Rapid Responses to:

LETTERS: C J Cates, Thomas P Shakespeare, Rachael C Bourke, Jessica Harris, Tom Fahey, and Nigel Stocks Systematic review of trials comparing antibiotic with placebo for acute cough in adults BMJ 1998; 317: 1014a [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

Two metaanalyses on acute bronchitis/cough with opposite conclusions about antibiotic treatment.

Morten Lindbęk, Per Hjortdahl   (5 November 1998)

Re: Two metaanalyses on acute bronchitis/cough with opposite conclusions about antibiotic treatment

Atle Klovning   (9 February 1999)

statistical calculation error

Helen E Hawkins   (18 September 2002)

Two metaanalyses on acute bronchitis/cough with opposite conclusions about antibiotic treatment. 5 November 1998
Morten Lindbęk, Associate professor Department of General Practice, University of Oslo, Per Hjortdahl Send response to journal: Re: Two metaanalyses on acute bronchitis/cough with opposite conclusions about antibiotic treatment.	We have read with great interest the discussion about the meta- analysis of patients with acute cough where Fahey et al concluded that treatment with antibiotic gave no significant differences in the resolution of acute cough (1,2), although they found a trend favouring antibiotics. The Cochrane meta-analysis on treatment of acute bronchitis concluded that antibiotics appear to have a modest beneficial effect (3). 8 of the 9 extracted studies were the same in both meta-analyses, and 90% of the 750 patients were used in both meta-analyses. How can two meta-analyses based on the same studies come to the opposite conclusions? In the Fahey study the outcome measures were resolution of cough and clinical improvement on re-examination. In the Cochrane meta-analysis they found four outcomes giving significant differences between antibiotic treatment and placebo. The patients were less likely to report feeling unwell at a follow up visit (two studies), to show no improvement on physician assessment (four studies), to have abnormal lung findings (two studies), and the patients had a more rapid return to work or usual activities (five studies). The two meta-analyses agreed in finding no significant difference in proportion of patients with resolution of cough at follow up. Both meta-analyses agreed that side- effects were significantly higher in the antibiotic groups. The search strategy and the inclusion/exclusion does not seem to explain the differences in conclusion as 90% of the patients were included in both meta-analyses. The answer may lie in the objective and the choice of outcome measures of the two analyses. Fahey et al had this objective: to assess whether antibiotic treatment for acute cough is effective. The Cochrane group had this objective: to determine whether antibiotic treatment is associated with any improved outcomes in patients with a clinical diagnosis of acute bronchitis. In our opinion the main questions in evaluations like this should be: Which outcome measures are the most clinical relevant? Are the differences found to be statistically significant also clinically significant? As the choice of outcome measures may be crucial for the main conclusion, the choice should be explicit early in the study and be well substantiated. Meta-analyses based on small studies are not easy to interpret, and conclusions should be drawn with caution. This example demonstrates that also in meta-analyses there is an element of subjectivity in the research question posed, the choice of outcome measures and the evaluation of whether statistical significant differences also are clinical significant. References: 1. Fahey T, Nigel Stocks N, Thomas T. Quantitative systematic review of randomised controlled trials comparing antibiotic with placebo for acute cough in adults BMJ 1998;316:906-10. 2. Cates CJ, Shakespeare TP, Bourke RC, Harris J, Fahey T, Stocks N. Systematic review of trials comparing antibiotic with placebo for acute cough in adults. BMJ 1998; 317:1014. 3. Becker L, Glazier R, McIsaac W, Smucny J. Antibiotics for Acute Bronchitis. In: Douglas R, Bridges-Webb C, Glasziou P, Lozano J, Steinhoff M, Wang E (eds.) Acute Respiratory Infections Module of the Cochrane Database of Systematic Reviews.
Re: Two metaanalyses on acute bronchitis/cough with opposite conclusions about antibiotic treatment 9 February 1999
Atle Klovning, Research Fellow University of Bergen Send response to journal: Re: Re: Two metaanalyses on acute bronchitis/cough with opposite conclusions about antibiotic treatment	Letter to the Editor In their electronic response (1) Morten Lindbęk and Per Hjortdahl had found that the authors of (2)had arrived at a different conclusion than the Cochrane Systematic Review using 8 of the 9 studies, comprising 90% of the same patients (3). I looked into the Cochrane review and discovered that for the same outcomes studied in both reviews, there were found different numbers of studies included for the meta-analysis, even though the Fahey paper also had used the Cochrane Handbook for their quality rating of included studies. Also, and more worrying, when citing the same studies, the numbers did not correspond. E.g. for proportion of productive cough at follow up, the Cochrane review used only one study, Dunlay et al, while the Fahey et al used 6 studies. The numbers cited by the Cochrane and Fahey reviews (antibiotic vs. placebo)for the Dunlay et al study were 14/23 vs. 18/22 and 10/21 vs. 17/24, respectively. For the outcome "Proportion of subjects who had not improved clinically at follow up", the difference was even greater, 2/37 vs. 1/32 and 16/37 vs. 11/32, respectively. It may therefore seem that the difference in conclusions in fact may be based on miscalculations. Atle Klovning, MD Div. for General Practice University of Bergen Norway No conflicts of interest. References 1. Two metaanalyses on acute bronchitis/cough with opposite conclusions about antibiotic treatment. Morten Lindbęk, Per Hjortdahl, Associate professor , Department of General Practice, University of Oslo. eBMJ, 4 Nov 1998. 2. Cates CJ, Shakespeare TP, Bourke RC, Harris J,Fahey T, Stocks N. Systematic review of trials comparing antibiotic with placebo for acute cough in adults. BMJ 1998; 317:1014. 3. Becker L, Glazier R, McIsaac W, Smucny J. Antibiotics for Acute Bronchitis. In: Douglas R, Bridges-Webb C, Glasziou P, Lozano J, Steinhoff M, Wang E (eds.) Acute Respiratory Infections Module of the Cochrane Database of Systematic Reviews.
statistical calculation error 18 September 2002
Helen E Hawkins, Research Fellow University of Queensland, Australia 4072 Send response to journal: Re: statistical calculation error	It appears that the value for the chi squared test for heterogeneity in the efficacy of antibiotic section is incorrect. It is given as 16.87, df=5, p<0.5. Perhaps the chi squred test should be 1.687. This would give a df =5 and p<0.5