Nighttime Respiratory Disturbance and Hypoxemia Associated with Impaired Linear Growth in Children with Obstructive Sleep Apnea

Abstract

Objectives

Although obstructive sleep apnea (OSA) related hypoxia shows stronger associations with multiple complications than respiratory disturbance, its impact on linear growth, a major pediatric concern, remains unclear. This study investigates independent effects of hypoxia and respiratory disturbance on the height of children with OSA.

Methods

This retrospective, cross-sectional study clinically evaluated children with suspicious OSA from January 2013 to June 2025. Individuals with other sleep disorders, severe neuropsychiatric disorders and physical disorders were excluded. Subjects underwent overnight polysomnography (PSG) and height measurement. OSA was defined as apnea-hypopnea index (AHI) ≥1 event/hour, and those with AHI <1 event/hour served as the non-OSA reference group. AHI and percentage of total sleep time with oxygen saturation < 90% (T90%), were separately used to evaluate the severity of respiratory disturbance (mild, 1≤AHI < 5 events/h; moderate, 5≤AHI < 10 events/h; severe, AHI ≥10 events/h) and hypoxia (T90% quartiles). Height z-scores ≤-1 SD defined subclinical growth impairment. Linear and logistic regression models analyzed the independent associations between AHI and T90% and height and subclinical growth impairment. Nutrition, endocrine disorders, chronic illnesses, and parental height were not collected and adjusted in this study.

Results

A total of 2443 children were included (348 children without OSA, 2095 children with OSA). In linear regression models, higher T90% was associated with a lower height z-score in the full sample (B = −0.005, 95% CI: −0.010 to −0.000, P = 0.033) and children with OSA (B = −0.005, 95% CI: −0.010 to −0.000, P = 0.034), whereas AHI was not (full sample: B = −0.003, 95% CI: −0.006 to 0.001, P = 0.111; children with OSA: B = −0.002, 95% CI: −0.005 to 0.002, P = 0.330).Logistic models showed higher odds of subclinical growth impairment in children with moderate (OR = 1.73, 95% CI: 1.05–2.84) or severe OSA (OR = 1.96, 95% CI: 1.19–3.21) and in those in the highest T90% quartile (OR = 2.17, 95% CI: 1.30–3.60), compared with children without OSA. These associations remained stable in females and children aged 2–9 years.

Conclusions

To our knowledge, this is the first study to show that greater respiratory disturbance and nocturnal hypoxemia are associated with a higher risk of short stature in children with OSA. These findings support routine height assessment in the clinical management of pediatric OSA.

Introduction

Growth and development are influenced by multiple factors including nutritional status, obesity, endocrine disorders, chronic illnesses, pubertal status, gene and socioeconomic status. In addition to these established determinants, sleep is an indispensable physiological requirement for children’s growth and development, and beneficial role of healthy sleep in height growth has been widely demonstrated. A substantial body of research evidence has confirmed that the secretion pattern of growth hormone (GH), a key hormone directly associated with height growth, is closely linked to the sleep cycle.1–3 It has long been recognized that healthy sleep, including sufficient sleep duration and quality, is a necessary factor in ensuring optimal height attainment in children. A recent study further elucidated the neuroendocrine circuit through which sleep promotes height growth,4 reigniting discussions on the relationship between sleep and childhood height development. This recent finding prompts us to further consider whether various types of sleep disturbances in children—along with their associated disruptions in sleep duration and quality—may increase the risk of impaired height growth.

Obstructive sleep apnea (OSA) is one of the most common sleep disorders in children, with a prevalence of approximately 1–2%, demonstrating peak incidence in preschool years.5 OSA is characterized by recurrent partial or complete upper airway obstruction during sleep, causing nocturnal oxygen desaturation and sleep fragmentation. The clinical manifestations of pediatric OSA differ from those of adult OSA. Excessive daytime sleepiness, a classical adult symptom, is less commonly reported in children, while many children with OSA can present with behavioral neurocognitive symptoms and growth disturbance.6–8 Considering linear growth as one of the most closely monitored physiological phenomena during childhood, among those complications, growth disturbance has garnered significant attention. Although OSA-related height impairment has been well-documented in major textbooks and guidelines,9–11 the direct evidence on how OSA impairs the height of children remains unclear.

Our previous study has demonstrated that slow wave sleep reduction was significantly associated with impaired linear growth in children with OSA.12 Although slow-wave sleep is temporally associated with sleep-dependent secretion of GH and insulin-like growth factor-1 (IGF-1),13,14 the mechanisms linking OSA to impaired linear growth are comprehensive and multifactorial. Beyond changes in sleep architecture, respiratory disturbance and nocturnal hypoxemia, as the core pathophysiological processes of OSA, may further impair linear growth by reducing GH/IGF-1 levels and inhibiting growth-plate chondrocyte proliferation and osteogenic differentiation.15 However, the independent associations of respiratory disturbance and nocturnal hypoxemia with height in children with OSA remain unclear. The apnea-hypopnea index (AHI) is the gold standard metric for OSA diagnosis and severity classification. Nevertheless, previous study demonstrated that hypoxia index (eg, oxygen desaturation index (ODI), percentage of total sleep time with oxygen saturation < 90% (T90%)) exhibited stronger associations than AHI with OSA-related comorbidities or complications, including hypertension, diabetes, mild cognitive impairment, and even cancer,16–20 which has not been sufficiently validated regarding height impairment of children with OSA. Current research on OSA and childhood height relationships focuses on assessing the severity of growth restriction and the prevalence of short stature among pediatric OSA patients.21–24 Only one study investigated the correlations between respiratory events and hypoxia parameters and height in children with OSA, yet without adequate adjustment for potential confounders.25 Elucidating the independent risk factors affecting height in pediatric OSA populations would provide crucial evidence for developing targeted prevention and intervention strategies for height impairment in those children.

Our study addresses this gap by analyzing PSG-measured respiratory events and hypoxia parameters (AHI and T90%) and physical examinations of children (height) in a large opportunistic clinical sample collected over a 12-year period, with adjustment for potential confounders. We hypothesized that in children with OSA, T90% demonstrates a stronger association with height than the AHI, and is significantly related to increased subclinical growth impairment.

Material and Methods

Study Design and Setting

This was a cross-sectional, observational study conducted at Sleep Medicine Center, West China Hospital.

Participants

All subjects visited the Sleep Medicine Center due to clinical suspicion of obstructive sleep apnea (OSA, primary symptoms including snoring, mouth breathing, hypertrophy of tonsils, etc). Well-trained staff additionally collected information on general health, medication use, and sleep complaint history via a comprehensive questionnaire.

We excluded subjects who had (1) duplicate data; (2) incomplete height or weight information; (3) other sleep disorders (eg, central disorder of hypersomnia, parasomnias, insomnia, sleep movement disorders and others); (4) severe neuropsychiatric disorders (eg, epilepsy, depression, anxiety, schizophrenia, Tourette’s syndrome, and others); (5) physical disorders (eg, other respiratory diseases). The flow chart of inclusion and exclusion of subjects is provided in Figure 1.

Figure 1.

Flow chart of the study.

Abbreviations: PSG, polysomnography; AHI, apnea-hypopnea index; OSA, obstructive sleep apnea.

Data Collection

Anthropometric measurements were performed by a trained physician on the night of PSG recording. Standing height was measured without shoes using a stadiometer, with the participant standing upright, and recorded to the nearest 0.5 cm. Height and body mass index (BMI) z-scores were calculated using the World Health Organization (WHO) 2006 Child Growth Standards26 and Cole’s LMS method27 to enable age- and sex-adjusted comparisons across the study sample.

Polysomnography (PSG)

An overnight PSG was conducted in our sleep laboratory (using the Alice 5 Diagnostic Sleep System prior to 2016 and the Alice 6 thereafter; Philips Respironics Bend, OR) by well-trained sleep technicians using the standardized procedures. Participants followed their habitual sleep schedule, with the recording time ranging from 21:00-22:00 to 06:00-07:00.

Sleep recordings were scored according to the latest version of the American Academy of Sleep Medicine (AASM) Manual for the Scoring of Sleep and Associated Events applicable in the corresponding year. The respiratory events were scored based on AASM pediatric criteria. AHI is defined as the average number of apnea and hypopnea events per hour of sleep. Pediatric obstructive apnea was scored when there was a ≥90% reduction in airflow measured by an oronasal thermal sensor, lasting for at least two breaths with continued respiratory effort. Pediatric hypopnea was scored when airflow decreased by ≥30% from baseline for at least the duration of two breaths accompanied by an arousal or a ≥3% oxygen desaturation.28

Severity of Respiratory Disturbance and Hypoxia Definitions

OSA was defined as AHI ≥1 event/hour, and those with AHI < 1 event/hour served as the non-OSA reference group.29–33 Respiratory disturbance and hypoxia were assessed using AHI (mild, 1≤AHI < 5 events/h; moderate, 5≤AHI < 10 events/h; severe, AHI ≥10 events/h) and T90% quartiles (Q1, T90% < 0.039; Q2, 0.039 < T90% < 0.045; Q3, 0.045 < T90% < 4.30; Q4, T90% ≥ 4.30). T90% was selected as the primary indicator of nocturnal hypoxia because it reflects the cumulative hypoxic burden—the total duration and depth of oxygen desaturation throughout the night—which is conceptually more aligned with our study aim of assessing the association between nocturnal hypoxia exposure and impaired linear growth. Unlike ODI or minimum SpO₂, T90% captures the overall hypoxic exposure rather than discrete events or extreme values, and has been shown to better predict adverse clinical outcomes in OSA populations.34,35

Outcome Measures

Short stature was commonly defined as a height z-score ≤ −2 SD. However, only 1.09% of children in our sample met this criterion, limiting statistical power. Therefore, a milder term, “subclinical growth impairment”, defined as height z-score ≤ −1 SD, was used to identify a broader group of children with height impairment. This cutoff threshold is described in previous studies as “mild stunting” or “at risk of stunting”.36,37 In the present study, this cutoff was used for analytical classification rather than as a clinical diagnosis of short stature.

Statistical Analysis

Sample characteristics comparisons between or among groups were examined using analysis of variance (ANOVA), independent-sample t-tests, or Mann–Whitney U-tests for normally distributed and skewed continuous variables, respectively. Chi-square tests were utilized for categorical variables.

Linear regression models were used to explore the association between height z-score and AHI and T90% in all participants and children with OSA. When examining the association between height z-score and AHI, we performed linear regression with sequential adjustment for covariates across three models. Candidate covariates were selected based on their clinical relevance based on previous studies as potential confounders14,38,39 and observed differences between the OSA and non-OSA groups. Before model fitting, we assessed multicollinearity among these variables, as reported in Table S1 of the Supplementary Materials. When substantial multicollinearity was identified, representative variables were retained to improve model stability. Model 1 was adjusted for demographic covariates: sex, age, and BMI z-score. Model 2 further adjusted for sleep architecture parameters: total sleep time (TST), arousal index (AI), percentage of non-rapid eye movement sleep stage 1(N1%), percentage of non-rapid eye movement sleep stage 3 (N3%), percentage of rapid eye movement sleep stage (REM%), while retaining all covariates from model 1. Model 3 additionally adjusted for nocturnal oxygen variables: mean oxygen saturation and minimum oxygen saturation, while retaining all covariates from model 1 and model 2. When exploring the association between height z-score and T90%, we also performed linear regression with sequential adjustment for covariates across three models. Model 1 was adjusted for demographic covariates: sex, age, and BMI z-score. Model 2 further adjusted for sleep architecture parameters: TST, AI, N1%, percentage of non-rapid eye movement sleep stage 2 (N2%), REM%, and AHI. Model 3 additionally adjusted for respiratory disturbance variable: AHI.

Independent associations between AHI and T90% and subclinical growth impairment were assessed by logistic regression models, using children without OSA as reference, and the same covariates as mentioned above were included in logistic regression models. The same logistical regression models were used in different age and sex subgroups. In the age-stratified subgroups, subjects were categorized as children (2–9 years) and adolescents (10–18 years).40 For subgroup analyses stratified by sex and age group, formal tests of interaction were performed by introducing multiplicative interaction terms (subgroup × each predictor) into the logistic regression models. The P values for interaction were obtained using likelihood ratio tests, comparing models with and without the interaction term. All subgroup analyses were considered exploratory; therefore, results were interpreted with caution, and primary conclusions were based on the main effect models.

Statistical analyses were performed with SPSS version 29.0. All statistical tests were two-sided, and P values of less than 0.05 were considered statistically significant.

Although a priori sample size calculation is not routinely performed for observational studies using existing clinical data, we conducted a post‑hoc power analysis to evaluate whether the final sample provided sufficient statistical power. For the linear regression model examining the association between AHI and height z‑score (11 predictors, overall multiple partial correlation = 0.154), and for the model examining T90% and height z‑score (10 predictors, overall multiple partial correlation = 0.150), the achieved power was 1.000 with α = 0.05 (n = 3866). These results indicate that the study was adequately powered to detect the associations of interest.

Ethical Considerations

This study complies with the Declaration of Helsinki. Ethical approval was granted by the University Review Board (NO. 1232). Prior to PSG monitoring, each participant (aged 2–18 years) and their guardian received standardized verbal information regarding the study objectives and the use of anonymized clinical data. Enrollment required agreement from both the participant and guardian, confirmed by a signed consent form. Only data from consented participants were included in the analyses.

Results

The sample comprised 2443 subjects, of whom 2095 were patients with OSA. Among all the participants, 66.4% were male. The mean age was 8.52±4.42 years, the mean BMI z-score was 0.34±1.42, and the mean height z-score was 0.19±0.99. Demographic and sleep characteristics of this sample of non-OSA and children with OSA are provided in Table S2. Compared to children without OSA, those with OSA had a higher proportion of males, SE, REM%, AI, AHI, T90%, and ODI, and lower age, height z-score, WASO, N3%, mean oxygen saturation, and minimum oxygen saturation (all P value <0.05). Children without OSA and children with OSA showed comparable BMI, TST, SOL, N1%, and N2% (all P value ≥0.05). The demographic and sleep characteristics of children with OSA stratified by AHI and T90% are provided in Table S3. Children with OSA with more severe respiratory events and hypoxia were younger, had lower height z-score, N2%, REM%, and higher AI.

As shown in Table 1, the inverse association between AHI and height z-score was significant in Model 1 but attenuated in Model 2. In the fully adjusted model (Model 3), AHI was no longer significantly associated with a lower height z-score in either the full sample (B = −0.003, 95% CI: −0.006 to 0.001, P = 0.111) or children with OSA (B = −0.002, 95% CI: −0.005 to 0.002, P = 0.330). In contrast, higher T90% was consistently associated with a lower height z-score across all models. In the fully adjusted model (Model 6), the association remained significant in both the full sample (B = −0.005, 95% CI: −0.010 to −0.000, P = 0.033) and children with OSA (B = −0.005, 95% CI: −0.010 to −0.000, P = 0.034).

Table 1.

Linear Regression Models Estimating the Association of Height Z-Score with AHI, and the Association of Height Z-Score with T90%

	Association of Height z-Score with AHI			Association of Height z-Score with T90%
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
All children (n=2443)
B (95% CI)	−0.006 (−0.009 to −0.003)	−0.004 (−0.008 to −0.001)	−0.003 (−0.006 to 0.001)	−0.008 (−0.012 to −0.003)	−0.006 (−0.011 to −0.002)	−0.005 (−0.010 to 0.000)
P	<0.001	0.011	0.111	<0.001	0.006	0.033
Children with OSA (n=2095)
B (95% CI)	−0.004 (−0.007 to −0.001)	−0.003 (−0.006 to 0.001)	−0.002 (−0.005 to 0.002)	−0.007 (−0.012 to −0.003)	−0.006 (−0.011 to −0.001)	−0.005 (−0.010 to 0.000)
P	0.004	0.094	0.330	0.002	0.012	0.034

Notes: Model 1 and Model 4 were adjusted for sex, age, and BMI z-score. Model 2 was adjusted for variables included in Model 1, total sleep time, arousal index, N1%, N3% and REM%. Model 3 was adjusted for variables included in Model 2, mean and minimum SpO2. Model 5 was adjusted for variables included in Model 4, total sleep time, arousal index, N1%, N2% and REM%. Model 6 was adjusted for variables included in Model 5 and AHI. Bold font indicates the P values less than 0.05.

Abbreviations: OSA, obstructive sleep apnea; BMI, body mass index; AHI, apnea-hypopnea index; T90%, percentage of total sleep time with oxygen saturation < 90%; N1, non-rapid eye movement sleep stage 1; N2, non-rapid eye movement sleep stage 2; N3, non-rapid eye movement sleep stage 3; REM, rapid eye movement sleep stage; SpO2, oxygen saturation.

Table 2 presents the associations of AHI and T90% with subclinical growth impairment, defined as height z-score ≤ −1 SD. When using AHI to define the respiratory event severity of OSA, in fully adjusted models, compared to children without OSA, the moderate and severe groups were associated with higher odds of subclinical growth impairment (moderate: OR = 1.73, 95% CI: 1.05–2.84; severe: OR = 1.96, 95% CI: 1.19–3.21), while the mild group was not (OR = 1.44, 95% CI: 0.91–2.28; P for trend = 0.005). In the groups defined by T90% quartiles, only the highest quartile was associated with higher odds of subclinical growth impairment (OR = 2.17, 95% CI: 1.30–3.60; P for trend = 0.006), when compared to children without OSA.

Table 2.

Adjusted ORs and 95% CI for the Association Between AHI and Subclinical Growth Impairment, and Between T90% and Subclinical Growth Impairment

	n	OR (95% CI)
Children Without OSA	348	Reference	Reference	Reference
Children with OSA stratified by AHI		Model 1	Model 2	Model 3
Mild	1004	1.53 (0.97–2.41)	1.47 (0.93–2.33)	1.44 (0.91–2.28)
Moderate	499	1.94 (1.19–3.16)*	1.76 (1.07–2.89)*	1.73 (1.05–2.84)*
Severe	592	2.53 (1.57–4.06)***	2.11 (1.30–3.44)*	1.96 (1.19–3.21)*
P for trend		<0.001***	0.001**	0.005*
Children with OSA stratified by T90% quartiles		Model 4	Model 5	Model 6
Q1	522	1.49 (0.91–2.46)	1.42 (0.86–2.34)	1.39 (0.84–2.30)
Q2	525	1.82 (1.12–2.97)*	1.68 (1.03–2.75)*	1.64 (1.00–2.69)
Q3	522	1.63 (0.99–2.69)	1.50 (0.90–2.47)	1.44 (0.86–2.39)
Q4	526	2.68 (1.67–4.32)***	2.31 (1.42–3.76)***	2.17 (1.30–3.60)*
P for trend		<0.001***	0.001**	0.006*

Notes: Subclinical growth impairment was defined as height z-score ≤ −1SD in children with or without OSA., Mild, children with 1≤AHI < 5 events/h; Moderate, children with 5≤AHI < 10 events/h; Severe, children with AHI ≥10 events/h. Q1, T90% < 0.039; Q2, 0.039< T90% <0.045; Q3, 0.045< T90% <4.30; Q4, 4.30≤T90%. Model 1 and Model 4 were adjusted for sex, age, and BMI z-score. Model 2 was adjusted for variables included in Model 1, total sleep time, arousal index, N1%, N3% and REM%. Model 3 was adjusted for variables included in Model 2, mean and minimum SpO2. Model 5 was adjusted for variables included in Model 4, total sleep time, arousal index, N1%, N2% and REM%. Model 6 was adjusted for variables included in Model 5 and AHI. Bold font indicates thePvalues less than 0.05. *P<0.05; **P<0.01; ***P<0.001.

Abbreviation: OSA, obstructive sleep apnea; BMI, body mass index; AHI, apnea-hypopnea index; T90%, percentage of total sleep time with oxygen saturation < 90%; N1, non-rapid eye movement sleep stage 1; N2, non-rapid eye movement sleep stage 2; N3, non-rapid eye movement sleep stage 3; REM, rapid eye movement sleep stage; SpO2, oxygen saturation.

We also conducted subgroup analysis stratified by sex and age (Figure 2). In exploratory subgroup analyses stratified by sex and age, the patterns of association were generally consistent across strata. Formal tests of interaction did not reveal statistically significant differences between subgroups (all P for interaction > 0.05), suggesting that the observed subgroup-specific estimates should be interpreted with caution and do not support differential effects by sex or age. Significantly higher odds of subclinical growth impairment were obtained in females with mild (n = 380; OR = 2.52, 95% CI: 1.15–5.52) and severe OSA (n = 143; OR = 4.77, 95% CI: 1.75–13.01), compared with females without OSA; and in children with moderate OSA (n = 335; OR = 2.14, 95% CI: 1.09–4.22) and severe OSA (n = 438; OR = 2.74, 95% CI: 1.33–5.65), compared with children without OSA. When hypoxia severity was classified by T90% quartiles, females in Q2 (n = 170; OR = 2.39, 95% CI: 1.01–5.68), Q3 (n = 139; OR = 2.80, 95% CI: 1.15–6.80), and Q4 (n = 162; OR = 3.56, 95% CI: 1.48–8.56), as well as children in Q2 (n = 345; OR = 1.99, 95% CI: 1.01–3.90) and Q4 (n = 424; OR = 2.37, 95% CI: 1.21–4.62), had higher odds of subclinical growth impairment than their corresponding non-OSA groups.

Figure 2.

Subgroup analysis.

Notes: (A) Adjusted ORs and 95% CI for the association between AHI and subclinical growth impairment in different subgroups. (B) Adjusted ORs and 95% CI for the association between T90% and subclinical growth impairment in different subgroups. Children were defined as those aged 2–9 years, and adolescents were defined as those aged 10–18 years. The sex-stratified models were adjusted for age; age-stratified models were adjusted for sex. When examining AHI/T90% associations with subclinical growth impairment, models were adjusted for BMI z-score, sleep efficiency, arousal index, N1%, N3%, REM%, mean and minimum SpO2 (AHI models only), and AHI (T90% models only). Bold font indicates thePvalues less than 0.05. *P<0.05; **P<0.01; ***P<0.001.

Abbreviations: OR, odds ratio; CI, confidence interval; OSA, obstructive sleep apnea; BMI, body mass index; AHI, apnea-hypopnea index; T90%, percentage of total sleep time with oxygen saturation < 90%; N1, non-rapid eye movement sleep stage 1; N3, non-rapid eye movement sleep stage 3; REM, rapid eye movement sleep stage; SpO2, oxygen saturation.

Discussion

To the best of our knowledge, our study provides the first large population-based evidence showing an association between OSA and subclinical growth impairment in children, and explores factors related to the subclinical growth impairment in children with OSA. Our findings reveal that children with OSA with most severe respiratory disturbance and hypoxia had 1.96- and 2.17-fold higher odds of subclinical growth impairment relative to those children without OSA, even after comprehensive adjustment for potential confounders; and this relationship existed significantly and stably across subgroups of females and prepubertal children. The height of children with OSA was independently linearly correlated with T90% rather than the AHI. These findings support an association between pediatric OSA and linear growth and suggest that nocturnal hypoxemia may be more consistently associated with impaired height development than AHI in this clinical sample.

Growth restriction has been mentioned as one of the symptoms of children with OSA;6–11 a few studies have verified this claim. Some studies found that children with OSA had lower height and a higher growth failure rate, compared to healthy controls.21,25 In addition, some researchers have indirectly verified the impact of OSA on children’s height by exploring the height changes of children with OSA before and after tonsillectomy and/or adenoidectomy.21,23,24 To the best of our knowledge, our study is the first to employ height and PSG assessments of a large clinic-based sample of children with OSA. By excluding the impact of comorbidities, we have directly confirmed the influence of OSA on height impairment in children, especially for those with moderate and severe OSA or/and severe hypoxia OSA, whose height development should be given full attention.

Although logistic regression found an independent correlation between subclinical growth impairment and AHI and T90%, the results of linear regression revealed a significant independent correlation between height z-score and T90%, but an insignificant linear relationship with AHI. A previous study reported a significant negative linear relationship between AHI and height, as well as between hypoxia indicators and height, which is consistent with our findings.25 We speculate that differences in confounding factor adjustment may be the primary explanation for the discrepancies between our findings and these previous results, as the previous study did not fully account for the effects of confounding variables. Our fully adjusted model, which accounted for a broader range of confounding factors, strengthens the finding that height impairment in children with OSA is independently associated with hypoxia rather than with AHI.

The linear growth of children and adolescents is promoted by GH, which stimulates the synthesis of IGF-1 in the liver and other target tissues. IGF-1 supports bone growth by stimulating chondrocyte proliferation in the epiphyseal plates.41,42 Additionally, IGF-1 levels reflect the daily mean GH levels, correlate with physiological changes in GH secretion and are considered as a major mediator of the growth-promoting effects of GH.43 The chronic intermittent associated with to OSA may negatively affect GH–IGF-1 axis, potentially contributing to lower GH and IGF-1 concentrations, while inflammatory mediators may be also involved in this pathway This speculation is consistent with a previous study showing that the release of inflammatory factors (eg, IL-5) regulates GH levels to different degrees, thus affecting the growth and development of children.22 Other studies have reported lower circulating levels of insulin-like growth factor-binding protein 3 (IGFBP-3) and higher levels of inflammatory markers, including C-reactive protein (CRP), in children with OSA than in healthy controls.44,45 Although supported by relevant studies, the role of the GH–IGF-1 axis and inflammatory pathways in the association of OSA-related respiratory disturbance and nocturnal hypoxemia with childhood height remains hypothetical, as these biomarkers were not measured in the present study. Future studies incorporating GH, IGF-1, and inflammatory cytokines are needed to clarify their potential roles in this association.

Subgroup analyses stratified by sex and age showed generally consistent associations across strata. Formal interaction tests did not reveal statistically significant heterogeneity between subgroups (all P for interaction > 0.05), indicating no strong evidence of effect modification by sex or age. These exploratory findings suggest that the observed associations between OSA-related respiratory disturbance, hypoxia, and short‑stature risk are robust across sexes and age groups.

Our findings support important clinical and public health education considerations. Textbooks and guidelines for children with OSA are suggested to emphasize the impact of OSA on growth and development, especially its adverse effect on the height of children before puberty. In clinical practice, greater attention to linear growth may be warranted in children with OSA. Periodic height assessment, for example every 6–12 months, may be considered. A decline in height z-score of >0.5 SD or a height z-score falling below −1 SD could prompt closer growth evaluation and consideration of referral to a pediatrician. Future prospective longitudinal studies are needed to validate the clinical utility of these proposed monitoring intervals and referral thresholds. For the treatment of OSA, clinicians should follow up on the improvement of respiratory disturbance and hypoxia related to OSA as well as the growth of height. Among children with impaired growth and development, it is recommended to screen for the existence of OSA, especially in those with concurrent risk factors of OSA, such as tonsillar and/or adenoid hypertrophy, obesity, etc.

Our study has several strengths, including its large sample size, drawn from consecutive clinical cases, and its structured assessment of height and use of PSG, and its comparison of OSA to children without OSA. Moreover, this is the first study to report direct evidence of the association between OSA and the height of children after fully adjusting for potential confounders. Nonetheless, our study has limitations. Due to the low prevalence of children meeting the diagnostic criteria for short stature (only 1%) in our study population, we employed the subclinical growth impairment rather than a binary diagnosis of short stature. Although the subclinical growth impairment cannot be directly used for clinical disease evaluation, it provides clinical value as an early indicator reflecting disease risk. Because participants were referred to a single sleep center for OSA-related symptoms, selection and referral biases cannot be excluded. This clinical sample also limited the generalizability of the findings to community-based or those evaluated in other clinical settings. Due to the observational design of our study, we cannot establish definitive causal relationships between OSA and impaired height growth and reverse causality cannot be excluded. Although we adjusted for numerous potential confounders, unmeasured factors such as nutritional status, physical activity, parental height, and genetic predisposition may still influence the results, resulting in residual confounding. Night-to-night variation and a first-night effect on the measures of interest cannot be ruled out by the one-night PSG protocol.

Conclusion

This study presents evidence that severe respiratory disturbance and greater nocturnal hypoxemia were associated with higher odds of subclinical growth impairment in children with OSA, with nocturnal hypoxemia showing a more consistent association with height z-score. These findings support attention to linear growth and routine height assessment in the clinical management of pediatric OSA, particularly in children with marked nocturnal hypoxemia. Future prospective and longitudinal studies conducted across multiple centers or in community-based populations are needed to confirm these findings and clarify the value of routine height assessment in pediatric OSA care.

Funding Statement

This work was supported by the Ministry of Science and Technology of the People’s Republic of China (2021ZD0201900), National Natural Science Foundation of China (82120108002 and U21A20335).

Abbreviations

GH, growth hormone; OSA, obstructive sleep apnea; AHI, apnea-hypopnea index; ODI, oxygen desaturation index; T90%, percentage of total sleep time with oxygen saturation < 90%; BMI, body mass index; WHO, World Health Organization; PSG, polysomnography; AASM, American Academy of Sleep Medicine; TST, total sleep time; AI, arousal index; N1%, percentage of non-rapid eye movement sleep stage 1; N3%, percentage of non-rapid eye movement sleep stage 3; REM%, percentage of rapid eye movement sleep stage; N2%, percentage of non-rapid eye movement sleep stage 2; IGF-1, insulin-like growth factor-1; PA, physical activity; MVPA, moderate-to-vigorous physical activity.

Data Sharing Statement

The ethical approval of this trial does not allow for the sharing of the individual-level data that supports the findings. However, synthetic data are available from the corresponding authors on reasonable request.

Ethics Approval and Informed Consent

This study complies with the Declaration of Helsinki and informed consent was obtained for all participants and their parents. The West China Hospital Research Ethics Board has approved the study (NO. 1232).

Consent for Publication

Not applicable.

Author Contributions

Yuan Shi: Conceptualization; Methodology; Software; Data curation; Investigation; Validation; Formal analysis; Visualization; Writing – original draft; Writing – review & editing. Zhaohua Chen: Conceptualization; Methodology; Software; Data curation; Investigation; Validation; Formal analysis; Visualization; Writing – original draft; Writing – review & editing. Xujun Feng: Conceptualization; Methodology; Software; Data curation; Investigation; Validation; Writing – original draft; Writing – review & editing. Fei Lei: Data curation; Validation; Supervision; Writing – review & editing. Lu Tan: Data curation; Validation; Supervision; Writing – review & editing. Taomei Li: Data curation; Validation; Supervision; Writing – review & editing. Junying Zhou: Data curation; Validation; Supervision; Writing – review & editing. Yuanfeng Sun: Data curation; Validation; Supervision; Writing – review & editing. Ye Zhang: Data curation; Validation; Supervision; Writing – review & editing. Rong Ren: Data curation; Validation; Supervision; Writing – review & editing. Jian Jiao: Data curation; Validation; Supervision; Writing – review & editing. Yun Li: Data curation; Validation; Supervision; Writing – review & editing. Yanyan Wang: Data curation; Validation; Supervision; Writing – review & editing. Larry D Sanford: Conceptualization; Investigation; Validation; Writing – review & editing. Michael V Vitiello: Conceptualization; Investigation; Validation; Writing – review & editing. Xiangdong Tang: Conceptualization; Methodology; Data curation; Investigation; Validation; Supervision; Funding acquisition; Project administration; Resources; Writing – review & editing.

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting or writing, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for the contents of this article.

Disclosure

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.