Sample description
This cross-sectional study was conducted at the department of orthodontics at the Hospital of Stomatology, Wuhan University. The study protocol was approved by the Ethics Committees of the Hospital of Stomatology, Wuhan University. Informed written consent was obtained from patients and their parents before data collection.
We recruited consecutive children who sought orthodontic treatment and had lateral cephalograms taken for routine diagnosis from January 2022 to October 2022. The inclusion criteria for the study were (1) children aged 3 to 12 years, (2) clear identification of oropharyngeal airway and tonsils on lateral cephalograms, and (3) body mass index (BMI) below cutoff points of obesity[20]. The exclusion criteria were (1) patients with acute upper airway infection, (2) a history of previous tonsillectomy, tonsillotomy, or orthodontic treatment, and (3) craniofacial syndromes.
Within a week of the cephalometric examinations, subjects received a clinical assessment of tonsil size at the department of otolaryngology-head and neck surgery, Zhongnan Hospital of Wuhan University. The data including age, sex, weight, and height were collected for each child. Patients were divided into several subgroups with subgrouping variables of sex, age, and sagittal skeletal patterns.
Cephalometric analysis
All lateral cephalograms were obtained with the same device (Soredex, Tuusula, Finland) and performed by the same operator, according to a standard protocol (73 kV, 10 mA). All cephalograms were taken in centric occlusion and natural head position without swallowing. The cephalograms were coded with numbers and digitized with Dolphin Imaging software (Version 11.7, Dolphin Imaging & Management Systems, Chatsworth, USA) by 1 researcher (C.L), who was unaware of each subject's characteristics and subjective tonsil size.
The reference points and lines used in the cephalometric analysis were shown in Figure 1. On lateral cephalograms, palatine tonsils appear as an oval-shaped shadow in the oropharyngeal space close to the root of the tongue. Image enhancement brightness and contrast can be adjusted to improve image quality and anatomic landmarks for measurement. A protocol for measuring the T/O ratio proposed by Baroni et al.[5] was followed. Sagittal skeletal patterns were characterized by ANB angle: Class I (1°< ANB ≤5°), Class II (ANB >5°), and Class III (ANB ≤1°)[21].
Clinical assessment of tonsil size
A clinical oropharyngeal examination was carried out by an otolaryngologist (X.C). Patients were asked to sit up straight, open their mouths wide, and continuously pronounce the phoneme /r/. When the patient had a Friedman palate position of 3 or 4[22] as the tonsils were not visualized, a tongue depressor was used to push the tongue against the floor of the mouth. During the examination, the investigator examined the oropharynx without activating the gag reflex, which could make the tonsils closer to the midline artificially.
The tonsil sizes of children were clinically graded on a scale of 1 to 4 according to the Brodsky scale (Figure 2). Grades 3 and 4 were deemed a diagnosis of TH in clinical settings. Grade 0 was reserved for postsurgical patients with no tonsils who were excluded from the study. We recorded the percentage of oropharyngeal obstruction by both tonsils to assign a grade instead of assessing the left and right tonsils separately, and the same goes for the asymmetric tonsils.
Statistical analysis
The collected data were analyzed using the SPSS software (version 26.0, Chicago, USA). The cephalograms of 30 randomly selected children were measured twice to calculate the measurement error, with a 2-week interval between the measurements. Spearman’s rank correlation coefficients (ρ) were calculated to evaluate the correlation between the T/O ratio and clinical tonsil size in the total sample and all subgroups. We classified the Spearman correlations in groups of negligible (0–0.09), weak (0.1–0.39), moderate (0.4–0.69), strong (0.7–0.89), and very strong (0.9–1.0) correlation[23]. The Fisher’s z-statistics with Zou’s confidence intervals (CI)[24] was used to test for differences in coefficients between two subgroups, implemented in the software package Cocor (version 1.1.3, Duesseldorf, Germany)[25].
The receiver operating characteristic (ROC) curve was performed to analyze the validity of T/O ratios for evaluating TH. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of different cutoff points were calculated. In all tests, P values of < 0.05 were considered statistically significant.