Importance and generaliseability of trial data
We congratulate Dr Zwart and colleagues for providing a very valuable
addition to the literature, but would like to highlight several important
issues raised by the trial.
Regarding generaliseability, the authors state that 1147 patients
contacted 55 GPs in two years, 21% did not fulfil the Centor criteria,
and 79% of those with the criteria had a positive throat swab. This is, we
are sure, unintentionally misleading, but it implies that the trial
population are representative of those presenting to GPs. Inevitably in
general practice trials some selection occurs, and this can be difficult
to document, but unless the patients in Holland are very different, 1147
patients represents a fraction of those presenting to GPs , and no data
is provided about potential selection bias compared to the larger
presenting population. Assuming 1) those with the Centor criteria who
were not randomised had a similar prevalence of positive throat swabs, and
2) none of those without the Centor criteria has a positive throat swab
(both conservative assumptions), then the 1147 patients had a prevalence
of positive throat swabs of at least 63% i.e. rather unrepresentative of
UK patients. Similarly the high prevalence of fever, and of quinsy in
untreated patients (1:60) - compared to relatively unselected patients in
a UK trial (1:400) [4; 5] or routine data  - indicate this is a rather
selected population. It would be helpful if the authors clarified
information about the presenting population from which their patients were
drawn, and if possible clarified the possible biases in pre-trial
Symptomatic benefit is a little difficult to assess from the trial
results as the severity of symptoms is not presented. Analgesic use was
similar in all trial groups until day four: symptom relief after four days
when symptoms are milder may not be very important to patients. Similarly
the lack of difference in time off school and work implies that the
magnitude of activity restriction in the placebo group was modest. It
would be very helpful if the authors provided information about severity,
not just when activity/symptoms become normal. Even if there is
significant benefit from severity of symptoms and activity restriction,
the estimated figure of 1 and a half days of benefit must be put in
context of the much larger data set from the systematic review of trials
which estimates the likely duration of symptoms is of the order of a 8 .
Perhaps the most important outcome of the study is that quinsy may be
prevented. This is important, since the systematic review of evidence to
date indicates that antibiotics prevent quinsy , but relies heavily (76%
of the weight) on a study where systemically unwell patients admitted to
hospital were given parenteral penicillin, and where quinsy was common
(1:18). This evidence cannot be safely extrapolated to healthier modern
patients given oral antibiotics, mostly not systemically unwell, and with
a much lower prevalence of quinsy. Although the Centor criteria are crude
since they are validated against the throat swab  - which is neither
sensitive nor specific compared to a rise in ASOT titres 8 - the authors 
provide evidence supporting a similar recent trial  that antibiotic
targeting (albeit crude) to a selected clinical subgroup of a modern
population may prevent quinsy. However, since most patients were unwell
with fever , targeting clinical subgroups may be no better than the GPs'
overall judgement that the patient are not unwell systemically. In a
recent trial this judgement identified the common group of patients who
are at low risk of complications [4 5] irrespective of their clinical signs:
20% of the 716 patients [4 5] had 3 out of 4 Centor criteria, and the only
patient who developed quinsy did not fulfil the Centor criteria.
Although double blind randomised 'efficacy' trials provide the most
important basis to make treatment decisions, care must still be taken
extrapolating results to the effectiveness in routine settings. Patients
who agree to efficacy trials may be unusual, and put themselves in the
unusual psychological position of being unsure about treatment. Since
both observational studies and trial evidence suggest that psychological
variables probably determine clinical outcomes [10-12] this may plausibly
modify effect size. A more specific issue about efficacy trials like Zwart
et al1 is that unrealistic manipulation of patient behaviour may occur:
patients had tablets counted, and repeated throat swabs and
examinations. This doesn't happen routinely, and is likely to artificially
improve compliance, altering estimates of efficacy. Regarding symptoms, in
an effectiveness trial (no repeated throat swabs/examinations, no tablet
counting) patients with a very similar cluster to the Centor Criteria, but
not very unwell systemically, showed a benefit from antibiotics of a
fraction of a day [4; 13] much less than reported in Zwart et al .
Regarding complications, the computerised notes for five years in two
contrasting practices (deprived urban and market town) with 16000 patients
documented 9 cases of quinsy/peritonsillar cellulitis following
presentation to the GP with sore throat - i.e. approximately 1:400 - 1;800
, similar to the effectiveness trial . 6 of these 9 patients had
penicillin which did not prevent their quinsy from developing i.e.
preventing 100% of quinsy following sore throat as implied by the efficacy
trial results of Zwart et al  and Dagnelie et al is perhaps unlikely in
We therefore very much welcome the results of this important study,
but the results must be put in context of the population from which they
are drawn, the other evidence available, and the possible difficulties of
the generalising evidence from efficacy trials to effectiveness in routine
Dr Paul Little GP and MRC Clinician Scientist 1,2
Dr Ian Williamson1 GP and Senior Lecturer
Dr Greg Warner2 GP
Dr Michael Moore3 GP
1 Community Clinical Sciences,
(Primary Medical Care Group),
University of Southampton,
Aldermoor Health Centre, Aldermoor Close,
Southampton, England,SO15 6ST
2 Nightingale Surgery, Greatwell drive, Romsey, Hants SO51 7QN
3 Three Swans Surgery, Rollerstone street, Salisbury, Wilts SP1 1DX
1 Zwart S, Sachs A, Ruijs G, Hoes A, DeMelker R. Penicllin for acute
sore throat: randomised double blind trial of seven days versus three days
treatment or placebo in adults. B.M.J. 2000;320:150-154.
2 Centor RM, Witherspoon JM, Dalton HP. The diagnosis of strep
throat in the emergency room. Med.Decis.Making 1981;1:239-246.
3 HMSO, OPCS. Morbidity statistics from general practice: Fourth
National study 1991. HMSO, London:1994;
4 Little PS, Williamson I, Warner G, Gould C, Gantley M, Kinmonth
AL. An open randomised trial of prescribing strategies for sore throat.
5 Little PS, Gould C, Williamson.I., Warner G, Gantley M, Kinmonth
AL. Reattendance and complications in a randomised trial of prescribing
strategies for sore throat: the medicalising effect of prescribing
antibiotics. B.M.J. 1997;315:350-352.
6 Simo R, Avinash P, Belloso A, et al. Complications of sore throat
are not rare. B.M.J. 1998;316 (21st February):631
7 Del Mar C, Glaziou P. Antibiotics for the symptoms and
complications of sore throat. In: The Cochrane Data base of systematic
reviews. The Cochrane Library 1998;last ammended 5/1998:
8 Valkenburg HA, Haverkorn MJ, Goslings WRO. Streptococcal
pharyngitis in the general population. II. The attack rate of rheumatic
fever and acute glomerulonephritis in patients not treated with
penicillin. J.Inf.Dis. 1971;124:348-358.
9 Dagnelie CF, Van der Graf Y, De Melker R, Touw-Otten FWMM. Do
patients with sore throat benefit from penecillin? A Randomised double
blind placebo controlled clinical trial with penicillin V in general
practice. B.J.G.P. 1996;46:589-593.
10 Olsson B, Tibblin G. Effect of patients' expectations on recovery
from acute tonsilltis. Fam.Pract. 1989;6:188-192.
11 Little P, Gould C, Williamson I, Warner G, Gantley M, Kinmonth A.
Clinical and Psychosocial predictors of illness duration from a randomised
controlled trial of prescribing strategies for sore throat. B.M.J.
12 Stewart M. Effective physician-patient communication and health
outcomes: a review. Can.Med.Assoc.J. 1995;152:1423-1433.
13 De Meyere M, Bogaert M, Verschraegen G, et al. Trial of
prescribing strategies in sore throat. B.M.J. 1997;314:1904
Competing interests: No competing interests