Meta-analysis of the timing of haemorrhage after tonsillectomy: an important factor in determining the safety of performing tonsillectomy as a day case procedure.

OBJECTIVES: To perform a meta-analysis of studies of the timing of primary tonsillectomy haemorrhage. In particular to compare the difference in risk between 0-8 and 8-24 h; that is whether overnight inpatient tonsillectomy is required. DESIGN: Medline search of all tonsillectomy studies to perform a meta-analysis of the timing of primary haemorrhages. SETTING: Literature-based study. PARTICIPANTS: All adult and paediatric tonsillectomy studies giving the absolute number and timing of all primary haemorrhages. MAIN OUTCOME MEASURES: The overall incidence of haemorrhage occurring between 0-8 and 8-24 h. The overall incidence of haemorrhage for each of the first 24 h after operation. Compare risk of a bleed occurring 0-8, 8-24 and >24 h where data were available. RESULTS: From a 1.4% overall risk of a primary haemorrhage only one in 14 occur after 8 h, i.e. 0.1% (95% CI=0.08-0.16%). A total of 833 patients would require to be kept overnight in order to identify one case of bleeding after 8 h. CONCLUSIONS: Little benefit was conferred from overnight admission from the point of view of monitoring for primary haemorrhage. A case can be made for either day-case tonsillectomy (hospital stay over the period in which 93% of primary haemorrhages would occur) or the 'belt-and-braces' approach of a 1-week stay (during which all haemorrhages would occur) but current 24-h admission appears illogical.