Obstructive sleep apnea (OSA) is common in children with Down syndrome (DS) and associated with significant morbidity. In the current study we examined polysomnographic outcomes of children with DS who underwent tonsillectomy.
A retrospective chart review of children with DS who underwent a tonsillectomy between 2009–2015 was performed. All children had either a concurrent adenoidectomy or had previously underwent an adenoidectomy. Children with preoperative and postoperative polysomnograms within 6 months of surgery were included in the analysis. Preoperative OSA severity was categorized by obstructive apnea-hypopnea index (OAHI) as follows: mild = 1.5–4.9 events/h; moderate = 5–9.9 events/h; severe ≥ 10 events/h.
Seventy-five children with DS met inclusion criteria. The cohort included 41 males and 34 females with mean age of 5.1 years (± 3.6 years), range of 0.51–16.60 years. Preoperative OSA severity was as follows, mild = 8/75; moderate = 16/75; severe = 51/75. Cure rates varied depending on definition: 12% for OAHI < 1 event/h and 21% for OAHI < 2 events/h. 48% had residual OAHI < 5 events/h. On postoperative PSG 16/75 saw resolution (OAHI < 2) in OSA; mild = 21/75; moderate = 20/75; severe = 18/75. 48% moderate/severe patients saw conversion to mild or cure. Overall, tonsillectomy resulted in significant improvements in multiple respiratory parameters, including OAHI (OAHI; 21.3 ± 19.7 to 8.0 ± 8.1, P < .001), percent sleep time with oxygen saturations < 90% (19.0 ± 25.0 to 6.1 ± 10.1, P < .001), and percent sleep time with end-tidal carbon dioxide above 50 mmHg (7.7 ± 18.0 to 1.8 ± 6.6, P = .001). Average asleep oxygen saturation was associated with postoperative OSA severity.
Children with DS and OSA who undergo tonsillectomy experience improvements in both respiratory event frequency and gas exchange but approximately half still have moderate to severe residual OSA.
Ingram DG, Ruiz AG, Gao D, Friedman NR. Success of tonsillectomy for obstructive sleep apnea in children with Down syndrome. J Clin Sleep Med. 2017;13(8):975–980.
Sleep disorders are common and may contribute to health problems in children with Down syndrome (DS).1–3 Obstructive sleep apnea (OSA) is especially common in this population because of a combination of factors: macroglossia, hypotonia, increased prevalence of obesity, and craniofacial anatomy. Identification and treatment of OSA is important because it has been implicated as a contributing factor to pulmonary hypertension1 as well as behavioral problems.2 Successful treatment of OSA has been shown to improve measures of attention in children with DS.3
Prevalence estimates of OSA in children with DS vary from 20% to 80%.4–7 In 2011, the American Academy of Pediatrics (AAP) published clinical guidelines for children with DS recommending universal screening for OSA with a polysomnogram (PSG) by age 4 years even in asymptomatic children.8 A recent study from Belgium further supports this recommendation because 54% of children who had no parent reported OSA symptoms and had an obstructive apnea-hypopnea index (OAHI) > 2 events/h.5 Of note, the American Academy of Otolaryngology/Head and Neck Surgery recommends obtaining a PSG for all children with DS prior to scheduling an adenotonsillectomy (T&A) to confirm the OSA diagnosis as well as define the severity of OSA to facilitate preoperative planning.9
Current Knowledge/Study Rationale: Children with Down syndrome (DS) have a high prevalence of obstructive sleep apnea (OSA), and adenotonsillectomy is the first-line therapy. To date, outcomes data are limited to small case series, and the current study is the largest study to date examining polysomnographic outcomes.
Study Impact: The outcome data are reliable because there was not a lengthy delay from the diagnosis of OSA to surgical intervention and subsequent postoperative polysomnographic assessment. Our results demonstrate that despite these children having significant improvements in OSA severity following surgery, approximately 50% have residual disease that is moderate/severe.
The AAP clinical guideline on OSA recommends T&A as first-line therapy for OSA.10 In the first randomized T&A trial in healthy children, surgical cure (apnea-hypopnea index [AHI] < 2 events/h) was 79% for the T&A group compared to 46% in a watchful waiting arm.11 Among children with DS, surgical success of T&A is substantially less, ranging from 16% to 66%3,12–17; these previous outcome data were based on relatively small cohorts of patients, ranging from 3 to 36 in sample size. Thottam and colleagues published the largest series to date, at 36 children, and they found that tonsillectomy (TO) decreased OSA severity, although only 16% of patients experienced complete resolution (OAHI < 1 event/h).17 They also reported a significant decrease in the frequency of central apneas following surgery.
Given the small sample sizes of prior studies, we performed a retrospective review of children with DS who underwent TO or T&A at our institution over a 5-year period. We examined the effectiveness of surgery on OSA, central sleep apnea, oxygenation, and hypoventilation.
Institutional review board approval was obtained. Children were identified retrospectively via electronic medical record search that combined the diagnostic code for DS with the procedure code for TO from 2009–2015. From the initial electronic medical record search, individual patient records were examined to ensure accuracy of diagnosis and surgery performed. Children were included if they had both preoperative and postoperative PSGs within 6 months of the date of surgery, at least 60 minutes of sleep duration during the baseline portion of their study without oxygen and/or positive airway pressure, and an OAHI > 1 event/h. Only children who underwent TO or T&A were included; all children had either a concurrent adenoidectomy or had previously undergone an adenoidectomy. Children who had any concomitant airway procedures were excluded from analysis.
For each patient, the following demographic information was recorded: age, sex, ethnicity, date of surgery, indication for surgery, height, weight, body mass index (BMI) percentile, weight category (BMI 5th to 85th percentile = normal, 86th to 95th = overweight, and > 95th = obese), tonsil size (0% to 25% obstructive = 1+, 26% to 50% = 2+, 51% to 75% = 3+, and 76% to 100% = 4+), adenoid size, and medical comorbidities. The tonsil/adenoid data were collected from surgical reports completed by a pediatric otolaryngologist. The preoperative/ postoperative PSG metrics collected were sleep architecture, OAHI, central apnea index, mean oxygen saturation, oxygen nadir, percent sleep time with saturations less than 90%, average oxygen saturations during wake, percent sleep time with end-tidal carbon dioxide (ETCO2) greater than 50 mmHg, mean heart rate, and periodic limb movements. Studies were scored according to the American Academy of Sleep Medicine (AASM) guidelines from when the study was collected.18,19 Significant hypoventilation was defined as ETCO2 > 50 mmHg for ≥ 25% of the total sleep time.20 However, because CO2 > 50 for at least 10% of the total sleep time is often considered an indication for a TO, we also reported this measure. All data were recorded in a secure REDCap database (Vanderbilt University, Nashville, Tennessee, United States).
Patient characteristics were summarized with descriptive statistics. The efficacy of TO for OSA was tested multiple ways. First, the change in PSG parameters was quantified with paired sample t tests. Second, the number of subjects with surgical “cure” was quantified by OAHI < 1 event/h, < 2 events/h, as well as improvement to < 5 events/h. Third, the distribution of OSA severity and significant hypoventilation was examined preoperatively (mild = OAHI 1.5–4.9 events/h; moderate = OAHI 5–9.9 events/h; severe-OAHI ≥ 10 events/h) and postoperatively (cure = OAHI < 2 events/h; mild = OAHI 2–4.9 events/h; moderate = OAHI 5–9.9 events/h; severe = OAHI ≥ 10 events/h). Multivariate analysis was performed using criteria OAHI > 5 events/h. Fourth, the effect of baseline characteristics on postoperative improvement (OAHI < 5 events/h) and change in OAHI was explored because so few children met the rigid definition of cure (OAHI < 1 event/h) to make that analysis worthwhile. A cumulative logit model was used to assess the potential association between baseline and clinical characteristics with postoperative OSA severity (OAHI < 5 events/h, OAHI 5–9.9 events/h = moderate, and OAHI ≥ 10 events/h = severe), where the proportional odds assumption was evaluated. Results are presented as mean (standard deviation) unless otherwise noted. All analyses were performed using SAS 9.4 (SAS Inc. Cary, North Carolina, United States). P < .05 was taken as statistically significant.
Between 2009 and 2015, 323 children with DS underwent TO. Of those, 23% (n = 75) met the inclusion criteria, Forty-five percent were female, and the average age was 5.1 (3.6) years. Forty-seven percent were Caucasian, 41% Hispanic, 4% African American, and 8% other. Significant comorbidities included congenital heart disease (81%), prematurity (25%), pulmonary hypertension (11%), and hypothyroidism (24%). Twenty-four percent were obese and 14% were overweight. Ninety-two percent had concurrent adenoidectomy, and 8% had prior adenoidectomy. Intraoperative tonsil size distribution was as follows: 1+ (4%), 2+ (30%), 3+ (53%), and 4+ (13%). The adenoid size distribution was: absent (5%), < 25% (47%), 26% to 50% (28%), 51% to 75% (12%), > 75% (7%), not reported (1%).
On preoperative PSG, OSA severity was as follows, mild = 8/75 (10.7%); moderate = 16/75 (21.3%); severe = 51/75 (68.0%). The overall cure rate varied depending on the cut point employed. Using the most stringent AASM criteria of OAHI < 1 event/h, only 12% of participants achieved cure.20 Using the CHAT (Childhood Adenotonsillectomy Trial) criteria11 of OAHI < 2 events/h, the cure rate was 21%. Of note, postoperatively 48% of children had an OAHI < 5 events/h. The number of children with severe (OAHI ≥ 10 events/h) disease decreased from 50/75 (67%) to 18/75 (24%) (P < .001). Similarly, the number of children with at least moderate disease (OAHI > 5 events/h) decreased from 67/75 (89%) to 32/67 (48%) (P < .001) (Figure 1).
Change in OSA severity following tonsillectomy.
* = 2 patients had OAHI < 2 events/h preoperatively. Preoperative OSA severity: mild = OAHI 1.5–4.9 events/h; moderate = OAHI 5–9.9 events/h; severe = OAHI ≥ 10 events/h. Postoperative OSA severity: cure = OAHI < 2 events/h; mild = OAHI 2–4.9 events/h; moderate = OAHI 5–9.9 events/h; severe = OAHI ≥ 10 events/h. OAHI = obstructive apnea-hypopnea index, OSA = obstructive sleep apnea.
Change in OSA severity following tonsillectomy.
Three of the 20 patients with moderate OSA on preoperative PSG saw a resolution of their OSA with OAHI < 2.0 events/h, 7/20 were converted to mild OSA, and 5/20 saw no change in severity category, whereas 1/20 saw worsening in severity category. Of the 51 patients with severe disease on preoperative PSG, 10/51 saw resolution, 13/51 converted to moderate, 12/51 converted to mild, and 16/51 saw no change (Figure 1). There were four patients who had worsening in the severity category postoperatively. Two children who were initially categorized as having mild OSA converted to moderate OSA but both interpretations mentioned that that the preoperative OAHI was an underestimate. One child had significant worsening in breathing with an OAHI that increased from 4.9 to 18 events/h. Postoperatively, there was less rapid eye movement sleep, about the same amount of supine sleep, and the videos confirmed that the respiratory pattern was worse. One child went from moderate to severe, but the mean oxygen saturation and ETCO2 both improved.
Changes in PSG parameters with surgery are presented in Table 1. TO resulted in significantly increased sleep efficiency but no other changes in sleep architecture. As already mentioned, the frequency of obstructive respiratory events (OAHI) decreased substantially. There were no significant changes in percentage of children with central apnea index > 5 events/h (7/75 versus 5/75, P = .765). Multiple measures of oxygenation improved, including mean wake (P = .004) and sleep saturations (P < .0001), percent sleep time with oxygen saturations less than 90% (P < .0001), and oxygen nadir (P = .0007). Oxygen nadir distribution is shown on box and whisker plot in Figure 2. When defined as ETCO2 > 50 mmHg for ≥ 10% of total sleep time, hypoventilation improved with a decrease in percent sleep time with ETCO2 above 50 mmHg but the number of children with significant hypoventilation had no statistical improvement (11/70 to 4/73, P = .0575). When hypoventilation was defined as ETCO2 > 50 mmHg for > 25% of the total sleep time, the number of patients decreased from 7/70 to 2/73, P = .0927. Heart rate during sleep decreased significantly (P < .001). There was no change in periodic limb movements.
Preoperative and postoperative changes in polysomnographic measures.
Preoperative and postoperative changes in polysomnographic measures.
Oxygen saturation nadir.
Baseline characteristics of sex, age, BMI Z score, tonsil size, adenoid size, mean asleep oxygen saturation, mean awake oxygen saturation, and preoperative OAHI as well as history of prematurity, congenital heart disease, pulmonary hypertension, and hypothyroidism were examined as potential predictors of postoperative OAHI. Average asleep oxygen saturation (P = .024) was significantly associated with postoperative OAHI in a univariate cumulative logit model. Multivariable ordinal logistic regression models were also carried out with forward selection and stepwise selection procedures. However, average asleep oxygen saturation (P = .024) remained the only significant predictor. The average oxygen saturation for the cohort was 92%. For every one-unit increase in average asleep oxygen saturation, the odds of having mild OSA versus the combined moderate/severe OSA are 25% (95% CI 3%, 52%) greater (ie, a child who has an asleep mean saturation of 93% is 1.25 [25%] more likely to have a postoperative OAHI of < 5 events/h than a child with an asleep mean saturation of 92%). Neither a child's age, weight, or any other variable collected was a significant predictor for the change in OAHI severity (results not shown).
OSA is a common condition with significant health risks. Unfortunately, outcomes data for the first-line therapy, T&A, is sparse. In the current study, we present the largest cohort to date of children with DS undergoing T&A with PSG obtained close to the time of surgery. A strength of this article is that there was not a lengthy delay from the diagnosis of OSA to the intervention and subsequent postoperative PSG assessment; therefore, the surgical outcomes are more likely to be representative of the procedure.
Overall, we found that T&A resulted in substantial improvements in multiple measures of OSA severity, including frequency of obstructive respiratory events and gas exchange. Rates of PSG “normalization” varied depending on the cut point employed, with 12% of children achieving postoperative OAHI < 1 event/h, 21% with OAHI < 2 events/h, and 48% of the cohort had a postoperative OAHI < 5 events/h. These rates of PSG improvement are substantially lower compared to generally healthy children,11,21,22 but similar to that observed in obese children and smaller cohorts of children with DS.15,16
Although the goal of any sleep surgery is to cure OSA, a positive outcome is to convert a child from moderate/severe obstruction to mild. Although an OAHI between 1 and 5 events/h is abnormal, it is a clinically meaningful threshold for several reasons. First, mild residual OSA can be treated with intra-nasal steroids and/or leukotriene inhibitor rather than positive airway pressure or advanced sleep apnea surgery.23–25 Second, an OAHI < 5 events/h may be acceptable if the child does not have significant residual symptoms or OSA-related comorbidity. A recent study examining neuropsychological measures and their association with OAHI in over 1,000 school-aged children found a clear deleterious effect at OAHI > 5 events/h compared to more mildly affected groups.26 Third, one would theoretically expect that by removing obstructing tonsils/adenoid and improving airway patency that if a child is referred for noninvasive ventilation, the airway pressure would be lower and/or better tolerated.
Our data indicate that T&A is likely to substantially decrease OSA severity in children with DS. The high rate of residual moderate to severe OSA seen in this study supports the use of routine postoperative sleep studies in children with DS.
Because surgical therapy is often not curative, identifying factors predictive of success has great clinical utility. In the current study, multiple analyses to identify predictive factors to reduce the OAHI < 5 events/h were performed. The only factor found to be associated with postoperative OAHI was preoperative mean asleep oxygen saturation (1.25, 95% confidence interval 1.03–1.52). It is interesting that the only predictive factor was asleep and not awake oxygen saturation. One would expect that a child with more severe OSA would have lower asleep saturations whereas a child with cardiopulmonary disease would have both lower asleep and awake saturations. Because neither disease severity (as operationalized by OAHI) nor cardiopulmonary status was predictive, one could speculate that it is a muscle tone issue, with those children having worse hypotonia exhibiting lower asleep oxygen saturations and poorer response to surgery. In contrast to T&A studies in healthy children, neither baseline disease severity as defined by OAHI, age, nor overweight/obesity status affected cure rate,11,22 which is consistent with more recent studies in children with DS.16,17 Hypotonia, craniofacial anatomy, and macroglossia are probably more influential in children with DS.
Beyond respiratory parameters, we also found that children had improvements in sleep efficiency and overnight heart rate with surgery. A recent meta-analysis of PSG changes with T&A in children also found increases in sleep efficiency following surgery.27 Interestingly, another study of children with DS found altered sleep architecture, including decreasing sleep efficiency in comparison with typically developing children28; the clinical significance of alterations in sleep architecture remain to be elucidated. The observed decrease in heart rate with T&A has previously been reported in generally healthy children.29,30 One possible explanation is that the intermittent hypoxemia and arousals of untreated sleep apnea result in increased sympathetic tone. With T&A, autonomic balance is improved, with decreased sympathetic and increased para-sympathetic components.30 It is possible that increased sympathetic activity may contribute to long-term cardiovascular morbidity associated with untreated sleep apnea, and this finding may have special importance in children with DS given their increased prevalence of structural and physiologic cardiac abnormalities. Normalization of sleep architecture and improvement in heart rates are an additional benefit of surgery.
The current study certainly has several limitations. First, although this represents the largest reported cohort of patients with DS undergoing T&A, the absolute number of patients was still relatively small. By only including children with both preoperative and postoperative PSGs within 6 months of surgery and excluding children with concomitant other airway surgeries, we sacrificed a larger sample size for a more homogenous comparison. The time restriction for PSG was to limit bias. If there was a longer window for the PSG to be obtained, the OAHI may no longer be representative of their preoperative or postoperative OAHI. Because children with mild OSA may be less likely to have a postoperative PSG, this possibly biased the results. One could assume that those children with mild OSA and persistent symptoms would be more likely to have a postoperative PSG. For an asymptomatic child postoperatively who had only mild OSA preoperatively, the parents may not see the benefit nor want to pay out of pocket for a repeat study. The physician may also not see the necessity of testing an asymptomatic child because they would not be recommending any intervention if the OSA was still mild. Second, the number of African-American children in the cohort is small, potentially influencing ethnicity's role on surgical success. Third, there was not uniform recording of tongue position, so we were unable to assess the effect of additional airway characteristics, which may be related to surgical success.31 Of note, the AASM scoring criteria underwent revision toward the end of data collection. The majority of studies utilized the 2007 guidelines. Because the 2015 guidelines relax the scoring criteria for hypopneas, outcomes studies that include children from 2015 may have less favorable results. A major limitation of this study is that it was retrospective and uncontrolled in nature. A prospective trial with universal postoperative sleep studies would provide a more accurate estimate of surgical efficacy. Although a randomized controlled trial comparing T&A to watchful waiting would be scientifically most valid, this would be difficult to perform in children with DS, as most have moderate to severe disease at baseline and comorbidities that can be exacerbated by untreated OSA.
In conclusion, our data demonstrate that T&A substantially decreases OSA severity in children with DS. However, few achieved surgical cure using the most stringent criteria of OAHI < 1 event/h. Approximately half achieved a postoperative OAHI < 5 events/h. These results highlight the need for prospective, multicenter studies of surgical outcomes in children with DS and OSA, with a particular emphasis on identifying patient characteristics predictive of residual OSA that would require additional medical and/or surgical interventions. A strength of this article is that there was not a lengthy delay from the diagnosis of OSA to the intervention and subsequent postoperative PSG assessment. The surgical outcomes are more likely to be representative of procedure because there was no prolonged delay in obtaining the postoperative PSG.
This study was supported in part by NIH/NCRR Colorado CTSI Grant Number UL1 TR001082. Contents are the authors' sole responsibility and do not necessarily represent official NIH views. The authors report no conflicts of interest. Dr. Norman Friedman is a member of the American Board of Internal Medicine (ABIM) Board of Directors and of the ABIM Internal Medicine Exam Committee. To protect the integrity of Board Certification, ABIM strictly enforces the confidentiality and its ownership of ABIM examination content, and Dr. Friedman has agreed to keep ABIM examination content confidential. No ABIM examination content is shared or otherwise disclosed in this article.