• Tonsils;
  • bleeding;
  • steroids;
  • dexamethasone;
  • Level of Evidence: 1a



Tonsillectomy and adenotonsillectomy are common pediatric surgical procedures. Although perioperative administration of dexamethasone is common, recent data indicate a possible association between dexamethasone and increased risk of postoperative hemorrhage.

Study Design:

Meta-analysis of prospective studies.


Meta-analysis using MEDLINE and EMBASE, including all placebo-controlled prospective studies with perioperative dexamethasone as the intervention and postoperative bleeding as the outcome, in children age <18 years undergoing tonsillectomy or adenotonsillectomy. The association between any dexamethasone dose and bleeding outcomes was quantified. Dexamethasone dose analyses were performed using meta-regression and stratified dose analysis models. Individuals were pooled using the random effects model, and Mantel Haenszel odds ratios (OR) were used to assess the associations. All P values were two-sided.


Twelve studies with 1,180 total participants were included in the meta-analysis. There was no significant association between dexamethasone at any dose and odds of bleeding compared to placebo (OR = 1.07; 95% confidence interval [CI], 0.58-1.98; P = .82). There was no significant association between increasing dexamethasone dose and bleeding (regression slope = −1.47, P = .10). Stratifying by dose ranges, studies comparing dexamethasone doses in 0.4 to 0.6 mg/kg range to placebo showed significantly increased odds of bleeding (OR = 3.14; 95% CI, 1.14-8.65; P = .03). Higher or lower doses were not associated with a difference in bleeding.


There was no overall association between dexamethasone administration and postoperative bleeding in children undergoing tonsillectomy or adenotonsillectomy. However, this study cannot exclude the possibility of an association between specific dexamethasone doses and increased odds of bleeding. The results underscore the need for more dedicated prospective studies of this very common intervention.