Does microbiological testing of foods and the food environment have a role in the control of foodborne disease in England and Wales?

This review looks at the contribution of microbiological sampling to the safety of retail foods in England and Wales. It compares sampling methods available and assesses the value of testing as part of outbreaks of foodborne disease, as part of routine management by local authorities, as part of work done or commissioned by the food industry, and as part of research. It confirms that microbiological testing has a role during outbreaks as it makes a significant contribution to help identify foods and other areas of greatest risk for future study. The review suggests that routine testing by local authorities is often of limited use and could be improved by more targeted surveillance. Testing could be better used to validate primary control methods, such as Hazard Analysis and Critical Control Point (HACCP) system. Any public health benefit from testing in the food industry is often restricted by client confidentiality. Microbial research on foods is important as it can lead to significant improvements in safety. Current microbiological methods are slow and, in future, rapid molecular methods may make an even bigger contribution to the control of foodborne disease.

Two separate outbreaks of Salmonella enteritidis phage type 14b food poisoning linked to the consumption of the same type of frozen food.

OBJECTIVES: To ascertain, using a combination of epidemiological, environmental and microbiological methods of investigation, a possible link between two outbreaks of salmonella food poisoning. METHODS: Case-control studies were carried out on the known at-risk populations. Environmental investigations took place in the food preparation areas used for the social functions and microbiological examinations were carried out on faecal specimens obtained from cases, environmental swabs, and food specimens when these were available. RESULTS: In both outbreaks, illness was associated with the consumption of sesame prawn toast (outbreak one P<0.004; outbreak two P<0.0001). Salmonella enteritidis phage type 14b was cultured from the faecal specimens of cases in both outbreaks and from a packet of sesame prawn toast used for the second outbreak function. Molecular typing methods indicated that the salmonella cultures obtained in both outbreaks were indistinguishable from each other and from cultures obtained from imported Spanish eggs in a previous survey. Imported Spanish eggs were used in the manufacture of the sesame prawn toast. CONCLUSIONS: Adequate cooking must take place of raw food products, which should be clearly labelled as such. Manufacturers should consider, when possible, the use of pasteurized egg in the preparation of food products.

Outbreak of Norwalk-like virus infection associated with salad provided in a restaurant.

Preliminary enquiries following the reporting of some cases of suspected food poisoning by members of the public revealed that they had all eaten at the same restaurant during the preceding weekend. Subsequent investigation, including a merged cohort study, identified 49 people with gastrointestinal symptoms, six of whom showed evidence of infection with Norwalk-like virus (NLV). Over a four day period all had eaten at the restaurant. Eating salad was strongly associated with infection. One of the chefs, who handled salad in the food preparation area and who had been absent from work with symptoms of gastrointestinal illness, had returned to work the day before the four-day period, reporting that he was 48-hours symptom free. He was subsequently shown to have been infected with NLV by reverse transcriptase polymerase chain reaction (RT-PCR) assay. The NLV belonged to the same genogroup as that infecting the cases who had eaten at the restaurant.

Outbreak of Salmonella enteritidis phage type 6 infection associated with food items provided at a buffet meal.

Preliminary enquiries following prompt notification of three cases of suspected food poisoning revealed that they had all attended the same three functions during the preceding weekend. Subsequent investigation identified 49 people with gastrointestinal symptoms, 13 of whom were infected with Salmonella enteritidis phage type 6. Forty-five of those with symptoms, including 11 with confirmed infection, had eaten a buffet meal at a public house. Eating egg sandwiches was strongly associated with infection. Defects in the kitchen structure and the storage and handling of the implicated food items provided the potential for cross contamination. Salmonella was isolated from several environmental sites, including a general purpose cleaning cloth. Two different quiches and pork pies, which were possible vehicles of infection were thought to have been contaminated after being brought into the kitchen. The investigation did not reveal whether or not shell eggs used in the sandwiches were the original source or whether they too had been contaminated during their preparation.

Outbreak of Salmonella enteritidis phage type 4 infection associated with cheese and onion quiche.

Twelve people developed gastrointestinal symptoms suggestive of food poisoning after a party attended by between 60 and 80 people. The outbreak went unrecognised, however, until general practitioners admitted three patients to hospital, where faecal specimens were examined. An epidemiological investigation identified 12 cases, and a case control study showed a significant association between illness and eating a cheese and onion quiche at the party. Salmonella enteritidis phage type 4 was cultured from three out of seven faecal specimens obtained from cases. Consultants in communicable disease control should continue to encourage general practitioners to investigate and notify suspected cases of food poisoning.

Detection of Clostridium perfringens and its enterotoxin in cases of sporadic diarrhoea.

AIMS: To determine the incidence of sporadic and apparently non-food related diarrhoea associated with Clostridium perfringens enterotoxin. METHODS: Enzyme linked immunosorbent assay (ELISA) and reversed phase latex agglutination (RPLA) were used to detect C perfringens enterotoxin in faecal specimens from 818 sporadic cases of diarrhoea. RESULTS: C perfringens enterotoxin was identified as a cause of sporadic diarrhoea in 56 of 818 (6.8%) cases. Diarrhoea was prolonged (three days or more) in most cases. Ages ranged from 3 months to 89 years, although most patients were over 60 years of age. CONCLUSIONS: These results suggest that C perfringens may be a cause of sporadic cases of diarrhoea when causes such as food consumption or cross-infection are absent, particularly in the elderly.

Development of standardized inspections in restaurants using visual assessments and microbiological sampling to quantify the risks.

The relationship between visual inspections carried out by environmental health officers and microbiological examination was studied in 89 restaurants. Using 30 variables a standardized inspection procedure was developed and each of the premises was assessed in six main areas-structure and design, cleaning and cleanliness, personal hygiene, risk of contamination, temperature control, and training and knowledge about food hygiene. Selected foods and specimens from hands, surfaces, and wiping cloths were examined. There were significant associations between all six areas of the inspections. The structure and design were significantly related to the combined score from all the other areas (P less than 0.001). There were no highly significant associations between microbiological examination and visual assessments. The microbial contamination of wiping cloths, however, was related to the cleaning and cleanliness (P = 0.005). Microbial sampling provided additional information to inspections and was a valuable aid. Further development of this risk-assessment approach could provide an effective system for monitoring potential health risks in high-risk food premises.

An assessment of cleaning and sampling methods for food-contact surfaces in premises preparing and selling high-risk foods.

The performance of agar-contact plates and an alginate-swab method for sampling food surfaces before and after cleaning was compared. Contact plates were more convenient, and were at least as sensitive as the swabbing method. To assess cleaning efficiency repeated sampling was carried out in selected premises, and several cleaning methods were introduced for trial periods. Some surfaces, notably wood and polypropylene, were particularly difficult to clean. For these scrubbing with a nylon brush was the best method. Other surfaces were more easily cleaned, and generally the methods introduced as part of this study were better than the original method used in the premises. Paper proved to be unpopular, and cleaning solutions applied with it did no better than those cleaned with a multiuse cloth kept soaking in a detergent and hypochlorite solution.

Comparative study of visual inspections and microbiological sampling in premises manufacturing and selling high-risk foods.

The possible relationship between the results of a microbiological sampling programme and visual inspections carried out in local food-manufacturing premises was examined. Using five main parameters - overall appearance, personal hygiene, risk of contamination, temperature control, and training and education - a visual inspection rating score was established for each of the premises. A variety of high-risk processed foods, and specimens from hands, wiping cloths and environmental swabs were examined. The results from two study periods indicated that there was an overall poor agreement between microbiological results and inspection ratings. On its own, neither sampling nor visual assessment reliably monitored the performance of the premises. A combined approach, using selective microbiological examination to support a system of standardized inspections, is suggested for monitoring food hygiene standards in premises selling high-risk foods.

Laboratory evaluation of disposable and reusable disinfectant cloths for cleaning food contact surfaces.

A comparison of five methods of cleaning Formica surfaces contaminated with bacteria dried in milk has been carried out. A standardized procedure was developed, and impression plates were found to be at least as sensitive as a swab-rinse method for detecting bacteria on the surfaces. The most satisfactory results were obtained with one type of disposable alcohol-impregnated wipe and with a detergent/hypochlorite solution applied with paper. A reusable cloth impregnated with disinfectant initially performed well against all test organisms, but was less reliable against Staphylococcus aureus and Streptococcus faecalis, after the cloth had been used and rinsed several times. The importance of introducing methods to reduce the high risk of cross-contamination presently associated with the use of wiping cloths in catering premises is stressed.

An evaluation of various working practices in shops selling raw and cooked meats.

Three groups of premises (butchers' shops, supermarkets and general dealers) which sell raw and cooked meats were compared. Salmonellas were not detected, but Escherichia coli, and to a lesser degree Staphylococcus aureus and Streptococcus faecalis, were widely distributed in all three groups of premises. Contamination of hands, towels and nail brushes was related to poor working practices. The presence of E. coli or Str. faecalis on slicing machines was associated with contamination of meat samples. A number of wiping cloths were heavily contaminated with E. coli, and many also contained Clostridium perfringens. Fewer premises provided written cleaning plans, and in many cases staff did not receive an adequate training in food hygiene. The use of disinfectants as part of the cleaning process did not necessarily reduce the level of bacterial contamination. In general there was poor correlation between microbiological results and a visual inspection made by an environmental health officer. The possible reasons for this finding are discussed.

Study of postoperative chest infections with particular emphasis on those caused by Haemophilus influenzae.

Fifteen per cent of patients admitted to a male general surgical ward were found to be carrying Haemophilus influenzae. Except for patients with chronic bronchitis, who developed an infection with the same micro-organism, carriage did not predispose patients to postoperative chest infection. Age, heavy smoking, and abdominal surgery increased the likelihood of respiratory infection after operation. H influenzae accounted for 58% of bacterial pathogens isolated from sputum, and most infections occurred within 48 hours of operation. Chemotyping showed that most infections were caused by different strains, and cross infection by H influenzae seemed to be rare. Cefuroxime given during anaesthesia did not prevent postoperative chest infection.

A microbiological study of various food premises with an assessment of cleaning and disinfection practices.

A study of cleaning and disinfection methods in a variety of types of catering premises has been carried out. The level of bacterial contamination of the hands and of equipment was related to cleaning methods and to the type of catering establishment. Wiping cloths were frequently contaminated with Escherichia coli, and these may be important reservoirs of bacteria for contamination of the hands of catering staff. Regular and efficient cleaning of food surfaces and equipment was found to be more important than the use of a disinfectant as part of the cleaning process. Methods for reducing the risks of cross-contamination in catering premises are discussed.

A chemotyping scheme for clinical isolates of Haemophilus influenzae.

A total of 464 Haemophilus influenzae strains, most of them fresh clinical isolates, have been classified by chemotyping --a combination of auxotyping and biotyping. Seven auxotests and four other biochemical tests allowed recognition of 56 types. These were to a degree site-specific. H. influenzae of capsular type b proved almost without exception to belong to one chemotype, and 24 of 33 strains assigned to this chemotype were capsulated. When surgical-ward isolates of H. influenzae were typed, the results suggested that some cross-infection had occurred.

Evaluation of some methods for the laboratory identification of Haemophilus influenzae.

Five tests--satellitism, synthesis of porphyrins, acid production from sucrose, beta-galactosidase activity (ONPG), and indole production--to differentiate between strains of Haemophilus influenzae and strains of V-dependent Haemophilus species were evaluated. Six per cent of strains of H influenzae were misidentified as H parainfluenzae by a test for satellitism using filter paper discs impregnated with X factor, V factor, or both, applied to Columbia Agar. None of seven nutrient agars tested grew Haemophilus species, and determined accurately the X factor requirement. Synthesis of porphyrins from delta-aminolaevulinic acid provided a reliable means of demonstrating that X factor was required. A test for the production of acid from sucrose discriminated successfully between strains of V-dependent Haemophilus species (positive) and H influenzae (negative). Most isolates were identified correctly by the ONPG test, but occasional V-dependent strains were negative and could be misidentified as H influenzae. The discriminative value of the indole test was unsatisfactorily low. The results of the tests are discussed in relation to the identification of H influenzae in the diagnostic laboratory.

Evaluation of some methods for the laboratory examination of sputum.

Methods for the quantitation of leucocytes, squamous epithelial cells, and potential pathogens in sputa are described. Microscopic examination showed that 58% of sputa tested (554/957) were purulent or moderately purulent and 48% were moderately or heavily contaminated by squamous epithelial cells. The presence of squamous epithelial cells indicated oropharyngeal contamination. A simple dilution technique was chosen to compare the isolation of potential pathogens from direct cultures and from dilutions of sputa (10(-7) per ml original sputum). The dilution technique permitted easier reading of sputum cultures and avoided the possible over-reporting of enterobacteria and Pseudomonas aeruginosa, which were frequently found on direct examination. Enterobacteria and Pseudomonas aeruginosa were more likely to be isolated from sputa moderately or heavily contaminated with squamous epithelial cells. The use of counter-current immunoelectrophoresis (CIE) and co-agglutination tests to detect pneumococcal antigen in sputa is described. The presence of antigen in sputum was a more reliable index of lower respiratory tract infection than a positive culture. Co-agglutination tests were simpler to perform and used smaller amounts of expensive antiserum than CIE.

Further studies on the reliability of the bacitracin inhibition test for the presumptive identification of Lancefield group A streptococci.

The reliability of the bacitracin inhibition test to differentiate between 125 Lancefield group A and 122 non-group A beta-haemolytic streptococci was studied. Bacitracin discs containing 0-02, 0-04, or 0-1 international units and the conditions recommended by both the Association of Clinical Pathologists and the Federal Drug Administration for routine sensitivity testing were used. The results suggest that the presence or absence of a zone of inhibition around a 0-04 unit disc can be used routinely for the presumptive identification of group A streptococci and that a specific zone size can be used for reading the test only with a 0-1 unit disc.

Grouping of beta-haemolytic streptococci with group-specific antibodies absorbed to staphylococcal protein A.

A total of 245 isolates of beta-haemolytic streptococci was grouped by the counter current immunoelectrophoresis method and by co-agglutination with antibody-coated staphylococci. Co-agglutination was found to be simple and accurate, and many commercially available grouping sera were suitable for use in this test, though some of them had to be absorbed to remove cross-reactions.

The adherence of pilate and non-pilate strains of Neisseria gonorrhoeae to human and guinea-pig epithelial tissues.

Strains of Neisseria gonorrhoeae adhered to pieces of human endocervix and appeared to be embedded in the surface mucus. Although a pilate strain adhered better than a non-pilate strain, the difference was small and pilation did not appear to be exclusively responsible for adherence. The pilate strain showed better adherence to pieces of human ectocervix and fallopian tube, but both strains were similarly adsorbed to human bronchus and guinea-pig uterus, cervix, male urethra and bladder, although to different degrees for different tissues. Since gonococci adhered to all tissues examined, their ability to infect human endocervix and fallopian tube and their failure to infect human ectocervix or guinea-pig urogenital tract mucosae are determined by factors other than a capacity for primary adherence to the tissue.