A, Severe lingual tonsil hypertrophy. The laryngoscope is displaced anteriorly to show the lingual tonsil filling the vallecula. B, Severe postglottic edema. The posterior laryngeal mucosa is more than twice the usual width. C, Arytenoid edema with loss of normal contour. D, Widespread tracheal cobblestoning. E, Severe carinal blunting.
Michele M. Carr, Mark L. Nagy, Michael P. Pizzuto, Christopher P. Poje, Linda S. Brodsky. Correlation of Findings at Direct Laryngoscopy and Bronchoscopy With Gastroesophageal Reflux Disease in ChildrenA Prospective Study. Arch Otolaryngol Head Neck Surg. 2001;127(4):369–374. doi:10.1001/archotol.127.4.369
To correlate direct laryngoscopic and bronchoscopic findings with the presence of positive test results for gastroesophageal reflux disease (GERD) in children.
Prospective collection of structured data.
An academic pediatric otolaryngology department.
Seventy-seven consecutive patients who underwent direct laryngoscopy and bronchoscopy between June and October 1999.
During direct laryngoscopy and bronchoscopy, descriptions of 7 laryngeal and 6 cricotracheal findings were recorded on a 3-point scale (ie, absent, mild, or severe). Medical records were later reviewed to obtain results of the following tests, if they were part of the record: gastric scintiscan, 24-hour pH probe monitoring, upper gastrointestinal tract series, and esophageal biopsy.
Main Outcome Measures
Correlation of mucosal abnormalities with the presence or absence of a positive test result for GERD.
Fifty (65%) of 77 patients had GERD diagnosed with at least 1 positive test result, 21 (27%) had no clinical symptoms and no positive GERD test results, and 5 (7%) had clinical symptoms but no positive test results. There were significant differences for total laryngeal and cricotracheal scores (P<.001) between the groups with positive and negative results. Significant differences were as follows: in the larynx—large lingual tonsil (P<.001), postglottic edema (P<.001), arytenoid edema (P<.001), ventricle obliteration (P = .03), and true vocal fold edema (P = .001), and in the cricotracheal region—general edema and erythema (P = .003) and blunting of the carina (P<.001). Severe arytenoid edema, postglottic edema, or enlargement of lingual tonsil were pathognomonic of GERD.
Many direct laryngoscopic and bronchoscopic findings correlate well with the diagnosis of GERD as determined by using other tests.
MANY TESTS are useful to diagnose pathologic gastroesophageal reflux disease (GERD). Even in the best hands, each test has a significant false-negative rate.1,2 The use of airway endoscopy in children with extraesophageal reflux has not been clearly established although it is well accepted that adults with GERD may have posterior laryngeal edema and erythema. This study compared airway endoscopic findings in pediatric patients with and without GERD to examine the usefulness in diagnosing GERD.
All patients who underwent direct laryngoscopy and bronchoscopy (DLB) in the Department of Pediatric Otolaryngology, Children's Hospital of Buffalo, Buffalo, NY, between June and October 1999 had descriptions of endoscopic appearance recorded on a checklist. The checklist was developed from a retrospective analysis of endoscopic findings in a series of patients positive for GERD, reported elsewhere.4 Findings were divided into a laryngeal group, which could be assessed by direct laryngoscopy or flexible laryngsocopy, and a cricotracheal group, which required bronchoscopy for evaluation. There were 7 laryngeal factors and 6 cricotracheal factors, each graded as absent, mild, or severe. A mild finding was assigned 1 point; a severe finding was assigned 2 points. A list of these factors and severity definitions is given in Table 1. Some representative photographs are shown in Figure 1. Points were summed weighing each factor equally to determine laryngeal and cricotracheal scores, with the total score being the sum of these 2 scores. In most cases, evaluators were not blinded to the patient's history.
Five patients were scored separately by 2 evaluators (M.M.C. and L.S.B.) using videotape and photographs to establish the reliability of the scoring system. The coefficient of correlation for the presence or absence of factors as judged in the small group of patients by 2 observers blinded to the other's scoring was 0.94 showing good reliability.
The following information was recorded for all patients: sex, age at bronchoscopy, reason for bronchoscopy, and whether a tracheotomy was performed. Medical records were reviewed at the study end point for the following results: gastric scintiscan, an upper gastrointestinal tract series, 24-hour pH probe monitoring, and esophageal biopsies. Any test positive for GERD resulted in the patient being included in the GERD-positive (GERD[+]) group. Patients who did not have symptoms suggestive of GERD and no positive test results composed the GERD-negative (GERD[−]) group. Patients who had GERD symptoms but no positive test results composed the GERD indeterminant group. Patients with tracheal mucosal abnormalities underwent bronchial washings for the presence of lipid-laden macrophages and elevated amylase levels.
Data were entered into an Excel 97 spreadsheet (Microsoft Corporation, Seattle, Wash) and analyzed with SPSS 8.0 (SPSS Inc, Chicago, Ill). Comparisons were made between the GERD(+) and GERD(−) groups for the number of laryngeal and cricotracheal findings, the number of mild and severe findings, and scores for each individual factor. Age, sex, and presence of tracheotomy were also compared. For continuous data, t tests were used; for categorical data, the Mann-Whitney tests were used.
Seventy-seven patients underwent DLB during the study period. There were 51 male and 26 female patients who had an average age of 4.2 years (age range, 0.01-15.1 years). Twenty-four (31%) had a tracheotomy present. Reasons for endoscopy are given in Table 2.
Fifty patients (65%) had at least 1 positive GERD test result and were included in the GERD(+) group (Table 3). Twenty-one patients (27%) had no positive tests and were believed to be GERD free on clinical grounds; they composed the GERD(−) group. Six patients had no positive GERD test results but were believed to have GERD symptoms; they made up the indeterminant group. The distribution of tracheotomies was similar between the GERD(+) and GERD(−) groups. The average age of the patient was 4.2 years (age range, 0.01-15.1 years), with the same age distribution among groups. Overall, the incidence of GERD in this population of children undergoing DLB was 65%, 69% in those younger than 2 years, and 63% in those older than 2 years.
No statistical difference was noted between the groups for the findings of lipid-laden macrophages and elevated amylase levels from tracheal washings, which represents lower power to detect a difference owing to small sample size. None of the patients in the GERD(−) group had positive tracheal washings, but only 3 patients had these washings done. In the GERD(+) group, 27 patients had tracheal washings sent to the laboratory for analyses; the results for 12 patients (44%) were positive for lipid-laden macrophages; 15 patients (55%) had elevated amylase levels.
The average number of laryngeal and cricotracheal findings is summarized in Table 4. In all cases, the number of findings was greater in the GERD(+) group than in the GERD(−) group, and the difference was statistically significant.
For laryngeal and cricotracheal factors, average scores on each were higher for the GERD(+) group than for the GERD(−) group (except for vocal fold lesions), and 5 of 7 laryngeal factors and 2 of 6 cricotracheal factors were significantly different between these groups (Table 5).
The presence of either severe arytenoid edema, severe postglottic edema, or a severely enlarged lingual tonsil was always associated with GERD. Twenty-five (50%) of the 77 patients with GERD had at least 1 severe finding from among these 3 factors. Forty-three (86%) of the 77 patients with GERD had at least 2 of these findings (mild or severe) present, and 49 patients (98%) had at least 1 factor. Seventeen (81%) of the patients who were GERD(−) had only 1 or none of these findings present, and none of these patients had severe scores for these factors.
Calculations of sensitivity and specificity were done. The presence of at least 1 severe finding among lingual tonsil enlargement, postglottic edema, or arytenoid edema had a sensitivity of 50% and a specificity of 100%. Finding at least 2 of these at a mild or severe level had a sensitivity of 87.5% and a specificity of 68%. A laryngeal score of 4 or more had a sensitivity of 74% and a specificity of 81%. A cricotracheal score of 2 or more had a sensitivity of 82% and a specificity of 67%. A total score of 7 or more had a sensitivity of 76% and a specificity of 86%.
The small indeterminant group (n = 6) was compared pairwise with the GERD(+) and GERD(−) groups. There were no statistically significant differences between the indeterminant group and the GERD(+) group. However, the following several factors were different (P<.05) between the indeterminant group and the GERD(−) group: the total number of laryngeal and cricotracheal findings, the number of severe cricotracheal findings, and findings of posterior laryngeal edema, arytenoid edema, generalized tracheal edema, and blunt carina. For these factors, and most of the others (except subglottic stenosis and ventricular obliteration), the indeterminant group had higher scores than did the GERD(−) group. One patient in the indeterminant group had significant levels of lipid-laden macrophages in a tracheal aspirate, suggesting aspiration.
Posterior laryngitis, ranging from erythema and edema to pachyderm changes in the interarytenoid region, and from contact granuloma formation to diffuse vocal fold edema,5 has had a recognized association with GERD in adults since Cherry and Margulies6 reported on this topic in 1968. The mucosa in this region does not have defenses against acid exposure and brief acidifications can result in significant pathology.7 Adult patients with GERD and laryngeal symptoms (ie, dysphonia, cough, globus, throat clearing, or sore throat) had significantly more proximal esophageal acidification than did patients who had GERD with no laryngeal symptoms, but only 40% of those patients with GERD laryngitis had laryngeal findings on flexible laryngsocopy.8 This may be related to variable definitions of posterior laryngitis. In another study,5 75% of the adult patients with symptomatic posterior laryngitis had pharyngeal acid reflux despite being free of typical GERD symptoms. Matthews et al9 recently reported the presence of pharyngeal acid reflux in all of a group of 24 children with laryngomalacia. Some of their patients would not have been considered to have pathologic condition using North American Society of Pediatric Gastroenterology and Nutrition10 criteria for evaluating pH monitoring since they had very few episodes of reflux. This evidence supports the concept that even a small amount of pharyngeal acid can cause edema of the posterior larynx and resultant symptoms. In the study by Deveney et al,11 gastroesophageal reflex–related posterior laryngitis and hoarseness resolved in 73% of the adult patients within 3 months of Nissen fundoplication, showing a relationship with the resolution of reflux. Our group of pediatric patients shows a very strong correlation between posterior laryngeal pathology and GERD; virtually all of the patients with GERD(+) exhibited this.
There are few reports of tracheal findings in patients with GERD. In this group of patients with respiratory symptoms, tracheal findings of GERD are probably more common than in a group without such symptoms. We found that diffuse tracheal edema including a blunt carina was associated with GERD. The mechanism of this may be repeated microaspiration of refluxate.12 About half of those in the GERD(+) group who had tracheal aspirates examined for lipid-laden macrophages or elevated amylase levels were positive, suggesting aspiration in about a quarter of this group.
In this group of pediatric patients undergoing DLB, we found the incidence of GERD to be 65%, much higher than that reported by Altman et al13 in their group of 174 children with congenital airway abnormalities who required hospitalization. They reported an incidence of 28%. We suspect that we may be still be underdiagnosing because of the limitations of existing GERD tests. The endoscopic appearance in most of our indeterminant cases was similar to that found in our GERD(+) group, suggesting that they may in fact have GERD with secondary airway symptoms. This study does not give evidence that allows us to conclude that GERD has a causal relationship with airway problems, but it does allow us to conclude that GERD is present in a significant proportion of pediatric patients with airway symptoms, more than we would expect by chance.
How useful is DLB to diagnose GERD? Commonly used tests for diagnosing GERD include barium esophagram, gastric scintiscan, 24-hour pH monitoring, and esophageal biopsy. Each has its limitations and drawbacks. Barium esophagram had a sensitivity of 86% and a specificity of 69%, extended 24-hour pH monitoring had a sensitivity of 88% and specificity of 94%, and esophagoscopy had a sensitivity of 54% and a specificity of 100% in one study.14 Compare these with the best values for gastric scintiscans with a sensitivity of 79% and a specificity of 93% when any reflux episode is considered a positive test, and where a true-positive test is defined according to 24-hour pH monitoring.15 In our study, finding at least 1 instance of severe arytenoid edema, postglottic edema, or enlargement of lingual tonsil was always associated with the presence of GERD, but only half of the patients displayed at least 1 of these findings. Finding at least 2 of these at a mild or severe level had a sensitivity of 87.5% and a specificity of 68%. Overall, laryngeal appearance is very useful in the diagnosis of GERD in patients with airway symptoms, but less useful to rule it out; this makes sense since some patients may not have refluxate up to the level of the larynx, but may have reflex-mediated respiratory symptoms, such as bronchospasm, laryngospasm, central apnea, or bradycardia.16 Direct laryngoscopy and bronchoscopy have the drawback of requiring a general anesthetic, but these pediatric patients with reports of respiratory symptoms usually require this study to characterize their disease. The findings with the strongest association with GERD can be evaluated with flexible laryngsocopy done at the bedside. We believe that careful observation of the airway appearance can provide strong evidence to make a diagnosis of GERD in pediatric patients with airway symptoms.
In summary, we conclude that 65% of children in this group who underwent DLB for any reason ultimately had GERD diagnosed by having at least 1 positive traditional test. There were significant differences for laryngeal and cricotracheal abnormalities between the GERD(+) and GERD(−) groups. Laryngeal findings associated with GERD were as follows: large lingual tonsil (P<.001), postglottic edema (P<.001), arytenoid edema (P<.001), ventricle obliteration (P = .03), and true vocal fold edema (P = .001). Severe arytenoid edema, postglottic edema, or enlargement of the lingual tonsil were pathognomonic of GERD. Cricotracheal findings associated with GERD were general edema and erythema (P = .003) and blunting of the carina (P<.001). Laryngeal findings associated with GERD can be assessed with flexible laryngoscopy. Bronchoscopy is less likely to have a high yield of information supporting a diagnosis of GERD except in the most severe cases. The incidence of GERD is high enough in children with airway symptoms that these specific mucosal abnormalities should be evaluated in each case.
Accepted for publication September 22, 2000.
Corresponding author: Michele M. Carr, DDS, MD, MEd, FRCSC, EN7-238, Department of Otolaryngology, Toronto General Hospital, 200 Elizabeth St, Toronto, Ontario, Canada M5G 2C4 (e-mail: firstname.lastname@example.org).