Lingual Tonsillectomy for Treatment of Pediatric Obstructive Sleep Apnea

Key Points

Question

What effect does lingual tonsillectomy have on polysomnography in children with lingual tonsil hypertrophy and obstructive sleep apnea?

Findings

In this meta-analysis, significant improvements in the apnea-hypopnea index and the minimum oxygen saturation were observed after lingual tonsillectomy. However, children frequently have residual obstructive sleep apnea after lingual tonsillectomy, and postoperative complications must be carefully managed.

Meaning

Lingual tonsillectomy is an effective surgical management for children with obstructive sleep apnea caused by lingual tonsil hypertrophy.

Abstract

Importance

Evidence indicates correlations between lingual tonsil hypertrophy and pediatric obstructive sleep apnea (OSA). However, to our knowledge, a meta-analysis of surgical outcomes for lingual tonsillectomy in children with OSA has not been conducted.

Objective

To evaluate the therapeutic outcomes of lingual tonsillectomy for treatment of pediatric OSA.

Data Sources

The study protocol was registered on PROSPERO (CRD42015027053). PubMed, MEDLINE, EMBASE, and the Cochrane Reviews databases were searched independently by 2 authors for relevant articles published by September 2016.

Study Selection

The literature search identified English-language studies that used polysomnography to evaluate children with lingual tonsil hypertrophy and OSA after lingual tonsillectomy alone. The search keywords were lingual tonsil, lingual tonsillectomy, sleep endoscopy, sleep apnea, and child.

Data Extraction and Synthesis

Polysomnographic data from each study were extracted. A random-effects model pooled postoperative sleep variable changes and success rates for lingual tonsillectomy in treating pediatric OSA.

Main Outcomes and Measures

Four outcomes for lingual tonsillectomy were analyzed. These included net postoperative changes in the apnea-hypopnea index (AHI), net postoperative changes in the minimum oxygen saturation, the overall success rate for a postoperative AHI less than 1, and the overall success rate for a postoperative AHI less than 5.

Results

This meta-analysis consisted of 4 studies (mean sample size, 18.25 patients), with a total of 73 unique patients (mean [SD] age, 8.3 [1.1] years). Fifty-nine percent (27 of 46) of the patients were male, and 1 of the 4 studies did not specify number of males. Lingual tonsillectomy was indicated for persistent OSA after adenotonsillectomy in all cases. Lingual tonsil hypertrophy was evaluated using computed tomography or magnetic resonance imaging in 1 study, sleep endoscopy in 2 studies, and cine magnetic resonance imaging in 1 study. The mean change in the AHI after lingual tonsillectomy was a reduction of 8.9 (95% CI, −12.6 to −5.2) events per hour. The mean change in the minimum oxygen saturation after lingual tonsillectomy was an increase of 6.0% (95% CI, 2.7%-9.2%). The overall success rate was 17% (95% CI, 7%-35%) for a postoperative AHI less than 1 and 51% (95% CI, 25%-76%) for a postoperative AHI less than 5. Postoperative complications that developed included airway obstruction, bleeding, and pneumonia.

Conclusions and Relevance

Lingual tonsillectomy is an effective surgical management for children with OSA caused by lingual tonsil hypertrophy, and it achieves significant improvement in the AHI and the minimum oxygen saturation. However, children frequently have residual OSA after lingual tonsillectomy, and postoperative complications must be carefully managed.

This meta-analysis evaluates the therapeutic outcomes of lingual tonsillectomy for treatment of pediatric obstructive sleep apnea.

Introduction

Obstructive sleep apnea (OSA) in children covers a spectrum of respiratory disorders characterized by upper airway collapse during sleep. The pathogenesis of childhood OSA is mainly due to enlarged adenotonsillar tissues. Adenotonsillectomy is widely considered the first-line therapy for childhood sleep apnea. Treatment outcomes for adenotonsillectomy have been studied extensively. In 2006, a meta-analysis by Brietzke and Gallagher demonstrated that the mean change in the apnea-hypopnea index (AHI) after adenotonsillectomy was a reduction of 13.92 events per hour, with a success rate of 82.9%. A review article by Friedman et al in 2009 found that the mean change in the AHI after adenotonsillectomy was a reduction of 12.42 events per hour, with an overall success rate lower than commonly believed (59.8% for an AHI<1). A meta-analysis in 2016 by our research group demonstrated that adenotonsillectomy results in notable improvement in a number of sleep variables, with an overall success rate of 51% for a postoperative AHI less than 1. Other meta-analyses have shown that postoperative residual OSA remained in approximately half of the children treated with adenotonsillectomy. Therefore, additional treatment strategies are desirable for children with persistent OSA after adenotonsillectomy.

Lingual tonsillar hypertrophy is a known cause of OSA and pediatric airway obstruction. The lingual tonsil is a component of the Waldeyer ring of lymphoid tissue located at the base of the tongue, and its hypertrophy may cause OSA. The diagnostic approach to detecting lingual tonsil hypertrophy varies. Conventionally, lingual tonsil size is evaluated using imaging studies, such as computed tomography (CT) or magnetic resonance imaging (MRI). More recently, the use of drug-induced sleep endoscopy enables physicians to identify sites of obstruction immediately and to direct surgical interventions accordingly. After identifying lingual tonsil hypertrophy as the main cause of OSA, lingual tonsillectomy is indicated, and studies have proved its effectiveness. However, to our knowledge, a meta-analysis of surgical outcomes for lingual tonsillectomy in children with OSA has not been conducted.

The objective of this study was to evaluate the changes in sleep variables (ie, the AHI and the minimum oxygen saturation) after lingual tonsillectomy for treatment of OSA in children. In addition, we aimed to assess the overall success rate of the procedure (ie, for a postoperative AHI<1 and for a postoperative AHI<5).

Methods

Search Strategy

This meta-analysis was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement and the recommendations of the Meta-Analysis of Observational Studies in Epidemiology group. The study protocol was registered on PROSPERO (CRD42015027053) before commencement. Two of us (K.-T.K. and C.-H.L.) independently searched databases, including PubMed, MEDLINE, EMBASE, and the Cochrane Reviews, for articles published by September 2016. Reference lists for identified studies were searched to yield additional articles. The search keywords were lingual tonsil, lingual tonsillectomy, sleep endoscopy, sleep apnea, and child. The literature search identified English-language studies. eTable 1 in the Supplement summarizes the literature search process and the keywords used.

Inclusion criteria were children younger than 18 years, preoperative polysomnography (PSG) studies confirming the diagnosis of OSA, and postoperative PSG studies for outcome comparisons. The diagnosis of pediatric OSA was defined as an AHI greater than 1 event per hour in the PSG studies. Therefore, articles included in this meta-analysis had to have information on the AHI, and those that failed to report the AHI were excluded. The surgical procedure included in this meta-analysis was the removal of hypertrophic lymphoid tissue at the base of the tongue (ie, lingual tonsillectomy alone). Articles in which patients had concurrent procedures performed at the time of lingual tonsillectomy were excluded from the meta-analysis. The details of each technique, instrument, and access performed varied based on surgeon preference and experience.

Exclusion criteria were based primarily on absence of one of the inclusion criteria. Case reports, abstracts, letters to the editor, and unpublished studies were excluded from the meta-analysis. The initial search was conducted by the 2 key reviewers (K.-T.K. and C.-H.L.) independently and was verified by 2 of us (P.J.K. and W.-C.H.).

The methodological assessment of article quality in this meta-analysis entailed using the relevant section of the Newcastle-Ottawa Scale (NOS). Quality scores ranged from 0 (lowest) to 9 (highest). The NOS tool set was adapted to each article separately by 2 of us (M.-T.L. and W.-C.H.), and disagreements were resolved by consensus among the authors.

Statistical Analysis

Data were analyzed using statistical software (Comprehensive Meta-Analysis, version 2; Biostat Inc). A random-effects model was used for calculating the overall effect of lingual tonsillectomy. The overall effect of net postoperative changes (raw score) in the PSG variables (ie, the AHI and the minimum oxygen saturation) was extracted for calculation in the meta-analysis. The overall success rate (ie, for a postoperative AHI<1 and for a postoperative AHI<5) for lingual tonsillectomy as a treatment for OSA was analyzed. Effect size and 95% CIs were used to describe the magnitude and precision of the difference in compared groups.

Statistical heterogeneity among studies was assessed using I² statistics that measured the proportion of overall variation attributable to between-study heterogeneity. An I² statistic exceeding 50% indicated moderate heterogeneity, whereas an I² statistic exceeding 75% indicated high heterogeneity. Potential publication bias was assessed using a funnel plot and the Egger intercept test.

Results

Literature Search

The initial web-based literature search yielded 265 results. Studies that were unpublished, were noninterventional, did not enroll patients or children with OSA, or did not perform lingual tonsillectomy procedures were excluded. In total, 16 potentially pertinent studies were retrieved. After a careful review of the full-text articles, 8 studies were excluded because of a lack of preoperative or postoperative PSG data, leaving 8 studies included in the meta-analysis. Of these articles, 3 studies were from the same institution and conducted by the same research team (at Lucile Packard Children’s Hospital, Stanford University). As agreed on by the senior author of the 3 studies (P.J.K.), 2 studies were considered duplicative for the purposes of our meta-analysis and were excluded. Two other studies were excluded because they documented concurrent procedures at the time of lingual tonsillectomy. Therefore, 4 studies were included in the quantitative analyses (Figure 1).

Figure 1. Flow Diagram of the Literature Search.

PSG indicates polysomnography.

Quality Assessment

The quality of the 4 studies included in the meta-analysis was assessed using the NOS, ranging from 0 (lowest) to 9 (highest) points. The Table and eTable 2 in the Supplement list the NOS results in detail. The NOS scores for the 4 studies ranged from 5 to 7; the mean and median scores were both 6.

Table. Characteristics of Included Studies^a.

Source	Country	Sample Size	Quality Assessmentb	Mean Patient Age, y	Male, No./Total No. (%)	Body Weight	Patients With Comorbidities, No./Total No. (%)	Diagnostic Approach to Lingual Tonsils	Operative Method	Postoperative Complications
Abdel-Aziz et al, 2011	Egypt	16	5	8.3	11/16 (69)	2/16 (12%) Obese	5/16 (19)	CT or MRI	Unipolar diathermy probe	3 Patients with airway obstruction caused by tongue base edema
Truong et al, 2012	United States	27	7	7	NR	NR	7/27 (27)	Sleep endoscopy	Endoscopic lingual tonsillectomy	NR
Thottam et al, 2015	United States	9	6	10.5	5/9 (56)	25.2 Mean BMI percentile	4/9 (44)	Sleep endoscopy	Transoral robotic surgery	1 Patient with pneumonia, 1 patient with bleeding
Prosser et al, 2017	United States	21	6	9.3	11/21 (52)	82.8 Mean BMI percentile	Down syndrome in 21/21 (100)	Cine MRI	Radiofrequency ablation	NR

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); CT, computed tomography; MRI, magnetic resonance imaging; NR, not reported.

^aThe study designs were all retrospective case series, with a level of evidence of 4 in all 4 studies.

^bNewcastle-Ottawa Scale score.

Basic Demographics

The Table lists the basic demographics for the 4 included studies, with a total of 73 unique patients. Sample size, age, sex, body weight, comorbidities, surgical indications, diagnostic approach to lingual tonsils, and operative methods were extracted from the articles. One study was conducted in Egypt, and the other 3 studies were conducted in the United States. The sample size of these studies ranged from 9 to 27 patients (mean, 18.25 patients). The study designs were all retrospective case series, with a level of evidence of 4 in all of them.

The mean (SD) age of all 73 included patients was 8.3 (1.1) years. Boys constituted 59% (27 of 46) of all patients, and 1 of the 4 studies did not specify number of males. Lingual tonsillectomy was performed in healthy patients and in those with medical comorbidities, such as Down syndrome, mucopolysaccharidosis, velocardiofacial syndrome, Beckwith-Wiedemann syndrome, and other craniofacial anomalies. The prevalence of comorbidities varied in each study, ranging from 19% to 100% (Table). Abdel-Aziz et al evaluated 1 patient with Down syndrome and 2 patients with mucopolysaccharidosis in their study. Truong et al reported the prevalence of comorbidities in their study and defined comorbidities as Down syndrome, craniofacial abnormalities, achondroplasia, Beckwith-Wiedemann syndrome, Duchenne muscular dystrophy, Mobius syndrome, Crouzon syndrome, seizure disorder, and Arnold-Chiari malformation. However, they did not report case numbers for each comorbidity. Thottam et al evaluated 3 patients with Down syndrome and 1 patient with Noonan syndrome. All patients in the study by Prosser et al had Down syndrome.

All patients had undergone lingual tonsillectomy because of persistent OSA after previous adenotonsillectomy. The diagnostic approach to evaluating lingual tonsils included the use of CT or MRI in 1 study, sleep endoscopy in 2 studies, and cine MRI in 1 study. Surgical instrument use varied, including a unipolar diathermy probe and a coblation wand to remove lingual tonsils. Surgical techniques involved an endoscopic approach or transoral robotic surgery in all 4 studies. Most patients are admitted overnight in a monitored hospital bed for observation after surgery, with administration of postoperative antibiotics and pain medications. All patients in the study by Thottam et al were monitored in the intensive care unit on postoperative day zero.

Prosser et al reported in their study that the duration between lingual tonsillectomy and postoperative PSG was 4 months. The other 3 studies did not state the timing of PSG after surgical procedures.

Surgical Outcomes of Lingual Tonsillectomy

Four outcomes for lingual tonsillectomy were analyzed: (1) net postoperative changes in the AHI (Figure 2), (2) net postoperative changes in the minimum oxygen saturation (Figure 3), (3) the overall success rate for a postoperative AHI less than 1 (eFigure 1 in the Supplement), and (4) the overall success rate for a postoperative AHI less than 5 (eFigure 2 in the Supplement).

Figure 2. Forest Plot for Change in the Apnea-Hypopnea Index (AHI) After Lingual Tonsillectomy.

The 4 studies had a total of 73 patients.

Figure 3. Forest Plot for Change in the Minimum Oxygen Saturation After Lingual Tonsillectomy.

The 3 studies had a total of 46 patients.

Changes in the AHI and the Minimum Oxygen Saturation After Lingual Tonsillectomy

All 4 studies reported data for changes in the AHI after lingual tonsillectomy. The mean change in the AHI after surgery was a reduction of 8.9 (95% CI, −12.6 to −5.2) events per hour (Figure 2).

Changes in the minimum oxygen saturation after lingual tonsillectomy were reported in 3 studies. The combined effect obtained from the random-effects model for the minimum oxygen saturation showed an increase of 6.0% (95% CI, 2.7%-9.2%) (Figure 3).

Success Rate of Lingual Tonsillectomy

All 4 studies reported their success rate after lingual tonsillectomy. Surgical success was usually defined as a postoperative AHI less than 1 or a postoperative AHI less than 5. Two studies reported data for a postoperative AHI less than 1, and the random-effects model estimate for the success rate was 17% (95% CI, 7%-35%) (eFigure 1 in the Supplement).

The success rate for a postoperative AHI less than 5 was reported by 2 studies. The random-effects model estimate for the success rate was 51% (95% CI, 25%-76%) when treatment success was defined as an AHI less than 5 (eFigure 2 in the Supplement).

Postoperative Complications

Complications after lingual tonsillectomy were reported, including airway obstruction caused by tongue base edema, intraoperative or postoperative bleeding, and pneumonia. Abdel-Aziz et al described 3 patients who developed postoperative airway obstruction caused by tongue base edema. All 3 children were successfully treated with oxygen therapy and administration of corticosteroids, and reintubation was not required. Thottam et al described 1 patient who had postoperative bleeding that required reoperation for control. Thottam et al also described 1 patient with pneumonia who required intubation, ventilator support, and a 14-day hospital stay.

Publication Bias

eFigure 3 in the Supplement shows a funnel plot of the SE according to the difference in the AHI means. The plot is generally symmetrical, suggesting no obvious publication bias. The results of the Egger intercept test also indicated no apparent publication bias.

Discussion

To our knowledge, the present study is the first meta-analysis to determine the effectiveness of lingual tonsillectomy in treating pediatric OSA. Meta-analysis results show that lingual tonsillectomy is an effective surgical management for children with OSA, resulting in an AHI reduction of 8.9 events per hour and a minimum oxygen saturation increase of 6.0%. The overall success rates identified in this study are 17% for a postoperative AHI less than 1 and 51% for a postoperative AHI less than 5. From a clinical perspective, this study confirms beneficial effects of lingual tonsillectomy in treating pediatric OSA, and physicians may consider it a treatment strategy in patients with OSA caused by lingual tonsil hypertrophy.

The lingual tonsils are a component of lymphoid tissue in the Waldeyer ring. Several causes may contribute to lingual tonsil hypertrophy, including reactive lymphoid hyperplasia due to previous adenotonsillectomy, obesity, and laryngopharyngeal reflux. Hypertrophy of the lingual tonsils has several clinical implications, including problems with dysphagia, difficult intubation, and upper airway obstruction. In particular, lingual tonsil hypertrophy is considered an important factor for the development of OSA, and lingual tonsillectomy is indicated for children with OSA caused by lingual tonsil hypertrophy.

Several surgical techniques for pediatric OSA have been proposed, including adenotonsillectomy, supraglottoplasty, lingual tonsillectomy, and tracheostomy. Persistent OSA has been observed in approximately half of all children after adenotonsillectomy. Laryngomalacia and lingual tonsillar hypertrophy are 2 major causes of residual OSA. Recently, emerging evidence indicates that supraglottoplasty is an effective surgery for children with OSA and laryngomalacia. Lee et al showed that supraglottoplasty resulted in an AHI reduction of 8.9 events per hour, and their success rate was 28% for a postoperative AHI less than 1. However, treatment outcomes of lingual tonsillectomy for children with OSA and lingual tonsil hypertrophy have never been clarified, to our knowledge. This meta-analysis identifies that lingual tonsillectomy results in an AHI reduction of 8.9 events per hour and that its overall success rate is 17% for children with persistent OSA caused by lingual tonsil hypertrophy.

A relevant question is the nature of the risk factors determining persistent OSA after lingual tonsillectomy. Obesity has been shown to be associated with poor surgical outcomes. A previous meta-analysis by our research group demonstrated that the postoperative AHI is positively correlated with the body mass index z score before surgery for children with OSA undergoing adenotonsillectomy. For children with OSA who underwent lingual tonsillectomy, Chan et al found that cure rates were significantly poorer for overweight children undergoing lingual tonsillectomy than for other children. Obesity is a risk factor for lingual tonsil hypertrophy. Adipose tissue in obese children around the pharynx and neck, along with hypertrophic adenoids and tonsils, compresses the pharynx and reduces its cross-sectional area. Obese children may also have a high preoperative AHI, which is less likely to be cured by surgery alone independent of obesity. However, data comparing surgical outcomes between obese and nonobese children after lingual tonsillectomy are limited, and the factors affecting treatment outcomes in children undergoing lingual tonsillectomy require further study.

In this meta-analysis, comorbidities in children with OSA who underwent lingual tonsillectomy included Down syndrome, mucopolysaccharidosis, velocardiofacial syndrome, Beckwith-Wiedemann syndrome, and other craniofacial anomalies. In particular, more than one-third (26 of 73) of the patients included in our meta-analysis had Down syndrome. Given the fact that lingual tonsil hypertrophy is a common condition in children with Down syndrome, it is not surprising that so many patients in this meta-analysis had this disorder. Disparities in surgical outcomes between children with and without comorbidities are of particular interest to clinicians. In children with Down syndrome, several contributing factors have been implicated for persistent OSA after surgical procedures, including muscular hypotonia and anatomic features like macroglossia, relative glossoptosis, midface hypoplasia, and hypopharyngeal collapse. The possibility is raised that children with comorbidities may have poor surgical outcomes after lingual tonsillectomy. However, the identified data are limited, and additional analysis in patients with and without comorbidities requires future study.

Diagnostic approaches for the evaluation of lingual tonsil size have shifted from imaging studies (ie, CT or MRI) to sleep endoscopy. Sleep endoscopy is a cost-effective method enabling physicians to study the dynamic airway in a sleeplike stage. Compared with awake endoscopy, sleep endoscopy encounters more sites of obstruction (eg, lateral pharyngeal wall or tongue base collapse), and it reliably identifies sites of obstruction in surgically naive children and in those with persistent OSA after adenotonsillectomy. There is growing consensus supporting the use of sleep endoscopy to identify sites of obstruction in children with OSA. In 2 studies included in this meta-analysis, sleep endoscopy to detect lingual tonsil hypertrophy was used.

Disparities in surgical techniques for lingual tonsillectomy have been reported. Historically, lingual tonsillectomy has been a challenge because of poor access, airway edema, postoperative pain, and bleeding during tissue removal. Studies included in this meta-analysis used various instruments and techniques for lingual tonsillectomy. Surgeons can use a unipolar diathermy probe or a coblation wand to remove hypertrophic lingual tonsils. Surgical techniques may involve suspension laryngoscopy for access, endoscopy, or transoral robotic surgery. There is no standard, best-practice lingual tonsillectomy technique that has been proved to provide optimal surgical outcomes with minimal postoperative complications. Consequently, the choice of surgical technique is based on the condition of the patient and the expertise of the surgeon. Lingual tonsillectomy technique is a critical issue. In 2009, endoscopic-assisted coblation lingual tonsillectomy was described as an effective method to treat lingual hypertrophy associated with OSA, in addition to providing improved visualization. The need for further enhanced visualization drove the development of the transoral robotic approach as a means to assist access and decrease intraoperative uncertainty. Transoral robotic surgery provides a 3-dimensional view and more freedom of motion than the previous endoscopic coblation method. Thottam et al demonstrated that children who underwent transoral robotic surgery for sleep apnea had short hospital stays and a low rate of complications. However, learning assembly of the robotic apparatus may require time, and the cost of the equipment may be prohibitive.

Data on the complications of lingual tonsillectomy are limited. Although the complication rate appears to be low, postoperative complications may be serious and must be carefully managed. Intraoperative or postoperative bleeding requires close observation and possible reoperation for control. Respiratory complications may require oxygen therapy, corticosteroid use, or intubation. Among the 73 patients included in this meta-analysis, 1 (1%) had postoperative bleeding, and 4 (5%) had respiratory complications. However, the data should be interpreted with caution because of possible selection bias. Compared with those children who have preoperative PSG data only, children who have both preoperative and postoperative PSG data may have a better clinical course and a lower complication rate. Leonardis et al reported 16 patients who underwent lingual tonsillectomy, and 2 of them (12.5%) had bleeding from the operative site. The learning curve associated with lingual tonsillectomy should also be taken into consideration. Postoperative complications may be high initially and diminish substantially thereafter. Furthermore, because of limited data, factors contributing to postoperative complications were not clearly identified in our meta-analysis.

Limitations and Future Directions

There were many limitations surrounding the current literature identified in this meta-analysis, which may influence future study. First, most studies were retrospective case series. The limitations associated with observational studies, including confounding factors and selection bias, have been documented. Second, evidence is lacking on changes in quality of life, blood pressure, or biomarkers in children after lingual tonsillectomy. Third, factors related to surgical outcomes and persistent OSA after lingual tonsillectomy are not well identified and require further study. Fourth, children with OSA often have multilevel obstructions. They may also experience occult laryngomalacia and be seen without stridor but have supraglottic collapse during sleep. Therefore, some children included in this meta-analysis required extension of lingual tonsillectomy to a midline glossectomy and performance of supraglottoplasty. Our research group plans future studies on the effectiveness of multilevel surgery and consensus regarding treatment strategies for children with OSA. Fifth, there is a lack of long-term follow-up data on children undergoing adenotonsillectomy for OSA, as well as little evidence regarding lingual tonsil regrowth after lingual tonsillectomy. Prospective, longitudinal studies are required to elucidate long-term outcomes of lingual tonsillectomy and lingual tonsil regrowth among children with OSA.

Conclusions

Lingual tonsillectomy has been shown to be an effective surgery for children with persistent OSA after adenotonsillectomy caused by lingual tonsil hypertrophy. This procedure resulted in an AHI reduction of 8.9 events per hour and a minimum oxygen saturation increase of 6.0%. After lingual tonsillectomy, 17% of all patients in this meta-analysis had an AHI less than 1, and 51% had an AHI less than 5. However, complete resolution of OSA is not achieved in most cases of lingual tonsillectomy, and postoperative complications, including respiratory and bleeding problems, must be managed carefully.

Supplement.

eTable 1. Literature Searches and Keywords

eTable 2. Newcastle-Ottawa Scale Quality Assessment of Included Studies

eFigure 1. Forest Plot of Success Rate (Postoperative AHI<1) After Surgery

eFigure 2. Forest Plot of Success Rate (Postoperative AHI<5) After Surgery

eFigure 3. Funnel Plot of Standard Error by Difference in Means of the AHI