Hospital Pharmacist Vol 7 No 2 p52-54
February 2000 Original Papers

Controlling Clostridium difficile infection by a restrictive antibiotic policy

By R. P. D. Cooke, MRCP(UK), FRCPath, D. Binyon, MRPharmS, MSc, and S. V. Goddard, RGN

Aim - To assess the long-term effect of a restrictive antibiotic policy on the incidence of Clostridium difficile infection (CDI).
Design - Early reporting of restricted antibiotic prescriptions was made by ward- based pharmacists to consultant medical microbiologists, following the hospital antibiotic policy. After discussion with the relevant physician, prescriptions were discontinued if considered inappropriate. The total number of days of restricted and freely available antibiotic treatment per 1,000 inpatient days was calculated using a defined daily dose. The standard error test was used for statistical analysis. CDI was diagnosed by testing liquid stool samples for C difficile toxin A (CDTA) in all inpatients over 60 years, in others on antibiotics or chemotherapy and in those with bloody diarrhoea.
Setting - A 650-bed district general hospital providing all acute facilities. Study performed between 1993 and 1998.
Outcome measures - Annual incidence of CDI and total treatment days for restricted (and freely available) antibiotics per 1,000 inpatient days, for each year of the study.
Results - There was no increase in CDI. Annual rate of CDTA-positive tests fell from 12 to 8.3 per cent between 1993 and 1998, but the change was not statistically significant. Restricted antibiotic usage followed the guidelines for specialist use. There was no significant change in restricted (or freely available) antibiotic usage during the study. Conclusions - Incidence of CDI can be controlled by strict application of a restrictive antibiotic policy, provided there is close liaison with ward-based pharmacists.

Clostridium difficile infection (CDI) remains the foremost identified cause of nosocomial diarrhoea.¹ It is responsible for virtually all cases of pseudomembranous colitis. Recurrent CDI may occur in up to 20 per cent of affected patients. Non-intestinal manifestations include C difficile-associated arthritis and septicaemia.²
Although the exact number of CDI cases in the United Kingdom (UK) is not known, the numbers of laboratory reports and clinical cases of CDI increased markedly in the 1990s. Whether this is a true increase due to more frequent use of antibiotics, or an apparent increase due to increased and improved diagnostic methods, is not known.
In England and Wales between 1990 and 1994, the number of CDI cases increased eight-fold.³ Annual figures for the detection of faecal toxins A or B, which form the basis of the laboratory diagnosis of CDI, rose by 32 per cent in 1997.⁴ Hospital outbreaks of CDI resulting in ward closures are also widespread though falling, despite the increasing number of laboratory reports.⁵
CDI is a major cause of morbidity, especially in the elderly population. The burden of morbidity, mortality and costs attributable to hospital-acquired infection has recently been highlighted by the Office of Health Economics.⁶ In the case of CDI, these costs probably account for millions of pounds each year.⁷ The major factor is the prolonged duration of hospital stay which averages 21-24 days.^8,9
The most important known risk factor for CDI is antibiotic exposure.¹⁰ Following a major outbreak of CDI in which 17 patients died,¹¹ the Public Health Laboratory Service and the Department of Health published guidelines for the prevention and management of CDI.¹⁰ These emphasised the importance of controlling antibiotic usage and, in particular, limiting the use of broad-spectrum antibiotic therapy. However, subsequent antibiotic audits have often shown a discrepancy between observed practice and policy.¹² Furthermore, some medical microbiologists appear to be unconvinced about the effectiveness of antibiotic guidelines.¹³The long-term effect of a restrictive antibiotic policy on CDI also remains unclear. The introduction of restrictive antibiotic policies has successfully reduced the rate of CDI when standard infection control precautions have failed, particularly in units for the elderly, where C difficile disease rates are associated with the use of broad- spectrum antibiotics.^14,15
However, such interventions have been uncontrolled and subject to limited follow- up. The components of enhanced infection control programmes that have most influence on CDI (eg, handwashing, antibiotic restriction, stop dates and feedback on infection rates) are not well defined.^16,17
In general, restrictive hospital antibiotic prescribing policies simplify the education of junior medical and paramedical staff as well as improving familiarity with freely available agents. We have adopted such a policy in our own hospital and have audited its effect on CDI over a five-year period.

Methods

Our general hospital has 650 beds and provides the vast majority of the district's acute facilities. A revised hospital antibiotic policy was established in 1993 after discussion with prescribers and approval by the hospital's drug and therapeutics committee. It was based on the general recommendations of the British National Formulary¹⁸ and local patterns of antibiotic resistance.
The policy promotes adherence by encouraging early discussion with consultant medical microbiologists as appropriate. The use of certain antibiotics (third-generation cephalosporins, co-amoxiclav, co-fluampicil, aztreonam, imipenem, antipseudomonal penicillins, 4-quinolones, clindamycin and vancomycin) are restricted on the basis of rational and economic prescribing.
The policy is closely monitored by ward-based pharmacists. Prescriptions for restricted antibiotics that fall outside locally agreed guidelines (eg, for neutropenic fever or neonatal sepsis) are reported by the ward-based pharmacist to the consultant medical microbiologist as soon as possible if an initial review fails to clarify the prescriber's rationale. After discussion with the relevant clinician, the prescription may then be changed to a more suitable antibiotic.
The microbiology laboratory has a standard operating procedure for diagnosing CDI, which has remained unchanged since 1993. Liquid stool samples are tested for C difficile toxin A (CDTA) by enzyme immunoassay (Vidas, BioMerieux) in all inpatients over 60 years of age and in other patients on antibiotics or chemotherapy and those with bloody diarrhoea. All patients with positive CDTA results were followed up by one of the authors (SVG).
Over the five-year study period (1993-98), the total number of antibiotic treatment days for both restricted and freely available antibiotics was calculated by dividing the total amount of each antibiotic issued to the wards per annum by the defined daily dose.¹⁹
Data were obtained retrospectively by reviewing computerised records held in the pharmacy department. Incidence rates for CDI were also determined annually. In addition, the number of CDI cases and antibiotic treatment days were expressed per 1,000 inpatient days for each year of the study to adjust for inpatient hospital activity. The standard error test was used to detect any statistically significant differences in results.

Results

All cases of CDI were sporadic. There was no evidence of a C clifficile outbreak during the study period. The annual number of stool samples tested increased from 298 to 563 during the study. This rise probably reflects a combination of higher inpatient activity and greater awareness by staff on the ward.
An average of 40 patients (range 37-47) were diagnosed each year with CDl. Annual rates of CDI fell from 12 per cent to 8.3 per cent over the period 1993-98 (Figure 1). However, there was no significant difference in the number of CDI cases per 1,000 in-patient days during the study period: figures ranged from 0.20 to 0.26 with a 95 per cent confidence interval (CI) of 0.16-0.27 (Figure 2).

Figure 1: Annual number of cases of CDI/percentage of positive tests, 1993-98

Figure 2: Annual antibiotic treatment days and number of cases of CDI per 1,000 inpatient days

The mean age of CDI patients was 77 years (range 18-94 years). Ninety-one per cent of patients were aged 60 years or more. The majority came from the medical and care for the elderly units.
No history of recent antibiotic exposure was noted on the microbiology request form in 24 per cent of cases. Laboratory criteria alone successfully identified 76 per cent of all CDTA positives. Twenty per cent of CDTA-positive patients died directly or indirectly from their infection.
Although mean antibiotic treatment days were determined for each year of the study, data for 1993-94 were incomplete and therefore not suitable for analysis. The total number of treatment days per annum for the most frequently used restricted antibiotics in 1994-98 is shown in Figure 3.

Figure 3: Annual total number of treatment days with restricted antibiotics, 1994-98

Generally, restricted antibiotic usage followed the antibiotic guidelines for specialist use within the antibiotic policy. Third-generation cephalosporins (eg, cefotaxime) co-amoxiclav and ciprofloxacin were the most commonly prescribed restricted agents, and these were subsequently changed to a narrower spectrum antibiotic after consultation. Piperacillin, and subsequently piperacillin/ tazobactam, were generally used only in the treatment of neutropenic fever and ventilator-associated pneumonia, following unit policies.
There was no significant change in restricted antibiotic usage per 1,000 inpatient days during the study: figures ranged from 25.9 to 31.4 (CI 23.8-33.0).
This was also the case for freely available antibiotics: figures here ranged from 332.7 to 408.4 (CI 300.5-450.2) These results are shown in Figure 2.

Discussion

Unlike the incidence of CDI in the UK, which showed a widespread increase over the five-year study period, the incidence of infection in our hospital did not change over this period. The effectiveness of the restrictive policy was primarily due to the close liaison between ward-based pharmacists and consultant medical microbiologists, which allowed early intervention when restricted antibiotics were prescribed. The majority of requests for, and treatments with, restricted antibiotics involved cefotaxime, co-amoxiclav and ciprofloxacin. This is not surprising since these antibiotics are often freely available in UK hospitals.
Third-generation cephalosporins have a particular propensity to cause CDI.^20-23 Indeed, in a recent review of risk factors for CDI, these antibiotics were considered to have the greatest risk for C difficile diarrhoea when compared with 17 other commonly used antibiotics.²⁴ Within the hospital's antibiotic policy, cefotaxime is usually preferred to cefuroxime only in the treatment of bacterial meningitis, owing to its superior penetration into cerebrospinal fluid and improved activity against penicillin-resistant pneumococci.¹⁸ Co-amoxiclav is ranked next after third-generation cephalosporins for CDI risk, although there is less supporting evidence.²⁴ Co-amoxiclav is marketed as a suitable antibiotic for a variety of infections, yet it is a first choice agent only for the treatment of bite injuries.¹⁸ At Eastbourne district general hospital, only one consultant surgeon regularly uses this antibiotic, preferring its oral formulation for diabetic foot and vascular infections.
Neither cefotaxime, ceftazidime or co-amoxiclav are listed by the Committee on Safety of Medicines as antimicrobial agents associated with antibiotic-associated colitis.²⁵ This list, which was last reviewed in 1994, clearly needs updating.
Ciprofloxacin is a relative C difficile-sparing antibiotic.²⁴ However, it is expensive and has relatively few first-choice indications.¹⁸ During the study, the use of ciprofloxacin was primarily restricted to patients in whom an amino-glycoside was contraindicated and to those requiring treatment for salmonellosis.
Reported prevalence rates for CDI range from 0.15 per cent to 10 per cent in hospitalised patients during non-outbreak situations.² This wide variation is probably due to different laboratory criteria for
C difficile toxin testing and diverse hospital populations. The mean annual rate of CDI, based on CDTA-positive results, was 10 per cent in this study, which is likely to reflect widespread use of freely available antibiotics. As up to 50 per cent of inpatient antibiotic prescriptions are considered to be of questionable therapeutic value,²⁶ sole reliance on a restrictive antibiotic policy will only partially control CDI.
Education programmes, automatic stop dates, written justification on prescriptions and appropriate laboratory reporting of antibiotic susceptibilities are equally important components of policies aimed at improving antibiotic usage.¹⁰ Nevertheless, with adherence to rational antibiotic guidelines, the incidence of CDI can be kept relatively low. An effective way to enforce antibiotic guidelines is through close liaison between ward-based pharmacists and consultant medical microbiologists.

Dr Cooke is consultant medical microbiologist, Mr Binyon is principal pharmacist - clinical services and Miss Goddard is senior infection control nurse at Eastbourne district general hospital, East Sussex

References