Parental History of Adenotonsillectomy Is Associated with Obstructive Sleep Apnea Severity in Children with Snoring Emmanouel I. Alexopoulos, MD1, George Charitos, MD1, Georgia Malakasioti, MD1,2, Vasiliki Varlami, MD1, Konstantinos Gourgoulianis, MD1, Elias Zintzaras, MS, PhD3,4, and Athanasios G. Kaditis, MD1,2 Objective To test the hypothesis that history of adenoidectomy and/or tonsillectomy (AT) in at least 1 of the parents during childhood, is a risk factor for moderate-to-severe obstructive sleep apnea (OSA) (apnea-hypopnea index [AHI] >5 episodes/hour) in the offspring with snoring. Study design Data of children with snoring who were referred for polysomnography over 12 years by primary care physicians were reviewed. Results Data of 798 children without history of prior AT, neuromuscular, or genetic disorders or craniofacial abnormalities were analyzed. Of these children, 69.3% had tonsillar hypertrophy, 25.8% were obese, 26.8% had at least 1 parent with history of AT, and 22.1% had AHI >5 episodes/hour. Parental history of AT was significantly associated with moderate-to-severe OSA (logit model including sex, tonsillar hypertrophy, obesity, and physiciandiagnosed wheezing; OR [95% CI], 1.70 [1.18-2.46]; P < .01). When significant variables from the logit model (tonsillar hypertrophy, obesity, parental history of AT) were considered independently or in combination, tonsillar hypertrophy combined with history of AT in at least 1 of the parents had high specificity (84.4%) and the highest positive likelihood ratio (1.78) for identifying children with AHI >5 episodes/hour. Conclusions Among children with snoring who are referred for polysomnography by primary care physicians, those with tonsillar hypertrophy and parental history of AT have increased risk of moderate-to-severe OSA and represent 1 of the subgroups that should be prioritized for a sleep study in settings with limited resources. (J Pediatr 2014;164:1352-7). See related article, p 1346
O
bstructive sleep-disordered breathing (SDB) in childhood represents an abnormal respiratory pattern during sleep resulting from increased upper airway resistance and/or pharyngeal collapsibility and manifested as snoring and increased work of breathing.1 When apnea, hypopnea, and gas exchange abnormalities also occur, the term obstructive sleep apnea (OSA) is applied which is the most severe form of obstructive SDB.2 Nocturnal polysomnography is the gold standard for the diagnosis of OSA, but fewer than 10% of children undergoing adenoidectomy and/or tonsillectomy (AT) for habitual snoring have a sleep study preoperatively to assess SDB severity because of the limited number of pediatric sleep laboratories in many health systems.3 Obesity, history of wheezing, prematurity, race, low socioeconomic status, and OSA diagnosed in first-degree relatives are clinical factors that predict severity of SDB in population-based studies and possibly contribute to the pathogenesis of the disorder either directly or indirectly.1,4-7 For instance, history of AT in parents has been recognized as a risk factor for pharyngeal lymphoid tissue hypertrophy and habitual snoring in the offspring.8 Because SDB is the most common indication for AT in childhood, especially during preadolescent years, parental history of AT can be used as a surrogate measure of clinically significant upper airway obstruction during sleep in the family.9,10 The primary aim of the current study was to assess the relationship between AT in parents and severity of OSA in children in an effort to make speculations about the pathogenesis of intermittent upper airway obstruction during sleep. Hence, we hypothesized that history of AT in at least 1 of the parents during their childhood was significantly associated with the presence of moderate-to-severe OSA (apnea-hypopnea index [AHI] >5 episodes/hour) in the offspring with snoring (>1 night/week). The secondary aim of this investigaFrom the Sleep Disorders Laboratory, School of tion was to evaluate the potential clinical usefulness of a positive parental history Medicine, University of Thessaly and Larissa University Hospital, Larissa, Greece; Pediatric Pulmonology Unit, of AT for the recognition of patients with symptoms of SDB who are at risk of Sleep Disorders Laboratory, First Department of moderate-to-severe OSA given that requests for polysomnography should be Pediatrics, School of Medicine, University of Athens and 1
2
AHI AT BMI OSA SDB
Apnea-hypopnea index Adenoidectomy and/or tonsillectomy Body mass index Obstructive sleep apnea Sleep-disordered breathing
Aghia Sophia Children’s Hospital, Athens, Greece; 3 Department of Biomathematics, School of Medicine, University of Thessaly, Larissa, Greece; and 4 Biostatistics Research Center, The Institute for Clinical Research and Health Policy Studies, Tufts Medical Center, School of Medicine, Tufts University, Boston, MA The authors declare no conflicts of interest. 0022-3476/$ - see front matter. Copyright ª 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jpeds.2014.01.021
1352
Vol. 164, No. 6 June 2014 prioritized in health systems with limited resources. It is important to recognize children with AHI >5 episodes/ hour because they have low rate of spontaneous resolution of OSA, and they are at risk of cardiovascular morbidity, development of respiratory complications following AT, and persistent disease postoperatively.2,11-13
Methods Data of children with snoring ($1 night/week) referred for polysomnography to the Larissa University Hospital by their primary care physician from January 2001 to December 2012 for snoring were considered for analysis. Only data of subjects without history of AT, neuromuscular or genetic disorders, or craniofacial abnormalities were included in the analysis. This retrospective research protocol was approved by the Larissa University Hospital Ethics Committee. Information regarding history and physical examination and results of polysomnography were recorded in a special hospital form (sleep record) and archived in an electronic database within 1 week following the sleep study. Any missing data in the database were completed during the follow-up clinic visit. Since the establishment of the pediatric sleep laboratory in our hospital, all children arrive in late afternoon prior to polysomnography and parents undergo an interview with questions about their child’s health including: (1) symptoms of SDB; (2) history of AT; (3) history of physician-diagnosed wheezing treated with inhaled medications over the past year; (4) history of any chronic disease; and (5) history of AT in parents during their childhood. All children have a physical examination with measurement of height and body weight and calculation of body mass index (BMI) z score.14 Evaluation of tonsillar size by visual inspection of the oropharynx is completed by a single physician (E.A.) using a score ranging from 1+ to 4+.15 During the study period, overnight polysomnography was performed (Alice 4 or 5; Healthdyne, Marietta, Georgia). Polysomnography During the study period, overnight polysomnography was performed using the Alice 4 or 5 computerized systems (Healthdyne), and the following variables were recorded: (1) electroencephalogram (C3/A2, C4/A1, O1/A2); (2) right and left oculogram; (3) submental and tibial electromyogram; (4) body position; (5) electrocardiogram; (6) thoracic and abdominal wall motion (piezoelectric transducers); (7) oronasal airflow (3-pronged thermistor and nasal pressure transducer); and (8) oxygen saturation of hemoglobin. Arousals and sleep stages were assessed using standard criteria.16,17 Obstructive apnea was defined as the absence of airflow for at least 2 breaths in the presence of chest/abdominal wall motion.18 Apneas were considered of mixed type when there was both an obstructive and a central component. Hypopnea was defined as a reduction in the airflow signal amplitude of at least 50% compared with baseline in the presence of chest/ abdominal wall motion and associated with oxygen desaturation of hemoglobin equal to or greater than 4%, or with an
arousal. AHI was the mean number of obstructive and mixed apneas and hypopneas per hour of total sleep time. All polysomnographies were scored by a single physician. Explanatory Variables of Interest and Outcome Variable AHI >5 episodes/hour (moderate-to-severe OSA) was the main outcome variable and parental history of AT (history of AT in at least 1 of the parents during childhood) was the main explanatory variable. Paternal history of AT, maternal history of AT, tonsillar hypertrophy (tonsillar size >2+), obesity (BMI in the 95th percentile or greater), sex, and physician-diagnosed wheezing over the past year treated with inhaled medications19 were other explanatory variables. Statistical Analyses Children without and with parental history of AT were compared in terms of demographic, clinical, and polysomnography variables. Children with both tonsillar hypertrophy and parental history of AT were also compared with all other participants regarding AHI and proportion of subjects with AHI >5 episodes/hour. Student t test was used for continuous variables that approached a normal distribution, Mann–Whitney U test for continuous variables that did not follow a normal distribution, and c2 test for categorical characteristics. The association between moderate-to-severe OSA and each explanatory variable was tested using Fisher exact test. The magnitude of association was expressed as OR and the corresponding 95% CI was calculated using Woolf estimate.20 Furthermore, the effect of parental history of AT and the other explanatory variables (sex, tonsillar hypertrophy, obesity, and physician-diagnosed wheezing) on OSA severity was assessed simultaneously by fitting a logit model.21 The significance of each variable in the model was evaluated using the respective change in deviance and the goodness-of-fit of the model was assessed using the residual deviance after fitting all explanatory variables.21 Two more logit models were fitted for evaluating the effect of paternal or maternal history of AT. The analysis was performed using R (www.r-project.com) and the logit models were fitted using GLIM 3.77 (Royal Statistical Society; London, United Kingdom). Sensitivity, specificity, positive and negative predictive values, and positive likelihood ratio of all explanatory variables that were significant in the logit models for prediction of moderate-to-severe OSA were calculated. Positive likelihood ratio was equal to sensitivity/(1 specificity). Similar calculations were carried out for combinations of parental history of AT with other significant explanatory variables.
Results A total of 798 children with history of snoring ($1 night/week) and without chronic disorders or prior AT underwent polysomnography during the study period. The median age of participants in the cohort was 5.8 years (range 2-15 years and IQR 4.5-7.7 years). Of all subjects, 553 (69.3%) had tonsillar hypertrophy, 206 (25.8%) were obese, and 214 (26.8%) had at least 1353
THE JOURNAL OF PEDIATRICS
www.jpeds.com
1 parent who underwent AT in childhood. One hundred seventysix children (22.1% of 798 subjects) had AHI >5 episodes/hour. Children with parental history of AT did not differ from subjects without parental history of AT in terms of age, BMI z score, proportion of male subjects and frequencies of obesity, tonsillar hypertrophy, prematurity, or history of physiciandiagnosed wheezing treated with inhaled medications over the past year (Table I). The study group with history of AT in parents had significantly higher AHI compared with the group without such history (Table I). In addition, children with both tonsillar hypertrophy and history of AT in parents (n = 146) had significantly higher AHI than all other participants (n = 652): median 3.4 (IQR 1.5-6.6) episodes/ hour vs 2.1 (1-4.1) episodes/hour; P < .01. The former group differed significantly from the latter in the proportion of subjects with AHI >5 episodes/hour: 33.6% vs 19.5%; P < .01. Association of Parental History of AT with Moderate-to-Severe OSA Paternal, but not maternal history of AT, was significantly associated with the presence of moderate-to-severe OSA: OR (95% CI) = 1.51 (1.02-2.23), P = .04 and OR = 1.65 (0.98-2.78), P = .08, respectively. Parental history of AT was significantly related to AHI >5 episodes/hour: OR = 1.63 (1.13-2.33), P < .01 (Table II). The significant associations of paternal and parental history of AT with moderate-to-severe OSA persisted after fitting the logit models (Table II). The residual deviance of models 1 and 3 was significant (P < .01) indicating that additional risk factors should be incorporated in the models. Speculative Analysis In the primary analysis, it was assumed that all parents with history of AT (100%) underwent surgery as treatment for
Vol. 164, No. 6 SDB and the OR (95% CI) for moderate-to-severe OSA in their children was 1.63 (1.13-2.33) and P < .01. If SDB were the indication for surgery in only 70% of parents with history of AT and recurrent tonsillitis were the indication in the remaining parents (30%), provided that the margins of patient numbers with and without moderate-to-severe OSA and their ratio remained fixed (ie, equal to the observed ones), then OR would be 1.56 (1.04-2.33) and P = .03. Also, if only 50% of parents with history of AT underwent surgery for SDB and the remaining for recurrent tonsillitis, the OR for moderate-to-severe OSA in their children would be 1.54 (0.97-2.42) and P = .06. Sensitivity, Specificity, Positive and Negative Predictive Values, and Positive Likelihood Ratio of Parental History of AT and Other Explanatory Variables Significantly Associated with AHI >5 Episodes/hour Among the 4 significant explanatory variables in the logit models (tonsillar hypertrophy, obesity, parental history of AT, paternal history of AT), tonsillar hypertrophy had the highest sensitivity (81.8%) for finding AHI >5 episodes/ hour and the lowest specificity and positive likelihood ratio (Table III). Paternal and parental history of AT had lower sensitivity but higher specificity than tonsillar hypertrophy. When tonsillar hypertrophy was combined with parental history of AT, sensitivity was low but specificity and positive likelihood ratio were higher than those of tonsillar hypertrophy or obesity alone (Table III).
Discussion In the present cohort of children with snoring who had mostly tonsillar hypertrophy and a fairly low frequency of
Table I. Subjects’ characteristics and statistical comparisons of children without and with parental history of AT Variables Age, y Sex, male (%) BMI z score Obesity (%) Tonsillar hypertrophy (%) Physician-diagnosed wheezing treated with inhaled medications over the past year (%) Prematurity (
O
bstructive sleep-disordered breathing (SDB) in childhood represents an abnormal respiratory pattern during sleep resulting from increased upper airway resistance and/or pharyngeal collapsibility and manifested as snoring and increased work of breathing.1 When apnea, hypopnea, and gas exchange abnormalities also occur, the term obstructive sleep apnea (OSA) is applied which is the most severe form of obstructive SDB.2 Nocturnal polysomnography is the gold standard for the diagnosis of OSA, but fewer than 10% of children undergoing adenoidectomy and/or tonsillectomy (AT) for habitual snoring have a sleep study preoperatively to assess SDB severity because of the limited number of pediatric sleep laboratories in many health systems.3 Obesity, history of wheezing, prematurity, race, low socioeconomic status, and OSA diagnosed in first-degree relatives are clinical factors that predict severity of SDB in population-based studies and possibly contribute to the pathogenesis of the disorder either directly or indirectly.1,4-7 For instance, history of AT in parents has been recognized as a risk factor for pharyngeal lymphoid tissue hypertrophy and habitual snoring in the offspring.8 Because SDB is the most common indication for AT in childhood, especially during preadolescent years, parental history of AT can be used as a surrogate measure of clinically significant upper airway obstruction during sleep in the family.9,10 The primary aim of the current study was to assess the relationship between AT in parents and severity of OSA in children in an effort to make speculations about the pathogenesis of intermittent upper airway obstruction during sleep. Hence, we hypothesized that history of AT in at least 1 of the parents during their childhood was significantly associated with the presence of moderate-to-severe OSA (apnea-hypopnea index [AHI] >5 episodes/hour) in the offspring with snoring (>1 night/week). The secondary aim of this investigaFrom the Sleep Disorders Laboratory, School of tion was to evaluate the potential clinical usefulness of a positive parental history Medicine, University of Thessaly and Larissa University Hospital, Larissa, Greece; Pediatric Pulmonology Unit, of AT for the recognition of patients with symptoms of SDB who are at risk of Sleep Disorders Laboratory, First Department of moderate-to-severe OSA given that requests for polysomnography should be Pediatrics, School of Medicine, University of Athens and 1
2
AHI AT BMI OSA SDB
Apnea-hypopnea index Adenoidectomy and/or tonsillectomy Body mass index Obstructive sleep apnea Sleep-disordered breathing
Aghia Sophia Children’s Hospital, Athens, Greece; 3 Department of Biomathematics, School of Medicine, University of Thessaly, Larissa, Greece; and 4 Biostatistics Research Center, The Institute for Clinical Research and Health Policy Studies, Tufts Medical Center, School of Medicine, Tufts University, Boston, MA The authors declare no conflicts of interest. 0022-3476/$ - see front matter. Copyright ª 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jpeds.2014.01.021
1352
Vol. 164, No. 6 June 2014 prioritized in health systems with limited resources. It is important to recognize children with AHI >5 episodes/ hour because they have low rate of spontaneous resolution of OSA, and they are at risk of cardiovascular morbidity, development of respiratory complications following AT, and persistent disease postoperatively.2,11-13
Methods Data of children with snoring ($1 night/week) referred for polysomnography to the Larissa University Hospital by their primary care physician from January 2001 to December 2012 for snoring were considered for analysis. Only data of subjects without history of AT, neuromuscular or genetic disorders, or craniofacial abnormalities were included in the analysis. This retrospective research protocol was approved by the Larissa University Hospital Ethics Committee. Information regarding history and physical examination and results of polysomnography were recorded in a special hospital form (sleep record) and archived in an electronic database within 1 week following the sleep study. Any missing data in the database were completed during the follow-up clinic visit. Since the establishment of the pediatric sleep laboratory in our hospital, all children arrive in late afternoon prior to polysomnography and parents undergo an interview with questions about their child’s health including: (1) symptoms of SDB; (2) history of AT; (3) history of physician-diagnosed wheezing treated with inhaled medications over the past year; (4) history of any chronic disease; and (5) history of AT in parents during their childhood. All children have a physical examination with measurement of height and body weight and calculation of body mass index (BMI) z score.14 Evaluation of tonsillar size by visual inspection of the oropharynx is completed by a single physician (E.A.) using a score ranging from 1+ to 4+.15 During the study period, overnight polysomnography was performed (Alice 4 or 5; Healthdyne, Marietta, Georgia). Polysomnography During the study period, overnight polysomnography was performed using the Alice 4 or 5 computerized systems (Healthdyne), and the following variables were recorded: (1) electroencephalogram (C3/A2, C4/A1, O1/A2); (2) right and left oculogram; (3) submental and tibial electromyogram; (4) body position; (5) electrocardiogram; (6) thoracic and abdominal wall motion (piezoelectric transducers); (7) oronasal airflow (3-pronged thermistor and nasal pressure transducer); and (8) oxygen saturation of hemoglobin. Arousals and sleep stages were assessed using standard criteria.16,17 Obstructive apnea was defined as the absence of airflow for at least 2 breaths in the presence of chest/abdominal wall motion.18 Apneas were considered of mixed type when there was both an obstructive and a central component. Hypopnea was defined as a reduction in the airflow signal amplitude of at least 50% compared with baseline in the presence of chest/ abdominal wall motion and associated with oxygen desaturation of hemoglobin equal to or greater than 4%, or with an
arousal. AHI was the mean number of obstructive and mixed apneas and hypopneas per hour of total sleep time. All polysomnographies were scored by a single physician. Explanatory Variables of Interest and Outcome Variable AHI >5 episodes/hour (moderate-to-severe OSA) was the main outcome variable and parental history of AT (history of AT in at least 1 of the parents during childhood) was the main explanatory variable. Paternal history of AT, maternal history of AT, tonsillar hypertrophy (tonsillar size >2+), obesity (BMI in the 95th percentile or greater), sex, and physician-diagnosed wheezing over the past year treated with inhaled medications19 were other explanatory variables. Statistical Analyses Children without and with parental history of AT were compared in terms of demographic, clinical, and polysomnography variables. Children with both tonsillar hypertrophy and parental history of AT were also compared with all other participants regarding AHI and proportion of subjects with AHI >5 episodes/hour. Student t test was used for continuous variables that approached a normal distribution, Mann–Whitney U test for continuous variables that did not follow a normal distribution, and c2 test for categorical characteristics. The association between moderate-to-severe OSA and each explanatory variable was tested using Fisher exact test. The magnitude of association was expressed as OR and the corresponding 95% CI was calculated using Woolf estimate.20 Furthermore, the effect of parental history of AT and the other explanatory variables (sex, tonsillar hypertrophy, obesity, and physician-diagnosed wheezing) on OSA severity was assessed simultaneously by fitting a logit model.21 The significance of each variable in the model was evaluated using the respective change in deviance and the goodness-of-fit of the model was assessed using the residual deviance after fitting all explanatory variables.21 Two more logit models were fitted for evaluating the effect of paternal or maternal history of AT. The analysis was performed using R (www.r-project.com) and the logit models were fitted using GLIM 3.77 (Royal Statistical Society; London, United Kingdom). Sensitivity, specificity, positive and negative predictive values, and positive likelihood ratio of all explanatory variables that were significant in the logit models for prediction of moderate-to-severe OSA were calculated. Positive likelihood ratio was equal to sensitivity/(1 specificity). Similar calculations were carried out for combinations of parental history of AT with other significant explanatory variables.
Results A total of 798 children with history of snoring ($1 night/week) and without chronic disorders or prior AT underwent polysomnography during the study period. The median age of participants in the cohort was 5.8 years (range 2-15 years and IQR 4.5-7.7 years). Of all subjects, 553 (69.3%) had tonsillar hypertrophy, 206 (25.8%) were obese, and 214 (26.8%) had at least 1353
THE JOURNAL OF PEDIATRICS
www.jpeds.com
1 parent who underwent AT in childhood. One hundred seventysix children (22.1% of 798 subjects) had AHI >5 episodes/hour. Children with parental history of AT did not differ from subjects without parental history of AT in terms of age, BMI z score, proportion of male subjects and frequencies of obesity, tonsillar hypertrophy, prematurity, or history of physiciandiagnosed wheezing treated with inhaled medications over the past year (Table I). The study group with history of AT in parents had significantly higher AHI compared with the group without such history (Table I). In addition, children with both tonsillar hypertrophy and history of AT in parents (n = 146) had significantly higher AHI than all other participants (n = 652): median 3.4 (IQR 1.5-6.6) episodes/ hour vs 2.1 (1-4.1) episodes/hour; P < .01. The former group differed significantly from the latter in the proportion of subjects with AHI >5 episodes/hour: 33.6% vs 19.5%; P < .01. Association of Parental History of AT with Moderate-to-Severe OSA Paternal, but not maternal history of AT, was significantly associated with the presence of moderate-to-severe OSA: OR (95% CI) = 1.51 (1.02-2.23), P = .04 and OR = 1.65 (0.98-2.78), P = .08, respectively. Parental history of AT was significantly related to AHI >5 episodes/hour: OR = 1.63 (1.13-2.33), P < .01 (Table II). The significant associations of paternal and parental history of AT with moderate-to-severe OSA persisted after fitting the logit models (Table II). The residual deviance of models 1 and 3 was significant (P < .01) indicating that additional risk factors should be incorporated in the models. Speculative Analysis In the primary analysis, it was assumed that all parents with history of AT (100%) underwent surgery as treatment for
Vol. 164, No. 6 SDB and the OR (95% CI) for moderate-to-severe OSA in their children was 1.63 (1.13-2.33) and P < .01. If SDB were the indication for surgery in only 70% of parents with history of AT and recurrent tonsillitis were the indication in the remaining parents (30%), provided that the margins of patient numbers with and without moderate-to-severe OSA and their ratio remained fixed (ie, equal to the observed ones), then OR would be 1.56 (1.04-2.33) and P = .03. Also, if only 50% of parents with history of AT underwent surgery for SDB and the remaining for recurrent tonsillitis, the OR for moderate-to-severe OSA in their children would be 1.54 (0.97-2.42) and P = .06. Sensitivity, Specificity, Positive and Negative Predictive Values, and Positive Likelihood Ratio of Parental History of AT and Other Explanatory Variables Significantly Associated with AHI >5 Episodes/hour Among the 4 significant explanatory variables in the logit models (tonsillar hypertrophy, obesity, parental history of AT, paternal history of AT), tonsillar hypertrophy had the highest sensitivity (81.8%) for finding AHI >5 episodes/ hour and the lowest specificity and positive likelihood ratio (Table III). Paternal and parental history of AT had lower sensitivity but higher specificity than tonsillar hypertrophy. When tonsillar hypertrophy was combined with parental history of AT, sensitivity was low but specificity and positive likelihood ratio were higher than those of tonsillar hypertrophy or obesity alone (Table III).
Discussion In the present cohort of children with snoring who had mostly tonsillar hypertrophy and a fairly low frequency of
Table I. Subjects’ characteristics and statistical comparisons of children without and with parental history of AT Variables Age, y Sex, male (%) BMI z score Obesity (%) Tonsillar hypertrophy (%) Physician-diagnosed wheezing treated with inhaled medications over the past year (%) Prematurity (
