The number of reported cases of food poisoning and food-borne disease continues to increase in most countries. The published figures are recognized as being only a small fraction of the true total and the problem is clearly both very large and international.

Of the variety of micro-organisms responsible for outbreaks, Salmonella spp. are by far the most frequently incriminated and in the United Kingdom these organisms cause over 90% of cases (Epidemiology, 1986). The almost universal presence of these organisms in certain common foods, their ability to grow in a wide variety of foodstuffs over a substantial temperature range, the ease with which dissemination occurs from person to person and the prolonged period of excretion following recovery are the properties which, taken together, distinguish Salmonella spp. from other food-poisoning organisms. It is because of these characteristics that salmonellas are really the only food-poisoning organisms in which human beings as carriers pose potential problems as sources of outbreaks. This review is, therefore, confined to a consideration of the practical significance of the faecal carriage of salmonellas by asymptomatic food handlers, to an evaluation of the degree of risk, if any, that such a person may pose and to an assessment as to whether the time and money devoted to the investigation and exclusion of such persons is well spent.

Coagulase-negative staphylococci isolated from blood or spinal fluid during a period of 1 year in a department of neurosurgery, were analysed by biotyping, antibiotic resistance pattern and plasmid profiles. Altogether 41 isolates from 19 patients were studied. About 90% of the isolates were Staphylococcus epidermidis. The antibiotic resistance pattern seemed to be closely related to antibiotic usage in the unit. Most common was resistance to penicillin (63%), trimethoprimsulphamethoxazole (49%) and cloxacillin (39%) while resistance to gentamicin was seen in only one strain.

In several cases species and antibiograms were identical in isolates from different patients. Plasmid pattern analysis could then be used for identification of different strains. In one instance, plasmid pattern and restriction enzyme analysis confirmed that two patients probably were infected by the same strain.

Restriction enzyme fingerprinting was applied to 72 transferable trimethoprim resistance plasmids to examine aspects of their epidemiology and molecular relatedness.

These plasmids had previously been divided into 25 groups according to differences in mol. wts and in antimicrobial resistance determinants. Restriction enzyme fingerprinting allowed the plasmids to be further divided into 44 different groups. The groups based on molecular weight and resistance patterns often, but not invariably, corresponded with those based on restriction enzyme fingerprints. Some plasmids with the same mol. wt and resistance pattern had different digest fingerprints and conversely, although more rarely, plasmids which differed in molecular weight by as much as 10 MDa or in resistance pattern by one resistance marker, had indistinguishable fingerprints.

The plasmids were initially divided into three broad categories according to which restriction enzymes gave fingerprints of 6–20 fragments. These categories differed in the molecular weights of the plasmids contained, the numbers of resistance markers, and the proportions of the plasmids which carried transposon Tn7.

Some plasmids were more widespread and persistent than others with the same mol. wt and resistance pattern but with a different restriction enzyme fingerprint.

Thus, application of this technique has shown the trimethoprim resistance plasmids studied to be more diverse than was indicated by determination of mol. wt and resistance pattern, and has indicated changes in the plasmid pool over the 3 years during which they were collected.

Swabs and water samples from a hospital water system were cultured for legionellae over an extended period. Legionella pneumophila serogroup 1, including outbreak associated strains, were isolated in small numbers from approximately 5% of these samples despite implementation of the current DHSS/Welsh Office regulations. No cases of nosocomial legionnaires' disease were proven during the study. Physical cleaning and chemical sterilization of taps, and replacement of washers with ‘approved’ brands did not eradicate the organisms. Eradication of legionellae in hospital water supplies appears to be unnecessary in preventing nosocomial legionnaires' disease provided the current DHSS/Welsh Office recommendations are implemented.

Twelve herds of dairy cows were examined by rectal swabbing for the presence of Campylobacter jejuni. Ten herds were positive with the incidence of colonized animals ranging from 10 to 72% of those tested. With the exception of the two negative herds where mains water only was consumed, all animals drank from rivers or streams when grazing. There was no relationship between total and coliform counts and the presence of C.jejuni in raw milk. However, milk from one farm that consistently gave positive results had significantly higher Escherichia coli counts than other samples.

A total of 181 samples of irrigation water from the farmlands of Granada were examined for the presence of Salmonellaspp. At the same time 849 samples of the crops from these farmlands and of vegetables sold in city market-places were studied. Sampling was done regularly over the period of study which ran from March 1981 to February 1983. Isolates from these sources were compared with 93 salmonellas isolated from human pathological material at various hospitals of the city of Granada from 1979–80, and again from 1981–3.

The most commonly isolated serotypes of human origin were S. typhimurium and S. enteritidis. In irrigation waters and in crops, S. typhimurium, S. kapemba, S. london and S. blockley were found to be the most common. The results indicate a close relationship between the isolates from the irrigation waters and those from the vegetables, but their relationship to prevalent human infections is less clear.

Isolations of salmonellas from humans and food products are recorded for the period 1981–5 and an attempt has been made to investigate the relationship between serotypes isolated from humans and those from meat products.

The predominant serotypes isolated from humans were Salmonella typhimuriunu S. enterilidis and S. virchow. S. typhimurium was commonly isolated from a range of meat products. S. derby was one of the most common serotypes isolated from tripe and sausages but was relatively uncommon in humans.

Salmonellas were found in < 0·5% of most cooked meat products apart from tripe and udder (3·2%) and pet foods (12·4%). Isolations from raw meats ranged from 3 % for pork to 28% for poultry.

Incidents of salmonella infection in humans in Manchester increased between 1981 and 1984 but decreased during 1985.

The bacterial populations of six freshwater streams in populated areas of the Hong Kong New Territories were studied. There is considerable faecal contamination of these streams, with coliform counts as high as 105 c.f.u./ml and the contaminating organisms show a high prevalence of antibiotic resistance and multiple resistance. With direct plating of water samples onto antibioticcontaining media, an average of 49% of the gram-negative bacteria were ampicillinresistant, 3% chloramphenicol-resistant and 1% gentamicin-resistant. At individual sites resistance to these three drugs was as high as 98%, 8% and 3% respectively. More than 70% of strains were resistant to two or more antibiotics, 29% to five or more and 2% to eight or more. A total of 98 patterns of antibiotic resistance were detected with no one pattern predominating. Twenty-eight gram-negative bacterial species were identified as stream contaminants. Escherichia coli was the commonest bacterial species isolated and other frequent isolates were Enterobacter sp., Klebsiella sp. and Citrobacter sp., but no enteric pathogens were detected. The greatest prevalence of resistance and multiple resistance was associated with the heaviest contamination by E. coli. Analysis of selected stream isolates revealed multiple plasmid bands arranged in many different patterns, but multiple antibiotic resistances were shown to be commonly mediated by single transferable plasmids. Faecally-contaminated freshwater streams in Hong Kong may be reservoirs of antibiotic resistance plasmids for clinically-important bacteria.

The Salmonella Bio-EnzaBead Screening Kit, in its modified form with both the MOPC 467 and the 6H4 antibodies, was used for the detection of salmonellas in naturally contaminated foods and animal feeding stuffs in parallel with a traditional cultural procedure.

Initial results showed an 82% agreement between the enzyme immunoassay (EIA) and cultural methods when using the criterion recommended by the manufacturer as a cut-off for all types of foods. By adjusting the cut-off for each type of food, the number of EIA positive, culture negative samples was reduced although the number of EIA negative, culture positive samples increased. The EIA may be more sensitive than the cultural methods as in many cases the EIA positive, culture negative results could be real positives which were not detected by the cultural methods.

The screening kit provides a simple and convenient method for the detection of salmonella in foods and feeds and a presumptive positive result can be reported within 48 h. The advantages and disadvantages of the method are discussed.

Oral administration of strains of food poisoning salmonellas to day-old chickens produced a profound inhibition in the subsequent colonization of the caeca by a strain of Salmonella typhimurium given one day later. Closely related genera were unable to produce a similar inhibition. The inhibition was not the result of bacteriophages produced by the first strain. Neither was it the result of an immunological response by the host induced by the first strain. In additional experiments in day-old chickens, inhibition of an Escherichia coli Nalr strain and of a Citrobacter sp. Nalr strain was produced by the antibiotic-sensitive forms of the homologous strains while strains from other genera did not produce any inhibition. When an avirulent mutant of S. typhimurium was used for pre-treatment a statistically significant reduction in the excretion of the super-infecting S. typhimurium Nalr strain over several weeks was produced. A genus specific inhibition was reproduced in vitro by mixed culture experiments. Live cultures were necessary for in vitro inhibition. Killed cells or a culture supernatant produced no inhibition.

A longitudinal study of the effect of time spent in lairage on salmonellas in the caecum and on the skin surface of 450 slaughter pigs from a single producer was conducted. Pigs were tested in 6 groups at 2 abattoirs, with one-third of a group being slaughtered after 18 h, one-third after 42 h and one-third after 66 h spent in lairage. The salmonella isolation rate from caeca and carcass surfaces increased significantly with increasing time spent in lairage. Salmonellas were isolated from the caeca of 18·5% of pigs held less than 24 h in lairage, 24·1% of pigs held a further 24 h and 47·7% of pigs held for 66 h in lairage before slaughter. The salmonella isolation rates from carcasses were 9·3%, 12·8% and 27·3% for the same groups. Thirteen salmonella serotypes were isolated from the caecal contents and carcasses over the 6 weeks of the trial. One abattoir had a higher salmonella isolation rate from pigs than the other and this was probably related to lairage management. It appeared that lairage is an important factor in the manipulation of the salmonella contamination of pig carcasses.

Aeromonas has recently been recognized as an important enteropathogen. In faecal specimens the organism is readily identified and differentiated from coliforms by the positive oxidase reaction. However, few media have proved useful for foods and environmental specimens. Bile salts brilliant green starch agar (BBGS) was studied for this purpose by testing artificially contaminated foods as well as natural potential sources. The results indicate the suitability of the medium for the isolation of Aeromonas from foods and from environmental sources.

The degerming effect of a 3 min handwash with 2% triclosan, or 4% chlorhexidine, in detergent and enhanced efficacy of either antiseptic in isopropyl alcohol, was evaluated in volunteers. Handwashing with either antiseptic preparation reduced the normal flora by a factor of 10; alcohol rubbing by approximately a factor of 1000. Both regimens eliminated Micrococcus roseus, artificially inoculated before every procedure. The sustained action of the same detergent preparations was further studied in gloved and ungloved hands by the Vinson's ‘finger imprint test’. In the gloved hand both antiseptics inhibited Staphylococcus epidermidis for 4 h. In the ungloved hand however, triclosan remained active longer than chlorhexidine. Whilst the activity of chlorhexidine was short-lived against a clinical isolate of S. aureus, particularly in the ungloved hand, the sustained effect of triclosan against the same strain persisted for 4 h on either hand.

Mice killed shortly after receiving 1300–3000 spores of Clostridium botulinum type C per os were incubated at one of four chosen temperatures together with bottles of cooked meat medium seeded with a similar inoculum. After incubation the rotting carcasses were homogenized. Sterile membrane filtrates of the homogenates (10–20.8%, w/v) and pure cultures were then titrated for toxicity. A temperature of 37 °C produced less toxicity in most carcasses than in cultures. At 30 °C, however, toxicity (often 2× 105 to 2× 106 mouse intraperitoneal LD/g or ml) was roughly similar in both systems, and some carcasses and cultures were still toxic (2× 104 to 2× 105 LD/g or ml) after 349 days. Surprisingly, at 23 °C, though greatly reduced in the cultures, toxicity was high in many carcasses for at least 28 days. Little or no toxin was produced in either system at 16 °C. Unfiltered homogenates (17·8–22·5%, w/v; dose 0·25 ml per os) of toxic carcasses incubated at 30 °C for 7 days invariably produced death from botulism, often within as little as 4 h, but a 1 in 10 dilution produced less than 100% mortality.

Ureaplasma urealyticum organisms (ureaplasmas) and Mycoplasma hominis organisms (mycoplasmas) were sought in mid-stream urines collected from 200 men and 200 women attending hospital with conditions of a non-venereal nature. In addition, the urines from 100 male and 100 female healthy volunteers were examined. Overall, ureaplasmas were isolated four times more often than mycoplasmas. In individuals less than 50 years of age, the organisms were found in about 20 % of men and about 40 % of women. In individuals 50 years or older, they were found about one-third to one-half as frequently. Centrifugation of urine and examination of the resuspended deposit did not increase the isolation rates. In men, the numbers of organisms in the urine were usually small (< 103 c.c.u./ml) with less than tenfold more in the urine of women. The occurrence of 51– > 1000 leucocytes per mm3 in some of the urines was not associated with either the presence or an increased number of ureaplasmas/mycoplasmas, whereas they were associated with the presence of 105 or more bacteria/ml. The significance of these findings in the context of defining the role of ureaplasmas/mycoplasmas in genital-tract disease is discussed.

The antibiotic resistance of Mycoplasma mycoides ssp. mycoides strain T1 was investigated. This strain was resistant to high levels ( > 100 μg ml−1) of rifampicin and nalidixic acid. It was sensitive to streptomycin, spectinomycin and novobiocin; however, single step mutants with high levels of resistance ( > 100 μg ml−1) were readily isolated. With erythromycin and tylosin for which the minimum inhibitory concentration (MIC) for the parent strain was < 0·1 μg ml−1, mutants resistant to > 100 μg ml−1 were obtained in two and three steps respectively. The MIC of tetracycline in single step resistant mutants (0·6 μg ml−1) was tenfold higher than the parent strain, but could not be increased further. There was only a twofold increase in resistance to chloramphenicol in single step mutants. The frequency of resistant mutants varied with the antibiotic and was between 4× 10minuss;6 and 2× 10−8. The mutation rate to antibiotic resistance to streptomycin, spectinomycin, novobiocin, erythromycin and tylosin was between 3× 10−8 and 5× 10−9 per cell per generation. There was a fivefold decrease in mutation rate to resistance to 60 μg ml−1 streptomycin compared to that to 20 μg ml−1.

A strain of Mycoplasma californicum successfully infected an experimentally inoculated ovine mammary gland causing a severe mastitis. The condition lasted for about 25 days, and resulted in atrophy and loss of milk production in the gland. Four experimentally infected ewes, treated over a 3-day period with various regimes of the antibiotics oxytetracycline or tylosin during the acute stage of infection, successfully eliminated the infection. Two others similarly treated with combined intramammary and intramuscular tiamulin or with intramammary Bay Vp2674, did not eliminate the infection; but another ewe treated with intramuscular as well as intramammary Bay Vp2674, did resolve the infection. The two ewes that were unsuccessfully treated with antibiotics at the acute stage did respond to tylosin or oxytetracycline at a later stage of infection. Measurement of antibiotic concentrations demonstrated that the persistence of inhibitory levels in the milk varied between the antibiotics and were influenced by the extent of parenteral treatment.

Of 2356 strains of Salmonella typhi isolated in Britain in the 8-year period 1978–85, 2345 (99·53%) were sensitive to all antibiotics tested and 11 (0·47%) were chloramphenicol-resistant; chloramphenicol resistance was plasmid-mediated in 6 strains. It is concluded that chloramphenicol remains a satisfactory first-line choice of drug for tyrphoid fever in Britain.

The administration per-orally to mice of the non-absorbable antibiotics polymyxin E, tobramycin and amphotericin B resulted in the elimination of detectable aerobic gram-negative rods from the faecal flora without affecting the total viable aerobic count. The addition of parental cefotaxime to the regime caused a fall in the number of aerobic lactobacilli and an increase in the number of enterococci. The rise was associated with the translocation of viable enterococci to the mesenteric lymph nodes and the spleen. The changes induced by cefotaxime were reversed when the antibiotic was withdrawn. Following withdrawal of all antibiotics the total aerobic faecal flora increased to above normal levels, but there was no associated diarrhoea. Attempts to implant exogenous enterobacteria into the digestive tract resulted in only low level colonization both in treated mice and in control mice. These results may have implications for the use of this antibiotic regime for selective decontamination of the digestive tract in humans, particularly those who are immunocompromised.

The consumption of bi-valve mollusean shellfish has been associated with outbreaks of viral gastroenteritis and hepatitis A. Investigations were undertaken to determine the heat inactivation conditions necessary to render shellfish such as cockles safe for the consumer. Conditions for the laboratory maintenance of live cockles are described. In preliminary experiments either poliovirus (106 TCID50/ml seawater) or hepatitis A virus (HAV) (approx. 104 RFU/ml seawater) was introduced into the shellfish tank. Following 48 h filter feeding, virus was recovered from cockles using an adsorption-elution extraction procedure. Titres of virus recovered ranged from 104 to 105 TCID50/ml of shellfish extract for poliovirus and from 103 to 105 RFU/ml of shellfish extract for HAV. Active ingestion of the virus from the seawater was demonstrated by recovering virus from within cockle guts. To quantify recovered HAV, end-point dilutions and an adaptation of a radioimmunofocus assay (RIFA) were compared. The tests were of similar sensitivity but the RIFA has the advantage of being relatively rapid, shortening the time taken to complete an experiment by as much as 4 weeks.