Intended for healthcare professionals

Rapid response to:

General Practice

Penicillin for acute sore throat: randomised double blind trial of seven days versus three days treatment or placebo in adults

BMJ 2000; 320 doi: (Published 15 January 2000) Cite this as: BMJ 2000;320:150

Rapid Response:

For sore throats, seven days of penicillin is not superior.

Based on the findings from a randomized double blind clinical trial,
Professor de Melker and colleagues conclude that seven days of penicillin
therapy is superior to three days of penicillin in treating acute sore
throat symptoms. [1] In the accompanying editorial, Professor Del Mar,
who co-authored the Cochrane review on the lack of efficacy of
antibiotics for sore throat (8 hours benefit), is gently puzzled by the
anomalous de Melker data. [2]

The primary outcome measure of the trial, resolution of symptoms or
resolution of impaired daily activities (figures 1 and 2), in fact do not
support those conclusions
reached by the authors. A cursory inspection of the Kaplan-Meier plots
shows that at day three, the resolution of symptoms among the group 1
(seven days penicillin) and group 2 (three days penicillin) subjects were
significantly different (37% v. 15%), although both groups had received
the identical three days of penicillin therapy. This unexplained two-fold
difference between the two groups is, at best, an unaccounted statistical
fluke, and it carries over, to the misleading conclusions at day seven.

Another way of looking into this data problem is by simply
reformulating the research question as “What is the benefit from an
additional four days of penicillin therapy, after
the initial three days of therapy?”. The data are clear and compelling.

This is how it goes. Day 3 through 7, among the treatment groups 1
and 2, the proportion of patients who were symptom free are 37% v. 15%,
50% v. 25%, 70% v. 43%, 80% v.
55%, and 88% v. 63%. Those figures lead to the cure rate differences of
22%, 25%, 27%, 25%, and 25%, respectively. Using day 3 as the reference
point (since additional
therapy in group 1 starts after this point), the relative benefits as
measured from the day 3 baseline are 3%, 5%, 3%, and 3%, respectively, for
days 4 through 7. This clearly demonstrates that there are no cumulative
effects from the additional four days of penicillin.

Similarly, the other primary outcome measure, resolution of impaired
activity (figure 2), shows relative benefits of -2%, 3%,-1%, and 3%,
respectively, for days 4 through 7.
Again seven days of penicillin has no additional effect over three days of
penicillin for sore throats in resolving impaired activity.

The authors, I think, have a scientific obligation to explain the
anomalous cure rate advantages reported for group 1 patients at the 3 day
baseline. My guess, among other things, is the unaccounted differences in
prognostic factors at entry among the three treatment groups.

For example, the onset of the sore throat is a very important
prognostic factor. The larger the “sore throat days” at entry, the
shorter would be the duration of symptoms. Table 1, the listing of
baseline characteristics, indicates some advantage favoring group1 with
respect to the “sore throat days”. If the authors reanalyze their data
and calculate the proper disease duration (day of onset, day of entry, and
day of resolution), and then correct the life-tables by accounting for
prognostic factors, I am confident that their
conclusions would change. If not, they at least would have an explanation
for the two-fold difference observed at day three among the two
identically treated patient groups.

Since the present results show that the placebo group is doing as
well as the three days of penicillin group in treating the symptoms of
acute sore throats, I see no reason that the Cochrane review of Del Mar
and colleagues should be remanded. [3]

It is somewhat astounding that the peer review of BMJ did not detect
this simple fallacy embodied in the data analysis before publication.
This is a well designed and carefully conducted randomized clinical trial
with health policy implications. It certainly deserves a better data


1. Zwart S, Sachs APE, Ruijs GJHM, et al. Penicillin for acute sore
throats: randomised double blind trial of seven days versus three days of
treatment or placebo in adults. BMJ 2000;320:150-154.

2. Del Mar C. Sore throats and antibiotics. BMJ 2000;320:130-131.

3. Del Mar C, Glasziou PP. Do antibiotics shorten the illness of sore
throat? The Cochrane Library. Acute Respiratory Infections Module at the
Cochrane Database of
Systematic Reviews (updated 02 December 1996). Oxford: Update Software,

Erdem I. Cantekin

Professor of Otolaryngology

University of Pittsburgh, School of Medicine
Pittsburgh, PA,

Competing interests: No competing interests

20 January 2000
Erdem I Cantekin
Professor of Otolaryngology
University of Pittsburgh School of Medicine