Post-tonsillectomy haemorrhage: a prospective, randomized, controlled clinical trial of cold dissection versus bipolar diathermy dissection

Abstract

To determine whether bipolar dissection tonsillectomy is associated with a higher post-operative haemorrhage rate than cold dissection tonsillectomy. Prospective, randomized, controlled trial. Otolaryngology department of a teaching hospital. Two hundred and forty-five patients undergoing elective tonsillectomy between July 2002 and November 2004. Patients were randomly assigned to either bipolar dissection or cold dissection (with bipolar haemostasis). Post-operative haemorrhage rates, management (conservative or surgical) and blood transfusion requirements were recorded. The grade of surgeon and history of quinsy were also recorded. One hundred and forty-one patients (58 per cent) were randomized to the bipolar dissection and 104 (42 per cent) to the cold dissection groups. Seventeen patients (12.1 per cent) in the bipolar dissection group and eight patients (7.7 per cent) in the cold dissection group suffered haemorrhage (p = 1.0; degrees of freedom (Df) 0.0; 95 per cent confidence intervals (CI) -0.1 to 0.0). The haemorrhage rates for procedures conducted by senior house officers, specialist registrars and consultants were 11.4 per cent, 10.3 per cent and 5.0 per cent, respectively. Two patients required surgical intervention, both from the bipolar dissection group. No patients required blood transfusion. A history of quinsy was not associated with an increased haemorrhage rate. The difference in haemorrhage rates between groups and surgeon grades did not reach statistical significance. Nonetheless the trend towards a greater incidence of haemorrhage in the bipolar group and in procedures conducted by more junior surgeons during the trial raised concerns. The results of the National Prospective Tonsillectomy Audit and our interim results have led us to abandon the trial and disallow the use of bipolar dissection in tonsillectomies performed by junior staff members.