate not only the tagging of goods but also the retrieval of data from all stages of the supply chain. Both codes follow the same ISO standard for unique identification.

Currently it is still common that identifiers are hand-written or printed on to a label in human-readable format only. This has a number of disadvantages: usually the production of such labels is not done under computer control, and so the link to other records is tenuous; manual transcription of the code onto a paper record or computer has a high error rate; and the productivity of the operation is low.

For productive and effective identifier production and reading, the technologies of AIDC (Automatic Identification and Data Capture) are used. The most common of these is the barcode, which is cheap to produce, very well proven and compatible with readers already installed or within the reach of nearly all companies.Unlike the barcodes on products sold in the supermarket, traceability IDs have to be different each time they are produced. They are therefore normally produced as required by special printers on the production line or at the dispatch point of food businesses, although pre-printed serialised labels can be used.

The barcode symbology commonly employed is Code 128 (used by the SSCC and RGID codes mentioned above), which provides an effective density for traceability codes (typically 15-20 characters long).

However, emerging barcode solutions are based around 2-dimensional formats such as Data Matrix, which can apply the same amount of data in a much smaller space.

Code 128 barcodes cannot be printed with 100% reliability on all types of corrugate (cardboard) surfaces, and so generally these barcodes are printed on a label rather than directly on the surface of a box.

RFID (Radio Frequency Identification) technology is now being applied for traceability identifiers, although this is still very much at the pilot phase despite the high-profile directive by Wall-Mart that their major suppliers have to use the technology. The benefits of RFID are that the tags do not have to be on the surface of a product (therefore less liable to damage), they do not need to be visible and orientated towards the reading equipment, and they can be read at high speed. The disadvantages are that they are relatively expensive (low end tags are perhaps $0.20 each in large quantities) and that they need controlled conditions for good performance (for example, they can be adversely affected by the presence of water or metal).

RFID tags come in different formats, varying in terms of whether they are active or passive, frequency range in which they operate and the memory capacity. There are also different coding formats. Active tags have built-in batteries and so they have a longer range performance, but their price makes them uneconomical except possibly for reusable containers. Frequency ranges affect the operating characteristics of the tags – HF tags at 13MHz are most commonly used today, but new UHF tags at around 900MHz are likely to be most popular in the future due to their increased reading distance. For traceability purposes, only a small memory capacity is required (eg, 96 bits) because only an identifier is needed, but some tags have larger memories in order to carry further information (which can be written to during the lifetime of the tag).

Natural and artificial “bio-tags” are likely to be increasingly popular in the future for food products, since these can be inside the food itself rather than on the packing. Already DNA testing is used as a method for complementing or corroborating other methods of identifying animal meat, although the cost is too high and the testing time too long for this yet to be a viable general method of identifying for traceability. Artificial bio-identifiers are under development, but mostly these are at the experimental phase.

The core technologies needed to implement traceability are not new, but practical implementations are still at an early stage of evolution and many companies, particular smaller companies, do not yet have a solution.

Also the industry is only just beginning to understand the possible benefits which will result from traceability. In addition to improving food safety and avoiding market disruption – the objectives of the legislation – traceability can bring significant benefits to individual companies and to trading partners.

The first observable benefit is the improvement in quality control and in logistics which results from more disciplined, and more-detailed, record-keeping procedures. Defective products which used to “slip through the net” are detected more easily, and feedback on any issues can be pinpointed to the individual lot or time of production. Warehouse management is automatically improved, moving from generic FIFO management to precise product picking.

Paradoxically, although traceability implies a high level of data entry, the total amount of data entry in a supply chain can be reduced as a result. Once data has been entered into a reliable traceability system it can be re-used both within the same company and downstream in the supply chain. For example, a company receiving goods can simply scan the ID and receive all information about the product recipe and the detailed history of the lot (including information on previous quality testing). This can imply major savings and a reduction of errors by food processors and distributors.

The detailed level of data provided by traceability systems makes it possible to study nearly all aspects of business operations, from analysis of productivity of a specific process or production line to history-based predictions of customer demand.

The ultimate benefit will be a new relationship of trust with the consumer. With an increase in the transparency of information, consumers can look up an ID and check the history of a product, including full details on the ingredients it contains and the farming methods used. Japanese supermarkets are already ahead on this – a machine is provided at meat counters, and at the touch of a button buyers can get information about the animal from whom the meat comes.

Full food-chain traceability, for all the challenges it offers, is probably the only way to re-establish the confidence of consumers in the food they eat.