The Welsh food microbiological forum and the all-Wales shopping basket sampling program: a model for the surveillance of microbiological quality in ready-to-eat foods. (International Perspectives).

Introduction

The prevention and investigation of foodborne disease and the monitoring of microbiological food quality are the responsibility of a number of different interlinked public bodies, the food industry, and other related organizations. Figure 1 outlines the general activities required to enhance food safety and reduce the incidence of foodborne disease. In the United Kingdom, monitoring and enhancement of food safety and microbiological food quality are the duty of local authorities, the Food Standards Agency (FSA) (since its creation in 2000), and the Public Health Laboratory Service (PHLS). Local-authority functions include periodic inspections of all registered food premises, enforcement, education, and response to public complaints. In recent years, education and advice to food producers and caterers have become a major function of local authorities. ESA also has a major role in the provision of food safety advice to food producers, caterers, and the public, and works in tandem with local authorities to ensur e concerted action in the provision of advice. PHLS examines samples of food routinely submitted by local authorities for microbiological quality and the presence of specific pathogens. It also provides local authorities and FSA with advice and training on microbiological food quality, provides reference laboratory facilities, carries out applied research, and participates in both food surveys and outbreak control teams as required.

Wales is a country with a population of approximately three million. It is part of the United Kingdom, but since 1999 has had a devolved local government called the Welsh Assembly Government that deals with legal and constitutional issues related to Wales. Prior to this date, local government issues in Wales were managed by the Welsh Office, an office of the U.K. parliament. The country is divided into 22 local-authority areas.

This paper describes the creation of a collaborative forum in Wales that represents organizations with an interest in enhancing food safety, including local authorities, FSA Wales, and PHLS in Wales. It highlights the advantages of having a multi-agency forum that coordinates sampling, examination, and the centralized collection, analysis, and utilization of results. The paper also discusses the structured, randomized shopping-basket program that was developed by the forum.

The Development, Constitution, Aims, and Activities of the Forum

A previous paper describes the development of structured ready to eat food-sampling programs in southeast Wales during the early 1990s (Widdows, Ribeiro, & Brown, 1996). These programs were organized by local food groups, comprising representatives of local-authority environmental health departments and PHLS in Wales. Experience with these local arrangements demonstrated the benefits of collaboration and the added value obtained from focusing the efforts of a number of local authorities.

It was suggested in 1993 that broader coordination of routine sampling could be beneficial, and the Welsh Office invited all local food groups in Wales to extend the initiative and consider collecting data for the whole of Wales. At an exploratory meeting in 1993, it was agreed that this collaboration be initiated and coordinated through the formation of a multi-disciplinary and multi-agency group, which would be named the Welsh Food Microbiological Forum (WFMF).

WFMF was formally created in 1993 by agreement between the then all-Wales Chief Environmental Health Officers Panel, the Directors of Public Health Laboratories in Wales, the Communicable Disease Surveillance Center (CDSC) Wales, and the Welsh Office.

WFMF's overall purpose is to promote coordination and support of agencies concerned with improving the microbiological quality of ready-to-eat food in Wales. Its specific aims are to use standardized methods of food sampling and testing across Wales, to propose common food types for the all-Wales sampling program, to develop an all-Wales database for monitoring and recording the microbiological quality of ready-to-eat food at the point of sale, to collate and analyze results of food testing throughout Wales, to propose criteria for foods not included in PHLS guidelines, and to produce and disseminate reports and papers on its findings.

Local-authority membership in WFMF initially included the secretaries of the local food groups in Wales, but following local-government reorganization in 1996, local food groups were abolished, and a Food Safety Technical Panel of the Society of Directors of Public Protection in Wales was created. The WFMF membership and reporting arrangements were amended accordingly. As well as a local-authority presence, WFMF currently has representatives from PHLS in Wales, CDSC Wales, FSA Wales, and the Local Authorities Coordinators of Regulatory Services, as well as an academic with expertise in hazard analysis critical control point (HACCP) methods. WFMF has a core of approximately 30 members at any one time. Observers from the National Assembly for Wales and the Society of Directors of Public Protection in Wales also may attend. Other relevant individuals can be recruited as members. A chair and secretary are elected from the members. Between them, they represent the professional backgrounds of the membership, and t heir terms of office last a maximum of three years. Meetings are held at intervals of not more than seven months, WFMF reports to each of the parent agencies represented by its members and communicates with other agencies and organizations as appropriate.

The Shopping-Basket Model

In general terms, there are six basic types of food survey: structured, unstructured, targeted, passive, shopping-basket, and enhanced. Each format has advantages and disadvantages in terms of cost, robustness, statistical significance, and time needed. The choice of survey depends on the outcome required, which is determined by assessing the reasons for doing the survey and the audience for which the results are intended. Structured surveys involve precise statistical sampling and are statistically significant and representative. The main disadvantages are that they can be expensive and may take months to set up and years to complete. The unstructured survey, with no predetermined sampling plan and perhaps very few samples taken, is the format with the fewest--if any--advantages and is unlikely to produce any conclusions that will enhance food safety or food quality. A targeted survey focuses on a product or sector for an extended period. In surveys of this type, resources are focused, but the studied area i s narrow The passive survey collects data from other surveys. The advantages of this method are that it is cheap and that data can be used for more than one purpose. The disadvantages are that the surveys from which data are culled can be unstructured and methods may differ. Shopping-basket surveys examine random samples according to a predetermined statistical plan. Advantages are that results are statistically significant and that a wide range of products can be covered, A disadvantage is that a large number of samples are required. Enhanced surveys combine data from all studies in a central database, which can cover the whole food chain and can be continuous. Surveys of this type may not, however, involve standard methodology and may require expertise in information technology

The shopping-basket model used by WEME in Wales relies on a nominated list of foods that are randomly sampled by all participating local authorities for 12 months, although foods of interest can be sampled for longer, as required. Foods to be sampled are regularly reviewed at WFMF meetings and are usually selected as being those of particular interest to local authorities, FSA Wales, and PHLS in Wales. Foods are excluded when it is considered that enough data have been collected, but may be included again at a later stage. Foods in the basket are primarily but not exclusively, foods of interest on a U.K. level. There have been six shopping baskets to date, as detailed in Table 1. Foods in the shopping basket are assigned differing priorities. Foods at the top of the list have priority over those below. The shopping-basket approach, with a number of different foods rather than just one in the sampling program, decreases the chance of a wasted sampling visit and increases the range of foods and premises sampled .

It was recognized early in the process that taking samples on an ad hoc basis from premises chosen by a sampling officer introduced sampling bias into the program and that some form of randomized sampling needed to be introduced. As well as preventing bias, random selection of premises has increased the variety of outlets sampled, to include restaurants, schools, gas station shops, and factory canteens. Sampling officers use a random-number generator to identify premises from their register of food premises. On a sampling day the sampling officer has the randomized list of premises to be visited and the prioritized list of foods to be sampled. At the first food business on the list, the sampling officer will aim to sample foods from the top of the list. The officer may take the maximum required number of samples or may find none to sample and be required to move on to the next premises on the randomized list. The intent is to ensure not only that fewer visits are wasted, but also that the number of samples ob tained from each premises is increased, thereby improving both the effectiveness and the efficiency of the exercise.

Samples are delivered to one of the four PHLS laboratories in Wales that examine foods or to one of two PHLS laboratories in neighboring parts of England. The choice of laboratory is a matter of preference for the local authority Over the period of the shopping-basket program, samples have been examined for a range of analytes, including aerobic colony count, Escherichia coli, Bacillus, Clostridium perfringens, Staphylococcus aureus, Vibrio, Salmonella, Campylobacter, and Listeria.

The shopping-basket model in Wales has evolved over the seven years that it has been in existence. As mentioned above, food types in the basket change annually in response to other survey results, outbreak information, or specific local interests. The range of organisms for which the samples are tested also has changed over the years. Tests for aerobic colony count, Escherichia coli, Bacillus, Staphylococcus aureus, Salmonella, and Listeria have been maintained as the standard range of tests over the years of the program, while tests for other organisms have been introduced or removed at WFMF's request. For example, Campylobacter testing was introduced in 2000, reflecting the increased importance of the organism, and Clostridium perfringens testing was phased out for routine shopping-basket samples. The changes to both the content of the shopping basket and the organisms reflect the dynamic and evolving nature of the shopping basket.

Provision and Utilization of Results

The results from the examination of foods by the PHLS laboratories are returned to the local authorities on a weekly basis. The local authorities currently have responsibility for inputting the microbiological results and associated sample information into the WFMF database and electronically transmitting the data to CDSC Wales, where it is collated for storage and analysis. Each sample is allotted a unique sample number to help identify any duplication. CDSC Wales now has a substantial database on the microbiological quality of food throughout Wales from 1995 to the present. As of the end of March 2002, results from testing of approximately 15,000 samples have been collected. To date, only the results from the period 1995-2000 have been fully analyzed and are available to local authorities, with the full analysis of the data from 2000 to present still pending at the time of writing.

The WFMF database is periodically audited for accuracy, a process that involves comparing the results in the forum database to the laboratory results held by PHLS in Wales. The audit process is carried out on a regular basis--usually when enough new data have been electronically transferred from the participating local authorities--and prior to any data analysis or publication. The auditing of results ensures that the results in WFMF database are the same as those held by PHLS in Wales and are not duplicated or incorrect. Any incorrect data entry is corrected, and any duplicate entries are removed. Samples which are either not in the shopping basket or have not been taken as part of the randomized program also are removed, as required.

The existence of an all-Wales database that is easily searched for particular analytes allows the sampling results from the whole of Wales to be broken down by food type, by month or year, by local authority, by organism type and number, or by laboratory. The database can be searched for results that can be compared to the existing guidelines to determine the proportion of unsatisfactory foods, broken down either by food type or by organism. An example of this type of analysis is an examination of the database for the genus Bacillus. From the four laboratories in Wales, there were 843 positive Bacillus results for the period 1995-2000, out of 10,904 foods tested for the genus. Of these 843 positive results, only 55 were equal to or greater than 10 + colony-forming units per gram (CFU/g), the level that is defined as unsatisfactory in current PHLS guidelines (Gilbert et al., 2000). These 55 results can be further analyzed by food type. Six were from fruit and vegetables, 12 from meat and meat products, 17 from bakery products (especially custard-based products), 14 from sandwiches, three from rice, and three from desserts. The results clearly demonstrate two points. First, Bacillus contamination in the ready-to-eat foods available to consumers in Wales is relatively low, with only 0.5-percent of samples being unsatisfactory for the period in question. Second, the results identify the foods that have with the highest frequency been contaminated with unsatisfactory levels of the genus, namely bakery products and sandwiches. This finding allows organizations involved with food safety in Wales to focus on these higher-frequency food products in an effort to further reduce the numbers of ready-to-eat foods containing unsatisfactory levels of Bacillus contamination.

Another area in which the WFMF data could be used is in the evolution of existing guidelines for ready-to-eat foods. Following the publication in the United Kingdom of the Richmond Report (The Richmond Committee, 1990, 1991) and the Food Safety Act (Ministry of Agriculture, Food and Fisheries, 1990), it was recognized that there was a need for microbiological guidelines for ready-to-eat foods that would assist in the interpretation of microbiological results and thereby help to pinpoint problem foods and outlets. It was clear that such guidelines would be useful in delineating microbiological-contamination limits that could be consistently applied across the United Kingdom. This need was answered, in part, by the production of guidelines for ready-to-eat foods by the PHLS food surveillance group (Public Health Laboratory Service, 1992; Public Health Laboratory Service, 1996). The guidelines have subsequently been updated (Gilbert et al., 2000). Although this step has been extremely useful, a need still remain s for robust data to validate the guidelines and generate additional ones for foods not yet covered. It may well be that the data collected by WFMF and the randomized nature of the shopping-basket model will help to meet that need.

Conclusion

It is important in food surveillance that the organizations involved work closely together to ensure that the sampling, examination, and reporting of results are consistent and provide some value. The establishment of the WFMF and the subsequent development of the shopping-basket approach to randomized food sampling in Wales provide an example of a multi-agency, multi-disciplinary approach to the sampling and examination of ready-to-eat foods. This approach could serve as a potential model for the development of similar programs and highlights the ability of organizations involved in food safety to work closely together to develop a coordinated and randomized approach to sampling that is spread across a relatively large geographical area served by approximately 20 local authorities and six public health laboratories.

A large amount of data has accumulated in the seven years during which the shopping-basket model has been applied. Future developments in guidelines for the microbiological quality of ready-to-eat foods in the United Kingdom should benefit from the existence of this statistically robust and fully audited database covering a wide range of ready-to-eat foods. The data also can help to pinpoint problem foods and problem types of premises, thus identifying areas in which specific solutions are needed and helping to enhance food safety and reduce the incidence of food poisoning in Wales.

Already, new developments for the shopping-basket model are being planned, and it is hoped that these improvements, which are primarily based on improvements in the information technology support for the movement, storage, and analysis of data, will ensure that the coordinated approach in Wales, successful for so long, continues for the foreseeable future.

TABLE 1

Overview of the Six Shopping Baskets


Basket 1,                  Plain boiled rice, coleslaw, fresh
April-September 1995       cream cakes, cooked burgers, cooked
                           mince dishes, loose unwrapped
                           confectionery, sliced ham, egg mayon
                           naise sandwiches, prawn sandwiches,
                           whole lettuce, bean sprouts

Basket 2,                  Plain boiled rice, fried rice,
October 1995-March 1996    fresh cream cakes, roast pork,
                           roast-pork sandwiches, loose unwrapped
                           confectionery, watercress, bean sprouts,
                           egg mayonnaise sandwiches, prawn
                           sandwiches, salami, custard slices,
                           pate, quiche, yogurt sauces,
                           cooked mince dishes

Basket 3,                  Ice cubes, fried rice, shellfish, roast
April 1996-March 1997      pork, roast-pork sandwiches, fresh
                           soft berries, watercress, bean sprouts,
                           yogurt sauces, quiche, egg mayonnaise
                           sandwiches, prawn sandwiches, salami,
                           custard slices, pate

Basket 4,                  Ice cubes, fried rice, custard slice,
April 1997-March 1999      roast pork, roast pork sandwiches,
                           fresh soft berries, watercress, bean
                           sprouts, yogurt sauces, chicken
                           sandwiches, cooked chicken pieces,
                           cooked beef sandwiches, cooked
                           beef burgers

Basket 5,                  Fresh cream cakes, chicken sandwiches,
April 1999-September 2001  locally produced desserts, egg
                           mayonnaise sandwiches, soft berries,
                           sliced ham, ham sandwiches, fresh
                           herbs, hot sauces stored in
                           bain-maries, meat pies, organic
                           fruit and vegetables, roast pork,
                           poultry meals, rice- or pasta-based
                           salads, tuna sandwiches

Basket 6,                  Fresh cream cakes, chicken
October 2001-May 2002      sandwiches, locally produced
                           desserts, egg mayonnaise sandwiches,
                           soft fruit, fresh  herbs, hot
                           sauces stored in bain-maries, meat
                           pies, poultry meals, rice- or
                           pasta-based salads, tuna
                           sandwiches, pate

Acknowledgements: The authors would like to thank the members of the Welsh Food Microbiological Forum for permission to publish this article and for help and advice given during its preparation.

Gilbert, R.J., de Louvois, J., Donovan, T., Little, C., Nye, K., Ribeiro, C.D., Richards, J., Roberts, D., & Bolton, F.J. (2000). Guidelines for the microbiological quality of some ready-to-eat foods sampled at the point of sale. Communicable Disease and Public Health, 3(3), 163-167.

Ministry of Agriculture, Food and Fisheries. (1990). The food safety act. London: Her Majesty's Stationary Office.

Public Health Laboratory Service. (1992). Provisional microbiological guidelines for some ready-to-eat foods sampled at point of sale. Notes for PHLS food examiners. PHLS Microbiology Digest, 9(3), 98-99.

Public Health Laboratory Service. (1996). Microbiological guidelines for some ready-to-eat foods sampled at the point of sale: An expert opinion from the PHLS. PHLS Microbiology Digest, 13(1), 41-43.

The Richmond Committee, (1990). The microbiological safety of food part I. London: Her Majesty's Stationary Office.

The Richmond Committee. (1991). The microbiological safety offood part II. London: Her Majesty's Stationary Office.

Widdows, G.D., Ribeiro, CD., & Brown, A. (1996). Proactive structured food sampling programmes in South East Wales. PHLS Microbiology Digest, 13(4), 219-222.

Corresponding Author: Richard Meldrum, Group Food Scientist, Public Health Laboratory Service, Llandough Hospital, Penlan Road, Penarth Vale of Glamorgan, CF64 2XXUK. E-mail: [e-mail].