OSA indicates obstructive sleep apnea.
IV indicates inverse variance.
Farhood Z, Ong AA, Nguyen SA, Gillespie MB, Discolo CM, White DR. Objective Outcomes of Supraglottoplasty for Children With Laryngomalacia and Obstructive Sleep ApneaA Meta-analysis. JAMA Otolaryngol Head Neck Surg. 2016;142(7):665-671. doi:10.1001/jamaoto.2016.0830
Surgical intervention is the main treatment alternative for patients with severe laryngomalacia. Supraglottoplasty offers effective treatment results not only for laryngomalacia but also for concurrent obstructive sleep apnea (OSA).
To quantify the objective outcomes of supraglottoplasty for laryngomalacia with OSA via polysomnography data in the pediatric population.
A comprehensive literature search of the PubMed database was performed on May 20, 2015, using the search terms supraglottoplasty, epiglottoplasty, aryepiglottoplasty, laryngomalacia, obstructive sleep apnea, Apnea-Hypopnea Index (AHI), children, and polysomnography. There were no date restrictions.
The literature search identified English-language studies that used polysomnography to evaluate patients with laryngomalacia and OSA after supraglottoplasty. Two reviewers screened titles and abstracts of the studies. The full texts of the studies were examined to assess their relevance to the meta-analysis.
Numerical polysomnography data were extracted and compared among studies where appropriate. A fixed- or random-effects model was used, when appropriate, to analyze the data and calculate effect sizes.
Four studies were included in various subsets of the meta-analysis. After supraglottoplasty, the Apnea-Hypopnea Index (AHI) improved by a mean of 12.5 points in 4 studies (95% CI, −21.14 to −3.78; P = .005), oxygen saturation as measured by pulse oximetry nadir by 9.49 in 4 studies (95% CI, 4.87–14.12; P < .001), and Obstructive AHI by 21 points in 2 studies (95% CI, −50.3 to −8.29; P = .16). Twenty-nine of 33 children (88%) had residual disease. Patients 7 months and older had significant improvement in the AHI (P = .03).
Conclusions and Relevance
Supraglottoplasty is an effective treatment modality for patients with laryngomalacia and OSA with objectively measurable benefits; however, patients will frequently have residual disease. Additional polysomnography after treatment is advised to ensure adequate resolution of the disorder.
Laryngomalacia is the most common cause of stridor in infants, accounting for 45% to 75% of all cases.1 Theories regarding its pathophysiologic mechanisms describe collapse of redundant supraglottic structures during inspiration that may be secondary to immature laryngeal cartilages or abnormal sensorimotor intergration.2,3 Others have suggested that mucosal inflammation caused by gastroesophageal reflux disease (GERD) is associated with laryngomalacia and plays a role in the severity of symptoms.4- 6 Fortunately, most symptoms resolve spontaneously within the first 2 years of life. Despite the benign-appearing natural history, patients with severe laryngomalacia may experience significant effects, including failure to thrive, feeding difficulties and aspiration, and obstructive sleep apnea (OSA). Thus, the spectrum of laryngomalacia ranges from mild to severe and is based on the association with feeding and obstructive symptoms.3 Three percent of infants aged 6 months to 3 years have OSA, and approximately 20% of infants with laryngomalacia will have failure to thrive, hypoxemia, OSA, or other complications.2,7
Surgical intervention is the main treatment alternative for patients with severe laryngomalacia. The mainstay of surgical treatment is supraglottoplasty, which includes the anatomically subdivided procedures epiglottoplasty, aryepiglottoplasty, and arytenoidoplasty. Supraglottoplasty can be performed using a variety of methods, including cold steel, microdebrider, and carbon dioxide laser, with equivalent success rates.8,9 A previous systematic review10 has addressed supraglottoplasty failure for laryngomalacia; however, to date, no comprehensive review or meta-analysis has assessed polysomnography outcomes for patients with laryngomalacia with OSA. The purpose of the present study was to conduct a systematic review of the subject, pool preoperative and postoperative polysomnography data, and quantify the effect of supraglottoplasty on patients with laryngomalacia with OSA. The null hypothesis was that there was no difference between preoperative and postoperative polysomnography values.
Question What effect does supraglottoplasty have on polysomnography in children with laryngomalacia and obstructive sleep apnea?
Findings In this systematic review and meta-analysis, significant improvements in the apnea-hypopnea index were observed, but patients frequently continued to have residual disease. Young age and severe disease may be factors that positively influence the degree of polysomnography improvement after supraglottoplasty.
Meaning Supraglottoplasty is a beneficial treatment for laryngomalacia with obstructive sleep apnea.
At the Medical University of South Carolina, institutional review board approval is not required for studies that examine previously institutional review board–approved and published data.
A comprehensive literature search of the PubMed database was performed by two of us (Z.F., A.A.O.) on May 20, 2015. The specific search terms included were supraglottoplasty, epiglottoplasty, aryepiglottoplasty, laryngomalacia, obstructive sleep apnea, Apnea-Hypopnea Index (AHI), children, and polysomnography. There were no date restrictions.
Inclusion criteria for studies were as follows: humans younger than 18 years, diagnosis of congenital laryngomalacia and OSA (AHI >1 on overnight polysomnography), supraglottoplasty performed as treatment, preoperative and postoperative polysomnography data, and published in the English language. The references of the included articles were also reviewed to identify additional articles. Articles were included regardless of surgical technique. Articles were excluded if the patients did not have both laryngomalacia and OSA. Articles that focused on state-dependent or late-onset laryngomalacia were also excluded. There were no limitations on the number of patients in any study. If discrepancies arose between the reviewers, they were discussed with the senior author (D.R.W.). Qualitative synthesis consisted of summarizing the results and characteristics of the relevant articles, including assessing level of evidence. Quantitative synthesis sought to pool numeric and equivalent data to determine effect sizes.
Results were extracted, summarized, collated, and analyzed, paying special attention to mean (SD) pretreatment and posttreatment data. If data of individual patients were listed, then these data were compiled and mean (SD) calculated. Attempts were made to contact authors of individual studies when more information was required.
The possibility of publication bias in each study was evaluated by using the Oxford Center for Evidence-Based Medicine criteria.11 According to this assessment tool, the quality of studies is determined on the basis of criteria, including systematic review, study randomization, and presence of a control group. The grading is divided into 5 levels as follows: level 1, systematic review of randomized clinical trials or n-of-1 trials; level 2, randomized clinical trial or observational study with substantial effect; level 3, nonrandomized controlled cohort or follow-up study; level 4, case series, case-control studies, or historically controlled studies; and level 5, mechanism-based reasoning.
The primary end points were changes in polysomnography parameters in children with congenital laryngomalacia and OSA after supraglottoplasty. The parameters included the AHI or Respiratory Disturbance Index (RDI); Obstructive Apnea-Hypopnea Index (OAHI), which excludes central apneas; peak carbon dioxide levels (Pco2, peak end-tidal carbon dioxide [PETco2], and peak transcutaneous carbon dioxide [Ptcco2] as if they were equivalent); and total sleep time (TST). The primary parameter of interest was the AHI; however, if studies had other polysomnography parameters that could be combined for analysis, they were included in this study.
Meta-analysis of selected studies with a continuous measure (comparison of means [SDs] between pretreatment and posttreatment groups) was performed with Cochrane Review Manager (RevMan), version 5.3 (The Nordic Cochrane Centre, The Cochrane Collaboration). Both the fixed-effects model and the random-effects model were used in this study. Under the fixed-effects model, it was assumed that all studies come from a common population and that the effect size (standardized mean difference) was not significantly different among the different trials. This assumption was tested by the heterogeneity test or I2 statistic. I2 is the percentage of observed total variation across studies that is due to real heterogeneity rather than chance. It is calculated as I2 = 100% × [(Q − df)]/Q, where Q is Cochran's heterogeneity statistic and df is the degrees of freedom. Negative values of I2 are put equal to zero so that I2 lies between 0% and 100%. A value of 0% indicates no observed heterogeneity, and larger values indicate increasing heterogeneity.12 If this test yielded a low P value (P < .05), then the fixed-effects model was invalid. In this case, the random-effects model was more appropriate, in which both the random variation within the studies and the variation among the different studies was incorporated. For all continuous data, the standardized mean difference was calculated for all outcomes because all analyzed outcomes used the same unit scale. A 95% CI was calculated for each standardized mean difference determined. P < .05 was considered statistically significant.
To examine the effect of age and preoperative AHI on supraglottoplasty success rate, articles that listed individual patient data were pooled, and weighted means (SDs) were calculated for younger and older patients using the median age as the age group–defining factor. The same was done for preoperative AHI data, using the median as a dividing factor. A t test for the comparison of means was then conducted, using MedCalc statistical software, version 184.108.40.206 (MedCalc Software), for both groups to determine the statistical significance of change in the AHI after supraglottoplasty.
The aforementioned search method yielded 217 articles (Figure 1). Six articles13- 18 met study inclusion and exclusion criteria and were selected for qualitative analysis. Four articles13,14,17,18 were ultimately selected for inclusion in our meta-analysis. One study15 did not contain mean (SD) values, and another16 used polysomnography values that could not be combined with the other 4 articles.13,14,17,18 Regardless, these articles15,16 were still included for completeness in the systematic review because they detail polysomnography outcomes in patients who met our inclusion criteria.
All studies in our systematic review were of level 4 evidence. All studies taken together had a total number of 38 patients. Patients’ ages ranged from 1 to 60 months. All articles diagnosed laryngomalacia by flexible fiberoptic examination of the airway. The Table summarizes the study details, including cohort ages, surgical technique, and comorbidities. Most studies had few patients with comorbidities. All studies were conducted at tertiary care hospitals. The following sections will summarize the calculated effect sizes.
Four articles13,14,17,18 were analyzed by meta-analysis of proportions to determine the effects of supraglottoplasty on AHI or RDI (Figure 2). The mean difference of AHI or RDI after supraglottoplasty was −12.46 (95% CI, −21.14 to −3.78; P = .005). One study15 reported median AHI data in patients with laryngomalacia with moderate to severe OSA and those with signs and symptoms of severe laryngomalacia but no OSA. These data could not be used in quantitative analysis.
Mean preoperative AHIs across studies ranged from 11.77 to 49.9, and the postoperative range was 2.2 to 8.4. All studies had statistically significant improvement in AHI. Because the study by O’Connor et al14 found a high mean preoperative AHI, a separate analysis excluding the article was performed. The results still revealed significant improvement (mean difference, −8.06; 95% CI, −11.91 to −4.2; P < .001). Four articles13,14,17,18 had individual patient AHI data with a median AHI of 12. For patients with AHIs less than 12, the mean difference after supraglottoplasty was −3.7 (P = .002). For patients with AHIs greater than 12, the mean difference after supraglottoplasty was −30.7 (P < .001).
Two articles13,18 had individual patient age data, and the median age for the patients with pooled data was 7 months. For patients 7 months and younger, the mean (SD) preoperative AHI was 17.6 (12.1) and the mean (SD) postoperative AHI was 7.6 (5.3) (P = .03). The patients older than 7 months had a mean (SD) preoperative AHI of 6.5 (3.5) and a mean (SD) postoperative AHI of 4.96 (6.58) (P = .07).
Four articles13,14,17,18 were analyzed to determine the effects of supraglottoplasty on oxygen saturation as measured by pulse oximetry (Spo2) nadir. The mean difference of Spo2 nadir after supraglottoplasty was 9.49 (95% CI, 4.87–14.12; P < .001) (Figure 3).
Two studies,14,18 totaling 20 patients, were analyzed to determine the effects of supraglottoplasty on OAHI. The mean difference for OAHI after supraglottoplasty was −21.0 (95% CI, −50.3 to −8.29; P = .16).
Two articles14,18 were analyzed to determine the effects of supraglottoplasty on Pco2 levels. The mean difference for Pco2 after supraglottoplasty was −3.8 (95% CI, −8.6 to 1.0; P = .12).
Two studies,14,18 totaling 20 patients, were analyzed to determine the effects of supraglottoplasty on TST. The mean difference for TST after supraglottoplasty was 0.8 hour (95% CI, 0.1 to 1.5; P = .03).
The principal goal of this study was to determine the magnitude of effect of supraglottoplasty outcomes via polysomnography in children with laryngomalacia and OSA through a comprehensive review of data in the existing literature. The results of this meta-analysis highlight the objective, measurable benefits for children with laryngomalacia and OSA who undergo supraglottoplasty. In our review, we analyzed data on 44 grouped patients. Statistically significant improvements were seen in AHI or RDI, Spo2 nadir, and TST. Of the measured outcomes, only OAHI and Pco2 failed to show statistically significant improvement, possibly because of a lack of power.
A total of 5% to 20% of laryngomalacia cases will have respiratory comorbidities, including OSA; however, the incidence of OSA in laryngomalacia is uncertain. Polysomnography is not routinely ordered in laryngomalacia, but 1 study19 noted a 43.6% prevalence of laryngomalacia in children with OSA confirmed by polysomnography. Another study20 found that the odds to undergo inpatient polysomnography were 17 times higher for patients with laryngomalacia vs those without laryngomalacia. Polysomnography should be considered when there is suspicion of nocturnal hypoxemia or OSA, although many children may present with severe disease that requires prompt surgical intervention.21 At our institution, polysomnography is ordered based on reported symptoms, such as witnessed apneic pauses or retractions during sleep. Polysomnography is also helpful in the evaluation of surgical failure.22
Not all studies reported the same polysomnography data in their outcomes, perhaps because the data were not significant or because the authors focused on specific aspects of the polysomnography data. Future studies analyzing sleep data ideally will include comprehensive descriptions or, at minimum, the AHI because it is a major component of diagnosing and stratifying OSA. Furthermore, although OAHI may be useful in focusing on obstructive events, the contribution of central apneas in laryngomalacia is not insignificant. Tanphaichitr et al23 noted a higher prevalence of central sleep apnea in patients with laryngomalacia who had neurologic disease, hypotonia, or syndromic comorbidities, but none of these factors were statistically significant. Zafereo et al18 reported Central Apnea Index data, but no conclusion could be drawn from this in light of their surgical intervention. Still, given the theory of abnormal sensorimotor integration in laryngomalacia, it is possible that residual disease may have neural contributions. Surgical alteration of the laryngeal mucosa and cartilage would not be expected to have an effect on sensorimotor integration.
Both AHI and RDI were considered equivalent for this study. Two studies14,18 that reported RDI as their primary outcome actually defined it as AHI would be defined (number of central, mixed, and obstructive apneas plus hypopneas per hour of sleep). Another study17 used RDI with OSA parameters defined by Katz et al24 and was therefore also considered to be equivalent to AHI. Powitzky et al15 also stated that the RDI was equivalent to the AHI in their study. Clearly, standardization in defining and reporting outcomes is needed in future studies, and the combination of these 2 parameters may limit our analysis.
The possible effect of age on the polysomnography success rate was examined in this study. The results revealed statistically significant improvement in the group 7 months or younger but not the older group. Given that each group comprised roughly 10 patients, a concrete conclusion cannot be drawn entirely from this analysis. Age may be a factor for polysomnography failure, however, and future studies would seek to further explore this association.
Patients with higher preoperative AHIs also had greater improvement compared with a low AHI cohort. This finding may be because these patients have more room for improvement compared with patients with less severe disease, and so not as great a change will be inherently seen for those with mild to moderate sleep apnea. Thus, the results suggest that supraglottoplasty is more beneficial to patients with more severe disease. However, our data were taken from groups of 14 and 17 patients, and a definitive conclusion cannot be made at this time.
Mean postprocedure AHIs still remained in the mild to moderate range, indicating residual disease. Therefore, although supraglottoplasty revealed objective improvement, disease cure was seldom. Observation and conservative management or multimodality treatment approaches may be necessary for patients with residual disease.
The severity of a patient’s laryngomalacia may influence the response to supraglottoplasty. Valera et al17 studied supraglottoplasty in patients with severe laryngomalacia; O’Connor et al,14 in patients with moderate to severe laryngomalacia; and Zafereo et al,18 in patients with moderate laryngomalacia. Lack of standardization regarding the classification of laryngomalacia severity may further affect outcome analyses based on this factor. In our study, we focused on the severity of the polysomnography findings.
For Pco2, data were pooled from one study18 that reported carbon dioxide levels via PETco2 and another study14 that used Ptcco2. Both PETco2 and Ptcco2 in the pediatric population are equivalent, especially with an AHI less than 10.25 Aggregation of these studies failed to reveal a statistically significant reduction in Pco2.
A meta-analysis by Preciado and Zalzal10 revealed a higher relative risk of polysomnography failure among patients with significant medical comorbidities. Chan et al26 revealed the reduced benefit of supraglottoplasty among comorbid patients with occult laryngomalacia, and Durvasula et al27 revealed higher success rates in neurologically comorbid patients when compared with syndromic patients. The presence of synchronous airway lesions; associated congenital anomalies; neurologic, cardiac, or other medical comorbidities (ie, GERD); and prematurity or younger age variably portends a worse prognosis than patients with isolated laryngomalacia.8,28,29 The mechanisms behind supraglottoplasty failure are not well understood, and there may not be a one-size-fits-all explanation given the multifactorial contributions of patients. Despite the heterogeneous populations pooled for our meta-analysis (different age groups, syndromic and neurologically comorbid patients), statistical significance was achieved for all but 1 polysomnography parameter. Supraglottoplasty may not benefit all patients with laryngomalacia. Powitzky et al15 observed a worsening of AHI in 4 patients with preoperative AHIs less than 5, ultimately improving after adenotonsillectomy. Thus, it is important to consider all sites of obstruction in patients who are being evaluated for airway surgery.30
This study is not without its limitations. The included articles were retrospective, and the overall number of patients was relatively small. Although prospective, double-blind randomized clinical trials would best evaluate the efficacy of supraglottoplasty, patients are often in need of prompt treatment, making such a study difficult to perform. Furthermore, because of the small aggregate sample size of our study, the conclusions should be interpreted cautiously.
The effects of GERD and its treatment were unable to be examined in each study, and the possibility exists for this to confound the success rate of supraglottoplasty. The 2 diseases are thought to contribute to one another: laryngomalacia causing increased work of breathing and therefore increased intra-abdominal pressure and GERD causing mucosal inflammation and the stimulation of secretions in the airway.13 Chan et al26 routinely treated their patients with a proton pump inhibitor for 1 month after surgery. McClurg and Evans13 observed a 42% prevalence of GERD (2 requiring fundoplication), with several patients experiencing significant improvement after laryngoplasty. Powitzky et al15 observed no difference in polysomnography outcomes when considering reflux treatment. Reflux treatment is useful for mild but not severe OSA in the pediatric population, which may explain this finding.31 The association between GERD and laryngomalacia is still, however, a weak one, and high-quality studies are needed to truly define the association and whether a causal effect exists.32
Finally, the risk of aspiration was not assessed because the purpose of this study was to quantify objective improvements in polysomnography. Previous data have revealed an increased risk of aspiration after supraglottoplasty; however, this risk may not be as great as previously thought.10,33 Richter et al34 found that those who undergo supraglottoplasty do not aspirate unless they have a preoperative history of aspiration. Anderson de Moreno et al35 found no significant association between postoperative aspiration and supraglottoplasty, concluding that it is a safe procedure.
The benefits of supraglottoplasty for OSA can be seen almost immediately, as early as 3 days postoperatively, and can result in normal sleep study results, significantly decreased stridor, and cessation of cyanotic spells.36 Treatment of OSA is essential to reducing the risk of comorbidities (ie, cardiovascular, metabolic). Furthermore, cognitive impairment has been found in children with AHIs of 5 or higher; treatment may alleviate this.37 Our analysis further bolsters the argument that surgical treatment of laryngomalacia should be considered as a primary treatment of infants with concomitant OSA. In addition, infants with laryngomalacia should be screened for signs and symptoms of OSA and undergo polysomnography when OSA is suspected.
Supraglottoplasty is beneficial in the treatment of concurrent laryngomalacia and OSA, although patients may have persistent disease. Younger patients (≤7 months) or those with more severe disease may see greater improvement after supraglottoplasty. As more evidence comes to light, definitive guidelines for the use of polysomnography and supraglottoplasty for these coexisting conditions may be established.
Correction: This article was corrected on July 21, 2016, to fix an error in Figure 3.
Accepted for Publication: March 16, 2016.
Corresponding Author: David R. White, MD, Department of Otolaryngology–Head and Neck Surgery, Medical University of South Carolina, 135 Rutledge Ave, Mailstop Code 550, Charleston, SC 29425-5500 (email@example.com).
Published Online: May 12, 2016. doi:10.1001/jamaoto.2016.0830.
Author Contributions: Drs Farhood and Nguyen had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Farhood, Nguyen, White.
Acquisition, analysis, or interpretation of data: Farhood, Ong, Gillespie, Discolo, White.
Drafting of the manuscript: Farhood, Ong, Nguyen, Gillespie, White.
Critical revision of the manuscript for important intellectual content: Farhood, Ong, Gillespie, Discolo, White.
Statistical analysis: Nguyen.
Administrative, technical, or material support: Gillespie, White.
Study supervision: Gillespie, Discolo, White.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.