Predictors of Obstructive Sleep Apnea Severity in Adolescents | Adolescent Medicine | JAMA Otolaryngology–Head & Neck Surgery | JAMA Network
[Skip to Navigation]
Sign In
Table 1.  Baseline Characteristics for Normal-Weight, Overweight, and Obese Adolescentsa
Baseline Characteristics for Normal-Weight, Overweight, and Obese Adolescentsa
Table 2.  Polysomnographic Data for Normal-Weight, Overweight, and Obese Adolescentsa
Polysomnographic Data for Normal-Weight, Overweight, and Obese Adolescentsa
Table 3.  ANOVA Analysis of Demographic and Clinical Parameters and the Apnea Hypopnea Index
ANOVA Analysis of Demographic and Clinical Parameters and the Apnea Hypopnea Index
Table 4.  Univariate Logistic Regression Model of Predictors of Severe OSA (AHI≥10)
Univariate Logistic Regression Model of Predictors of Severe OSA (AHI≥10)
Table 5.  Multiple Logistic Regression Model of Predictors of Severe OSA (AHI≥10)
Multiple Logistic Regression Model of Predictors of Severe OSA (AHI≥10)
Original Investigation
May 2017

Predictors of Obstructive Sleep Apnea Severity in Adolescents

Author Affiliations
  • 1University of Texas Southwestern Medical Center, Dallas
  • 2Children’s Health, Children’s Medical Center Dallas, Dallas, Texas
JAMA Otolaryngol Head Neck Surg. 2017;143(5):494-499. doi:10.1001/jamaoto.2016.4130
Key Points

Question  Which demographic and clinical variables are correlated with severe obstructive sleep apnea in adolescents?

Findings  In this retrospective case series of 224 adolescents, male sex, body mass index z-score, and tonsillar hypertrophy were significantly associated with severe sleep apnea as measured by the apnea hypopnea index. Age and ethnicity were not associated with objective sleep dysfunction.

Meaning  A low threshold for obtaining polysomnography to screen for sleep apnea is warranted in obese, male adolescents with tonsillar hypertrophy and symptoms of sleep-disordered breathing.


Importance  Untreated obstructive sleep apnea has severe health consequences, yet little is known about predictors of sleep apnea severity in the adolescent population.

Objective  To evaluate clinical and demographic factors associated with obstructive sleep apnea severity in adolescents.

Design, Setting, and Participants  A retrospective case series of 224 children (53% male), ages 12 to 17 years who underwent polysomnography from January 1, 2013, to June 4, 2015. The study was carried out in a large tertiary referral children’s hospital associated with an academic medical center in Dallas, Texas. Children were excluded if they were missing clinical data (eg, tonsil size), had major comorbidities (eg, chromosomal abnormalities), or had previously undergone tonsillectomy and adenoidectomy. The mean (SD) age was 14.6 (1.7) years (range, 12.0-17.9 years), and the patients were 55% Hispanic, 30% African American, 13% white, and 2% other.

Exposures  Electronic medical records were reviewed for demographic, clinical, and polysomnographic parameters.

Main Outcomes and Measures  Correlation between demographic and clinical characteristics and the apnea hypopnea index.

Results  In 224 adolescents (53% male) aged 12 to 17 years, the mean (SD) apnea hypopnea index was 14.9 (28.7) (range, 0.0-187.7) and was positively correlated with CDC-defined weight categories (P = .04) and tonsillar hypertrophy (P < .001). Sex, ethnicity, and age were not associated with the apnea hypopnea index. Severe obstructive sleep apnea (apnea hypopnea index >10) was more common in males (OR, 1.8; 95% CI, 1.0-3.2), patients with tonsillar hypertrophy (OR, 3.2; 95% CI, 1.8-5.8), and patients in a heavier CDC weight class (OR, 2.0; 95% CI, 1.3-3.2). Age and ethnicity did not predict severe obstructive sleep apnea.

Conclusions and Relevance  Obstructive sleep apnea in adolescents is associated with obesity and tonsillar hypertrophy in this study. Severe obstructive sleep apnea is more likely in adolescents who are male or obese, or who have tonsillar hypertrophy. This study supports routine polysomnography in obese male adolescents with tonsillar hypertrophy and symptoms of sleep-disordered breathing to screen for and treat severe obstructive sleep apnea.


Obstructive sleep apnea (OSA) is a sleep disorder characterized by periodic reductions in airflow during sleep secondary to partial or complete airway obstruction.1 Poor airflow leads to oxygen desaturation and subsequent arousal from sleep with spontaneous recovery. The frequent arousals from sleep and hypoxia are associated with daytime somnolence, neurocognitive deficits, cardiovascular disease, and a reduced quality of life.1

Obstructive sleep apnea is a relatively common disorder that affects 2% to 5% of children.2 While there is some data on risk factors for severe OSA in children and adults, there is a paucity of research on predictors of OSA in adolescents. Several studies3-6 have shown a positive correlation between obesity and OSA severity in children of all ages. Other studies7,8 have reported a correlation between obesity and OSA severity in older, but not younger children. However, because the majority of children in these studies were younger than 12 years, the applicability of these findings to adolescents is unknown. Given the increase in prevalence of obesity in adolescents over the recent decades and the negative health consequences of untreated OSA, defining the correlation between obesity and OSA severity has important clinical implications. Similarly, male sex and African American ethnicity have been shown to correlate with OSA severity in adults,9 but this relationship is less well-established in children and adolescents. The objective of this study was to establish the correlation between demographic and clinical parameters and OSA severity in adolescents.


This study was approved by the UT Southwestern Medical Center institutional review board and the need for consent was exempted owing to the retrospective nature of the study. All children who underwent polysomnography (PSG) for suspected OSA at the Children’s Medical Center Sleep Disorders Center of Dallas between January 1, 2013, and June 4, 2015, were considered for inclusion. Data was collected using the electronic medical records (EPIC). Inclusion criteria were age 12 to 17 years with height, weight, tonsil size, and clinical and polysomnographic data available. Children were excluded if they had major comorbidities including chromosomal syndromes, craniofacial abnormalities, Chiari malformations, central nervous system masses, or hypoxic brain damage. Because the study also evaluated the role of tonsil size in predicting OSA severity, children who were missing data on tonsil size or who had previously undergone a tonsillectomy and adenoidectomy were excluded.

All children had previously undergone full-night in-laboratory PSG following the guidelines established by the American Academy of Sleep Medicine.10 The following measurements were recorded: apnea hypopnea index (AHI), which was defined as the mean number of obstructive and central apnea and hypopnea events per hour. Mild OSA was defined as an AHI score of 1 to 4.9, moderate OSA was an AHI score of 5 to 9.9, and severe OSA was an AHI score of 10 or higher. Although there is no agreement on the definition of OSA severity in adolescents, we used the current criteria in our pediatric sleep facility. We recognize that adult criteria may better apply to older adolescents. An obstructive apnea was defined as at least a 90% reduction in oronasal thermal airflow signal lasting at least the duration of 2 breaths during baseline breathing despite respiratory effort. A hypopnea was a decrease in airflow of at least 30% for the duration of at least 2 breaths with either an arousal or a 3% decrease in oxygen saturation. A central apnea met the criteria for an apnea but either lacked inspiratory effort for at least 20 seconds, led to an arousal, or was associated with at least a 3% oxygen desaturation.11 Sleep efficiency was defined as the percentage of total study time spent asleep. The arousal index was defined as the average number of arousals per hour of sleep. The percentage of total sleep time spent in the rapid eye movement (REM) stage of sleep was defined as REM sleep. Oxygen desaturation (oxygen saturation nadir) was defined as the lowest hemoglobin oxygen saturation recorded by pulse oximetry. Peak carbon dioxide was the highest carbon dioxide pressure in mm Hg recorded.

From each patient’s electronic medical record, the following information was collected: age, sex, ethnicity, height, weight, tonsil size, and prior diagnoses of asthma, allergies, or gastroesophageal reflux disease (GERD). Ethnicity was self-selected by the caregiver as Hispanic, African American, white, or other. Children without an ethnicity listed in the medical record were included in the other category. The BMI (calculated as weight in kilograms divided by height in meters squared) z-score (number of standard deviations from population mean, controlled for sex and age) was calculated for each patient using CDC data charts for boys and girls ages 2 to 17 years.12 Children were placed into 1 of 3 age- and sex-adjusted BMI percentile categories, based on CDC classifications (normal weight, 5th-85th percentile; overweight, 85th-95th percentile; and obese, ≥95th percentile).13

Tonsil size was obtained from the sleep study report or prior otolaryngology clinic notes and was graded 1 to 4 according to the grading scale by Brodsky.14 Grade 1 tonsils were hidden behind the tonsillar pillars; grade 2 tonsils extended beyond the tonsillar pillars but occupied less than 50% of the pharyngeal space; grade 3 tonsils occupied 50% to 75% of the pharyngeal space; and grade 4 where tonsils occupied more than 75% of the pharyngeal space. Adenotonsillar hypertrophy was defined as grade 3 or 4 tonsils.

Categorical data was presented as counts with percentages. Continuous data was presented as mean with standard deviations (SD). To test for differences in baseline characteristics between normal weight, overweight, and obese adolescents, ANOVA was used for continuous variables and the Pearson χ2 test or Fisher exact test was used for categorical variables.

An analysis of variance (ANOVA) model was created to assess the correlation between each demographic/clinical variable (sex, age, ethnicity, weight classification, etc) and the AHI. Ethnicity was modeled as a binary variable of African American race and all other races. Tonsillar hypertrophy was also modeled as a binary (3/4 vs 1/2 tonsillar size), as was the presence of asthma, allergic rhinitis, and GERD. Weight classification was modeled as normal weight, overweight, and obese. The statistically significant variables from the univariate analysis (weight classification and tonsillar size) were included in a multivariable model which controlled for the effects of the other variable.

A similar approach was taken to assess predictors of severe OSA using a logistic regression model followed by a multivariable model that controlled for the other variables that were found to be significant in the univariate model (male sex, weight classification, adenotonsillar hypertrophy).

All statistics were performed with Stata statistical software (version 14, Stata Corp). Statistical significance was set at P ≤ .05.


A total of 625 adolescents underwent PSG over the study period. Of the initial group, 401 children were excluded from the study, resulting in a final study population of 224 patients. The most common reasons for exclusion were previous tonsillectomy and adenoidectomy (163 patients), neurological disorders (71 patients), respiratory and/or muscular abnormalities (43 patients), missing data for tonsillar size (34 patients), Down syndrome (28 patients), and craniofacial abnormalities (20 patients).

The baseline characteristics of the study population are presented in Table 1. The mean (SD) age was 14.6 (1.7) years and 119 (53%) were male. The mean (SD) BMI was 33.4 (10.4). Most were either Hispanic or African American (191, 85.3%). A total of 148 (66.1%) were obese. Most (153 [68%]), had OSA (AHI ≥1). Normal-weight adolescents were least likely to have OSA at 48%, while obese children were most likely at 77%. Obese adolescents had the highest percentage with severe OSA (41%) and normal weight had the lowest (16%). There were no significant differences in age, sex, ethnicity, tonsil size, or presence of reflux, asthma, or allergic rhinitis between normal weight, overweight, and obese adolescents.

Table 2 presents the polysomnographic data for normal weight, overweight, and obese adolescents. There were significant differences between the highest and lowest values that were all worst in obese and best in normal-weight adolescents. These included in obese adolescents: a higher AHI of 11 (95% CI, 0.02-22.0); the sleep efficiency was 6.0 lower (95% CI, 0.65-11.4); and the oxygen saturation nadir was 3.8% (95% CI, 0.98-6.6) lower. There were no significant differences in the time spent in central apnea index, REM sleep, arousal index, the peak carbon dioxide, and time spent in sleep with carbon dioxide greater than 50% between normal weight, overweight, and obese adolescents.

Table 3 is a summary of the ANOVA analysis of demographic and clinical parameters and the AHI. There was a positive correlation between the weight classification and the AHI. The AHI was also positively correlated with tonsillar hypertrophy. The positive correlation between the weight classification and tonsillar hypertrophy and the AHI remained after controlling for the effects of each other with multivariable analysis. This correlation also remained significant after controlling for age, sex, African American ethnicity, allergic rhinitis, asthma, and GERD.

Table 4 is a summary of univariate logistic regression of demographic and clinical parameters for severe OSA (AHI ≥10). Male sex nearly doubled the odds ratio for severe OSA (OR, 1.8; 95% CI, 1.0-3.2). Similarly, increasing weight classification increased the odds ratio for severe OSA (OR, 2.0; 95% CI, 1.3-3.2). Adolescents with tonsillar hypertrophy were more than 3 times as likely to have severe OSA (AHI ≥10) compared with those without tonsillar hypertrophy (OR, 3.2; 95% CI, −8 to 5.8) (Table 5). After controlling for the effects of the other variables (weight classification and tonsillar hypertrophy when assessing male sex), sex, BMI z score, and tonsillar hypertrophy continued to predict severe OSA with odds ratios of 2.1, 2.2, and 4.0 respectively. Age and African American ethnicity were not significant predictors of severe OSA, nor were the presence of asthma, allergic rhinitis, or GERD.


In this study of 224 adolescents between the ages of 12 and 17 years who underwent PSG, obesity and tonsillar hypertrophy were associated with increasing severity of OSA as measured by AHI. Severe OSA (AHI >10) was predicted by obesity, male sex, and tonsillar hypertrophy. Adolescents with higher obesity also had a lower blood oxygen nadir and sleep efficiency. African American ethnicity and age were not correlated with OSA severity.

There are few prior studies that reported on the association between weight classification and OSA in adolescents,15 and most published studies have focused on children younger than 12 years.2,8,15-21 In addition, direct comparison of study results is difficult because measures of OSA severity and definitions of overweight and obesity vary between studies. However, our finding of a positive correlation between increasing obesity and OSA severity in adolescents is consistent with prior reported studies. Redline et al15 studied 270 children aged 13 to 16 years and found that obesity (BMI >95th percentile) was strongly correlated with OSA as defined by an AHI greater or equal to 5 (P < .001). Likewise, Kohler et al7 studied 234 children between the ages of 2 and 18 years and found that the risk of OSA was 3.5 times greater with each standard deviation increase in BMI z score in children over 12; no association was found in children under 12.

In this study, male sex was a strong predictor of OSA severity. In some previous studies of OSA in young children, male sex has been shown to correlate with severe OSA.2,19,20 However, there are also multiple studies that failed to find a correlation between male sex and OSA severity in children.17,18 While there are only a few studies on predictors of OSA severity in adolescents, our findings agree with Redline et al,15 who found increased rates of OSA in adolescent males. Similarly, we report that tonsillar hypertrophy was associated with OSA severity. In a study of 69 obese children aged 10 to 18 years, Verhulst et al22 reported an association between tonsillar hypertrophy and AHI in children with mild (AHI <2), but not moderate or severe OSA (AHI ≥2). At least 3 other studies have also reported a correlation between tonsillar hypertrophy and OSA severity in children and adolescents.6,21,23

We did not find a correlation between African American ethnicity or age and OSA severity in adolescents. These findings are consistent with Redline et al,15 who found no significant difference in OSA severity in adolescents or between white and African-American adolescents. These findings are also consistent with a number of studies in younger children that have failed to find differences in OSA severity between children of different ethnicities,16,17,20 or ages.6,7,17 However, African American ethnicity has been strongly linked to OSA severity in adults.9 Because the mean age of the patients in our study was 14.6 years, a plausible explanation for the discord between adolescents and adults is that there are changes in airway physiology or fat distribution that occur in late adolescence or adulthood that were not detected in our comparatively younger population.

There are a number of strengths to this study. It included a large and heterogeneous population of adolescents who all had PSG. All PSG was performed at the same facility over a set time period, which may reduce interfacility variation in technique and analysis. In addition, to our knowledge, this is 1 of a few studies that focused on OSA in adolescents. Limiting the age range for subjects may remove an important potential source of confounding.


There are limitations that need further discussion. The study was retrospective, and some patients were excluded owing to incomplete data availability that introduces a potential source of confounding. While whites, African Americans, and Hispanics were well represented, Asians and Native Americans were poorly represented, so our results may be less applicable to those ethnicities. Our study population was also composed of only adolescents referred to a tertiary pediatric center for PSG for suspected OSA and, as such, may not represent the general population of adolescents. In addition, adolescents with considerable comorbidities were excluded from this study and predictors of OSA severity may be different if considerable comorbidities are present.


In this study of 224 adolescents aged 12 to 17 years with subjective sleep disturbances, increasing AHI was associated with obesity and tonsillar hypertrophy while severe OSA was predicted by obesity, male sex, and tonsillar hypertrophy. A low threshold for obtaining PSG to screen for OSA is warranted in obese, male adolescents with tonsillar hypertrophy and symptoms of sleep disordered breathing. Additional large prospective studies focusing on OSA in adolescents are needed.

Back to top
Article Information

Corresponding Author: Ron B. Mitchell, MD, Department of Pediatric Otolaryngology, UT Southwestern Medical Center, 2350 N Stemmons Freeway, F6212, Dallas, TX 75207 (

Accepted for Publication: January 10, 2017.

Published Online: February 23, 2017. doi:10.1001/jamaoto.2016.4130

Author Contributions: Dr Mitchell had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: All authors.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: All authors.

Critical revision of the manuscript for important intellectual content: Scott, Johnson, Mitchell.

Statistical analysis: Scott, Johnson.

Administrative, technical, or material support: Johnson, Mitchell.

Supervision: Johnson.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Jackson  ML, Howard  ME, Barnes  M.  Cognition and daytime functioning in sleep-related breathing disorders.  Prog Brain Res. 2011;190:53-68.PubMedGoogle Scholar
Lumeng  JC, Chervin  RD.  Epidemiology of pediatric obstructive sleep apnea.  Proc Am Thorac Soc. 2008;5(2):242-252.PubMedGoogle ScholarCrossref
Gabbay  IE, Lavie  P.  Age- and gender-related characteristics of obstructive sleep apnea.  Sleep Breath. 2012;16(2):453-460.PubMedGoogle ScholarCrossref
Mitchell  RB, Garetz  S, Moore  RH,  et al.  The use of clinical parameters to predict obstructive sleep apnea syndrome severity in children: the Childhood Adenotonsillectomy (CHAT) study randomized clinical trial.  JAMA Otolaryngol Head Neck Surg. 2015;141(2):130-136.PubMedGoogle ScholarCrossref
Redline  S, Tishler  PV, Schluchter  M, Aylor  J, Clark  K, Graham  G.  Risk factors for sleep-disordered breathing in children. Associations with obesity, race, and respiratory problems.  Am J Respir Crit Care Med. 1999;159(5 Pt 1):1527-1532.PubMedGoogle ScholarCrossref
Lam  YY, Chan  EY, Ng  DK,  et al.  The correlation among obesity, apnea-hypopnea index, and tonsil size in children.  Chest. 2006;130(6):1751-1756.PubMedGoogle ScholarCrossref
Kohler  MJ, Thormaehlen  S, Kennedy  JD,  et al.  Differences in the association between obesity and obstructive sleep apnea among children and adolescents.  J Clin Sleep Med. 2009;5(6):506-511.PubMedGoogle Scholar
Graw-Panzer  K, Muzumdar  H, Jambhekar  S, Goldstein  N, Rao  M.  Effect of increasing body mass index on obstructive sleep apnea in children.  Open Sleep J. 2010;3:19-23.Google ScholarCrossref
Pranathiageswaran  S, Badr  MS, Severson  R, Rowley  JA.  The influence of race on the severity of sleep disordered breathing.  J Clin Sleep Med. 2013;9(4):303-309.PubMedGoogle Scholar
Medicine AAoS. Standards for Accreditation. Accessed September 15, 2016.
Medicine AAoS. The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications, version 2.1. 2015. Accessed September 13, 2015.
Prevention CfDCa. BMI Percentile Calculator for Child and Teen. Accessed September 13, 2015.
Brodsky  L.  Modern assessment of tonsils and adenoids.  Pediatr Clin North Am. 1989;36(6):1551-1569.PubMedGoogle ScholarCrossref
Redline  S, Storfer-Isser  A, Rosen  CL,  et al.  Association between metabolic syndrome and sleep-disordered breathing in adolescents.  Am J Respir Crit Care Med. 2007;176(4):401-408.PubMedGoogle ScholarCrossref
Scott  B, Johnson  RF, Mitchell Md  RB.  Obstructive sleep apnea: differences between normal-weight, overweight, obese, and morbidly obese children.  Otolaryngol Head Neck Surg. 2016;154(5):936-943.PubMedGoogle ScholarCrossref
Carno  MA, Ellis  E, Anson  E,  et al.  Symptoms of sleep apnea and polysomnography as predictors of poor quality of life in overweight children and adolescents.  J Pediatr Psychol. 2008;33(3):269-278.PubMedGoogle ScholarCrossref
Tauman  R, O’Brien  LM, Ivanenko  A, Gozal  D.  Obesity rather than severity of sleep-disordered breathing as the major determinant of insulin resistance and altered lipidemia in snoring children.  Pediatrics. 2005;116(1):e66-e73.PubMedGoogle ScholarCrossref
Li  AM, Au  CT, Sung  RY,  et al.  Ambulatory blood pressure in children with obstructive sleep apnoea: a community based study.  Thorax. 2008;63(9):803-809.PubMedGoogle ScholarCrossref
Beebe  DW, Ris  MD, Kramer  ME, Long  E, Amin  R.  The association between sleep disordered breathing, academic grades, and cognitive and behavioral functioning among overweight subjects during middle to late childhood.  Sleep. 2010;33(11):1447-1456.PubMedGoogle ScholarCrossref
Xu  Z, Cheuk  DK, Lee  SL.  Clinical evaluation in predicting childhood obstructive sleep apnea.  Chest. 2006;130(6):1765-1771.PubMedGoogle ScholarCrossref
Verhulst  SL, Franckx  H, Van Gaal  L, De Backer  W, Desager  K.  The effect of weight loss on sleep-disordered breathing in obese teenagers.  Obesity (Silver Spring). 2009;17(6):1178-1183.PubMedGoogle Scholar
Verhulst  SL, Schrauwen  N, Haentjens  D,  et al.  Sleep-disordered breathing in overweight and obese children and adolescents: prevalence, characteristics and the role of fat distribution.  Arch Dis Child. 2007;92(3):205-208.PubMedGoogle ScholarCrossref