For children with trisomy 21, polysomnography at age 4 to assess obstructive sleep disordered breathing (OSDB) is the standard of care. Oximetry alone has been used to screen for disease among children without trisomy 21. This study evaluates the potential usefulness of oximetry scoring in diagnosing OSDB among children with trisomy 21.
A McGill oximetry score from 1 to 4 was derived from a full overnight PSG done on 119 consecutive pediatric subjects with trisomy 21. Most were referred to the sleep laboratory because of suspicion for OSDB. Oximetry scorers were blinded to the child's full PSG and clinical course. Results of the complete PSG were then compared to oximetry scores.
Obstructive apnea-hypopnea index (OAHI) was ≥ 2.5 for 50% of all subjects. Fifty-nine subjects (49.6%) had McGill Score 1 (“inconclusive”); median OAHI was 1.0 (IQR 0.4–3.3). McGill Score was 2 for 43 subjects (36.1%); median OAHI was 4.5 (IQR 1.3-8.8). Seventeen subjects (14.3%) had McGill Scores of 3 or 4; median OAHI was 16.1 (IQR 9.3–45.5, range 2.1 to 101.1). Ten percent of subjects had a considerable number of central events (≥ 2.5 respiratory events/h but OAHI < 2.5), including 7 with McGill Score 2.
In a retrospective cohort of children with trisomy 21, McGill oximetry scores of 3 or 4 reliably identified patients with marked OSDB. The possibility of central apneas causing hypoxemia must be considered in those with McGill Score 2. With these caveats, oximetry screening should be considered when developing streamlined protocols for early intervention to treat OSDB in this population.
Coverstone AM, Bird M, Sicard M, Tao Y, Grange DK, Cleveland C, Molter D, Kemp JS. Overnight pulse oximetry for evaluation of sleep apnea among children with trisomy 21. J Clin Sleep Med 2014;10(12):1309-1315.
Obstructive sleep disordered breathing (OSDB) has an estimated prevalence of 1% to 3% in the pediatric population.1–3 Risk factors for development of OSDB include hypotonia, obesity, adenotonsillar hypertrophy, and craniofacial abnormalities such as macroglossia and retrognathia.2
Children with trisomy 21 are at a particularly increased risk for OSDB, perhaps due to their relative macroglossia and tendency towards obesity.2 An initial study has shown that 57% of 4-year-olds with trisomy 21 have abnormal polysomnograms (PSG), as defined by hypercarbia, hypoxemia, and/or an abnormal obstructive index ≥ 1.3 Other studies suggest a prevalence of sleep disordered breathing of 30% to 90% at any age among children with trisomy 21.3–5 As in the general population of children, OSDB may affect neurocognitive potential for children with trisomy 21.6,7
The American Academy of Pediatrics (AAP) recommends that children with trisomy 21 should be assessed for OSDB by age 4 with overnight polysomnography.8 Unfortunately, PSGs can at times be particularly challenging for children with trisomy 21. The apparent high prevalence of OSDB among children with trisomy 21, the likelihood of significant morbidity, and the occasional challenge in performing the “gold standard” PSG makes validation of a simple, physiologic test particularly attractive.
Current Knowledge/Study Rationale: Children with trisomy 21 have a high prevalence of sleep disordered breathing. Screening for SDB by polysomnography is the standard of care for these children. This study was performed to evaluate the usefulness of oximetry scoring alone in diagnosing obstructive sleep disordered breathing in children with trisomy 21 in lieu of a full overnight polysomnogram.
Study Impact: Oximetry screening alone may be a useful tool in developing streamlined protocols for early intervention to treat OSDB in children with trisomy 21, who may be difficult to assess using full PSG. This study also indicates the importance of evaluating for central apnea in trisomy 21.
Oximetry alone has been used to identify adult and pediatric patients with obstructive sleep disordered breathing.9–12 The McGill oximetry score, developed at Montreal Children's Hospital, uses the frequency and depth of oxyhemoglobin desaturations throughout a single night of sleep to assign a score of 1 through 4 to the desaturation profile. The score, with 4 indicating the most severe desaturations, has been used to direct surgical or medical interventions or suggest the need for full PSG.1,13
This technique has not been evaluated among a large number of children with trisomy 21.14 Our study used the McGill oximetry score as a hypothetical tool for diagnosing patients with trisomy 21 with sleep disordered breathing. The results suggest that oximetry alone could be used to identify those patients most urgently requiring intervention for OSDB, a finding that has potentially important implications both for AAP policy and the clinical management of children with trisomy 21.
This was a retrospective cohort study of 119 consecutive subjects with trisomy 21 who had a PSG done to evaluate for OSDB and also had PSG oximetry results for which a score could be assigned.
This study was done using de-identified data. No informed consent was obtained. The study was approved by the Washington University Human Research Protection Office.
Subjects ranged in age from 1.75 to 257 months (21 years, 3 months) (median age 71 months, nearly 6 years old). They had PSG performed at St. Louis Children's Hospital, St. Louis, MO, between January 2009 and August 2012. If a patient had more than one sleep study during that time, the first one was used for data collection and McGill scoring. Age at time of study, BMI, history of previous tonsillectomy and/or adenoidectomy, and comorbidities such as cardiac, thyroid, or neurological disease were recorded for each subject. No subject had additional malformations or comorbidities that were exclusion criteria for this study of their PSG results.
With the exception of the 2 oldest subjects (ages 20 and 21 years), the protocols for performing and scoring overnight polysomnograms at St Louis Children's Hospital Sleep Center adhered to AASM criteria (AASM Manual for the Scoring of Sleep and Associated Events, 1st ed., 2007) for subjects younger than 18 years of age; the oldest subjects' PSGs were scored using adult criteria. The standard AASM recording montage was attempted, but several subjects (n = 34, 28.6%) did not have EEG, EMG, or EOG recorded because they did not tolerate the lead placement. For these studies, the occurrence of sleep within a 30-sec epoch was assigned by the technician doing the recording by observation and behavioral criteria: closed eyes, intermittent movements only, and absence of verbal interactions. All PSGs had been previously interpreted for sleep stage (if EEG, EMG, and EOG available), respiratory events, and cardiac events by a physician board-certified in sleep medicine. Apnea-hypopnea index and lowest oxyhemoglobin saturation were recorded from each sleep study. Apnea was identified by a drop in thermistor signal amplitude ≥ 90% from baseline for the entire event. An obstructive apnea was defined as apnea with continued or increased respiratory effort as measured by chest and abdominal excursion. A central apnea was defined by absent inspiratory effort lasting ≥ 20 sec or 2 missed breaths with an associated arousal or desaturation ≥ 3%. Hypopnea was defined as a drop ≥ 50% in nasal pressure signal for a minimum of 2 missed breaths, with an associated arousal or minimum 3% desaturation.
The oximeter's averaging time for SpO2% was 3 to 5 sec, depending on the child's heart rate (Nonin OEM III, Nonin Medical, Inc; Plymouth, MN). The tracing for the oximetry scoring was extracted from that used for the full polysomnography, and viewed alone to assign a McGill Score.
A single McGill score was assigned to each study.1,13 In brief, patients received a McGill score of 1 if there were < 3 desaturations below 90%; 2 if there were ≥ 3 desaturations below 90% but ≤ 3 desaturations below 85%; 3 if there were > 3 desaturations below 85% but ≤ 3 desaturations below 80%; and finally a score of 4 was given to those with > 3 desaturations less than 80%. Having ≥ 3 clusters of 5 desaturations by ≥ 3% was also a criterion for scoring ≥ 2, regardless of whether oxyhemoglobin saturations decreased below 90%, 85%, or 80%. Investigators assigning McGill scores (AC, MB, MS) were blinded to results of the full PSG, and only saw the continuous oximetry recording. Each PSG received a McGill score from 2 different investigators, and scores were checked for agreement. If there was any disagreement in score, a third scorer was used to negotiate a consensus.
Descriptive statistics are presented as mean with standard deviation, or as median with interquartile range, as appropriate. Kruskal-Wallis statistic was used to compare continuous variables among the 4 McGill score groups. Categorical variables (gender, presence of comorbidities, prior adenoidectomy and/ or tonsillectomy, and eventual use of noninvasive positive pressure ventilation) were compared using χ2 and Fisher exact test as appropriate. Independent samples t-tests or Mann-Whitney U Tests as appropriate were used to compare differences in BMI between groups. Strengths of associations between age, BMI, McGill scores, and AHI were calculated using Spearman correlation. The total AHI was log transformed due to lack of normal distribution, and then Proc GLM was applied to determine whether there was a significant relationship between Mc-Gill score and the total AHI after adjusting for age and BMI. All tests for significance were 2-tailed. P values < 0.05 were considered significant. IBM SPSS Statistics version 21.0 (IBM Corporation, Armonk, NY) and SAS version 9.3 (SAS Institute, Cary, NC) were used for all analyses.
There were a total of 131 subjects with trisomy 21 who underwent PSG at our center between January 2009 and August 2012. The indication for PSG for most children in the study was concern for OSDB; however, we estimated approximately 10% of patients underwent PSG based on the 2011 AAP Guidelines on the Health Supervision for Children with Down Syndrome.8
McGill scores were not assigned to 10 patients because the PSG had been transferred off site. One study was excluded from the analyses because the child slept < 6 h. Another was excluded because the clusters of potential desaturation events could not be clearly visualized. Results from a total of 119 subjects were available for final analysis.
Table 1 summarizes the demographic data and clinical characteristics of the subjects. Over half of subjects were male, and the average age was slightly under 7 years old. For all subjects, the median AHI was 4.6/h (range 0-101.8 events/h). Obstructive apnea-hypopnea index (OAHI) was ≥ 2.5/h for 50% of all subjects.
Demographic and clinical characteristics of 119 children with trisomy 21 undergoing overnight polysomnography.
Demographic and clinical characteristics of 119 children with trisomy 21 undergoing overnight polysomnography.
AHI and McGill Scores
Table 2 and Figure 1 summarize the findings from the polysomnograms according to the assigned McGill score. Fifty-nine subjects (49.6%) had a McGill Score of 1. Of this group, the median total AHI was 2.8/h (IQR 0.9-4.4) and the median OAHI was 1.0/h (IQR 0.4-3.3). Forty-three subjects (36.1%) had a McGill Score of 2. The median total AHI for these subjects was 6.5 (IQR 3.2-12.9) and the median OAHI was 4.5 (IQR 1.3-8.8). Seventeen of the 119 subjects (14.3%) had McGill Scores of 3 or 4. The median total AHI of subjects with McGill score > 2 was 17.1 (IQR 12.0-46.8), while the median OAHI was 16.1 (IQR 9.3-45.5). There were highly significant correlations between McGill score and both total AHI (Figure 2A) and obstructive AHI (Figure 2B) (Spearman correlation, rho = 0.649, p < 0.001, and rho = 0.605, p < 0.001, respectively).
Median Age, BMI, and PSG Results for McGill Groups 1-4.
Median Age, BMI, and PSG Results for McGill Groups 1-4.
Flow chart describing 119 consecutive pediatric subjects with trisomy 21 who had McGill oximetry scoring on PSG.
PSG, polysomnogram; CPAP, continuous positive airway pressure; O2, supplemental oxygen.
Flow chart describing 119 consecutive pediatric subjects with trisomy 21 who had McGill oximetry scoring on PSG. PSG, polysomnogram; CPAP, continuous positive airway pressure; O2, supplemental oxygen.
Total, obstructive and central apnea-hypopnea indices by assigned McGill score.
Box plots of apnea-hypopnea indices for each McGill score with median (horizontal line), 25th and 75th interquartile values (box ends), and total range without outliers (whiskers). Circles represent the outliers of 1.5-3 times the interquartile range (box length). Asterisks represent extreme outliers 3 times the box length. (A) Positive correlation between McGill Oximetry Score and total apnea-hypopnea index. Subjects with higher McGill scores had higher total AHI, Spearman correlation rho = 0.65, p < 0.001. (B) Positive correlation between McGill oximetry score and obstructive apnea-hypopnea index. Subjects with higher McGill Scores had higher OAHI, Spearman correlation rho = 0.60, p < 0.001. (C) Positive correlation between McGill score and central apnea index, Spearman correlation, rho = 0.25, p = 0.006. Extreme outlier for one subject with central apnea index of 35.2 excluded from this figure for purposes of the image scale.
Total, obstructive and central apnea-hypopnea indices by assigned McGill score. Box plots of apnea-hypopnea indices for each McGill score with median (horizontal line), 25th and 75th interquartile values (box ends), and total range without outliers (whiskers). Circles represent the outliers of 1.5-3 times the interquartile range (box length). Asterisks represent extreme...
Central Apneas and McGill Score
In addition to obstructive events, several subjects had central apnea events contributing to their total AHI (Figure 2C). Thirty-two subjects (26.9%) had central apnea indices ≥ 2.5/h. Furthermore, 10% of subjects (n = 13) had several central events but fewer obstructive events; that is, ≥ 2.5 central respiratory events/h but an OAHI < 2.5. Of patients with more central events than obstructive events, only one had a McGill score of 3 or 4; 5 had scores of 1, and 7 had scores of 2. One child with a McGill score of 2 had 35.2 central events/h and much periodic breathing (6.9% of sleep time), although his oxygen saturation nadir was only 85%. He was referred to the neurology attending physician, who requested spine and brainstem MRI, but his family decided not to have the test performed.
Clinical Factors and McGill Score
There was no significant difference in age of the subjects by assigned McGill score (ANOVA, F value 1.8, df 3, p = 0.15). The 2 subjects whose PSGs were scored using adult criteria had obstructive AHIs of 17.1 and 13.3, with McGill scores of 3 and 2, respectively. Age correlated with BMI, with older subjects having higher BMI (Spearman correlation, rho = 0.530, p < 0.001). McGill score also correlated with BMI, with subjects assigned McGill scores ≥ 2 having higher BMI values (see Figure 3; Spearman correlation, rho = 0.363, p < 0.001). With respect to BMI, the 13 subjects with more central events did not differ from subjects with McGill scores of 1 and 2 with more obstructive events (mean BMI 18.8 ± 3.6 vs. 21.9 ± 7.3, Mann-Whitney U Test, p = 0.225).
BMI values by assigned McGill Score for all subjects and for those with and without prior adenotonsillectomy.
Box plots of BMI values for each McGill score with median (horizontal line), 25th and 75th interquartile values (box ends), and total range (whiskers). Circles represent the outliers of 1.5-3 times the interquartile range (box length). Asterisks represent extreme outliers 3 times the box length. (A) Positive correlation of McGill score with BMI. Patients with trisomy 21 with higher assigned McGill scores had larger BMI (Spearman correlation rho = 0.36, p < 0.001). (B) BMI with McGill oximetry score for those who had a prior adenoidectomy and/or tonsillectomy compared to those who did not.
BMI values by assigned McGill Score for all subjects and for those with and without prior adenotonsillectomy. Box plots of BMI values for each McGill score with median (horizontal line), 25th and 75th interquartile values (box ends), and total range (whiskers). Circles represent the outliers of 1.5-3 times the interquartile range...
A history of congenital heart disease (Fisher exact test, p = 0.07) or prior tonsillectomy and/or adenoidectomy (Fisher exact test, p = 0.21) was not statistically associated with McGill scores. Overall, the BMI was similar between those that had prior tonsillectomy and/or adenoidectomy and those that did not (mean BMI 22.7 ± 8.2 vs. 19.9 ± 7.1, Independent samples t-test, p = 0.065), although those with McGill score of 4 with prior surgery had higher BMI than those that had not had surgery (mean BMI 31.9 ± 8.1 [n = 6] vs. 21.6 ± 3.9 [n = 5], Independent samples t-test p = 0.029) (see Figure 3B).
By analysis of covariance, McGill score itself was predictive of the logarithm of total AHI (F value 27.0, df 3, p < 0.001) after adjusting for age and BMI.
Otolaryngology Referrals and Respiratory Support
All subjects with a McGill score of 3 or 4 were either referred for surgical consultation or started on supplemental oxygen or noninvasive ventilation based on results of their polysomnograms, at the discretion of the physician ordering the PSG. For most subjects with a McGill score of 1, the recommendation was to repeat the PSG in approximately one year.
Fifty-five subjects (46.2% of total) were referred to the otolaryngology service (ENT) following their PSG by the physicians ordering the sleep study, based on the results for the entire PSG. Twelve of 17 subjects (70.6%) with a McGill score of 3 or 4 were referred for possible surgery. Nineteen of 59 (32.2%) with a McGill score of 1 and 24 of 43 (55.8%) with a McGill score of 2 were referred to ENT for possible surgery (Fisher exact test, p = 0.01).
Only one subject underwent immediate tonsillectomy and adenoidectomy. He had a McGill score 4, OAHI of 61.1, and lowest oxygen saturation of 57%. Another child with McGill score 4 was referred for immediate tonsillectomy and adenoidectomy, but parents chose not to perform surgery and he was begun on CPAP.
Fifteen subjects were started on noninvasive ventilation (CPAP or BPAP) or supplemental oxygen. Eight of these subjects had a McGill score of 3 or 4, 6 had a McGill score of 2, and only one had a McGill score of 1 (Fisher exact test, p < 0.001).
Specificity and Predictive Value of McGill Score for Obstructive Sleep Disordered Breathing
An AHI ≥ 2.5 was chosen arbitrarily as the “cut point” for categorizing subjects as likely abnormal and having OSDB versus possibly normal. Though higher than the typical threshold of 1.0 used in analysis of pediatric SDB, it is arguable that a different threshold is appropriate for patients with trisomy 21. Using OAHI ≥ 2.5 as indicative of abnormality and McGill Scores of 3 or 4 as a positive test, the oximetry score was 98% specific with a positive predictive value of 94% for obstructive SDB. This method yields a Likelihood Ratio of OSDB with McGill Score of 3 or 4 of 13.5. If a McGill score ≥ 2 is used as the criterion for a positive test, the specificity and positive predictive value for OSDB falls to 71% each. Likelihood ratio for OSDB with McGill score ≥ 2 is 2.4.
We have shown that the positive predictive value of McGill score 3 or 4 for an OAHI ≥ 2.5 is 94% in patients with trisomy 21. We did not use detailed follow-up to evaluate the sensitivity of the McGill oximetry score in this population,14 but we contend that with, a few caveats, oximetry scores will prove useful in the identification of the highest risk patients for early evaluation and possible treatment, particularly within clinical research protocols.
Preliminary studies indicate a high prevalence of OSDB, near 50% or more, among children with trisomy 21.3–5 OSDB may contribute to their greater propensity for pulmonary hypertension, a particular concern because many have had surgical repairs of cyanotic congenital heart disease with pulmonary overperfusion. It is also likely that OSDB further increases risk for behavioral and neurocognitive impairment among those with trisomy 21,6,7 a group of children with particular educational challenges. One small study has linked the number of apneas during sleep to specific deficiencies in visual learning among children with trisomy 21.6
Identifying patients with trisomy 21 for early interventions to treat OSDB should be a clinical research priority. Using oximetry scores among children with trisomy 21 appears to be as effective in identifying children with marked OSDB as it is for children with adenotonsillar hypertrophy, who have similar “pre-test” probability of having OSBD.1,13 Indeed, the greatest usefulness of oximetry and similar modalities is not as a screening test, per se, but in identifying those children most likely to have serious OSDB from among those with a high “pre-test” probability for OSDB because of, for example, adenotonsillar hypertrophy or craniofacial malformations.13 That being said, the efficacy of surgical intervention such as adenotonsillectomy for children with trisomy 21 is uncertain, with some studies indicating that over half require CPAP or supplemental oxygen after surgery.15,16 However, these studies do not indicate the degree of tonsillar hypertrophy, nasal obstruction, or obesity in the subjects; in our study, 35% of children with trisomy 21 already had an adenoidectomy or adenotonsillectomy prior to their evaluation with polysomnography. The McGill scores of these subjects did not differ from those that had not had prior surgery, therefore, we cannot reasonably speculate about the benefit of surgery.
It could be argued that the value of a simpler diagnostic test is diminished in the absence of an established, highly successful intervention. Nevertheless, the cardiovascular, behavioral and cognitive risks make it imperative that effective intervention protocols be tailored for patients with trisomy 21. Oximetry alone seems to be a promising and more efficient tool than full PSG for identifying children to enroll in research studies of treatment protocols, particularly those that evaluate the change in oximetry scores pre- and post-surgical intervention.17
Pediatric sleep laboratories and technicians are particularly adept at obtaining PSG studies in children with trisomy 21,18 but over a quarter of our subjects were unable to be studied using the entire montage with EEG, EMG, and EOG, because of lack of cooperation. Familiar surroundings and a simpler recording scheme make it likely that pulse oximetry may be successfully performed in a child's home rather than requiring the child to come to the sleep laboratory. It also seems likely that serial evaluation for OSDB will eventually be recommended for both younger and older children with trisomy 21, and a less expensive diagnostic test is an attractive option.
Ten percent of our subjects with a total AHI ≥ 2.5 had more central apneas than obstructive events. In general, the falls in SpO2% for these events among subjects with McGill score of 2 were greater than during typical short central apneas occurring after a sigh breath. Central apnea during sleep in trisomy 21 has been described in case reports, though its occurrence is not understood or widely appreciated.19,20 Obviously, a significant number of patients with desaturations but predominately central apneas and few obstructive events might present problems when using the McGill Score to devise treatment protocols to address OSDB. Baldassari et al. showed significant reduction in central apneas following tonsillectomy in children with both OSA and mild central apnea, a study that included some children with trisomy 21.21 In contrast, Clark et al. reported Cheyne-Stokes-like central apneas that persisted despite correction of the obstruction in a small cohort of subjects with trisomy 21 and OSDB.19 While noninvasive ventilation with a mandatory rate might be expected to improve ventilation, it is not known whether further investigation, for example, with brain imaging to evaluate for abnormalities akin to Chiari malformation, is also necessary.
In our cohort, only 1 of 17 subjects with a McGill score of 3 or 4 had a significant number of central events but fewer than 2.5 obstructive events per hour. Seven of 43 (16.3%) children with McGill score of 2 had more central apneas than obstructive events; 3 of these 7 also had periodic breathing (0.2%, 2.4%, and 6.9% of total sleep time), perhaps suggesting a propensity among these children for immature ventilatory control. The physiologic basis and the clinical importance of these central events deserve further investigation. The relative lack of variability in the amplitude of the pulse oximetry waveform during central events compared to obstructive events in infants, as suggested by Wertheim et al. may allow subjects with predominantly central events to be identified in future studies evaluating oximetry as a screening tool for SDB in trisomy 21.22
Based on these early results, we propose that a child with trisomy 21 and a McGill Score of 3 or 4 warrants early ENT referral for possible surgery. If adenotonsillar hypertrophy is absent or the child has previously undergone adenotonsillectomy, we would recommend noninvasive positive pressure ventilation, although its introduction and efficacy in patients with trisomy 21 deserve more study. For patients with a McGill score of 2, with elevated BMI and adenotonsillar hypertrophy, referral to ENT is also likely warranted, as prior studies suggest that BMI may be an indicator of OSDB in children with trisomy 21.23 Patients with McGill Scores of 1 or 2, without associated obesity or adenotonsillar hypertrophy, will likely require a full overnight polysomnogram in accordance with the general AAP guidelines to establish the level of disease and distinguish obstructive and central events.
In summary, the use of overnight pulse oximetry in children with trisomy 21 may be a convenient diagnostic tool to evaluate for obstructive sleep disordered breathing and help guide the process for referral for surgical intervention, both for them and for other groups of children who have a high prevalence of OSDB but for whom laboratory testing is challenging and often a burden for their parents. Future prospective studies are needed to evaluate the sensitivity of the test and its ability to serve as a screening tool to simplify the post-surgical follow-up.
This was not an industry supported study. Funding for data analysis was provided through the Just-In-Time Core Usage Funding Program of the Washington University in Saint Louis Institute of Clinical and Translational Sciences (ICTS ID# JIT259). Dr. Grange receives financial grant/research support from Biomarin Pharmaceuticals, Inc and Edimer Pharmaceuticals, Inc, neither of which are related to this manuscript. Dr. Molter is a consultant for Shire Pharmaceuticals; this relationship is unrelated to this manuscript. The other authors have indicated no financial conflicts of interest. The work was performed at Washington University School of Medicine in Saint Louis, St. Louis Children's Hospital.