A cross-sectional survey on height was conducted among children aged 7–12 in Chaozhou, Guangdong Province, between June and September 2023. This work was part of the 2023 Chaozhou Student Growth and Development Abnormality Monitoring and Intervention Project, jointly carried out by the Chaozhou Health Bureau, the Chaozhou Education Bureau, and local medical institutions. Height was measured for children of this age group using standardized tools. In addition, a random sample from the survey group was chosen for a questionnaire on short stature influencing factors.

All children aged 7–12 were eligible for this study, except those with disabilities, congenital diseases, or unwillingness to participate. In the influencing factor survey, questionnaires with incomplete information or obvious logical errors were excluded.

Height and weight measurements were taken by project members with standardized training. They used a calibrated stadiometer and weight scale (Brand: Ruke; Manufacturer: Wujin Instrument Co., Ltd.; Model: SZ-200; precision ±0.1 cm/kg). Children wore light clothing and stood barefoot with heels together and looking straight ahead. The measurers read the height at the bottom edge of the sliding board, with both height (cm) and weight (kg) recorded to one decimal place. Each measurement was taken three times and averaged. Data was double-checked and recorded in a summary table.

The questionnaire was developed online using the SoJump platform, a widely used online survey system in China, and was distributed to student guardians via QR code for completion. To investigate factors associated with short stature, stratified random sampling was applied using age and sex as key stratification variables. Subgroups were formed according to the proportional allocation principle, and approximately one-quarter of the population was randomly selected as the questionnaire sample. The electronic questionnaire was administered through an online platform, and all responses were verified by two independent reviewers with logic checks applied to ensure data quality. After collection, a third party uniformly exported, organized, and analyzed the final validated questionnaires.

The questionnaire collected information in the following domains: ①Basic demographic characteristics, including sex, age, height, weight, parental height, birth weight, and preterm birth status; ②Birth status and early-life factors, including birth weight, preterm birth status, feeding method during infancy; ③Lifestyle and physical activity, including daily sleep duration, bedtime, exercise frequency, weekly exercise duration and daily screen time; ④Dietary habits, including dietary regularity, food preferences, weekly snack consumption frequency, meal duration, and frequency of eating takeout per week.

Short stature is defined as a height more than 2 standard deviations (SD) below the mean for children of the same age and sex, or below the 3rd percentile (P3, −1.88 SD), based on the Chinese national growth reference standards (11). This study used the percentile values of height and weight for Chinese children and adolescents aged 0–18 to screen for short stature (4, 5).

The primary endpoints were the height characteristics and prevalence of short stature among children aged 7–12 in Chaozhou. The secondary endpoints included the influencing factors of short stature and the construction of a predictive model.

The statistical analysis was performed using SPSS version 26.0 (IBM Corp., Armonk, NY, USA) and R version 4.2.2 (R Foundation for Statistical Computing, Vienna, Austria). Continuous variables were summarized as mean ± SD or median with interquartile range (IQR), whereas categorical variables were presented as frequencies and percentages. Differences between the height of children in Chaozhou and the 2005 national average were assessed using one-sample t-tests. Chi-square tests were used to compare the prevalence of short stature across genders and age groups. Height standard deviation scores (SDS) were calculated using age- and sex-specific Chinese growth standard curves (12), and are shown in Supplementary Table 1; Supplementary Figure 1.

Univariate logistic regression was conducted to screen potential factors associated with short stature, and variables with P < 0.10 were included in the multivariate logistic regression model to identify independent risk factors. A risk prediction model was developed based on significant predictors and visualized using a nomogram.

Model discrimination was evaluated using the area under the receiver operating characteristic (ROC) curve (AUC), with higher values indicating better discriminatory ability. An AUC > 0.7 is generally considered acceptable (13). Model calibration was assessed using calibration plots and the Hosmer-Lemeshow goodness-of-fit test, with P > 0.05 indicating adequate model calibration. A two-sided P < 0.05 was considered statistically significant.

Before formal measurements, researchers explained the study’s purpose and methods in detail to schools, parents, and students. Measurements were conducted and questionnaires were distributed only after obtaining guardians’ informed consent. To protect participants’ privacy, all survey information was strictly confidential. The study was approved by the Ethics Committee of the Chaozhou Central Hospital (approval No. 2023005).

A total of 8,078 students aged 7–12 years were initially enrolled in Chaozhou, among whom 7,799 were included in the final analysis after excluding those with missing information or inability to cooperate. Of the analyzed population, 52.4% were male. The survey covered three administrative districts—Chao’an (25.5%), Raoping (30.2%), and Xiangqiao (44.3%). The age distribution for children aged 7 to 12 years was 17.9%, 15.9%, 16.4%, 17.3%, 18.4%, and 13.9%, respectively.wwc.

Analysis of height by age and sex showed that girls had significantly greater mean height than boys at ages 8 (133.0 ± 3.8 cm vs. 131.9 ± 3.7 cm, P < 0.001), 11 (146.6 ± 9.7 cm vs. 144.8 ± 9.8 cm, P = 0.001), and 12 (154.0 ± 9.0 cm vs. 151.5 ± 9.3 cm, P < 0.001), while no significant differences were observed at other ages. The study flow is presented in Figure 1, and detailed demographic and height characteristics are summarized in Table 1; Figure 2.

Among the 7,799 children included in this study, 287 were identified as having short stature, resulting in an overall prevalence of 3.7%. The prevalence of short stature growth among children aged 7 to 12 years was 2.2%, 1.8%, 4.4%, 7.3%, 8.4%, and 7.9%, respectively. There were significant differences in prevalence across age groups (χ² = 111.858, P < 0.001), with the highest prevalence observed in children aged 9–12 years.

In the overall population, the prevalence of short stature growth was 3.4% in boys and 4.0% in girls. Although the prevalence was higher in girls than in boys, the difference was not statistically significant (OR = 1.181, 95% CI: 0.933–1.495, P = 0.167). The prevalence among boys by age was 1.6%, 2.0%, 4.1%, 3.1%, 4.8%, and 4.9%, with a peak occurring between ages 9 and 12. Similarly, the prevalence among girls was 2.9%, 1.5%, 4.6%, 5.4%, 4.7%, and 4.9%, also peaking between ages 9 and 12.

At age 10, the prevalence of stunted growth was significantly higher in girls than in boys (OR = 1.813, 95% CI: 1.049–3.133, P = 0.031), while no statistically significant differences were observed between genders at other ages. Detailed data were presented in Table 2; Figure 3.

In this study, 2000 questionnaires were randomly distributed, and 1907 valid ones were analyzed after excluding those with incomplete data or logical errors. Of these, 52.5% were from male respondents, and 4.9% reported short stature cases. Baseline data for the analyzed group are in Table 3.

Univariate analysis showed that several factors were associated with short stature, including paternal height < 160 cm, maternal height < 150 cm, low birth weight (< 2.5 kg), preterm birth, exercising fewer than three times per week, weekly exercise time < 120 minutes, sleep duration < 8 hours, irregular eating patterns, preference for meat and dairy products, taking more than 1 hour per meal, and consuming takeout food more than three times per week.

Variables with P < 0.10 were entered into a multivariate logistic regression model. The results indicated that paternal height < 160 cm (OR = 10.677, 95% CI: 5.694-20.022), maternal height < 150 cm (OR = 8.071, 95% CI: 3.975-16.389), birth weight < 2.5 kg (OR = 8.750, 95% CI: 4.210-18.185), preterm birth (OR = 3.724, 95% CI: 2.012-6.891), exercising fewer than three times per week (OR = 2.815, 95% CI: 1.682-4.711), sleep duration <8 hours (OR = 2.294, 95% CI: 1.352-3.892), and irregular diet (OR = 3.264, 95% CI: 1.837-5.801) were independent risk factors for short stature. Conversely, a preference for meat and dairy products was independently associated with a reduced risk (OR = 0.519, 95% CI: 0.303-0.888). Details are shown in Table 4.

A nomogram model was developed based on the eight identified risk factors (Figure 4a). The diagnostic equation derived from multivariable logistic regression was: ln(p/(1-p)) = 2.3839 × paternal height + 2.0875 × maternal height + 2.2107 × birth weight + 1.2718 × prematurity + 1.0400 × weekly exercise frequency + 0.7789 × sleep duration + 1.2085 × regular diet − 0.6314 × preference for meat/dairy − 4.6828. All predictors were entered as binary variables, coded as follows: paternal height (<160 cm = 1, ≥160 cm = 0), maternal height (<150 cm = 1, ≥150 cm = 0), birth weight (<2.5 kg = 1, ≥2.5 kg = 0), prematurity (yes = 1, no = 0), weekly exercise frequency (<3 times/week = 1, ≥3 times/week = 0), sleep duration (<8 h = 1, ≥8 h = 0), regular diet (no = 1, yes = 0), and preference for meat/dairy (no = 1, yes = 0).

The AUC of the nomogram model for predicting short stature occurrence was 0.858 (95% CI: 0.815-0.900), which is shown in Figure 4b. Calibration assessed by the Hosmer–Lemeshow test showed no significant deviation between predicted and observed probabilities (P = 0.104), indicating satisfactory fit (Figure 5a). Decision curve analysis further showed that the nomogram provided greater net clinical benefit across a broad range of threshold probabilities compared with strategies assuming all or no children at risk, supporting its value in short stature risk prediction and clinical decision-making. The DCA curve is displayed in Figure 5b.