Best Evidence Topics
  • Send this BET as an Email
  • Make a Comment on this BET

Is prescription of antibiotics or surgical intervention best for recurrent tonsillitis?

Three Part Question

In [an patient with recurrent tonsillitis] are [antibiotics better than surgical intervention] at [reducing total duration of symptoms]?

Clinical Scenario

A 7 year old child presents to A&E with severe tonsillitis for the fourth time in 10 months. As they are discharged with antibiotics the SHO wonders if she should write to the GP suggesting ENT referral.

Search Strategy

Medline 1966 to June Week 1 2005
Embase 1980 to 2005 Week 24
Cinahl Cumulative Index to Nursing & Allied Health Literature 1982 to June Week 1 2005
The Cochrane Library Issue 2 2005
{ OR exp TONSILLITIS/} AND { OR exp Anti-Bacterial Agents/} AND { OR exp TONSILLECTOMY/ OR exp SURGERY/} limit to (humans and English language)

Search Outcome

medline 126 papers found, 2 relevant papers found including 1 systematic review.
embase 285 papers found, no extra relevant papers
cinahl -3 papers found, none relevant
cochrane 1 systematic review found

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
Burton MJ, Towler B, Glasziou P
April 1999
The search stategy included Cochrane,Medline and Embase, which were searched for RCT's and controlled clinical trials. Quality was assessed and data extracted and analysed blindly by 2 reviewers. 6 studies were identified which included children. 5 of these were RCT's however in 3 of these the either all the surgical patients or an unknown number underwent adeno-tonsillectomy. This may affect the tendency to have sore throats after the operationsystematic review of high quality RCT'sIs antibiotic prescription or surgical intervention more effective?No results presented. The studies assessed were of insufficient quality to draw conclusions.This systematic review was essentially negative. However it is important to know that no high quality trials have been carried out.
Stafford N, von Haacke N, Sene A
40 patients referred to ENT by their GP's were included provided they were; age 16 years or over, had a history of at least 4 severe episodes of tonsillitis per year for at least 2 years and good health between attacks. A normal FBC and negative Paul Bunnel test. Group 1. 20 patients underwent tonsillectomy and were reviewed at 6 months, 1 yr and 18 months post-op. Group 2 were treated with 8 day courses of antibiotics for each episode of tonsillitis over the year. Patients were reviewed every 3 months and GP's returned questionaires each time the patient was seen.Randomised controlled trialTonsillectomy patients who felt 'cured'18/20 (p=0.05)The study had a high risk of bias. Very small study, with only 40 patients. There was some baseline differences between the 2 groups. Females were over represented with 17/20 in group 1 and 14/20 in group 2. Also group 1 patients had an average of 6 episodes of tonsillitis per year whereas group 2 had an average of 5. The measurement of outcome is not assessed by the subsequent number of throat infections but by whether the patient felt cured. This may make the outcome measure invalid and unreliable. The patients and assessors were unable to be blinded.No side effects were reported.
non-surgical patients who felt 'cured'14/20 (p=0.05)
Marshall T
5 studies were found which met the inclusion criteria:participants had suffered from recurrent throat infections, the experimental intervention was tonsillectomy or adenotonsillectomy and the study was randomised with control groups. 4 of these were published, 1 was published only as an abstract but the data was made available to the authorsystematic review of RCT'sMean reduction in annual incidence of throat infections avoided as a result of tonsillectomy year 11.9,1.6,1.6,1.7, 1.5The author concludes there was little evidence found. The data was not analysed for statistical significance. The entry criteria into the studies included was varied or unclear. Together the studies included less than 350 patients. The patients were presenting to GP's not A&E this makes it less applicable to the A&E setting. Although the number of throat infections post op is relevant to my question, this needs to be balanced against the morbidity of the operation wich this review doesnt address.
year 21.1,1.1,0.7,1.6,0.8
annual incidence of throat infections tonsillectomy vs control1.24 vs 3.09, 1.45 vs 3.01, 0.36 vs 1.96, 0.56 vs 2.25, 0.64 vs 2.15 no p values


The evidence in the Marshall systematic review and the Stafford trial suggests that tonsillectomy is effective at reducing the number of episodes of throat infection. However these trials have a high risk of bias. The risks, side effects and costs of the operation must also be considered.

Clinical Bottom Line

More trials are needed but from the available evidence it seems tonsillectomy is more effective than antibiotic prescription at reducing total duration of symptoms.


  1. Burton MJ, Towler B, Glasziou Tonsillectomy versus non-surgical treatment for chronic/recurrent acute tonsillitis The cochrane database of systematic reviews 1999 Issue 2 April 1999
  2. Stafford N, von Haacke N, Sene A The treatment of recurrent tonsillitis in adults The journal of laryngology and otology Feb 1986 vol100 pg175-7
  3. Marshall T A review of tonsillectomy for recurrent throat infection British Journal of General practice June 1998, 48 1331-1335