Review of case definitions for nosocomial infection--towards a consensus. Presentation by the Nosocomial Infection Surveillance Unit (NISU) to the Hospital Infection Liaison Group, subcommittee of the Federation of Infection Societies (FIS).

As part of the preparatory work for a national surveillance scheme, consensus was sought on the case definitions for nosocomial infection. We compare six sets of case definitions for nosocomial urinary tract infection, surgical wound infection, bloodstream infection and pneumonia, and highlight areas of agreement and variation. We hope this will stimulate discussion among those with expertise and interest in surveillance and so contribute to the development of nationally agreed case definitions.

Pneumococci causing invasive disease in Britain 1982-1990.

A total of 5348 isolates of Streptococcus pneumoniae was serotyped and screened for insusceptibility to tetracycline, penicillin, erythromycin and chloramphenicol. Of these, 4238 (79%) were isolated from patients who had pneumonia or meningitis or were bacteraemic. Altogether, 3948 (74%) of the isolates belonged to one or other of the serotypes 1, 3, 4, 6, 8, 9, 14, 19 or 23 with serotypes 6, 14, 18, 19 and 23 being frequent causes of invasive disease in young children. Many isolates of type 1 were isolated from pneumonia and few from meningitis. Some 768 (14%) isolates were insusceptible to one or more antibiotic and 591 of these belonged to serotypes 6, 9, 14, 19 or 23. Representatives of type 14 resistant to erythromycin were prominent from 1986 onwards. There was an increase in the number of multi-resistant pneumococci from 1985. Among these were isolates of type 23 insusceptible to penicillin, chloramphenicol and tetracycline and cultures of type 6 resistant additionally to erythromycin.

Gentamicin resistance in clinical isolates of Escherichia coli encoded by genes of veterinary origin.

Seven (27%) of 26 gentamicin-resistant human clinical isolates of Escherichia coli were resistant to the veterinary aminoglycoside antibiotic apramycin. A gentamicin-resistant Klebsiella pneumoniae isolate from a patient infected with gentamicin/apramycin-resistant E. coli was also resistant to apramycin. DNA hybridisation studies showed that all gentamicin/apramycin-resistant isolates contained a gene encoding the enzyme 3-N-aminoglycoside acetyltransferase type IV (AAC[3]IV) that mediates resistance to gentamicin and apramycin in bacteria isolated from animals. Seven of the eight gentamicin/apramycin-resistant isolates were also resistant to the veterinary antihelminthic agent hygromycin B, a phenomenon observed previously in gentamicin/apramycin-resistant Enterobacteriaceae isolated from animals. Resistance to gentamicin/apramycin and hygromycin B was co-transferable in six of the isolates. Restriction enzyme analysis of plasmids in apramycin-resistant transconjugants derived from E. coli and K. pneumoniae isolates from the same patient were virtually identical, suggesting that inter-generic transfer of plasmids encoding apramycin resistance had occurred in vivo. These findings support the view that resistance to gentamicin and apramycin in clinical isolates of E. coli results from the spread of resistant organisms from animals to man, with subsequent inter-strain or inter-species spread, or both, of resistance genes on transferable plasmids.

The cost of infection in surgical patients: a case-control study.

To determine the excess hospital cost attributable to hospital acquired infection in a UK hospital 67 surgical patients with hospital acquired infection (HAI) were matched with uninfected controls on the primary features of the first operative procedure and primary diagnosis, and on the secondary features of sex, age and surgical service. Costs were calculated from the hospital's unit costs for pathology, radiology and for the cost of one day's extra stay. The mean cost of one day of antibiotic therapy was also measured. In infected patients there was a significant increase in the length of hospital stay of 8.2 days with a mean extra cost per patient of 1041 pounds (P < 0.001). Microbiology, haematology, chemical pathology and radiology requests were all significantly increased with a mean extra cost per infected patient of 10.4 pounds, 7.8 pounds, 96. pounds, and 3.3 pounds, respectively. Antibiotic therapy contributed significantly to the extra costs (44 pounds per infected patient). The mean extra cost per patient was highest in orthopaedic patients (2646 pounds) and least in gynaecology patients (404 pounds). For the infections with significantly increased cost, multiple infections carried the greatest (3362 pounds), and urinary tract infections the least (467 pounds) cost. Hospital length of stay was the greatest contributor to the cost and accounted for 95% of the extra cost in orthopaedics, 94% in gynaecology and 92% in general surgery and urology. Antibiotic therapy was the second most significant contributor to cost and, with the exception of urinary tract infection and infections in gynaecology, was at least five times more per patient than requests for microbiology, haematology, chemical pathology or radiology.

Introduction of laboratory based ward liaison surveillance of hospital infection into six district general hospitals.

A previous study demonstrated that laboratory based ward liaison surveillance (LBWLS) of hospital infection was an effective and efficient method. The method involved the follow-up of positive microbiology reports by the review of patient records and liaison with ward nursing staff to consider whether any patients had infection. Here we report the introduction of LBWLS into six district general hospitals to determine whether it is feasible to use this method on an everyday basis. The time required for data collection was assessed and the method was compared with a reference method in one hospital to check its ability to detect infections. To assess reproducibility two infection control nurses (ICNs) performed LBWLS independently, but concurrently, for 5 weeks. The method could be used in all hospitals studied; however, the time for data collection ranged from 3.0 to 6.8 h/100 beds per week. In comparison with the reference method, LBWLS detected 15/41 (37%) of community acquired infections and 30/43 (70%) of hospital acquired infections. In the reproducibility assessment 72 patients were identified by both ICNs. There was agreement about the infected/non-infected status of 65 of these patients. The mean pair agreement and Kappa statistic were 0.88 and 0.72. Laboratory based ward liaison was readily used in all hospitals and was reproducible.

An evaluation of surveillance methods for detecting infections in hospital inpatients.

Eight selective surveillance methods were compared with a reference method for their ability to detect hospital infections in patients was also assessed. In the reference method, case records were reviewed three times a week, and during the 11-month period of study, 668 infections were identified amongst 3326 patients. Three hundred and thirty-eight were community acquired infections (CAI) and 330 were hospital acquired infections (HAI). The time for data collection was 18.1 h per 100 beds per week. Of the selective surveillance methods, those based on the review of treatment and temperature charts detected the highest proportion (70%) of CAI; and the review of microbiology reports with regular ward liaison identified the highest proportion (71%) of HAI. The time for data collection in the eight methods ranged from 1.2 h per 100 beds per week to 6.5 h per 100 beds per week. After considering the sensitivity for identifying patients with HAI and time for data collection, the review of microbiology reports with regular ward liaison was judged to be an effective and efficient method of surveillance.

An 11-month incidence study of infections in wards of a district general hospital.

Between March 1988 and January 1989, an incidence study of infections in patients occupying 122 beds in a district general hospital was undertaken. Nursing notes, medical notes, temperature charts, drug prescription charts and laboratory information were reviewed three times a week to determine if patients had infection which met strict case definitions. In addition, the surveyor consulted with ward nursing and medical staff for clarification of symptoms and signs indicative of infection. During the study, 668 infections were identified amongst 3326 patients. Three hundred and thirty-eight (51%) were community-acquired infections (CAI) and 330 hospital-acquired infections (HAI). Excluding 24 HAI acquired in other hospitals, the incidence rates were 9.2 HAI per 100 discharges, and 1.1 HAI per 100 patient days. The common types of CAI were pneumonia, abdominal infection and urinary tract infection. The main types of HAI were urinary tract infection, surgical wound infection and pneumonia. The microorganisms most frequently associated with CAI and HAI were Gram-negative bacilli.

Acquired immune deficiency syndrome: recommendations of a Working Party of the Hospital Infection Society.

Unified procedures to control those infections that are transmitted by inoculation of blood are recommended. These should be applied to patients with acquired immune-deficiency syndrome (AIDS), AIDS-related complex, persistent generalized lymphadenopathy or hepatitis B, those with serological evidence of infection by human immunodeficiency virus or hepatitis B virus, and those in medical and social categories with a higher than average prevalence of such infections. When the identification of these categories of patient would be particularly difficult, the precautions should be applied to all patients, in situations of high risk for inoculation incidents. Rational infection-control measures, based on the known mode of spread, permit efficient management of infected patients, with satisfactory protection of staff and other patients.

Control of infection in general practice: a survey and recommendations.

Twenty general practices in four areas in Britain were surveyed to establish their needs for and practices of sterilising and disinfecting equipment. Of the 327 items of equipment and instruments examined in the survey, 190 were satisfactorily decontaminated, 100 were treated in a way judged to result in doubtful decontamination, and in 37 cases treatment was considered unsatisfactory. Decontamination apparatuses (autoclaves, hot air ovens, and hot water disinfectors) were generally in good working order, but the use of chemical disinfectants was often inappropriate. Recommendations were made on appropriate methods of decontamination for various items in common use in general practice. By virtue of the large numbers of patients treated by general practitioners there is a substantial possibility of transmitting infection; having appropriate methods for decontaminating instruments and equipment is therefore imperative.

Evaluation of electrophoretic methods for typing methicillin-resistant Staphylococcus aureus.

Three electrophoretic methods of typing methicillin-resistant Staphylococcus aureus (MRSA) strains--plasmid profiles (PP), whole-cell protein profiles (WCPP) and immunoblotting profiles (IP)--were evaluated and compared with phage typing. The results obtained with isolates from 12 outbreaks were compared both within the outbreaks, to determine the consistency of results, and between outbreaks. There was generally good agreement between the typing methods but in only six outbreaks did all four methods indicate the same relationship between isolates. WCPP comprised more than 50 bands; when differences occurred, they were seen in only a few bands. In contrast, IP comprised only one or two major bands and the differences were much easier to interpret. The PPs of many of the isolates were similar; many isolates contained a plasmid of mol. wt (18-25) x 10(6). In several outbreaks both WCPP and IP showed minor differences between isolates that were not apparent with phage typing. When comparisons were made between the 12 index strains and an isolate representing the London epidemic MRSA strain, phage typing and WCPP were the most discriminatory methods; both gave nine distinct patterns, whereas there were eight IPs and only six PPs amongst the 13 strains. It was concluded that both WCPP and IP could provide valuable epidemiological data on MRSA and that IP was the easiest of the three methods to interpret.

A comparison of the effects of preoperative whole-body bathing with detergent alone and with detergent containing chlorhexidine gluconate on the frequency of wound infections after clean surgery. The European Working Party on Control of Hospital Infections.

In a prospective, randomized, double-blind, placebo-controlled study involving 27 surgical units in six European countries, the effect of preoperative whole-body bathing on two occasions with a detergent containing chlorhexidine (CHX+) on the incidence of wound infection in elective, clean surgery was compared with two bathings with a detergent without chlorhexidine (CHX-). In the CHX+ group 2.62% of 1413 patients and in the CHX- group 2.36% of 1400 patients subsequently became infected. The infection rate in the CHX+ group was 1.11 times that in the CHX- group with 95% confidence limits ranging between 0.69 and 1.82. Consequently, bathing patients twice preoperatively with chlorhexidine-detergent did not reduce the incidence of infection of clean wounds.

Comparison of faecal florae in patients with rheumatoid arthritis and controls.

The faecal flora of 25 out-patients with active rheumatoid arthritis (RA) was compared with that of 25 age- and sex-matched controls. A comprehensive survey revealed a significantly higher carriage rate of Clostridium perfringens in the RA population (88%) than controls (48%) (p less than 0.01). Coliform counts also tended to be higher, but there were no other significant differences between patients and controls. When the study was enlarged to include a further 113 RA patients with variable disease activity and a further 38 controls, clostridia were again more frequently carried by those with RA (70%) than controls (45%) (p less than 0.01) and clostridial counts were significantly higher in the patient group (p = 0.006). Moreover, counts in patients with active or moderately active disease were significantly higher than in those with inactive disease (p less than 0.001). These data are consistent with the hypothesis that Cl. perfringens plays a role in triggering or is otherwise associated with disease activity in RA. The findings may be alternatively an effect of the disease or its treatment with, for example, anti-inflammatory drugs.