Making sense out of the tonsillectomy literature

With the increase in new technology and changing indications for tonsillectomy, the literature has exploded during the last 20 years with scores of publications reporting the relative effectiveness of many different techniques. Despite this "wealth of information," no single technique has been adopted by most surgeons.

To systematically analyze the usefulness of this literature of the past 20 years concerning tonsillectomy technique in children. To propose the use of specific study parameters that could optimize clinical decision-making and future research.

Detailed review of the methodologies and findings in articles which compared one or more tonsillectomy techniques in clinical trials of children, ages 1-23 years from 1987 through 2007.

The Medline search revealed 255 papers of which 89 studies were suitable for inclusion in our review. In these 89 studies (found in 87 papers), 9 dissection techniques, 3 planes of dissection, 8 methods of hemostasis, and 41 different outcome measures were reported. Forty-four (49%) were described as randomized, 63 (71%) prospective, 25 retrospective (28%), 1 case report, 1 matched pair, and 9 case series papers. Sixteen (18%) trials were non-blind, 23 (26%) were single blind, 17 (19%) were double blind, and 7 (8%) were not stated. Seventy-five (84%) were comparative and 14 (16%) non-comparative. Eight (9%) studies reported power analyses. Twelve (13%) had no follow-up; 67 (75%) of the studies performed had short-term follow-up in the peri-operative period; 10 (11%) had follow-up for greater than 1 year. Eleven (12%) mentioned outcomes related to the effectiveness of the procedure itself in relieving symptoms for which the surgery was done.

Tonsillectomy technique research is of obvious interest to the otolaryngologist. We found deficits in: the precise reporting of surgical techniques, adequate study design and useful outcome measures, all of which make the literature less useful than it could be. Guidelines for study design parameters which could lead to more valuable information for the clinician are suggested.