Tonsillectomy for adult obstructive sleep apnea: A systematic review and meta-analysis

Objectives/Hypothesis To determine if sleepiness and sleep study variables (e.g., Apnea‐Hypopnea Index [AHI] and lowest oxygen saturation) improve following isolated tonsillectomy for adult obstructive sleep apnea (OSA). Study Design Systematic review and meta‐analysis. Methods Nine databases (PubMed/MEDLINE included) were searched through November 24, 2015. Results Seventeen studies (n = 216 patients, 34.4 ± 10.0 years and body mass index: 29.0 ± 6.1 kg/m2) met criteria. Tonsils sizes were hypertrophied, large, enlarged, extremely enlarged, or grades 2 to 4. Apnea‐Hypopnea Index decreased by 65.2% (from 40.5 ± 28.9/hour to 14.1 ± 17.1/hour) (n = 203). The AHI mean difference (MD) was −30.2 per hour (95% confidence interval [CI] −39.3, −21.1) (P value < 0.00001). The AHI SMD was −1.37 (−1.65, −1.09) (large effect). Lowest oxygen saturation improved from 77.7 ± 11.9% to 85.5 ± 8.2% (n = 186). Lowest oxygen saturation MD was 8.5% (95% CI 5.2, 11.8) (P value < 0.00001). The Epworth Sleepiness Scale decreased from 11.6 ± 3.7 to 6.1 ± 3.9 (P value < 0.00001) (n = 125). Individual patient outcomes (n = 54) demonstrated an 85.2% success rate (AHI < 20/hour and ≥ 50% reduction) and a 57.4% cure rate. Individual patient data meta‐analysis showed preoperative AHI < 30 per hour to be a significant predictor of surgical success (P value < 0.001) and cure (P value = 0.043); among patients with preoperative AHI < 30 per hour, tonsillectomy success rate was 100% (25 of 25) and cure rate was 84% (21 of 25) with a mean postoperative AHI of 2.4 ± 2.1 per hour; this compares to tonsillectomy success rate of 72.4% (21 of 29), cure rate of 10 of 29 (34.4%), and mean postoperative AHI of 14.3 ± 13.9 per hour for patients with preoperative AHI ≥ 30 per hour. Conclusion Isolated tonsillectomy can be successful as treatment for adult OSA, especially among patients with large tonsils and mild to moderate OSA (AHI < 30/hour). Laryngoscope, 2016 Laryngoscope, 126:2176–2186, 2016