Selective decontamination of the digestive tract was developed following the observation that the digestive tract normally has considerable resistance to colonization by newly ingested bacteria. The research that eventually led to selective decontamination was performed because in the late 1960s and early 1970s, the need for prophylaxis against Gram-negative infections in immuno-compromized patients became evident. At that time, the relatively small number of antibiotics available for therapy of serious infections often lead to treatment failure. To introduce the subject of selective decontamination, this paper, therefore, starts with a short historical overview of the kind of infectious agents as well as the antibiotics available in the 1960s; particularly regarding the type and treatment of infections in severely compromised patients. The fact that the possibilities in infectious diseases were limited was the reason for our experimental search for ways of successful prophylactic treatment with minimal risk of development of resistance.

It is well known that severe neutropenia, as usually seen in patients with acute leukaemia, aplastic anaemia or secondary to aggressive chemotherapy, predisposes to infections with Gram-negative enteric bacilli, Pseudomonas aeruginosa, Staphylococcus aureus, and to fungal infections. Infection with anaerobes, in contrast, is rare in patients with haematologic malignancy [1]. The spectrum of bacterial pathogens in this patient population has recently broadened, and now includes coagulase-negative staphylococci, viridans group streptococci, and, occasionally, coryneforms and other rather unusual opportunistic organisms. All these microorganisms originate either from the patient's own microflora, especially from the digestive tract, or from the hospital environment after having colonized the patient during the hospital stay [2]. Studies have shown that the incidence of fever during periods of severe neutropenia approaches 100%, and most of these fever episodes actually represent bacterial infection. For more than 20 years, methods for the prevention of bacterial and fungal infections have been under investigation in patients with profound neutropenia. These included decontamination trails, oral or systemic antimicrobial prophylaxis, strict reverse isolation and maintenance of germ-free conditions [3–8], prophylactic granulocyte transfusions [9], and, more recently, the application of haemopoietic growth factors [10, 11]. The method which remains the most widely used is oral antimicrobial prophylaxis, especially with agents for so-called selective decontamination of the intestinal tract.

Nosocomial infection in intensive care unit (ICU) practice is a common problem and is associated with abnormal carriage of Gram-negative aerobic bacilli in the gastrointestinal tract, resulting in endogenous infections. Selective decontamination of the digestive tract (SDD) is a regimen aimed at preventing or eradicating this abnormal carriage.

A large number of trials examining SDD in ICU practice have been published, the vast majority showing a significant reduction in the incidence of nosocomial, Gram-negative infection. However, the impact on morbidity and mortality is much less certain. A recent meta-analysis has suggested a 10–20% reduction in mortality (3–6% absolute difference) with SDD. A discussion of these results is presented together with potential criticisms of SDD.

Patients undergoing bone marrow transplantation become immuno-compromised for various reasons. Deep granulocytopenia, induced by conditioning (chemotherapy and total body irradiation), renders the patient at risk for serious bacterial and fungal infections. Our strategy for prevention of these infections by selective decontamination (SD) is the result of more than 15 years of clinical experience and research. The combination of antibiotics, used as standard SD (neomycin, polymyxin B, pipemidic acid and amphotericin B), with the application of local antimicrobial agents eliminates aerobic Gram-negative rods, Staphylococcus aureus and Candida spp. from the mucosal surfaces of the digestive tract, while the majority of the anaerobic flora persist and support colonization resistance (CR). The antibiotics used either are not resorbed or do not yield therapeutic serum concentrations. Antibiotics which induce therapeutic serum concentrations, such as ciprofloxacin and cotrimoxazole, are only used for SD on a limited scale. When Gram-negative rods persist despite intake of the standard regimen, ciprofloxacin is given until these persisting rods are eliminated. If the patients cannot swallow the oral regimen, i.v. cotrimoxazole is given temporarily. Streptococcal infections are prevented by the i.v. adminstration of penicillin for 14 days starting on the first day after cytotoxic treatment (conditioning for bone marrow transplantation). The combination of SD and systemic prophylaxis has been shown to be adequate; the major problem then remaining is a relatively mild catheter-associated infection with coagulase-negative staphylococci.

Large increases in the number of cases of human food-poisoning caused by salmonella have occurred in several countries in recent years. In England and Wales the annual number of bacteriologically confirmed cases rose from 10665 in 1981 to 30112 in 1990 and it is generally accepted that these figures are underestimates. The reasons for the unprecedented increase are largely unknown but may include increases in the consumption of convenience foods, poultry, and poultry products, together with a dramatic increase in Salmonella enteritidis infections in poultry. S. enteritidis and S. typhimurium are now the predominant serotypes both in human disease and in poultry.

The relationship between size of the infecting dose and severity of the resulting disease has been investigated for salmonella infections by reanalysis of data within epidemics for 32 outbreaks, and comparing data between outbreaks for 68 typhoid epidemics and 49 food-poisoning outbreaks due to salmonellas. Attack rate, incubation period, amount of infected food consumed and type of vehicle are used as proxy measures of infecting dose, while case fatality rates for typhoid and case hospitalization rates for food poisoning salmonellas were used to assess severity. Limitations of the data are discussed. Both unweighted and logit analysis models are used.

There is no evidence for a dose-severity relationship for Salmonella typhi, but evidence of a correlation between dose and severity is available from within-epidemic or between-epidemic analysis, or both, for Salmonella typhimurium, S. enteritidis, S. infantis, S. newport, and S. thompson. The presence of such a relationship affects the way in which control interventions should be assessed.

An outbreak of Salmonella dublin infection occurred in England and Wales in October to December 1989. Forty-two people were affected, mainly adults, and most lived in south-east England. Microbiological and epidemiological investigations implicated an imported Irish soft unpasteurized cows' milk cheese as the vehicle of infection. A case-control study showed a statistically significant association between infection and consumption of the suspect cheese (p = 0·001). Salmonella dublin was subsequently isolated from cheeses obtained from the manufacturer's premises. Initial control measures included the withdrawal of the cheese from retail sale and a Food Hazard Warning to Environmental Health Departments, as well as a press release, from the Department of Health. Subsequently, a decision was taken by the manufacturer to pasteurize milk used in the production of cheese for the UK market and importation of the cheese resumed in June 1990.

Epidemiological investigation into an outbreak of food poisoning in 17 patients caused by Salmonella enteritidis phage-type 4 demonstrated a highly significant association with consumption of custard, retailed in custard slices and trifles from a bakery on one day. The bakery had changed their recipe for custard 2 weeks earlier to include fresh shell eggs and had not followed earlier national advice on cooking eggs for human consumption. The case-control study supports earlier work associating Salmonella enteritidis phage-type 4 infection with consumption of uncooked or lightly cooked shell eggs.

In the summer of 1991 a human outbreak of Salmonella enteritidis infection occurred following a barbecue in which about 100 persons were involved. Eggs, supplied by one or more of 10 different layer farms, were the most probable source of the infection. To identify the S. enteritidis-positive flocks, an immunoassay was used to detect salmonella serogroup D-specific antibodies in the yolk of hens eggs. Antibody titres in the eggs from two layer farms, farm A and B, clearly exceeded the titres found in randomly collected eggs. Further investigations on farm A and B yielded high antibody titres in the eggs from flocks A1, A2 and B2, and low titres in the eggs from flock B1. S. enteritidis was isolated from the faecal samples of flocks A1, A2 and B2, whereas no salmonella was detected in the faecal samples of flock B1. The flocks present on both farms originated from the same breeder flock.

Among 63 Staphylococcus aureus isolates (one isolate per one patient) counted from infections (from August to November 1991) in hospital T., eight exhibited resistance to fluoroquinolones. Seven of these quinolone-resistant isolates were multiply- and methicillin-resistant S. aureus (QR-MRSA). The results of phage-, plasmid- and genotyping (pulsed field electrophoresis) revealed that six different strain-clones of these MRSA were spread in the hospital. In vitro spontaneous mutants resistant to fluoroquinolones are 1–-100-fold more frequent in MRSA than in other S. aureus when selected on isosensitest-agar containing 1 μg/ml of ciprofloxacin. However, the same mutant frequencies were found in strain 8325–4 with and without the mecA-determinant. The resistance phenotype was stable over 30 generations of subculture in nutrient broth as well in natural quinolone resistant MRSA as in mutants of other types of S.aureus selected in vitro. The phenotypic association of quinolone resistance and MRSA is rather likely due to a higher frequency of spontaneous resistant mutants which are present in natural populations of MRSA. Data of chemotherapy prior to the isolation of 8. aureus show that three of seven patients from whom QR-MRSA were isolated were treated with a quinolone. In eight cases of infections with non-MRSA and quinolone treatment the isolated S. aureus strains were in vitro sensitive to quinolones.

This study examines a series of phenotypic variants of Staphylococcus epidermidis that were generated from a pair of parent variants, isolated from valvular tissue of a patient with prosthetic valve endocarditis. The variants were initially classified by examining their colonial morphology on Congo red agar. In addition to differences in Congo red binding and colonial morphology, they differed in the expression of several surface components and enzymes. Despite these phenotypic differences, all variants had the same restriction endonuclease profile of plasmid DNA. Examination of a collection of clinical isolates demonstrated that phenotypic variation is a common property of S. epidermidis. The ability to express different combinations of surface components and enzymes could contribute to the virulence of S. epidermidis strains by enabling these organisms to colonize a range of diverse environments.

High Staphylococcus hyicus colonization rates were found in vaginal samples of healthy breeding sows and in skin samples of their offspring. Twenty-two different phage types were identified among the 720 isolates of S. hyicus examined. Two to 13 different phage types were isolated per herd. Phage typing, as well as characterization of about 10% of the isolates by plasmid profiles and antibiogram patterns, showed that, several different clones of S. hyicus could be present simultaneously in vagina of gilts and also on skin of piglets. Generally isolates from the vagina of one animal were identical as regards to phage types, plasmid profiles, and antibiogram patterns during the entire investigation period. Isolates from the skin of piglets were of the same type as their mothers, indicating that vertical transmission had taken place. S. hyicus strains isolated from the skin of piglets within 24 h after birth were identical to strains isolated 3 weeks after birth from the same litter, indicating that the vaginal strains became part of a stable skin flora.

Non-secretors of ABO blood group antigens are over-represented among patients with meningococcal diseases. Lower levels of secretory IgA reported for non-secretors have been suggested to compromise mucosal defences. Total serum and salivary IgG, IgA and IgM and levels of these isotypes specific for Neisseria lactamica and five isolates of meningococci were determined by ELISA for 357 pupils and staff of a secondary school in which an outbreak of meningitis occurred. There were no differences in total or specific levels of serum IgG, IgA or IgM or salivary IgG or IgA of secretors compared with non-secretors. Non-secretors had significantly lower levels of salivary IgM (P=0·022) A similar pattern was observed for levels of IgM specific for N. lactamica and five meningococcal isolates. The significance of these results is discussed with reference to the role of secretory IgM in protection of mucosal surfaces in infants.

Between July 1987 and June 1989, 1054 urinary isolates of enterobacteria from Kaohsiung, Taiwan were studied for their trimethoprim resistance. Trimethoprim resistance was defined as MIC greater than 4 μg/ml and high-level resistance by MIC greater than 1000 μg/ml. The incidence of trimethoprim resistance increased from 33·6% in 1987 to 42·1% in 1989. Among the resistant strains studied, 90% were resistant to high levels of trimethoprim. An increase in the proportion of resistant strains (33·9–46·3%) exhibiting high-level non-transferable trimethoprim resistance was noted. The distribution of the dihydrofolate reductase (DHFR) genes by colony hybridization in 374 trimethoprim-resistant isolates revealed the presence of type I and type V DHFR genes in most of these isolates (45·4% and 10·4% respectively). Type I was predominant in Escherichia coli whereas type V was frequently seen in Enterobacter spp. None showed homology with the type II and type III DHFR probe DNA. In addition, transposon Tn7 was present in 7·8% of 374 trimethoprim-resistant enterobacteria.

A total of 598 isolates of Shigella species (24 S. dysenteriae, 254 S. flexneri, 30 S. boydii, 290 S. sonnei) submitted to the Ontario Public Health Laboratories in 1990 were tested for their susceptibility to 14 antimicrobial agents by the agar dilution method. Overall 79·6% of isolates were resistant to one or more antimicrobial agents and 52·0% were resistant to four or more. Trimethoprim resistance ranged from 26·7% among isolates of S. boydii to 39·4% among S. flexneri strains. The majority of the 224 TMP resistant isolates (88·8%) demonstrated high level resistance (MIC > 1000 mg/1) to trimethoprim. Resistance to cotrimoxazole increased from 3% in 1978 to between 26·7 and 37·6% in 1990. MICs for 90% of isolates (MIC90S) for ampicillin, ticarcillin and piperacillin were 128 to > 256 mg/1, > 256 for tetracycline and chloramphenicol, and > 2·0/38·0 for cotrimoxazole. These results from the Canadian Province of Ontario emphasize the need for prudent use of antimicrobial agents in the treatment of shigellosis.

Of 1171 isolates of Escherichia coli isolated from urine samples at the Public Health Laboratory, Toronto, Ontario, Canada, between May 1990 and December 1991, 120 (10·3%) were resistant to trimethoprim (TMP), cotrimoxazole (TMP/SMX), sulfamethoxazole (SMX) and other antimicrobial agents; 110 of the 120 isolates (91·7%) were resistant to four or more agents. The majority of resistant isolates (91·7%) exhibited high-level resistance (MIC > 1000 mg/L) to TMP. The MIC of TMP/SMX for all 120 isolates was > 2·0/38·0 mg/L and for SMX > 1024 mg/L. High-level resistances were also present among the β-lactam antimicrobials with MICs ranging from 16- > 256 mg/L. Forty-three of 120 TMP-resistant (35·8%) isolates conjugally transferred TMP-resistance to E. coli K-12. Co-transfer of several other resistances was observed. SMX cotransferred from 86% of the 43 donors and β-lactams together with SMX cotransferred from 70%. Nalidixic acid resistance was present among 22 (18·3%) of the 120 resistant isolates, however, nalidixic acid resistance was not transferred to E. coli K-12.

The potential risk of bacterial dissemination due to the presence of cockroaches (Blattella germanica, Blattellidae) in low-income flats was investigated. Cockroaches can carry a great variety of bacterial species; we identified 30 different species from 52 different flats. Klebsiella oxycytoca, K. pneumoniae and Enterobacter cloacae were the most frequently found. Pathogenic and potentially pathogenic bacteria represented 54% of all the bacterial identifications. Bacteria were carried either on the cuticle or in the gut. Contamination through external contact is sufficient to insure bacterial diffusion. There was a very low level of overlap estimated by Pianka's index (a) between the bacterial flora of neighbouring blocks of flats, and (b) between bacterial flora of different flats in the same block.

Strains of different bacterial genera: Escherichia coli (7), Salmonella serotypes (7), Enteropathogenic E. coli O 157 (2), Aeromonas hydrophila (4), Yersinia enterocolytica (1), Pseudomonas fragi (5), and Listeria monocytogenes (5), inoculated onto the surface tissues of pieces of fresh beef were found to be susceptible to the lethal action of hot water. On average, more than 3 log10 (99·9%) of these organisms were destroyed when water at 80 °C was applied for 10 or 20 s. Therefore, a hot water decontamination cabinet would be effective for destroying in situ any viable cells of these bacteria which might be present on the surface tissues of fresh beef carcasses.

A new technique to study the prevalence of circulating antibodies directed against different morphological groups (‘morphotypes’) of bacteria in fresh faeces is presented. The technique combines quantitative indirect immunofluorescence with digital image analysis. Plasma antibody titres and patterns of IgA, IgG and IgM isotype against morphotypes of faecal bacteria were determined in ten healthy individuals.

A total of 439 individuals with diabetes mellitus were examined for carriage of yeasts by the oral rinse and palatal swab techniques. Eighteen genetic or environment variables were assessed for their contribution to carriage of yeasts. The factor contributing to palatal and oral carriage of yeasts among individuals with insulin dependent diabetes mellitus (IDDM) was age (P < 0·01). The factor contributing to palatal carriage of yeasts among individuals with non-insulin dependent diabetes mellitus (NIDDM) was poor glycaemic control (glycosuria P < 0·01); carriage in the oral cavity as a whole was influenced additionally by non-secretion of ABH blood group antigens (P < 0·05). Introduction of a denture altered the above risk factors. For individuals with IDDM, oral carriage was associated with the presence of retinopathy (P < 0·05); palatal carriage was influenced by poor glycaemic control (HbA1P < 0·01, plasma glucose levels P < 0·05) and age (P < 0·05). For those with NIDDM, palatal carriage was associated with continuous presence of the denture in the mouth (P < 0·01); oral carriage was associated with plasma glucose levels (P < 0·05).