Objectives: To systematically review evidence addressing tonsillectomy in children with obstructive sleep-disordered breathing (OSDB) or recurrent throat infections.
Data sources: Multiple databases from January 1980 through June 2016.
Review methods: We included comparative studies of tonsillectomy, perioperative medications to improve outcomes, and postoperative medications for pain. We also included case series and database studies with ≥1,000 children to address harms. Two investigators independently screened studies and rated risk of bias. We extracted and summarized data qualitatively and quantitatively via Bayesian meta-analyses. We also assessed strength of the evidence (SOE).
Results: We identified 218 unique studies (141 randomized controlled trials [RCTs], 12 nonrandomized trials, 7 prospective and 5 retrospective cohort studies, and 53 database or registry studies or case series [67 low, 110 moderate, and 41 high risk of bias]). Populations; surgical approaches; anesthetic, analgesic, and antiemetic regimens varied across studies. For children with OSDB, most studies reported better sleep-related outcomes in those who had a tonsillectomy versus no surgery. For children with recurrent throat infections, tonsillectomy improved the number of infections, associated utilization (clinician visits), and work/school absences in the first postsurgical year. These benefits did not persist over time, however, and longer term outcomes are limited. Partial tonsillectomy was associated with faster return to normal diet or activity versus total tonsillectomy but also with a risk of tonsillar regrowth requiring reoperation. Commonly used “hot” techniques were generally associated with faster return to normal diet and activity than was cold dissection. In meta-analyses, frequency of post-tonsillectomy hemorrhage (PTH) was less than 4 percent, and frequency of bleeding-associated revisits or reoperations was less than 8 percent. Meta-analysis of nine RCTs reporting bleeding associated with perioperative dexamethasone compared with placebo did not indicate a significantly increased risk of bleeding with steroids, although confidence bounds were wide. Studies of perioperative medications were heterogeneous, but dexamethasone was consistently associated with less need for rescue analgesia than placebo. Preemptive perioperative 5-hydroxytryptamine (5-HT) antiemetics were associated with less need for postoperative antiemetics than placebo. Few studies of postoperative medications addressed the same agents or outcomes.
Conclusions: Tonsillectomy can produce short-term improvement in sleep outcomes compared with no surgery in children with OSDB (moderate SOE). In children with recurrent throat infections undergoing tonsillectomy, number of throat infections (moderate SOE) and associated health care utilization and work/school absences (low SOE) improved in the first postsurgical year. These benefits did not persist, and data on longer term results are lacking. Short-term improvements must be weighed against the risk of PTH (high SOE for low frequency of PTH). Surgical technique had little bearing on return to normal diet or activity (low SOE). Perioperative dexamethasone and pre-emptive 5-HT receptor antagonist antiemetics reduced the need for additional analgesics or antiemetics (low SOE). Dexamethasone did not increase risk of PTH compared with placebo, but estimates had wide confidence bounds (low SOE). Little evidence addressed the use of postoperative medications for pain-related outcomes (insufficient SOE).