Abstract
Objective
To determine common polysomnographic (PSG) diagnoses for children referred by otolaryngologists.
Study Design
Retrospective case series with chart review.
Setting
Single tertiary pediatric hospital (2010-2015).
Subjects and Methods
Review of the medical records of 1258 patients undergoing PSG by otolaryngology referral. Patients who underwent previous otolaryngologic surgery were excluded. Data distributions were evaluated using means with standard deviations for continuous variables and frequencies with percentages for categorical variables.
Results
A total of 1258 patients were included; 55.9% were male, 64.5% were Caucasian, 16.6% had Down syndrome, and 48% had public insurance. The median age at the time of PSG was 5.2 years (range = 0.2-18.94). Indications for PSG were sleep-disordered breathing (SDB; 69.4%), restless sleep (12.7%), airway anomalies (7.5%), and laryngomalacia (7.2%). SDB was seen in 73.4%, obstructive sleep apnea (OSA) in 53.2%, OSA + central sleep apnea (CSA) in 4.5%, CSA in 0.9%, and non-OSA snoring in 15%. Other diagnoses included periodic limb movements of sleep (PLMS; 7.4%), hypoventilation (6.8%), and nonapneic hypoxemia (2.6%). SDB was more common in younger children and seen in 91.4% of children <12 months and in 69.2% of children ≥24 months, while non-OSA snoring was more common with increasing age (3.7% in children <12 months, 17.7% of children ≥24 months). PLMS were seen in 8.9% of children ≥24 months and in no children <12 months.
Conclusion
While OSA and snoring were the most common diagnoses reported, PLMS, alveolar hypoventilation, and CSA occurred in 7.4%, 6.8%, and 5.4%, respectively. These findings indicate that additional diagnoses other than OSA should be considered for children seen in an otolaryngology clinic setting who undergo PSG for sleep disturbances.
Obstructive sleep apnea (OSA) is a sleep-related breathing disorder characterized by prolonged or intermittent, partial or complete, upper airway obstruction that disrupts normal ventilation during sleep and sleep patterns. This disorder reportedly affects 1.2% to 5.7% of children in the United States. 1 Snoring occurs more commonly in 10% to 20% of children. 2 Overnight, in-laboratory polysomnography (PSG) is considered the gold standard for diagnosing pediatric OSA, but cost and availability often limit its usage. Clinical history and physical examination have been found to be unreliable in diagnosing OSA, and there is no universal screening tool for OSA in children.2,3 Although the Pediatric Sleep Questionnaire has been shown to be a reliable instrument for the identification of OSA in children from 2 to 18 years, it is primarily used for research and not as a clinical tool. 4 Similarly, there is no current consensus or guideline as to when to obtain PSG in children who are candidates for adenotonsillectomy. Recommendations from the American Academy of Otolaryngology—Head and Neck Surgery (AAO-HNS), American Academy of Pediatrics (AAP), and American Academy of Sleep Medicine (AASM) are similar but differ regarding absolute indications for PSG. 2
The AAO-HNSF published a clinical practice guideline in 2011 regarding PSG for pediatric sleep-disordered breathing (SDB) prior to tonsillectomy. This guideline recommended preoperative PSG in children with comorbid conditions such as obesity, Down syndrome, craniofacial anomalies, neuromuscular disorders, sickle cell, or mucopolysaccharidoses. PSG was also suggested in children with discordance between tonsil size and reported severity of SDB and when the need for surgery was uncertain. 5
The guidelines published by the AAP recommend PSG for children with snoring and symptoms or signs of OSA ( Table 1 ). In lieu of PSG, referral to a sleep specialist or otolaryngologist is an option. PSG or evaluation by a sleep specialist was also recommended after adenotonsillectomy in those with significantly abnormal baseline PSG, health conditions secondary to OSA, obesity, and persistent symptoms. 1
Symptoms and Signs of Obstructive Sleep Apnea as per the American Academy of Pediatrics Clinical Practice Guideline. 1
The AASM practice parameters recommend PSG before adenotonsillectomy when history and physical examination suggests OSA. 6
The indications put forth by these organizations are widely debated by otolaryngologists, and a survey of practice in 2006 indicated <10% of children who underwent adenotonsillectomy for SDB had a preoperative PSG. 7 However, with increasing availability of PSG, increasing childhood obesity in the United States, and heightened awareness of the health consequences of OSA, many pediatric otolaryngologists are using PSG as an important diagnostic tool. Therefore, it is important to evaluate the utility of these studies. The primary objectives of this study were (1) to determine common diagnoses found on PSGs for children referred by otolaryngologists for suspected SDB or OSA and (2) evaluate age-specific diagnoses.
Materials and Methods
Patient Population
We performed a retrospective case series with chart review of PSGs performed by otolaryngology referral at Cincinnati Children’s Hospital Medical Center (CCHMC) between 2010 and 2015. The database included demographic information, type of sleep study, ordering diagnosis, final sleep study diagnosis, PSG parameters and findings, and the impression of the study. We excluded PSGs that were performed after an otolaryngologic surgery of the upper or lower airway, including adenotonsillectomy, adenoidectomy, tonsillectomy, midline posterior glossectomy, lingual tonsillectomy, and supraglottoplasty. Patients with tracheostomy tubes and those undergoing titration of oxygen or positive pressure were also excluded. PSGs with sleep time <2 hours were excluded, as were repeat studies. This study was approved by the CCHMC Institutional Review Board.
Overnight PSG
All patients underwent an overnight PSG (up to 12 hours) in our pediatric sleep laboratory. Patients went to bed at the time of their preference, and studies were terminated when they awoke spontaneously, typically between 5 and 6
Clinical Definitions
Subjects were considered to have OSA if their PSG showed an oAHI ≥1 event/h, with an oAHI ≥1 and <5 classified as mild OSA, oAHI ≥5 and <10 classified as moderate, and oAHI ≥10 classified as severe. Subjects were considered to have central sleep apnea (CSA) if their PSG showed ≥5 central apneas or central hypopnea/h. Alveolar hypoventilation syndrome was diagnosed when >25% of sleep time was spent with end-tidal CO2 was >50 mm Hg. Subjects were considered to have periodic limb movements of sleep (PLMS) if their PSG showed >5 leg movements/h. Nonapneic hypoxemia was diagnosed in patients whose PSG showed baseline oxygen saturations <92%, not associated with respiratory events such as apneas or hypopneas. Non-OSA snoring, previously referred to as primary snoring, was seen in the absence of apneas or hypopneas.
Identification of Diagnosis
PSG diagnoses were assigned to patients based on both the PSG clinical impression, as reported by the sleep specialist, and the sleep study diagnosis/ICD-9 code. Because of the large number of observations, natural language processing (NLP) methods were used to abstract specific terms from the notes, which were then cross-referenced with the ICD-9 code. More specifically, text fields within the final diagnostic impression were searched for specified terms to define the final diagnosis. Examples of terms included were OSA and obstructive sleep coupled with terms such as demonstrates, reveals, and presence of. Negation terms were used to rule out the diagnosis (not demonstrate, does not reveal, normal polysomnogram). This process was used for all relevant diagnoses reported. Discrepancies between the NLP results and the ICD-9 codes were reviewed manually. Further review of the medical chart was performed for discrepancies that could not be resolved. This method was also used to abstract the indication for PSG or ordering diagnosis. Medical chart review was necessary in <1% of the sample. All indications were reviewed by the primary author, and assignments were made into the following categories: SDB, restless sleep, anomalies of larynx/trachea, and laryngomalacia. A patient could have more than 1 indication for PSG. Children with Down syndrome could have an indication of “Down syndrome” as the only indication. SDB diagnosis was divided into mutually exclusive subgroups: OSA, OSA+CSA, CSA, and non-OSA snoring. PLMS were diagnosed if the sleep study variable PLM index was >5/h, which typically correlated to an impression or ICD-9 code of PLMS.
Statistical Analysis
Data distributions were evaluated using means with standard deviations for continuous variables and frequencies with percentages for categorical variables. Indication for PSG was assessed and associated with the sleep study diagnosis. The diagnoses were then assessed as a function of age (<12 months, 12 months to <24 months, and ≥24 months); these age classifications were based on typical differences in treatment options based on age. Significant differences in frequencies of diagnoses by age were tested using Pearson χ 2 . Because we conducted multiple tests across diagnoses, results (P values) were adjusted using Bonferroni procedures to reduce the potential for type I error.
Results
Demographics and Clinical Characteristics
A total of 2579 PSGs were performed; 1129 PSGs were excluded because of previous otolaryngologic surgery. Of the 1450 remaining PSGs, 92 were repeat studies and excluded; other exclusions included 60 PSGs performed in children with tracheotomy, 32 performed in patients >18 years old, and 8 with insufficient sleep time. A total of 1258 patients were included in the final analysis.
Among these 1258, 703 (55.9%) were male, 811 (64.5%) were Caucasian, 262 (20.8%) were African American, 26 (2.1%) were Asian, 36 (2.9%) were mixed (white and African American), 94 (7.5%) were other, and 29 (2.3%) were unknown. Public insurance was used for 604 (48%). The median age of the study population was 5.2 years (range = 0.2-18.94). There were 187 (14.9%) studies performed in children <12 months; the median age was 3.7 months (range = 0.2-11.9). In children 12 to 24 months of age, 123 (9.8%) studies were performed; the median age was 17.9 months (range = 12.3-23.9). For children >24 months, 948 (75.3%) studies were performed; the median age was 7.6 years (range = 2-18.94). For the entire population, 256 (20.4%) had a syndrome, with Down syndrome being most common (16.6%). In terms of craniofacial anomalies identified for the entire study population, 12 (4.7%) had Pierre-Robin Sequence, 7 (2.7%) had unspecified craniofacial anomaly, 6 (2.3%) had Coloboma of the eye, Heart defects, Atresia of the choanae, Retardation of growth and development, and Ear anomalies and deafness (CHARGE) syndrome, 3 (1.2%) had achondroplasia, and less than 1% of patients had either Beckwith Wiedemann, Treacher Collins, Williams, or Crouzon syndrome. For children <12 months, Pierre-Robin Sequence was seen in 2 (10.5%), micrognathia in 2 (10.5%), cleft palate in 1 (5.2%), and Beckwith Wiedemann in 1 (5.2%). Table 2 shows demographic and clinical characteristics.
Demographics and Clinical Characteristics of the Study Population.
Does not include cerebral palsy (CP), developmental delay, autism, or other developmental disorders that are not syndromic.
SDB was the most common indication (873 [69.4%]) for PSG. The next most common indications were restless sleep in 12.7%, laryngomalacia in 7.2%, and anomalies of the larynx or trachea in 7.5%. Trends were similar across the age groups, except for children <12 months, for whom laryngomalacia was the indication for the PSG in 38.5% and SDB in 44.4%. Indications are summarized in Table 3 and include breakdowns for each age group.
Indication for Polysomnography.
SDB was diagnosed in 947 patients (73.4%). OSA was seen in 669 (53.2%), OSA+CSA in 56 (4.5%), CSA in 11 (0.9%), and non-OSA snoring in 189 (15%). In terms of severity of OSA based on the oAHI, 395 patients (54.5%) had mild disease, 133 (18.3%) had moderate disease, and 179 (27.2%) had severe disease. The next most common diagnoses in this cohort were PLMS in 93 (7.4%), hypoventilation in 86 (6.8%), and nonapneic hypoxemia in 33 (2.6%). These diagnoses were not mutually exclusive; of the 86 patients with hypoventilation, 63 (73.3%) also had OSA. However, 36 of 93 (38.7%) patients had isolated PLMS. A normal study was seen in 275 patients (21.9%). Diagnostic findings are shown in Table 4 .
Polysomnographic Diagnoses of the Total Patient Population by Age Group. a
Abbreviations: CSA, central sleep apnea; OSA, obstructive sleep apnea.
Critical significance level set at .006 using Bonferroni adjustments to account for multiple testing.
Includes OSA, CSA, OSA+CSA, and non-OSA snoring as mutually exclusive categories.
Among all 725 OSA cases.
Thirty-six of 93 (38.7%) had periodic limb movements as the only diagnosis on record.
For children <12 months, SDB was seen in 171 (91.4%). OSA was seen in 118 (63.1%), OSA+CSA in 42 (22.5%), CSA in 4 (2.1%), and non-OSA snoring in 7 (3.7%). In terms of severity of OSA based on the oAHI, 65 (40.6%) patients had mild disease, 35 (21.9%) had moderate disease, and 60 (37.5%) had severe disease. The next most common diagnoses were hypoventilation in 6 (3.2%) and nonapneic hypoxemia in 4 (2.1%). There were no PLMS. A normal study was seen in 15 (8%).
In children 12 to 24 months of age, SDB was seen in 98 (79.7%). OSA was seen in 79 (64.2%), OSA+CSA in 4 (3.3%), CSA in 1 (0.8%), and non-OSA snoring in 14 (11.4%) children. In terms of severity of OSA based on the oAHI, 48 (57.8%) had mild disease, 16 (19.3%) had moderate disease, and 19 (22.9%) had severe disease. The next most common diagnoses were PLMS in 9 (7.3%), hypoventilation in 2 (1.6%), and nonapneic hypoxemia in 7 (5.7%). A normal study was seen in 24 (19.5%).
For children >24 months of age, SDB was seen in 678 (69.2%). OSA was seen in 472 (49.8%), OSA+CSA in 10 (1.1%), CSA in 6 (0.6%), and non-OSA snoring in 168 (17.7%) children. In terms of severity of OSA based on the oAHI, 282 (58.5%) had mild disease, 82 (17%) had moderate disease, and 118 (24.5%) had severe disease. The next most common diagnoses were PLMS in 84 (8.9%), hypoventilation in 78 (8.2%), and nonapneic hypoxemia in 22 (2.3%). A normal study was seen in 236 (24.9%).
Figure 1 shows trends in diagnosis by age. SDB was more common in younger children and seen in 91.4% of children <12 months, while non-OSA snoring was more common with increasing age (3.7% in those <12 months, 17.7% in children ≥24 months). PLMS were not seen in children <12 months but were present in 8.9% of children ≥24 months. The differences in frequencies of diagnoses across age groups were statistically significant (P < .01 after multiple testing adjustment) for all diagnoses listed except nonapneic hypoxemia and CSA. Figure 2 shows severity of OSA by age.
Trends in polysomnography diagnoses by age. OSA, obstructive sleep apnea; CSA, central sleep apnea; PLMS, periodic limb movements of sleep; AHS, alveolar hypoventilation syndrome; NAH, nonapneic hypoxemia.
Severity of obstructive sleep apnea (OSA) by age. OSA severity is defined by the obstructive apnea hypopnea index (oAHI): mild ≥1 and <5, moderate ≥5 and <10, and severe ≥10.
Discussion
In this cohort, we found SDB, which included OSA, OSA+CSA, CSA, and snoring, in 73.4% of children. Isolated OSA was seen in 53.2%, with non-OSA snoring in 15%. However, there was a significant proportion of children with PLMS, hypoventilation, and nonapneic hypoxemia. A normal study was seen in 22%. This indicates that sleep disorders beyond OSA should be considered in the workup of children who present with sleep complaints. In addition, we saw trends based on age. OSA was more common in younger children (85.6% of children <12 months, 51% ≥24 months), while non-OSA snoring was more common with increasing age (3.7% of children <12 months and 17.7% in children ≥24 months). Similarly, PLMS were seen in 8.9% of children >24 months and in 0% of children <12 months. Trends for severity of OSA were similar for all age groups, with mild OSA most frequent.
Although there is a tendency to focus on OSA when ordering a PSG, additional sleep disorders were common: PLMS, alveolar hypoventilation, CSA, and nonapneic hypoxemia. PLMS may occur in isolation or in conjunction with restless leg syndrome. As a result, identification of PLMS should be followed by an assessment of symptoms with the patient and/or caregiver. Most children with this condition will not complain of sleepiness but may have increased daytime irritability. Children typically complain of leg discomfort with motor restlessness and excessive activity that is worse in the evening or at bedtime; getting out of bed and moving typically relieves symptoms. 9 A family history is common. When PLMS occur in the context of untreated OSA, the patient’s OSA needs to be treated first. The necessity of a repeat PSG was suggested in a recent study by Chervin et al 10 of 137 patients with median age of 7.8 years. They showed a baseline level of PLMS ≥5 in 10% of subjects, with an increase to 15% after adenotonsillectomy. However, there was no significant neurobehavioral morbidity seen in patients with increased postoperative PLMS. 10
Second, hypoventilation is an abnormal finding on PSG. If seen in association with OSA, it typically resolves with treatment of OSA. However, when seen in isolation, it raises the question of an underlying neuromuscular disease, chest wall disorder, or obesity. These patients merit referral to a sleep specialist, as they often need further work up with a daytime arterial blood gas and pulmonary function studies.
CSA is an important finding, and its identification typically leads to evaluation for Chiari malformation in children. In addition to the association between CSA and Chari malformation, mainly type 1, recent studies have shown that it can be also be seen in children with OSA, particularly when there is an absence of adenotonsillar hypetrophy.11,12 Regardless, when CSA >5 events/h is found, a brain magnetic resonance image is typically considered. In our study, OSA+CSA was found in 4.5% of the children, and isolated CSA was seen in nearly 1%. OSA+CSA was much more common (22.5%) in children <12 months as the second most common diagnosis after OSA.
Last, nonapneic hypoxemia was seen in 2.6% of our cohort and likely represents children with disease of their lower airway (eg, cystic fibrosis or alveolar simplification in children with Down syndrome). If nonapneic hypoxemia is found, referral for further evaluation of underlying pulmonary or cardiac disease is recommended.
Our data suggest that sleep disorders seen in children <12 months are different than those seen in older children. The most common indications for PSG in this group were SDB (44.4%) and laryngomalacia (38.5%), with the vast majority (76%) having 1 or both. Of those with an indication of SDB or laryngomalacia, 83% had OSA, an additional 4.9% had non-OSA snoring, and 2.8% had CSA; thereby, 90.7% had SDB. This suggests that infants with either SDB or laryngomalacia have a very high likelihood of having OSA. In addition, of all infants <12 months, >50% had moderate to severe disease. In contrast to children >12 months, we did not find any infants with PLMS, indicating that screening for this sleep disorder may be less important in this age group. Although the data on PLMS in infants are limited, our findings differ from those previously published. In a study by Qubty et al, 13 139 patients aged 0 to 17 months with OSA underwent PSG. The severity of OSA was similar to that which was seen in our series; they reported 30% with mild disease and 70% with moderate-severe disease, which parallels our results, in which 59.4% had moderate-severe disease. In contrast to our findings, they reported that PLMS were seen in 59 of 139 (42%) of their patients. This may be due to our age cutoff of 12 months, whereas they included children up to 17 months. However, in our group of children ≥12 to <24 months, we observed PLMS in 7.3% (9/124) of children, which is much fewer than they reported. There are very little data on this in the literature, particularly in infants; further studies are needed.
Our study has several limitations. First, because of its retrospective nature, there is a lack of clinical data including body mass index, which is important when evaluating children with OSA, particularly in regard to hypoventilation. Another limitation is potential selection bias since these patients were referred from an otolaryngology clinic and not a sleep clinic. PSGs were ordered based on clinical grounds when there was concern for OSA. However, the large size of the study at 1258 patients is formidable, particularly since all PSGs were first diagnostic studies; no repeat studies were included. Last, the indication for PSG may not include all diagnoses. Despite this limitation, this is the first study to report common diagnoses found on PSG aside from OSA and non-OSA snoring and suggests that further evaluation of these conditions in children is warranted.
Conclusion
OSA and non-OSA snoring were the most common diagnoses seen on PSG evaluation for children referred from otolaryngology for suspected OSA; mild OSA was the most common degree of OSA severity. Additional sleep diagnoses were seen in the following order: PLMS, hypoventilation, OSA+CSA, nonapneic hypoxemia, and CSA. In addition, trends in diagnosis were seen for each age, with snoring more common in patients ≥24 months (17.7%) than those <12 months (3.7%). Similarly, PLMS were found in 8.9% of children ≥24 months and not seen in children <12 months. These findings indicate that additional diagnoses other than OSA should be considered for children seen in an otolaryngology clinic setting who undergo PSG for sleep disturbances.
Author Contributions
Disclosures
Footnotes
Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.
This article was presented at the 2016 AAO-HNSF Annual Meeting & OTO EXPO; September 18-21, 2016; San Diego, California.