Predictors of overnight postoperative respiratory complications in obese children undergoing adenotonsillectomy for obstructive sleep apnea.

Abstract

Current clinical guidelines from the American Academy of Otolaryngology Head & Neck Surgery (AAO-HNS) recommend a preoperative polysomnogram (PSG) for obese patients prior to adenotonsillectomy (T&A). An overnight admission is recommended for children with severe (AHI >10) obstructive sleep apnea (OSA), citing a higher incidence of post-operative respiratory complications (PRCs) and need for respiratory support. Routine admission of obese children based on AHI >10 alone after T&A may place undue strain on hospital resources and increase healthcare costs, especially considering that many of these children have uncomplicated postoperative courses. In this study, we sought to identify variables from the pre-operative PSG and post-anesthesia care unit (PACU) that could more accurately predict overnight PRCs and indicate the need for a post-surgical admission after T&A. A single-center retrospective chart review was performed on a cohort of 155 obese children who underwent adenotonsillectomy for OSA. Inclusion criteria included patients 17 years of age and younger who had BMI 95th percentile or greater, underwent preoperative polysomnography, and were admitted overnight after T&A. Overnight respiratory complications were defined as an O2 desaturation under 92%, the need for overnight airway support, a respiratory support regression, respiratory depression, and bronchospasm/laryngospasm. Multivariable binary logistic regression analysis, point-biserial correlation, and Chi-square tests were performed to assess relationship of BMI z-score, polysomnography parameters, and PACU events with overnight respiratory complications. Lower O2 saturation nadirs on polysomnography were an independent predictor of respiratory complications overnight (OR=0.953, 95% CI=0.91-0.99, P=0.021), as was sleep time with O2 saturation less than 90% (OR=1.04, 95% CI=1.00-1.07, P=0.048). A prediction model with preoperative and postoperative variables significant on simple logistic regression yielded a ROC curve with AUC 0.89 (95% CI 0.82, 0.96). At a cutoff point of O2 saturation nadir less than 80%, overnight PRCs were predicted with 70.8% sensitivity and 75.2% specificity. At a cutoff point of greater than 0.5% of sleep time spent with O2 < 90% on PSG, overnight PRCs were predicted with 82.6% sensitivity and 62% specificity. Obstructive apneas (OAI) was not predictive of PRCs. BMI percentile was not significantly correlated with overnight respiratory complications, but BMI z-score was significantly correlated with overnight respiratory depression and an overnight airway event. O2 saturation nadir on PSG and time spent with oxygen saturation <90% (TST90) on PSG were found to be independent predictors of overnight postoperative respiratory complications after adenotonsillectomy in obese children. In addition to reaffirming existing guidelines for postoperative admission of patients with O2 saturation nadir on PSG <80%, these findings also suggest considering postoperative admission for obese patients who experience >0.5% sleep time with O2 sat <90% during PSG due to increased risk of overnight postoperative respiratory complications.