A number of treatments for cystic fibrosis have evolved over the past four decades, based on the experience of clinicians involved in the care of these patients. Some of these treatments were developed without the benefit of large randomised controlled trials which would have been difficult to perform at the time. The value of prophylactic antibiotic treatment againstStaphylococcus aureus in the management of infants and children is an example of a logical practice which has developed on the basis of experience, but which requires careful review as to its efficacy and potential deleterious effects.

Pulmonary infection with S aureus is a frequent problem in patients with cystic fibrosis, particularly during the first decade of life.1 Cross sectional studies show that in this age group, 25−30% of patients cultureS aureus from sputum.2 This may be an underestimate as cough swabs in children unable to expectorate are often negative and infection may only be detected by bronchoalveolar lavage in such circumstances.3 Infection with S aureus is usually associated with symptoms, but asymptomatic carriage is also common.

The approach to the treatment of patients with cystic fibrosis withS aureus infection of the airways varies.4 In some centres patients are started on oral antistaphylococcal medication from diagnosis,5 while in others continuous antimicrobial treatment is started when the first infection with S aureusoccurs.6 Treatment is then usually continued into adulthood and is not adjusted whenPseudomonas aeruginosa or other chronic Gram negative infection occurs.5 ,6 Some centres only treat patients with an antistaphylococcal antibiotic for symptomatic exacerbations or if a sputum culture is positive, and treatment is continued until there is symptomatic improvement and eradication of the organism from sputum culture.7 In these centres long term antibiotics are not used and fewer than 10% of patients become chronically colonised withS aureus.1

In the systematic review by McCaffery et al 8 in this issue ofThorax these approaches to treatment are explored. They conclusively confirm that antistaphylococcal treatment consistently achieves sputum clearance of S aureus in patients with cystic fibrosis. Several antibiotics appear to be effective in eradicating S aureus, though none of the studies compared antibiotic treatment with a placebo. McCaffery et al also conclude that prophylactic antistaphylococcal treatment in young children with cystic fibrosis is likely to be of clinical benefit. This conclusion is based mainly on a single study performed in 38 patients over two years by Weaver et al.9 Long term prophylactic antibiotic treatment reduced the frequency of isolates of S aureus from sputum culture compared with intermittent therapy. The only clinical improvements in this study were a reduction in cough frequency and in the number of antibiotic courses and hospital admissions. Measurements of pulmonary function, which are difficult to perform in infants, were not significantly different.10Two other studies with similar numbers, though of a shorter duration, also failed to demonstrate any important clinical advantage in continuous over intermittent antimicrobial therapy.

A potential disadvantage of prophylactic antistaphylococcal treatment is the suggestion of early acquisition ofP aeruginosa reported in two studies included in the review, though this was not seen in the study by Weaveret al. This organism is a key factor in the amplification of pulmonary inflammation and lung injury and is associated with a much worse prognosis than intermittent infection withS aureus. The evidence for a predisposition to P aeruginosa infection in patients on prophylaxis is weak but, if confirmed in an adequately powered study, the value of long term prophylactic antistaphylococcal treatment would be in considerable doubt. In addition, prophylaxis with cephalexin may result in a change from non-mucoid P aeruginosa to the more virulent mucoid phenotype which is associated with a poorer prognosis. There may therefore be a case for stopping such treatment after isolation of P aeruginosa from sputum.

A second problem with long term prophylaxis is the development of resistant strains. This is confirmed in the review by McCafferyet al.8 Treatment with cephalosporins, macrolides, and tetracycline lead to increased resistance but this does not seem to be such a problem with flucloxacillin. Intermittent treatment is not associated with an increase in resistant organisms.7

It is therefore important for an adequately powered, randomised, placebo controlled trial to be performed comparing prophylactic treatment with careful intermittent antistaphylococcal therapy in patients with cystic fibrosis during the first five years of life. Preservation of lung function is the most relevant clinical end point currently available for short term assessment of treatment of lung disease in cystic fibrosis. A positive effect on lung function has not been reported for antibiotic prophylaxis against S aureus. The most effective and safe antibiotic should be chosen, and the evidence from this review suggests that flucloxacillin is likely to be the most appropriate one. Clinical and microbiological end points—particularly lung function, antimicrobial resistance, and rate of acquisition of P aeruginosa—would be important outcome measures. Such a study has been performed with cephalexin but unfortunately it was not reported in a peer reviewed journal and so is not included in the systematic review. It is quoted as a personal communication in a review of the management of cystic fibrosis in which it is indicated that prophylaxis for 5–7 years results in no clinical advantage compared with intermittent therapy other than a reduction in S aureusinfection, but at the cost of an increase in P aeruginosa infection.11

Antibiotic prophylaxis for S aureus has not been shown conclusively to be more effective than prompt treatment of symptoms or positive sputum culture and may have important detrimental effects. We should aim to keep all patients with cystic fibrosis free from pulmonary infection with S aureus, but this should not be at the expense of early acquisition ofP aeruginosa which may worsen prognosis.