Association between adenoid-to-tonsil size ratio and craniofacial characteristics in children with adenotonsillar hypertrophy: a cross-sectional study Authors: Liang Li, Peter Ngan, Tingting Zhao, Hong He Published: 2026 ABSTRACT AbstractObjectivesAdenotonsillar hypertrophy (ATH) is a major cause of upper airway obstruction and is associated with specific craniofacial characteristics. However, the craniofacial morphology of children with ATH is controversial. This study aimed to clarify these discrepancies by investigating the association between the adenoid-to-tonsil (A/T) size ratio and craniofacial characteristics in children with ATH, and examining whether age modifies these associations.MethodsChildren aged 6–10 years were recruited and classified into ATH or no hypertrophy (NH) groups based on lateral cephalograms. For children with ATH, the relative size of the adenoids and tonsils was quantified using the A/T length ratio measured on lateral cephalograms. Based on the median A/T ratio, the ATH group was further divided into two equal-sized groups: a tonsil-dominant (TD) and an adenoid-dominant (AD) group. Craniofacial differences between groups were evaluated. In addition, associations between the A/T ratio and craniofacial morphology were assessed in the ATH cohort. We further examined whether age (6–8 vs. 8–10 years) acted as a moderator of these associations using interaction analyses.ResultsA total of 563 children (207 AD, 207 TD, and 149 NH) were included. TD group exhibited greater SNB (B = 0.65, p = 0.028) and Go-Me (B = 0.69, p = 0.038), along with smaller Ar-Go-Me (B = − 0.98, p = 0.029) and Wits (B = − 0.68, p = 0.026) than AD group. In addition, the A/T ratio was positively associated with SGn/FH (B = 0.74, p = 0.048). Moreover, interactions between age and the A/T ratio were observed for MP-SN (B = 2.45, p = 0.044) and Ar-Go-Me (B = 2.86, p = 0.045), and Go-Me (B = − 2.12, p = 0.013).ConclusionThe relative size of the adenoids and tonsils may be associated with craniofacial morphology in children with ATH, with lower A/T ratios tending to be associated with mandibular protrusion and overgrowth, whereas higher ratios appearing to be associated with clockwise mandibular rotation. Notably, age may modify these associations, suggesting the potential importance of timely airway evaluation and management in orthodontic practice.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40510-026-00631-4.Keywords: Adenoids, Tonsils, Adenotonsillar hypertrophy, Craniofacial characteristics, Lateral cephalography Introduction According to the MOSS functional matrix theory, breathing mode and upper airway function may be related to craniofacial morphology [1–4]. Among children, enlargement of adenoids and/or tonsils is a common cause of airway obstruction and is often associated with specific craniofacial characteristics [5, 6]. Adenoid hypertrophy (AH) is typically associated with habitual mouth breathing and clockwise mandibular rotation, and a skeletal Class II malocclusion pattern [7–9]. In contrast, tonsillar hypertrophy (TH) was reported to be associated with mandibular overgrowth and prognathism, and a higher prevalence of skeletal Class III relationships [10–12]. However, this association was not observed in some studies [13, 14]. In fact, both the adenoids and the palatine tonsils are components of Waldeyer’s ring; the former is located in the posterior nasopharynx, whereas the latter resides in the fossa between the palatoglossal and pharyngopalatine arches [15]. Meanwhile, among children seeking orthodontic treatment, children with adenotonsillar hypertrophy (ATH) are more commonly observed than those with isolated AH or TH [16]. However, the craniofacial characteristics of the ATH children remain controversial. Huang et al. [17] and Singh et al. [18] reported that children with ATH exhibited craniofacial patterns similar to those with isolated AH. In contrast, another study reported that children with ATH resembled those with isolated TH in craniofacial morphology [19]. These discrepancies may complicate the orthodontic treatment planning for children with ATH. Therefore, identifying the potential sources of these inconsistencies is important. ATH is assessed primarily through clinical and endoscopic examinations [20, 21]. However, for children seeking orthodontic treatment, their routine radiographic examinations, such as lateral cephalograms, may also provide information for assessing the size of the adenoids and tonsils [10, 22]. It is generally accepted that children with both an adenoid-to-nasopharynx (A/N) ratio and a tonsil-to-oropharynx (T/O) ratio exceeding 50% on cephalometric radiographs are diagnosed with ATH [16, 17, 19, 23]. However, this criterion does not consider the relative size of the adenoids and tonsils in the ATH population. Specifically, the relative contribution of adenoid and tonsillar enlargement may vary across children with ATH, with some showing adenoid-dominant enlargement and others showing tonsil-dominant enlargement. Such differences may lead to distinct subgroups with different craniofacial features in the ATH population, which may explain the inconsistent results of previous studies. Therefore, we hypothesized that craniofacial morphology in children with ATH is associated with the relative size of the adenoids and tonsils. Given that ATH-related craniofacial characteristics have been widely reported to vary with age [19, 24], we further hypothesized that age may act as a moderator of these associations. Accordingly, the aim of this study was to investigate the associations between craniofacial characteristics and the adenoid-to-tonsil (A/T) size ratio in the ATH population, and to examine whether age modifies these associations. Our findings may contribute to a better understanding of the association between ATH and craniofacial morphology, and provide clinically relevant insights for both airway management and orthodontic practice. Materials and methods Ethical approval The study protocol was reviewed in accordance with the principles outlined in the Helsinki Declaration [25]. This study was designed following the research guidelines of Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) [26]. The Ethics Committee of the Stomatology Hospital of Wuhan University approved this cross-sectional study (Approval number: WDKQ2025- B92). Sample description Children admitted to the Orthodontics Department of the Stomatology Hospital of Wuhan University were recruited from January 2023 to November 2023. The inclusion criteria for the study group were: (1) children aged 6–10 years; (2) availability of routine pre-treatment lateral cephalograms; (3) ATH or no-hypertrophy (NH) confirmed on lateral cephalometric radiographs. The exclusion criteria for all participants were: (1) poor quality cephalometric image inadequate for identification of adenoid, tonsil or craniofacial structures; (2) history of severe craniomaxillofacial trauma or previous orthodontic treatment; (3) history of adenoidectomy or tonsillectomy; (4) acute systemic or upper respiratory tract infection at the time of examination, such as the common cold, fever, acute adenoiditis, or acute tonsillitis; (5) medical treatment that might affect lymphoid tissue size, such as topical corticosteroid therapy; (6) other upper airway obstruction related diseases, such as nasal polyps, nasal tumor, or severe allergic rhinitis; (7) history of bad oral habits such as thumb sucking, tongue thrusting, lip biting, or atypical swallowing patterns; and (8) craniofacial anomalies such as the presence of cleft lip and palate. Lateral cephalograms were obtained for all included participants, along with demographic information including gender and age. Sample size calculation We performed an a priori sample size calculation using G*Power (version 3.1.9.7), with the test power (1 − β) set at 0.95 and the significance level (α) established at 0.05. When based on a one-way analysis of variance (ANOVA), the estimated effect size was derived from group differences in the SNB angle reported in a previous study [16]. Specifically, the reported mean (standard deviation) SNB values were 76.74 (3.40) for the AH group (n = 88), 78.72 (4.22) for the TH group (n = 124), 77.09 (3.64) for the ATH group (n = 274), and 76.91 (3.63) for the NH group (n = 112). The corresponding Cohen’s f was calculated to be 0.19. Given that a total of three groups were involved in our study, the minimum required total sample size was calculated to be 432. When based on multiple linear regression, as no prior studies were available, a medium effect size was assumed (f² = 0.15) [27]. The number of tested predictors was set to 1, corresponding to the A/T ratio, and the total number of predictors was set to 4, including A/T ratio, age group, sex and age group*A/T ratio (interaction). Accordingly, the minimum required total sample size was estimated to be 89. Cephalometric analysis All lateral cephalograms were taken using the same cephalostat (OP200D Soredex, Instrumentarium, Finland), with the patient positioned in a natural head posture and maintaining the lips in a relaxed state. Prior to image acquisition, the children were instructed to swallow and then bite in maximum intercuspation, with the lips relaxed and the facial muscles at rest, under the direct supervision of an experienced technician to ensure a proper jaw position [28,29]. Dolphin Image Software (version 11.95, Dolphin Imaging & Management Solutions, Chatsworth, Calif.) was utilized as the tool for cephalometric measurement. The cephalometric variables analyzed in our study were selected based on previous studies on ATH and craniofacial morphology, with the SNB angle predefined as the primary outcome [16,17,19]. The definition and schematic diagram of the cephalometric landmarks and variables utilized in this study are presented in Supplementary Fig. 1 and Supplementary Table 1. In addition, according to the individual dentition stage, children were assigned to three sagittal skeletal malocclusion patterns based on the cephalometric norm of Chinese children: skeletal class I (3.3° ≤ ANB ≤ 6.1° in mixed dentition, 0.7° ≤ ANB ≤ 4.7° in permanent dentition), skeletal class II (ANB > 6.1° in mixed dentition, ANB > 4.7° in permanent dentition) and skeletal class III (ANB < 3.3° in mixed dentition, ANB < 0.7° in permanent dentition) [17, 19]. Assessment of adenoid and tonsil sizes The diagnoses of AH and TH were developed based on Fujioka’s and Baroni’s methods [23, 30]. Fujioka’s method, widely employed for assessing AH, was validated with 95% specificity and a 94% positive predictive value [31]. Similarly, Baroni’s method for evaluating tonsillar hypertrophy has shown strong overall accuracy (AUC: 0.90) [32]. The reference points and lines used for adenoid and tonsil measurements on the lateral cephalogram are depicted in Fig. 1. Cephalometric landmarks and reference lines for adenoid and tonsil measurement. Line 1, the line parallel to the anterior margin of the basiocciput; Line 2, the line tangent to the posterior wall of the oropharynx. Ad (Adenoidal point), the most anterior point on the anterior outline of the adenoid shadow (nearest point to the soft palate); Np (Nasopharynx posterior), the line perpendicular to line 1 and passing through Ad point intersects line 1 at Np point; PNS (posterior nasal spine), the posterior tip of the hard palate; To (Tonsillar point): the nearest point on the posterior outline of tonsil shadow to the posterior wall of the oropharynx; Op (Oropharynx posterior), the line perpendicular to line 2 and passing through To point intersects line 2 at Op point; Oa (Oropharynx anterior), the line perpendicular to line 2 and passing through To point intersects the posterior outline of the tongue at Oa point. Line segment A is the distance between point Np and Ad; line segment N is the distance between point Np and PNS. Line segment T is the distance between point To and Oa; line segment O is the distance between point Oa and Op Line segment A indicated the size of the adenoid; line segment N indicated the size of the nasopharyngeal space; line segment T indicated the size of the tonsil and line segment O indicated the size of the oropharyngeal space. The percentage of adenoid or tonsillar obstruction in the pharyngeal airway space was calculated mathematically as follows: (A/N) × 100% and (T/O) × 100%. Children were assigned to the ATH group when both percentages exceeded 50%. Those with neither percentage greater than 50% were included in the NH group. For children in the ATH group, the relative sizes of the adenoids and tonsils were further quantified using the adenoid-to-tonsil length ratio on lateral cephalograms, calculated as A/T. Subsequently, the ATH children were divided into two equal-sized groups according to the median A/T ratio of the children within the ATH cohort. Children with A/T ratios below the median were assigned to the tonsil-dominant (TD) group, while those with A/T ratios above the median were classified into the adenoid-dominant (AD) group. Quality control and evaluation of the error All cephalometric measurements were conducted by a proficient orthodontist graduate student (L.L.), who received systematic training in the software prior to commencing measurements. Prior to formal measurement, the primary (L.L.) and secondary (T.Z.) assessors independently evaluated 10 lateral cephalometric radiographs to assess inter-examiner reliability. If the inter-examiner intraclass correlation coefficient (ICC) was < 0.75, discrepancies were discussed and the procedure was repeated until acceptable agreement was achieved. The primary assessor then independently proceeded with all measurements [9]. After the initial measurements, 20% of the lateral cephalometric radiographs were randomly selected by the primary assessor for re-measurement after a one-month interval. The intra-examiner ICC was employed to assess the consistency of measurement results. An ICC value greater than 0.9 indicated excellent consistency, and statistical analysis would be performed on the initially measured data [33]. Angular and linear measurement errors were assessed using the method of moments estimator (MME) formula [34]. Data analysis Statistical analyses were performed using SPSS software (version 21.0; IBM, Armonk, NY, USA). Continuous variables were presented as mean (standard deviation), and categorical variables were expressed as n (%). The Shapiro-Wilk test was used to assess the normality of the data, and Levene’s test was applied to evaluate the homogeneity of variances. When both normality and homogeneity assumptions were satisfied, multiple-group comparisons were conducted using ANOVA; otherwise, the Kruskal-Wallis test was employed. Pairwise comparisons were performed using the Bonferroni correction where applicable. Differences in distribution of gender and sagittal skeletal malocclusion patterns among the groups were examined using the chi-square test. In addition, children were divided into two age groups (6–8 years and 8–10 years) to evaluate the moderation effect of age. Multiple linear regression analyses were conducted in the ATH cohort to investigate sex- and age-adjusted craniofacial differences between the AD and TD subgroups and to assess whether age acted as an effect modifier. In the initial model, each cephalometric variable was entered as the dependent variable, with sex, age group (dummy-coded), ATH subgroup (dummy-coded), and the interaction term (age group × ATH subgroup) included as independent variables. If the interaction term was not statistically significant, it was removed to simplify the model. We also repeated the multiple linear regression analyses using the A/T ratio in place of the ATH subgroups as a sensitivity analysis to further examine the association between continuous variations in adenotonsillar relative size and craniofacial characteristics. To visualize the regression results, marginal plots were generated using Python. Restricted cubic spline analyses were performed to assess potential nonlinearity in these associations. A p-value < 0.05 was considered statistically significant. Results Demographic characteristics of the study population The demographic profiles of the included children are summarized in Table 1. After applying the predefined inclusion and exclusion criteria, a total of 563 children were enrolled in this study, including 149 children in the NH group (72 boys, 77 girls; 7.31 ± 0.98 years old) and 414 children in the ATH group. The A/T ratios of the 414 ATH children ranged from 0.80 to 2.51, with a median of 1.48. Using this median as the cutoff, 207 children with lower A/T values were assigned to the TD group (102 boys, 105 girls; 7.37 ± 1.04 years old), and 207 with higher A/T ratios were assigned to the AD group (110 boys, 97 girls; 7.36 ± 1.03 years old). No significant differences in gender or age distribution were found among the three groups. Table 1Demographic characteristics of the study groupsVariableTDADNHp value Total n = 207n = 207n = 149Age7.37 ± 1.047.36 ± 1.037.31 ± 0.980.81Male, n (%)102 (49.3%)110 (53.1%)72 (48.3%)0.63 6–8 years n = 120n = 115n = 88Age6.62 ± 0.516.57 ± 0.506.49 ± 0.470.27Male, n (%)56 (46.7%)64 (55.7%)42 (47.7%)0.34 8–10 years n = 87n = 92n = 61Age8.40 ± 0.628.35 ± 0.568.49 ± 0.590.36Male, n (%)46 (52.9%)46 (50.0%)30 (49.2%)0.91TD, tonsil-dominant group; AD, adenoid-dominant group; NH, no-hypertrophy group Demographic characteristics of the study groups TD, tonsil-dominant group; AD, adenoid-dominant group; NH, no-hypertrophy group The method error ranged from 0.57° to 0.83° for angular measurements and from 0.64 to 0.96 mm for linear measurements. The ICCs between the initial and repeated measurements ranged from 0.91 to 0.95, demonstrating excellent intra-examiner reliability. Therefore, in accordance with the study protocol, the initial measurements were used for all subsequent statistical analyses. Craniofacial differences among groups Table 2 and Supplementary Table 2 present the craniofacial characteristics and multigroup comparison results among the TD, AD, and NH groups. Both the TD and AD groups exhibited significantly greater SNB, as well as smaller ANB and Wits, compared with the NH group. In addition, compared with the NH group (skeletal class III = 38.3%), the proportions of skeletal class III in the AD (57.5%, p < 0.001) and TD (65.7%, p < 0.001) groups were significantly increased (Supplementary Fig. 2). Table 2Comparison of craniofacial variables among the TD, AD and NH groupsCraniofacialvariablesTD (n = 207)Mean (SD)AD (n = 207)Mean (SD)NH (n = 149)Mean (SD)P valueSNA (°)79.37 (3.39)a78.73 (3.31)b79.66 (3.39)a 0.026 SNB (°)77.21 (3.20)a76.53 (3.53)a75.57 (3.45)b < 0.001 ANB (°)2.16 (3.10)b2.19 (3.67)b4.09 (2.82)a < 0.001 MP-SN (°)37.26 (4.61)37.95 (4.98)37.62 (5.19)0.355Ar-Go-Me (°)129.16 (5.36)ab130.16 (5.87)a128.41 (5.34)b 0.041 SGn/FH (°)61.14 (2.94)61.58 (3.00)61.86 (3.28)0.077S-N (mm)59.09 (3.16)b59.69 (2.75)a60.06 (2.77)a 0.002 Go-Me (mm)57.92 (3.74)57.34 (3.59)57.46 (3.71)0.245Wits (mm)− 2.39 (3.33)b− 1.60 (4.46)b− 0.05 (3.50)a < 0.001 TD, tonsil-dominant group; AD, adenoid-dominant group; NH, no-hypertrophy groupDifferent lowercase superscript letters (a, b, c) indicate statistically significant differences between groups (Bonferroni-corrected p < 0.05). Groups sharing at least one letter do not differ significantly Comparison of craniofacial variables among the TD, AD and NH groups TD, tonsil-dominant group; AD, adenoid-dominant group; NH, no-hypertrophy group Different lowercase superscript letters (a, b, c) indicate statistically significant differences between groups (Bonferroni-corrected p < 0.05). Groups sharing at least one letter do not differ significantly Table 3 presents age group- and sex-adjusted cephalometric differences between ATH subgroups, as the interaction term (age group × ATH subgroup) was not statistically significant for any cephalometric variable and therefore removed from the final model. TD group exhibited higher SNA (B = 0.61, 95% CI: 0.14–1.08, p = 0.011), SNB (B = 0.65, 95% CI: 0.08–1.23, p = 0.028) and Go-Me (B = 0.69, 95% CI: 0.04–1.34, p = 0.038), but lower Ar-Go-Me (B = -0.98, 95% CI: -1.86–-0.10, p = 0.029) and Wits (B = -0.68, 95% CI: − 1.28 to − 0.08, p = 0.026) compared with AD group. Supplementary Table 3 presents subgroup differences in other cephalometric variables. Table 3Multiple linear regression analyses of cephalometric variables in the ATH cohort, with ATH subgroup, age group and sex as independent variablesCraniofacial variablesATH subgroup (AD as reference)B95% CIP valueSNA (°)0.61(0.14, 1.08) 0.011 SNB (°)0.65(0.08, 1.23) 0.028 ANB (°)− 0.04(− 0.49, 0.41)0.862MP-SN (°)− 0.70(− 1.62, 0.22)0.136Ar-Go-Me (°)− 0.98(− 1.86, − 0.10) 0.029 SGn/FH (°)− 0.43(− 1.00, 0.14)0.141S-N (mm)− 0.42(− 0.93, 0.10)0.115Go-Me (mm)0.69(0.04, 1.34) 0.038 Wits (mm)− 0.68(− 1.28, − 0.08) 0.026 B : unstandardized regression coefficients for the ATH subgroup, representing age group- and sex-adjusted craniofacial differences between TD and AD groups. AD, adenoid-dominant group; CI, confidence interval Multiple linear regression analyses of cephalometric variables in the ATH cohort, with ATH subgroup, age group and sex as independent variables B : unstandardized regression coefficients for the ATH subgroup, representing age group- and sex-adjusted craniofacial differences between TD and AD groups. AD, adenoid-dominant group; CI, confidence interval Associations between craniofacial variables and the A/T ratio with age as a moderator Table 4 demonstrates age group- and sex-adjusted associations between the A/T ratio and craniofacial variables in the ATH cohort, as the interaction term was not statistically significant for these variables. In the ATH cohort, a higher A/T ratio was associated with increased SGn/FH (B = 0.74, 95% CI: 0.01–1.47, p = 0.048), as well as decreased SNA (B = − 0.84, 95% CI: − 1.64 to − 0.03, p = 0.041) and SNB (B = − 0.68, 95% CI: − 1.29 to − 0.07, p = 0.029). Associations between A/T ratio and other cephalometric variables are summarized in Supplementary Table 4. Table 4Multiple linear regression analyses of cephalometric variables in the ATH cohort, with A/T ratio, age group and sex as independent variablesCraniofacial variablesA/TB95% CIP valueSNA (°)− 0.84(− 1.64, − 0.03) 0.041 SNB (°)− 0.68(− 1.29, − 0.07) 0.029 ANB (°)− 0.15(− 0.97, 0.68)0.724SGn/FH (°)0.74(0.01, 1.47) 0.048 S-N (mm)0.14(− 0.53, 0.81)0.687Wits (mm)0.75(− 0.24, 1.74)0.136All models were initially specified as: y = β₀ + β₁(A/T ratio) + β₂(age group) + β₃(sex) + β₄(age group × A/T ratio). In this table, all outcomes were analyzed using a simplified model: y = β₀ + β₁(A/T ratio) + β₂(age group) + β₃(sex), as the interaction term (age group × A/T ratio) was not statistically significant (p-interaction > 0.05)B values represent unstandardized regression coefficients for the A/T ratio, indicating age group- and sex-adjusted changes in craniofacial variables per unit increase in the A/T ratio. CI, confidence interval Multiple linear regression analyses of cephalometric variables in the ATH cohort, with A/T ratio, age group and sex as independent variables All models were initially specified as: y = β₀ + β₁(A/T ratio) + β₂(age group) + β₃(sex) + β₄(age group × A/T ratio). In this table, all outcomes were analyzed using a simplified model: y = β₀ + β₁(A/T ratio) + β₂(age group) + β₃(sex), as the interaction term (age group × A/T ratio) was not statistically significant (p-interaction > 0.05) B values represent unstandardized regression coefficients for the A/T ratio, indicating age group- and sex-adjusted changes in craniofacial variables per unit increase in the A/T ratio. CI, confidence interval In addition, Table 5 shows a significant positive interaction term between A/T ratio and age group for MP-SN (B = 2.45, 95% CI: 0.07–4.84, p = 0.044) and Ar-Go-Me (B = 2.86, 95% CI: 0.06–5.67, p = 0.045), along with a negative interaction term for Go-Me (B = − 2.12, 95% CI: − 3.80 to − 0.44, p = 0.013). In addition, in the ATH children aged 6–8 years, the A/T ratio showed no significant association with these cephalometric variables. However, among children aged 8–10 years, the A/T ratio was positively associated with Ar-Go-Me (B = 1.88, p = 0.034) and negatively associated with Go-Me (B = − 1.82, p = 0.003). The age-modified associations between the A/T ratio and these cephalometric variables were illustrated in Fig. 2, which differed in both direction and magnitude between the 6–8 and 8–10 age groups. Meanwhile, restricted cubic spline analyses did not reveal any statistically significant nonlinearity in these associations (Supplementary Fig. 3). Table 5Multiple linear regression analyses of cephalometric variables in the ATH cohort, with A/T ratio, age group, sex and A/T ratio*age group (interaction) as independent variablesCraniofacial variablesA/T ratio (6–8 years)A/T ratio (8–10 year)Interaction: A/T * Age groupB95% CIP valueB95% CIP valueB95% CIP valueMP-SN (°)− 0.99(− 2.65, 0.66)0.2401.46(− 0.26, 3.18)0.0962.45(0.07, 4.84)0.044Ar-Go-Me (°)− 0.98(− 2.93, 0.97)0.3221.88(0.14, 3.62)0.0342.86(0.06, 5.67)0.045Go-Me (mm)0.30(− 0.87, 1.46)0.615− 1.82(− 3.03, − 0.61)0.003− 2.12(− 3.80, − 0.44)0.013All models were initially specified as: y = β₀ + β₁(A/T ratio) + β₂(age group) + β₃(sex) + β₄(age group × A/T ratio). All outcomes presented in this table were analyzed using this full model, as the interaction term was statistically significant (p-interaction < 0.05)B values of the A/T ratio (6–8 years) indicated sex-adjusted changes in craniofacial variables per unit increase in the A/T ratio in children aged 6–8 years, with this age group treated as the reference (coded as 0) in the modelB values of the A/T ratio (8–10 years) indicated sex-adjusted changes in craniofacial variables per unit increase in the A/T ratio in children aged 8–10 years, with this age group treated as the reference (coded as 0) in the model. CI, confidence interval Multiple linear regression analyses of cephalometric variables in the ATH cohort, with A/T ratio, age group, sex and A/T ratio*age group (interaction) as independent variables All models were initially specified as: y = β₀ + β₁(A/T ratio) + β₂(age group) + β₃(sex) + β₄(age group × A/T ratio). All outcomes presented in this table were analyzed using this full model, as the interaction term was statistically significant (p-interaction < 0.05) B values of the A/T ratio (6–8 years) indicated sex-adjusted changes in craniofacial variables per unit increase in the A/T ratio in children aged 6–8 years, with this age group treated as the reference (coded as 0) in the model B values of the A/T ratio (8–10 years) indicated sex-adjusted changes in craniofacial variables per unit increase in the A/T ratio in children aged 8–10 years, with this age group treated as the reference (coded as 0) in the model. CI, confidence interval Marginal plots illustrating the effects of the adenoid-to-tonsil (A/T) ratio on craniofacial variables in different age groups. Models included the A/T ratio, age group, sex and the interaction between the A/T ratio and age group. Shaded areas represent 95% confidence intervals. Only models with statistically significant interaction terms are shown Discussion Previous studies reported conflicting findings regarding craniofacial patterns in children with ATH. Some suggested that children with ATH resemble those with TH, characterized by a smaller ANB and a greater SNB compared with NH children [19]. However, other studies reported that children with ATH showed a superimposed craniofacial pattern of children with AH and TH [23]. Specifically, although these children also exhibited a reduced ANB similar to children with TH, this reduction was primarily driven by a decreased SNA, resembling the pattern seen in children with AH [8, 35]. In our study, these distinct craniofacial patterns were simultaneously observed in two subgroups of children with ATH. Both the TD and AD groups exhibited significantly smaller ANB compared with NH children. However, in the TD group, the reduced ANB was primarily attributable to an increased SNB, whereas in the AD subgroup, it was mainly driven by a decreased SNA (Table 2). These findings suggested that, in children with ATH, relatively greater adenoid size may be more closely associated with alterations in maxillary position, whereas relatively greater tonsillar size may be more closely associated with mandibular position. Moreover, these findings may explain the inconsistent craniofacial patterns reported in previous studies of ATH populations, as the relative proportions of TD and AD subgroups were neither considered nor balanced in those studies. However, these findings should be interpreted with caution, as a meta-analysis showed that the level of evidence for an association between adenoid or tonsillar hypertrophy and sagittal craniofacial characteristics in children remains low [35]. Moreover, the association between TH and skeletal Class III malocclusion remains controversial [13, 36]. Therefore, further studies are needed. Meanwhile, some cranial base morphology variables in the ATH group also differed from those in the NH group. For example, S-N was significantly shorter in the TD group than in the NH group, while the AD group showed a similar but non-significant trend. This was consistent with a previous finding, which reported that S-N length was significantly negatively correlated with tonsillar hypertrophy grades in children (r = − 0.239, p = 0.033) [37]. One possible hypothesis is that the anterior cranial base may serve as a structural template for the growth of the nasomaxillary complex, and its morphology may be associated with impaired respiratory function in children with ATH [38, 39]. In addition, some studies reported that children with obstructive sleep apnea (OSA) have a significantly shorter S-N than non-OSA children [40], whereas other studies found no significant difference between the two groups [41, 42]. Variations in ethnicity, age, and sample size may partly explain the discrepancies across studies. A recent systematic review concluded that only low-level evidence suggested a shorter cranial base length in children with sleep disordered breathing (SDB) [43]. Overall, the evidence supporting an association between cranial base morphology and ATH or SDB remains limited, and the underlying mechanisms are still unclear. Further studies are therefore needed to clarify these associations. In addition, several craniofacial features significantly differed between the AD and TD groups, suggesting the value of distinguishing subgroups in the ATH population. In terms of the sagittal dimension, the TD group exhibited a more anterior mandible (SNB, B = 0.65, p = 0.028) than the AD group. This may be attributable to a relatively longer mandibular body (Go-Me, B = 0.69, p = 0.038) or compensatory mandibular protrusion associated with TH-related airway obstruction [35, 44]. Moreover, compared with their TD counterparts, the AD group demonstrated a more retrusive maxilla (SNA, B = − 0.61, p = 0.011). Consistently, the A/T ratio was negatively associated with SNA (B = − 0.84, p = 0.041). This finding may be attributable to AH-related mouth breathing, which is associated with a low tongue posture and disrupted perioral muscular balance, thereby potentially restricting transverse and anteroposterior maxillary growth [2, 45]. Regarding vertical craniofacial characteristics, the AD group exhibited significantly greater clockwise rotation of the mandible (Ar-Go-Me, B = 0.98, p = 0.029) compared with the TD group. In addition, the A/T ratio was positively associated with SGn/FH (B = 0.74, p = 0.048). This may be because children in the AD group are more prone to mouth breathing and clockwise mandibular rotation, which have been previously associated with nasopharyngeal obstruction [46, 47], whereas those in the TD group may be more likely to exhibit compensatory mandibular advancement, a pattern previously reported in relation to oropharyngeal obstruction, rather than rotational changes [15]. Consequently, differences in the direction of mandibular movement may explain these vertical craniofacial discrepancies between the two groups. Although these subgroups within the ATH populations may explain many of the inconsistencies reported in previous studies, exceptions remained. Huang et al. [17 ] and Singh et al. [18] suggested that children with ATH resembled those with AH, characterized by smaller SNB and greater ANB values compared with NH children. However, in the present study, both the TD and AD subgroups exhibited greater SNB and smaller ANB compared with the NH group. Even when the AD subgroup predominates in the ATH population, children with ATH do not exhibit a skeletal Class II craniofacial pattern. Therefore, we proposed that age may represent another critical factor contributing to the discrepancies across studies. Adenoids and tonsils follow different growth and involution patterns, with adenoids occupying a relatively larger proportion of the upper airway over a longer developmental period [17, 48]. Consequently, studies with broader age ranges (e.g., children aged 3–18 years) [17, 18], may find children with ATH more similar to those with AH. However, our study focused on children aged 6–10 years, as this age range is generally considered to correspond to the peak prevalence of adenotonsillar hypertrophy. This may explain the closer alignment of our findings with those of Huang et al. [19], who stratified participants into subgroups with a 3-year age span. In addition, our study demonstrated that age may act as a moderator in the associations between the A/T ratio and craniofacial characteristics. Compared with children aged 6–8 years, children aged 8–10 years with a lower A/T ratio were more likely to exhibit a relatively longer mandibular body (Go-Me, B = 2.12, p-interaction = 0.013). One possible explanation for this observation is that the establishment of an anterior crossbite may be associated with reduced maxillary constraint on mandibular growth, although this remains speculative and requires further investigation. In addition, with increasing age, children with a higher A/T ratio showed greater vertical facial height (MP-SN, B = 2.45, p-interaction = 0.044; Ar-Go-Me, B = 2.86, p-interaction = 0.045). This may be associated with prolonged adenoid hypertrophy and habitual mouth breathing, which may stabilize the mandible in a clockwise-rotated position. These findings may indicate the hypothesis that ATH children with a higher A/T ratio may initially exhibit mandibular protrusion associated with tonsil-related airway obstruction; however, with increasing age, progressive clockwise mandibular rotation may offset or even reverse this protrusive tendency, ultimately contributing to a skeletal Class II craniofacial pattern, as reported in previous studies [17]. Moreover, age may modify both the direction and magnitude of these associations. Specifically, in children aged 8–10 years, a higher A/T ratio was significantly associated with increased Ar-Go-Me and decreased Go-Me, whereas in those aged 6–8 years, the associations tended in the opposite direction and did not reach statistical significance. These findings suggest that, from a clinical perspective, identification of enlarged adenoids or tonsils on lateral cephalograms in orthodontic practice may prompt referral to an otolaryngologist for further airway evaluation. However, lateral cephalograms should not be recommended as a routine method for evaluating ATH in children, as such use would be inconsistent with the As Low As Reasonably Achievable (ALARA) principle and may lead to unnecessary radiation exposure [49]. Timely adenoidectomy or other appropriate interventions may mitigate the unfavorable craniofacial characteristics and even enhance the effectiveness of subsequent orthodontic treatment [50, 51]. However, because the onset of ATH may vary among individuals, increasing age does not necessarily indicate longer exposure; therefore, age-related modification should be interpreted with caution. This study has several limitations. First, the cross-sectional design limits causal inference. In addition, given the retrospective nature of the study, the exclusion criteria were based on medical records; therefore, potential residual confounding factors may not be completely eliminated. Meanwhile, recruitment from the orthodontic department of a hospital may introduce selection bias to limit the generalizability of the findings. In addition, the assessment of adenoid and tonsil size in the present study was based on two-dimensional lateral cephalograms, which cannot reflect the dynamic behavior of the airway or provide information on the duration of the ATH. Furthermore, this method does not account for potentially important influencing factors, such as physiological variability or seasonal fluctuations in lymphoid tissue size. However, lateral cephalograms are widely accepted for airway evaluation because they are simple, accessible, and low-radiation [52], and their accuracy in evaluating adenoidal and tonsillar hypertrophy has been well validated [31, 32]. Moreover, the relative size of the adenoids and tonsils was quantified using absolute dimensions (A/T) rather than proportional indices calculated as (A/N)/(T/O). The latter approach was not adopted because children with ATH often exhibit mandibular advancement or rotation due to airway obstruction, which can alter oropharyngeal dimensions and compromise the accuracy of the T/O ratio. Future research should adopt longitudinal designs with larger sample sizes, incorporate information on the duration of ATH or lateral cephalograms at multiple time points, and investigate the reversibility of craniofacial characteristics following adenotonsillectomy. Conclusion The relative size of the adenoids and tonsils may be associated with craniofacial morphology in children with ATH. Children with a lower A/T ratio tended to be associated with a longer mandibular body and a more anterior mandibular position, whereas those with a higher A/T ratio appeared to be associated with greater clockwise mandibular rotation. Notably, these associations appeared to be modified by age. As the clinical significance of these differences may be limited, these findings should be interpreted with caution. Taken together, these findings may suggest the potential importance of timely airway evaluation and management in orthodontic practice. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Author contributions L.L. contributed to study design, result interpretation and manuscript writing. P.N. contributed to study design, result interpretation and manuscript revision. T.Z. contributed to study design, result interpretation and manuscript revision. H.H. contributed to conception, study design, result interpretation and manuscript revision. All authors reviewed and approved the final manuscript. Funding This work was supported by the International Orthodontics Foundation (No. IOF2022C01), the Sanming Project of Medicine in Shenzhen Nanshan (No. SZSM202103005), and the Wuhan University School & Hospital of Stomatology Clinical Research Project (No. LYZX202101). Data availability All the data used and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate This study was filed and approved by the Ethics Committee of the Hospital of Stomatology, Wuhan University, Wuhan, China (No. WDKQ2025-B92). Due to the retrospective and observational design of the study, informed consent was waived by the Ethics Committee of the Hospital of Stomatology, Wuhan University, Wuhan, China. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Tingting Zhao, Email: zhaott1991@whu.edu.cn. Hong He, Email: drhehong@whu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement All the data used and/or analyzed during the current study are available from the corresponding author on reasonable request.