Association among craniofacial morphology, ethnicity, and risk of pediatric sleep-related breathing disorders: A multicenter study

Introduction

Craniofacial morphology and ethnicity may be risk factors for sleep-related breathing disorder (SRBD) in children but have not yet been assessed in an international multicenter study. The objectives of this study were to assess the association among craniofacial features, self-reported ethnicity, and risk of SRBD in children undergoing orthodontic treatment.

Methods

Children aged 5-18 years who presented for orthodontic evaluation were enrolled in the United States, South Africa, South Korea, Saudi Arabia, and Japan. The risk of SRBD was defined as answering ≥0.33 positive responses to the Pediatric Sleep Questionnaire. Craniofacial features included measurements in sagittal and vertical dimensions to evaluate the cranial base, maxillomandibular and dental relationships, and nasopharyngeal airway dimensions. Logistic regression was performed to assess the association among craniofacial features, ethnicity, age, body mass index, and risk of SRBD.

Results

Data were obtained from 602 patients from 5 sites. A total of 76 patients (12.6%) had a risk of SRBD. The mean age was 12.5 years. Male gender (odds ratio [OR], 1.9; 95% confidence interval [CI], 1.0-3.4; P = 0.041), Middle Eastern ethnicity (OR, 10.2; 95% CI, 4.1-25.4; P = 0.001), body mass index (OR, 1.1; 95% CI, 1.04-1.10; P = 0.001), gonial angle (OR, 0.91; 95% CI, 0.85-0.98; P = 0.011), and inferiorly positioned hyoid (OR, 1.1; 95% CI, 1.0-1.2; P = 0.002) were significantly associated with the risk of SRBD.

Conclusions

In an ambidirectional cohort study across 5 sites, male gender, Middle Eastern ethnicity, body mass index, gonial angle, and inferiorly positioned hyoid were associated with the risk of SRBD in children undergoing orthodontic treatment.

Highlights

•

Older children with higher body mass index (BMI) are at risk for sleep-related breathing disorder (SRBD).
•

Children with retrusive jaws, hyperdivegence, retroclined incisors are at risk for SRBD.
•

In Middle Eastern children with SRBD, BMI, maxillary height, and Hy-MP were higher.

Sleep-related breathing disorder (SRBD) is characterized by upper airway dysfunction during sleep that results in apneas, hypopneas, respiratory effort-related arousals, or hypoventilation. ^, Rather than being a single defined illness, SRBD is a syndrome that occurs along a spectrum of severity, ranging from milder primary snoring to more serious obstructive sleep apnea (OSA), in which one or more segments of the extrathoracic airway collapse during sleep. SRBD, as estimated via the Pediatric Sleep Questionnaire (PSQ), affects 9.7% of children aged 2-16 years. Pediatric OSA impacts 1-5% of children worldwide and can occur at any age but appears most commonly between the ages of 2 and 6 years. In OSA, airway collapse is caused by an anatomically small upper airway and decreased neuromuscular tone of the pharyngeal dilator muscles relative to normal controls. Children with SRBD manifest deleterious changes to physiology, including impaired growth, systemic or pulmonary hypertension, ^, and myocardial remodeling. ^, Untreated pediatric OSA is also associated with cardiovascular complications, hypertension, cerebrovascular disease, and neurobehavioral problems, such as attention deficit disorder and poor academic performance. ^, Consequently, children with OSA use 226% of health care resources than do controls. Thus, SRBD warrants apt screening and diagnosis by clinicians to facilitate early intervention, which can include orthodontic treatment.

The severity and expression of untreated SRBD vary by ethnicity in adults; some studies have investigated if this variation is mediated by craniofacial features. Mehra et al, recruited patients of different ethnic backgrounds who resided in the United States and reported that the risk of SRBD, as defined as having a respiratory disturbance index score ≥15, was higher in Asians than in whites, African-Americans, and Hispanics, despite lower rates of obesity. Chen et al, also observed a higher risk of SRBD in Chinese and Hispanics than in whites after adjustment for sex, age, and study site. However, these studies evaluated adult patients residing within the United States. The generalizability of these findings to children in other communities around the world remains to be demonstrated. To our knowledge, no study has simultaneously assessed both SRBD risk and craniofacial features, specifically in diverse pediatric populations across multiple ethnicities from different international sites. Hence, the objective of this study was to investigate the associations between SRBD risk (as assessed by PSQ score) and craniofacial anatomic features (as assessed by cephalometric measurements) in children of different ethnicities (white, African, Hispanic, Asian, and Middle Eastern) from 5 global clinical sites.

Material and methods

In this ambidirectional study, we assessed children who presented for orthodontic screening at Boston University (Boston, Mass), 1 Military Hospital (Pretoria, South Africa), Yonsei University (Seoul, South Korea), Nippon Dental University (Tokyo, Japan), and King Abdul Aziz University (Jeddah, Saudi Arabia). This study was approved by the Boston University Medical Center Institutional Review Board under protocol (no. H-39464). Each institution obtained approval from its respective Institutional Review Board. At the start of routine orthodontic treatment, the parent or guardian was asked to complete a demographic and medical history form and PSQ on behalf of their child.

The sample size was estimated using χ ² statistics (version 3.1.9.4; G∗ Power, Franz Faul University, Kiel, Germany) for a 5 × 2-factor analysis (5 self-reported ethnic groups consisting of white, African, Hispanic, Asian, and Middle Eastern, which culminate in 4 degrees of freedom) and a binary outcome (presence or absence of risk of SRBD). Power was set to 80%, and α was set to 0.05. A conservative effect size was estimated by setting Cohen’s w to 0.3. The required sample size was 133 patients per ethnic group. We assumed ethnic groups were concentrated at particular sites and thus aimed to enroll 133 patients per site; with 5 sites, we set a goal of enrolling 665 total patients.

Inclusion criteria were children aged 5-18 years who were screened for their first orthodontic treatment. Exclusion criteria included (1) a history of chronic disease such as diabetes, heart disease, and/or cancer; (2) a history of trauma and/or surgery that could affect occlusion; (3) diagnosed respiratory conditions such as asthma, respiratory allergies, tonsillitis, or nasal septal deviation; (4) a history of adenotonsillectomy or nasal surgery; and (5) a history of craniofacial anomalies (eg, cleft lip and/or palate, Down, Crouzon, hemifacial microsomia, Treacher–Collins, or Pierre–Robin syndrome). The following data were collected for each patient: medical history, biological sex, self-identified ethnicity, height, and weight. The 22-item PSQ was given to the parent or guardian as a routine part of the initial orthodontic records examination that was performed for all new patients. The PSQ is used to assess the child’s snoring frequency, loud snoring, daytime sleepiness, observed apneas, difficulty breathing during sleep, inattentive or hyperactive behavior, and other pediatric SRBD features. If a patient answers >33% of the PSQ affirmatively, they are considered at risk for SRBD. The other patients served as controls. Pretreatment cephalometric radiographic images obtained during routine orthodontic screening were extracted from the dental record. Lateral cephalograms were digitally traced by the same operator using OrthoDx software (Phimentum, Boston, Mass) to ensure consistency.

The cephalometric analysis included common linear and angular measurements to evaluate the cranial base, hyoid-mandibular, maxillomandibular, and vertical-skeletal relationships, dental measurements, and nasopharyngeal and oropharyngeal airway dimensions. These measurements and their respective definitions are depicted in the Figure . Intrarater reliability was assessed by randomly selecting and retracing 60 cephalograms from across all sites; the mean intraclass correlation coefficient of 20 cephalometric measurements differed <2% between repeat and original analyses. Wits appraisal differed by an average of <10% because of the challenges posed by some radiographs that had superimposed molars in the mixed dentition stage.

Data analyses, visualization, and statistical tests were conducted using Python 3.8, Pandas, Seaborn, and StatsPy packages. Within each ethnicity, a univariate analysis was performed to compare cephalometric measurements in patients at risk of SRBD vs control patients. Multivariate logistic regression was conducted to estimate the effect of age, body mass index (BMI), ethnicity, and select craniofacial features on the risk of SRBD. Craniofacial features that significantly differed in univariate analysis of SRBD vs those of healthy controls were included in the subsequent logistic regression analysis, along with age, gender, and ethnicity as covariates. Statistical significance was defined as P <0.05.

Results

A total of 662 patients from 5 distinct sites and countries met the inclusion criteria. Only 1 white patient met the criteria for SRBD; thus, all 60 white patients were removed from subsequent analyses because of insufficient statistical power. Consequently, Asian ethnicity was used as the reference for subsequent statistical analyses of ethnicity. The final cohort for analysis comprised 602 patients. We enrolled 65 retrospective subjects from Boston University during the 2017-2019 time period and 537 prospective subjects from all sites during the 2019-2022 time period. A total of 341 patients were females (57%) and 261 were males (43%) ( Table I ). The mean age was 12.5 ± 2.7 years. 76 (12.6% of 602) patients were at risk of SRBD. 37 (11%) of females and 39 (15%) of males were at risk of SRBD ( Table I ). The ethnicity with the highest number of total patients was Asian (n = 210), and the highest prevalence of risk for SRBD was Middle Eastern (n = 50 of 135; 37% prevalence) ( Table I ). The distribution of patients per site was 150 from Boston University (United States), 133 from King Abdul Aziz University (Saudi Arabia), 126 from Yonsei University (South Korea), 125 from 1 Military Hospital (South Africa), and 68 from Nippon Dental University (Japan) ( Table I ).

Table I

Subjects stratified by sex, ethnicity, and site, as well as divided into risk of SRBD

Variables	SRBD, n (%)	Healthy, n (%)	Total (n)
Sex
Male	39 (14.9)	222 (85.1)	261
Female	37 (10.9)	304 (89.1)	341
Total	76 (12.6)	526 (87.4)	602
Ethnicity
Hispanic	6 (10.3)	52 (89.7)	58
African	7 (3.5)	192 (96.5)	199
Asian	13 (6.2)	197 (93.8)	210
Middle Eastern	50 (37.0)	85 (63.0)	135
Total	76 (12.6)	526 (87.4)	602
Site
Boston University (United States)	13 (8.7)	137 (91.3)	150
King Abdul Aziz University (Saudi Arabia)	49 (36.8)	84 (63.2)	133
Nippon Dental University (Japan)	4 (5.9)	64 (94.1)	68
Yonsei University (South Korea)	8 (6.3)	118 (93.7)	126
1 Military Hospital (South Africa)	2 (1.6)	123 (98.4)	125
Total	76 (12.6)	526 (87.4)	602

Note. The total in each subcategory is the sum of subjects at risk for SRBD and healthy controls.

Moreover, 21 cephalometric measurements, age, and BMI were obtained from all patients and stratified by SRBD vs healthy via the Mann–Whitney U test. Among the cephalometric measurements, SNA angle, SNB angle, gonial angle, lower gonial angle, U1 to SN angle, IMPA, SN/GoGn, lower facial height, posterior-to-anterior face height ratio, Hy-Mp, and PNS-AD1 significantly differed in patients at risk of SRBD vs in controls via univariate analyses ( Table II ). In the Asian patient population, only 13 of the 210 were considered at risk of SRBD. In the African patient population, 7 of the 199, and in the Hispanic group, 6 of the 58 were considered at risk of SRBD. The small number of patients at risk of SRBD within each of these 3 groups did not allow an accurate analysis of the differences without introducing potential errors. However, in the Middle Eastern group, 50 of the total 135 patients included were at risk of SRBD. In this group, BMI score was higher ( P <0.05); maxillary height was higher (59.48 ± 3.52 vs 58.26 ± 3.75; P = 0.035); and Hy-Mp distance was higher (16.49 ± 4.51 vs 13.38 ± 4.50; P <0.001) in patients at risk of SRBD vs in controls ( Table III ).

Table II

Covariates in all patients at risk for SRBD vs in healthy controls

Characteristics	SRBD (n = 76)	Control (n = 525)	P value
Age (y) ^∗	13.30 ± 1.90	12.39 ± 2.80	0.004
BMI ^∗	22.90 ± 7.44	20.29 ± 4.78	0.003
Facial depth (°)	88.19 ± 4.93	88.12 ± 4.16	0.869
SNA (°) ^∗	79.87 ± 4.94	81.45 ± 4.85	0.023
SNB (°) ^∗	77.27 ± 4.96	78.62 ± 4.89	0.034
ANB (°)	2.60 ± 2.99	2.83 ± 3.11	0.561
Wits (mm)	−0.24 ± 5.20	−0.35 ± 4.74	0.769
Anterior cranial base length (mm)	64.28 ± 5.33	63.22 ± 8.58	0.259
Maxillary height (°)	60.25 ± 4.73	59.50 ± 4.83	0.159
Gonial angle (°) ^∗	130.95 ± 7.65	129.35 ± 6.06	0.032
Upper gonial angle (°)	52.34 ± 4.32	52.85 ± 4.23	0.235
Lower gonial angle (°) ^∗	78.61 ± 6.27	76.49 ± 4.99	0.001
Mandibular corpus length (mm)	64.76 ± 7.60	64.20 ± 9.83	0.329
U1-SN (°) ^∗	84.36 ± 5.03	86.14 ± 5.39	0.013
IMPA (°) ^∗	81.73 ± 6.15	85.13 ± 6.37	0.001
Facial axis (°)	88.85 ± 5.60	89.71 ± 5.20	0.152
SN/GoGn (°) ^∗	36.02 ± 7.43	33.55 ± 6.45	0.003
Lower anterior facial height (°) ^∗	51.37 ± 6.14	49.86 ± 4.65	0.04
Posterior-to-anterior facial height (%) ^∗	61.84 ± 5.60	63.29 ± 5.26	0.028
Hy-Mp (mm) ^∗	15.90 ± 5.17	12.81 ± 5.16	0.001
PNS-AD1 (mm) ^∗	21.65 ± 4.11	22.56 ± 4.15	0.028
AD1-Ba (mm)	19.72 ± 3.31	19.72 ± 4.07	0.783
PNS-Ba (mm)	40.62 ± 5.24	41.55 ± 6.03	0.085