1. Fahey T, Stocks N, Thomas T. Quantitative systematic review of randomised controlled trials comparing antibiotic with placebo for acute cough in adults. BMJ 1998; 316: 906-910. (21 March.)
2. MacKay DN. Treatment of acute bronchitis in adults without underlying lung disease. J Gen Intern Med 1996; 11: 557-562[Medline].
3. Mulrow CD, Oxman AD. Cochrane collaboration handbook. Cochrane Library [database on disk and CD ROM]. Issue 2. Oxford: Update Software , 1996.
4. In: Chalmers I, Altman D, eds. Systematic reviews. London: BMJ Publishing, 1995.
5. Sackett D, Richardson WS, Rosenburg W, Haynes RB. Evidence-based medicine. London: Churchill Livingstone , 1997.

Authors' reply

EDITORCates bases his criticism on manipulation of the pooled effect estimates and our attributing the non-significant trend towards antibiotic as evidence of efficacy. Because of the substantial likelihood of bias we were deliberately cautious in attributing benefit to antibiotic.¹ The danger in systematic reviews, particularly those based exclusively on small trials, is not of statistical precision but of systematic bias and the production of false positive results. ^{2 3} Unfortunately, there is no statistical solution to this problem.¹ Funnel plot asymmetry from our systematic review (figure) shows that estimates of efficacy were far greater in the two smaller trials that contributed to the meta-analyses. ^{4 5} Thus we believe that we are correct in our cautious interpretation of the pooled results.

View larger version (11K): [in this window] [in a new window]   Funnel plot of six trials that contributed to outcome of productive cough at follow up. Pooled odds ratio (fixed effects model) is 0.78 (95% confidence interval 0.56 to 1.08)

In suggesting that there is a trend favouring the use of antibiotics over placebo, Shakespeare and Bourke fail to account for all the evidence that we presented. They are right that the number of subjects (700) in which the outcome of productive cough was based is small. They ignore, however, the 829 patients in the trial by Howie and Clark, which reported no difference between antibiotic and placebo but which we did not include in the pooled estimate because analysis was by episode of illness. Furthermore, as we reported, the trials with more positive results were smaller trials with substantial losses to follow up. ^{4 5}

We would have liked to include Howie and Clark's trial in the pooled analysis; we reported its results in detail because we wanted to emphasise its importance in the context of the evidence presented in our review. We agree with Harris's anxieties about the quality of individual studies, and this accounts for our cautious conclusions concerning pooled estimates. We presented losses to follow up because this allows readers to judge for themselves the quality of trials that contributed to the review. We agree with Harris that quality criteria and scoring systems may not distinguish high and low quality trials.

The results of our review should be seen in the context of the high prescribing of antibiotics for acute chest infection in otherwise healthy people throughout the developed world. We reported the number needed to treat (n=11) and number needed to harm (n=15) for clinical improvementeven if we accept the trend towards antibiotic as evidence of efficacyto show that the clinical benefit of antibiotic is probably marginal. Further trials in higher risk groups are more likely to show important benefits for patients.

In collaboration with American colleagues we will update this review for the Cochrane Library. We will make explicit our concerns about potential biases in the updated version.