SGP indicates supraglottoplasty.
Erickson B, Cooper T, El-Hakim H. Factors Associated With the Morphological Type of Laryngomalacia and Prognostic Value for Surgical Outcomes. JAMA Otolaryngol Head Neck Surg. 2014;140(10):927–933. doi:10.1001/jamaoto.2014.1843
The classic presentation of laryngomalacia (LM) is stridor, but alternate presentations include snoring and/or sleep-disordered breathing (S-SDB) and swallowing dysfunction (SWD). Several classification schemes have been developed for LM, but to our knowledge, none have been successfully investigated as to the ability to predict parameters of patients with LM or surgical outcomes.
To compare parameters of patients with different types of LM and determine whether the type has prognostic value for surgical outcomes and to explore if any variable predicts or correlates with the type of LM.
Design, Setting, and Participants
Retrospective case series from a single tertiary pediatric otolaryngology practice. Patients with LM treated with supraglottoplasty (SGP) were eligible. We included patients with confirmed diagnosis of LM who underwent a cold steel SGP and had complete resolution of symptoms or at least 3 months of follow-up, with complete data.
Investigations and treatment of the patients were followed as per routine practice for the senior author (H.E.).
Main Outcomes and Measures
Demographics, type of LM, secondary airway lesions, secondary diagnosis, primary presentation (stridor, S-SDB, SWD), and outcome of SGP were collected. Correlation and multiple regression analysis were performed.
A total of 125 children with LM who underwent SGP for LM were identified. Of these procedures, 119 were cold steel technique, and 8 were repeated procedures. Ninety patients met criteria and were included (mean [SD] age, 1.46 [2.34] years [range, <6 months to 15 years]; male to female ratio, 1.9:1). The primary presentation was stridor in 66 children, S-SDB in 14, and SWD in 10. The type of LM correlated significantly with age (−0.9), and presentation (0.49). Sex and presence of neurological diagnosis (correlation coefficient [SE], −0.317 [0.136], P = .02; and −0.968 [0.361], P <.01, respectively) were associated with outcome. Presentation and obesity were associated with type of LM (−0.251 [0.071], P <.001; and 0.593 [0.296], P = .048, respectively).
Conclusions and Relevance
Type of LM varies by age and primary presentation. Outcome of management is poorer for males and in the presence of a neurological diagnosis. The findings of the present study may help in counseling parents on the risks and benefits of SGP surgery as well as on expected outcomes postoperatively. Further work is required in validating an existing classification scheme for LM or developing a new, validated classification system with may be used for future outcomes research.
Quiz Ref IDLaryngomalacia (LM) is a common congenital laryngeal condition that most often presents with inspiratory stridor. Accurate diagnosis relies on endoscopy. This may be performed at the outpatient clinic or by the bedside while the child is awake, or intraoperatively with the child under sedation with spontaneous respiration.
Relatively recently, different primary presentations have been reported in the literature. These include sleep-disordered breathing, swallowing dysfunction, and exercise-induced LM.1- 4 Owing to the atypical presentations and unusually advanced age at the time of presentation, terms such as state dependent, late onset, or occult LM have been coined.5- 9 However, to our knowledge there have not been epidemiological reports with robust size or design that can justify the connotation of causation or determine the prevalence of these observations.
Meanwhile, it is believed that approximately 20% of children with LM require supraglottoplasty (SGP).10,11 This procedure has been reported to be effective, but, as expected, some patients have poor outcomes.12 Reports of attempts to predict effectiveness have been published, in which patients were stratified according to a variety of variables (medical comorbidities and congenital anomalies as risk factors, prematurity, and obesity9,13- 15). However, very few have used multivariable analysis or validated classifications of LM that can help prognostic research.
Several schemes of LM based on the anatomical subsite of collapse have been proposed.11,16- 21 These all classify LM based on permutations of dysfunction of the epiglottis, aryepiglottic folds, and supra-arytenoid tissue. Quiz Ref IDOlney et al21 described 1 such scheme in 1999. In this scheme, type 1 is described as prolapse of mucosa overlying the arytenoid cartilages. Type 2 involves foreshortened aryepiglottic folds, and type 3 is described as posterior displacement of the epiglottis. None of these schemes have proved consistently useful in grading the disease, documenting its natural history, or prognostication after interventions.
In this study, the primary aims were to determine whether the morphological type of LM has any prognostic value for surgical outcome and whether it is associated with certain parameters of the patients treated.
This was a retrospective observational case series. It involved a medical record review of a consecutive series of children diagnosed as having, and treated surgically for, LM at the Stollery Children’s Hospital (Edmonton, Canada, Alberta) in a single pediatric otolaryngology practice. All children who undergo surgery in this practice are entered into a prospectively kept surgical database by one of us (H.E.). The database includes demographics, diagnoses (otolaryngological and nonotolaryngological, ≥6), procedures performed, special instrumentation, and complications if any. Some of this information has been reported in works by Cooper et al,22 Tevasagayam et al,1 and Rifai et al.23 Institutional review board ethics approval was obtained through the Human Ethics Research Online system (Pro00035516).
All children who had undergone SGP with cold steel technique were eligible. They were identified from the surgical database using both electronic and manual searches. The patients were diagnosed as having LM using a 2.2-mm flexible endoscope. This examination was performed both while the patient was awake and sedated (using intravenous propofol and remifentanyl) while spontaneously breathing. The latter examination was supplemented by full rigid laryngoscopy and bronchoscopy to identify other large airway lesions. These endoscopies are routinely recorded digitally. All of these patients were evaluated in a systematic fashion, in which inquiries about symptoms, including life-threatening events, swallowing dysfunction (SWD), and snoring and/or sleep-disordered breathing (S-SDB) are made. Accordingly, investigations are ordered, and other clinicians and allied health professionals are consulted.
Our inclusion criteria consisted of an endoscopically confirmed LM diagnosis, performance of a cold steel SGP, and adequate follow-up (either follow-up until discharge or >3 months). We excluded those who did not have a confirmed diagnosis from all 3 sources, were diagnosed elsewhere, had a SGP using a different technique, and had incomplete follow-up or had incomplete data from the medical chart. Only data relating to the first operation were included for those patients who required revision procedures.
Sources of data collection included the private practice and hospital charts, and the prospectively kept surgical database previously described. We collected age at time of SGP, sex, primary presentation (stridor, S-SDB, and SWD), investigations performed (including sleep and swallowing studies), secondary diagnoses, and outcome of surgery based on resolution of symptoms.
The term primary presentation refers to the main reason for concern by the referring physician or parent. Stridor was defined as noisy breathing while awake, not primarily feeding-related. S-SDB was defined as snoring, arousals, apneas, or nocturnal or diurnal symptoms. Children with these symptoms were further assessed by overnight oximetry study (PO) or polysomnography (full or abbreviated). The results from the PO studies were considered abnormal if scored 2 or more according to the criteria by Nixon et al24 and polysomnography if the apnea–hypopnea index (AHI) was greater than 1/h.25 The third primary presentation, SWD, was determined by symptoms and signs of feeding difficulties, including choking or coughing, prolonged feeding time, recurrent chest infections, or cyanosis related to feeding. This was further evaluated, if indicated, with clinical swallowing assessment, functional endoscopic evaluation of swallowing and/or video fluoroscopic swallowing study (VFSS). Abnormal swallows were recorded as penetration or aspiration.
Criteria for surgical intervention included 1 or a combination of the following: stridor with signs of increased work of breathing (eg, nasal flaring, body posture, tracheal tugging, chest wall retractions); persistent snoring and sleep-disordered breathing (McGill score grades of 3 or 4 on pulse oximetry); swallowing dysfunction (aspiration or penetration on fiber-optic endoscopic evaluation of swallowing or videofluoroscopic swallowing study). In addition, cyanotic episodes and failure to thrive were considered absolute indications for surgical intervention.
Resolution of symptoms was considered complete if there was total resolution of all the patient’s symptoms allowing for discharge from the practice and successful weaning or cessation of all medical treatment. The other categories were “no resolution” if the patient had no improvement of symptoms and “partial” if the patient had some, but not complete, resolution.
The diagnosis of LM and assignment of type were according to Olney et al.21 The type of LM was determined (1, 2, or 3) and verified through 3 sources (clinical chart, operative report, and video recording).
Secondary diagnoses potentially affecting respiration were documented as 0 or 1 for absence or presence, respectively. A diagnosis of gastroesophageal reflux disease (GERD) was made using medical history, response to proton pump inhibitors, pH probe studies, or esophagoscopy and/or gastroscopy with biopsy. Obesity was defined as a documented body mass index, calculated as weight in kilograms divided by height in meters squared, greater than the 97th percentile for age and sex. Prematurity was documented when gestational age was less than or equal to 36 weeks. Other comorbidities were grouped into the following categories: syndromes and associations, cardiovascular diagnoses, neurological diagnoses (all according to documentation by respective specialists), and secondary airway lesions.
Descriptive statistics were calculated (mean [SD], and range). We used χ2 and Fisher exact tests (as appropriate) for comparison of proportions. Mann-Whitney U test was used for comparison of nonparametric data. Pearson correlation was used for correlation of the various parameters and backward stepwise regression for multivariate analysis. In 2 separate multivariate analyses, the dependent variables were outcome (complete, partial, or no resolution) and type (1, 2, and 3), with the independent variables including age (in years), obesity (absent or present), GERD (absent or present), syndromes or associations (absent or present), cardiovascular diagnoses (absent or present), neurological diagnoses (absent or present), primary presentation (1, 2, and 3), prematurity (absent or present), and secondary airway lesions (absent or present). The statistics were performed using SigmaStat and SigmaPlot software (version 3; Systat Software Inc).
A total of 125 procedures were initially identified as consecutive supraglottoplasties that took place from August 2002 through October 2012. Of these, 6 were found to be erroneously documented and did not receive SGP. Of the 119 remaining, 113 were performed with cold steel instrumentation. Eight of these were revision procedures and therefore were excluded. This left 105 initial supraglottoplasty patients. Finally, 15 patients were excluded owing to insufficient follow-up (Figure).
Ninety patients were included. There were 59 males (66%) and 31 females (34%), resulting in a 1.9:1 male preponderance. The mean (SD) age at time of surgery was 1.5 (2.3) years (range, <6 months to 15 years). Sixty-six children (73%) presented primarily with awake stridor and variable respiratory distress, 14 (16%) with S-SDB, and 10 (11%) with SWD. Multiple patients exhibited symptoms other than the primary complaint. Therefore, cumulatively, stridor was documented in 78, SWD in 73, and S-SDB in 68. Sixty-seven patients had swallowing assessments, and 30 had VFSS. Findings from 44 of these assessments were abnormal, including 31 patients with aspiration and 2 with penetration. Sixty patients had PO sleep studies, and 17 had PSG studies. Fifty-one children had abnormal results from sleep studies, with a McGill score greater than 2. Of the 90 included patients, 38 were admitted for at least 1 night in the pediatric intensive care unit postoperatively. The remaining patients were admitted to either a close observation bed (routine nursing bed with continuous cardiorespiratory monitoring) if older than 1 year, or to a high-dependency bed (1:2 nursing to patient coverage with continuous cardiorespiratory monitoring) if younger than 1 year and older than 3 months, unless results from sleep studies were severely abnormal (grade 4 McGill score).
Complete resolution of all symptoms was achieved in 52 of the children (58%) and partial response in 31 (34%); 7 (8%) had no resolution. No patients required a tracheostomy, although seven patients had revision supraglottoplasty. One of these patients required 2 revisions. The mean (SD) duration of follow-up was 7.7 (6.5) months (range, 2-29 months).
There were 20 patients with GERD, 3 patients with obesity, and 3 patients who were born prematurely. Seven patients had cardiovascular diagnoses (tetralogy of Fallot, transposition of great vessels, atrioventricular septal defect, aortic stenosis), 3 had neurological diagnoses (2, developmental delay; 1, cerebral palsy), and 7 had named syndromes or associations (asymmetric crying facies, Noonan syndrome, dysmorphia, chromosomal abnormality, microcephaly, synostosis). Five patients had 1 or more secondary airway lesions (3, subglottic stenosis; 1, anterior larynx; 2, tracheomalacia).
The proportions of types of LM 1, 2, and 3 were 33:55:2. Because there were only 2 patients with type 3 LM, we focused further analyses on types 1 and 2. Quiz Ref IDType correlated negatively with primary presentation (−0.313; P = .003), indicating that stridor as a primary presentation was more likely to be associated with type 2, whereas S-SDB was likely to be associated with type 1. Age also correlated with primary presentation (0.5; P < .001). It did not, however, correlate with the rest of the variables. The results of the correlation analysis are shown in Table 1.
The median ages were significantly different (Mann Whitney rank sum test; P < .001). Quiz Ref IDThe median age for type 1 (1.6 years) was greater than for type 2 (<6 months). There was also a significant difference in the proportion of primary presenting symptoms (P < .001), with S-SDB more pronounced in patients with type 1 LM (Table 2). Finally, although all the children who had undergone a second SGP had type 2 LM (P = .05), there was no difference in the distribution of outcomes, sex, GERD, prematurity, neurological diagnosis, or obesity within this group.
On multivariable analysis (Table 3), Quiz Ref IDmale sex (−0.317; P = .02), and neurologic diagnosis (−0.968; P = .009) predicted significantly worse outcome of surgery. Type of LM was not a significant predictor of surgical outcome. The significant predictors for type of LM were the primary presentation (−0.251; P < .001), and obesity (0.593; P = .048) (Table 3). The power of both of these tests was greater than 90%.
In this study, infants and children with LM who presented primarily with stridor were more likely to exhibit type 2 LM, whereas those who presented atypically were more likely to exhibit types 1. Patients with type 2 LM were more likely to be younger at presentation.
The age and sex demographics of the patients in our study are similar to those reported in several previous articles,26,27 although the mean age of 1.5 years and upper range of 15 years are higher than studies describing traditional LM. These older ages may reflect the increased diagnosis of LM in atypically presenting patients in more recent publications courtesy of the increased availability and use of flexible fiber-optic scopes. The predominance of males compared with females, 1.9:1 in our study, is also consistent with findings of previously published large series.26,27
We did not find any predictive value for the type of LM with respect to outcome of SGP. To date, 7 studies11,16- 21 have proposed classifications for LM. Only 2 of these20,21 attempted to test for a prognostic value. Lee et al20 in 2007 performed a retrospective study of 138 patients. They20 proposed a classification composed of 3 types (A, B, and C) based on the supraglottic subsite affected and used permutations of these types to test for the prognostic ability of the scheme. They clustered these permutations into 3 groups: group I (type A only), group II (type B or B + A), and group III (type C, C + A, or C + B + A), and used both multivariable and univariable regression analysis. Type of LM, age, intensive care duration, GERD, cerebral palsy, epilepsy, cardiac diagnoses, laryngeal paralysis, nasogastric tube feeding, and postoperative days were included as independent variables in the analysis. Group 3 was associated with statistically significant rates of resolution (B coefficient = 0.79; 95% CI, 0.01-1.59). In addition, patients with cerebral palsy faced a worse outcome (B coefficient = −1.02; 95% CI, −1.80 to −0.25; P < .05). Strengths of this study include the use of multivariable analysis and input of comorbid conditions individually into the statistical analysis. However, the permutations used for types of LM were not justified on a biological basis. Therefore, the validity and applicability of their results is uncertain.
Olney et al21 also attempted to use a classification system to predict outcomes in LM. Their study included 54 patients diagnosed as having LM, of whom only 9 (16%) required surgical treatment. The group21 performed univariable correlation analysis that did not demonstrate a significant association between age or type of LM to spontaneous resolution. They21 did not test their impact on outcome after surgery owing to the limited sample size. This population represents a more homogenous group from the perspective of age and presentation.21 To our knowledge, no other studies have used classification by Olney et al21 for correlation with demographics of patients.
Finally, our results found that male sex and the presence of neurological diagnosis predicted a worse outcome of surgery. Neurological comorbidities specifically have been shown to affect outcome repeatedly, and this is consistent with our data.20,28,29 Some previous studies9,30 claimed that patients with comorbidities develop a poorer outcome after SGP. However, according to the definition of comorbidities, multiple unrelated diagnoses (cardiac, neurological, metabolic, syndromic, and others) were included.9,14,28 By grouping multiple heterogeneous conditions in the performed analysis, the clinical significance of any of the individual conditions cannot be determined, limiting the applicability of these results. In addition, there were limited sample sizes, with the exception of the study by Denoyelle et al,14 and use of univariable analysis. This further limits any conclusions regarding the interaction of the included variables.
The sex difference that we have documented is a novel finding in LM literature. Males had worse SGP outcomes, but the mechanism for this difference based on sex is uncertain. Other researchers have documented higher prevalence of SDB in boys,31 higher incidence of apnea in male infants,32 and worse exacerbations of asthma in boys.33 These researchers are in the same position, unable to offer an explanation, although a hormonal basis has been discussed.
Our study is limited by its retrospective design from claiming any etiological basis. Despite the large sample size (compared with many prior series), certain subgroups may be underrepresented to test for effect on outcome. This pertains in particular to obese patients, and children with secondary airway lesions. Furthermore, our series was from a single pediatric otolaryngology practice, allowing for homogeneous treatment, but this may reduce its applicability to other practices. Herein, we used multivariate analysis, whereas most previous authors largely used univariable tests. This method credits our results in factoring interactions between independent variables.
On a fundamental note, our use of scheme by Olney et al21 is limited by its lack of validation. Classification of disease is important for better definition, description of natural course, and identification of predictive variables of outcome and response to treatment. None of the schemes described thus far have been validated internally or externally. Shah and Wetmore17 stated that their scheme was easy to identify by professionals, but no rigid validation study was performed. In the context of LM, the importance of a valid classification cannot be overemphasized. It may allow some better use of the newly introduced terms linking LM to SDB or swallowing disorders, or explain the phenomenon of late presentation. Our findings may serve in hypothesis generation for future research on atypical presentations of LM and in determining whether this entity represents a separate category of disease based on anatomical location of dysfunction.
The type of LM varies by age and primary presentation. Outcome of surgical treatment seems to be poorer for males and in the presence of a neurological diagnosis. The findings of the present study may help in counseling parents on the risks and benefits of SGP surgery as well as on expected outcomes postoperatively. Further work is required in validating an existing classification scheme for LM or developing a new, validated classification system that may be used for future outcomes research.
Submitted for Publication: April 3, 2014; final revision received June 26, 2014; accepted June 29, 2014.
Corresponding Author: Hamdy El-Hakim, FRCS (ORL-HNS), Division of Pediatric Surgery, Department of Pediatrics, Stollery Children’s Hospital, 2C3.57 Walter MacKenzie Centre, Edmonton, AB T6G 2R7, Canada (email@example.com).
Published Online: September 4, 2014. doi:10.1001/jamaoto.2014.1843.
Author Contributions: Drs Erickson and El-Hakim had full access to all of 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: Erickson, El-Hakim.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Erickson, El-Hakim.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Erickson, El-Hakim.
Administrative, technical, or material support: Erickson, Cooper.
Study supervision: El-Hakim.
Conflict of Interest Disclosures: None reported.
Previous Presentation: This study was presented at the annual meeting of the Canadian Society of Otolaryngology Head and Neck Surgery; June 2, 2013; Banff, Alberta, Canada.