This cross-sectional study consecutively enrolled adults who underwent routine health examinations at the Health Management Center of Jiaozuo Second People’s Hospital between January 1 and December 31, 2024. The study protocol received approval from the Medical Ethics Committee of Jiaozuo Second People’s Hospital (Approval No: KY2025-11-010) and was conducted in accordance with the principles of the Declaration of Helsinki. Due to the retrospective design, which relied exclusively on anonymized electronic health record (EHR) data, the requirement for informed consent was waived.

Eligible participants were required to be aged 18 years or older and to have completed a standardized check-up protocol that included thyroid ultrasonography, anthropometric measurements, and essential laboratory tests—specifically thyroid function, thyroid autoantibodies, and urinary iodine concentration. Additionally, individuals were only included if their EHRs contained complete data on demographic characteristics, medical history, and lifestyle factors, defined as having less than 20% missing information across these domains. Key exclusion criteria comprised a history of thyroid surgery, radioiodine therapy, or neck radiotherapy; a prior pathological diagnosis of thyroid nodules (e.g., via fine-needle aspiration or surgery); current pregnancy or lactation; severe hepatic or renal dysfunction (Child-Pugh class C or eGFR <30 mL/min/1.73m²); and current use of medications known to affect thyroid function, such as levothyroxine, methimazole, lithium, or amiodarone.

After applying these criteria, a total of 12,468 participants were included in the final analysis. An a priori sample size calculation was performed to ensure statistical robustness. Based on an anticipated thyroid nodule prevalence of 35% (3), with α = 0.05, β = 0.2 (power = 80%), and a minimum detectable odds ratio of 1.5 for key exposures such as insufficient iodine intake, a minimum of 10,800 participants was required. The final sample size of 12,468 exceeded this threshold, ensuring sufficient power to detect clinically meaningful associations.

All study data were systematically extracted from the hospital’s integrated electronic health record (EHR) system, which consolidates standardized information from health examinations, laboratory testing, medication records, and clinical diagnoses into a unified platform. The primary outcome of this study was the presence of thyroid nodules, defined as the sonographic identification of one or more focal lesions with a diameter of 3 mm or greater. All detected nodules were classified according to the Chinese Thyroid Imaging Reporting and Data System (TI-RADS) (1). Under this system, TI-RADS 2 indicates benign nodules (e.g., simple cysts or spongiform nodules); TI-RADS 3 indicates probably benign nodules with a malignancy risk below 2%; TI-RADS 4A reflects mildly suspicious nodules (malignancy risk 2%–10%); and TI-RADS 4B-5 denotes highly suspicious nodules with a malignancy risk exceeding 10%. For analytical purposes, nodules categorized as TI-RADS 4A or higher were collectively defined as “suspicious malignant nodules.”

A comprehensive set of explanatory variables was pre-specified based on clinical relevance and prior literature. Demographic variables included sex, age (analyzed continuously and categorized into <30, 30–39, 40–49, and ≥50 years for descriptive summaries), educational attainment (primary school or below, junior high school, high school/technical secondary school, college or above), and occupation type (e.g., government/institutional employee, enterprise employee, self-employed, retired, other). Iodine nutrition status was assessed via urinary iodine concentration (UIC), measured using arsenic-cerium catalytic spectrophotometry, and classified as insufficient (<100 μg/L), adequate (100–299 μg/L), or excessive (≥300 μg/L). Key metabolic indicators comprised body mass index (BMI), calculated as weight in kilograms divided by height in meters squared and categorized per Chinese guidelines as underweight (<18.5 kg/m²), normal (18.5–23.9 kg/m²), overweight (24.0–27.9 kg/m²), or obese (≥28.0 kg/m²). Hypertension was defined as systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg or a prior physician diagnosis; diabetes was defined as fasting blood glucose ≥7.0 mmol/L or a previous clinical diagnosis.

Thyroid function was evaluated using serum levels of thyroid-stimulating hormone (TSH; reference range: 0.27–4.2 mIU/L), free triiodothyronine (FT3; 3.1–6.8 pmol/L), and free thyroxine (FT4; 12.0–22.0 pmol/L). Abnormal TSH was defined as values outside the reference interval, encompassing both subclinical and overt thyroid dysfunction. Thyroid autoimmunity was assessed via thyroid peroxidase antibody (TPOAb; positive ≥34 IU/mL) and thyroglobulin antibody (TgAb; positive ≥115 IU/mL). Autoimmune thyroiditis (AIT) was defined as seropositivity for TPOAb and/or TgAb, while a clinical history of thyroiditis (e.g., Hashimoto’s thyroiditis, subacute thyroiditis) was ascertained from EHR documentation (10–12).

Lifestyle variables included smoking status (never vs. ever—the latter defined as having smoked for at least 1 year or having quit within the past year), alcohol consumption (never vs. regular drinking, the latter defined as consuming two or more alcoholic drinks per week for five or more years), and physical activity level, classified as either <150 min or ≥150 min per week of moderate-to-vigorous intensity exercise (≥3 METs) (13). Relevant medical history variables included family history of thyroid disease among first- or second-degree relatives and history of non-neck radiation exposure, encompassing both medical (e.g., CT or MRI scans) and occupational sources.

Thyroid ultrasonography was uniformly performed using a Logiq E9 system (GE Healthcare, USA) equipped with a 7.5–10 MHz linear array transducer. All scans were conducted by certified physicians with at least 5 years of specialized experience in thyroid imaging, adhering to a standardized protocol to systematically evaluate nodule characteristics including size, margin, shape, internal echogenicity, presence of calcifications, and vascularity in accordance with established guidelines (1). To ensure diagnostic consistency, inter-observer agreement was formally assessed through independent re-evaluation of a randomly selected 10% of participants (n = 1,247) by a second senior physician, yielding a Kappa statistic of 0.89 (95% CI, 0.86–0.92), which reflects excellent reproducibility.

A multi-layered quality control framework was implemented throughout the study. During the data extraction phase, two trained researchers independently abstracted information from the electronic health records, with any discrepancies adjudicated by a third senior researcher. Double data entry was employed for key variables—including TI-RADS classification, TSH levels, TPOAb status, and medication history—to minimize transcription errors. The overall rate of missing data was 3.2%, primarily affecting lifestyle-related variables. To address this, multiple imputation using the MICE algorithm was applied, creating five imputed datasets with 10 iterations. The imputation model incorporated all explanatory and outcome variables to preserve statistical validity and reduce bias (14). Distributions of key variables were compared before and after imputation, confirming the robustness of the imputation process.

Laboratory measurements—including urinary iodine concentration, thyroid function tests, and autoantibody assays—were performed in the hospital’s accredited clinical laboratory, which maintains rigorous internal quality control procedures and participates in external quality assessment schemes to ensure analytical accuracy. Furthermore, all sonographers involved in the study received uniform training on the application of the TI-RADS classification criteria prior to the study’s initiation, thereby standardizing diagnostic assessments across operators.

All statistical analyses were conducted using SPSS Statistics 26.0 (IBM Corp., USA) and R 4.2.1 (R Foundation for Statistical Computing). Continuous variables are summarized as mean±standard deviation if normally distributed, or as median with interquartile range otherwise; group comparisons were performed using Student’s t-test, ANOVA, or the Kruskal–Wallis test as appropriate. Categorical variables are presented as frequencies with percentages, and compared using the chi-square test or Fisher’s exact test for small cell counts.

To identify factors independently associated with thyroid nodules, multivariable logistic regression was applied. Potential collinearity among independent variables was carefully considered during model construction. In particular, variables with known biological correlations (e.g., urinary iodine, TSH, and TPOAb) were rigorously evaluated. Given their irreplaceable clinical and pathophysiological significance, all were retained in the final model. For missing data (overall missing rate 3.2%), we performed multiple imputation using the Multivariate Imputation by Chained Equations (MICE) algorithm. Five imputed datasets were created with 10 iterations to ensure convergence, and the imputation model included all analytical variables. A comparison of key variable distributions before and after imputation confirmed the robustness of the procedure (14). Variable selection was conducted a priori based on established biological plausibility and previous literature, and included sex, age (as a continuous variable), iodine nutrition status, BMI category, TSH status (normal/abnormal), TPOAb status (negative/positive), history of thyroiditis, physical activity level, family history of thyroid disease, and history of non-neck radiation exposure. Results are reported as adjusted odds ratios (ORs) with corresponding 95% confidence intervals. Owing to the limited number of cases classified as suspicious malignant nodules (n = 176), correlates for this subgroup were analyzed descriptively and presented as crude ORs without multivariable adjustment.

A two-tailed p-value <0.05 was considered statistically significant. In line with the cross-sectional design of the study, all results are interpreted as statistical associations rather than evidence of causation, and caution is exercised to avoid overinterpretation.

The final analytical cohort comprised 12,468 adults, of whom 5,823 (46.70%) were male and 6,645 (53.30%) were female, with a mean age of 42.3 ± 11.5 years. The majority of participants exhibited adequate iodine nutrition (68.23%) and fell within the normal body mass index range (41.25%). Thyroid function, as reflected by TSH levels, was within the normal range in 79.36% of the population. Evidence of thyroid autoimmunity—defined as positivity for TPOAb and/or TgAb—was present in 12.86% of subjects, while 8.76% had a documented clinical history of thyroiditis. A family history of thyroid disease was reported by 1,205 individuals (9.66%), and 3,740 participants (29.99%) engaged regularly in moderate-to-vigorous physical activity, as defined by ≥150 min per week. Consistent with the study’s exclusion criteria, no participants were using medications known to influence thyroid function at the time of assessment. The overall profile reflects a generally middle-aged health-seeking population with a low prevalence of major thyroid-related comorbidities. A comprehensive summary of baseline characteristics, stratified by thyroid nodule status, is presented in Table 1.

Thyroid nodules were sonographically identified in 4,409 participants, yielding an overall prevalence of 35.36% within this health examination cohort. Among the detected nodules, the vast majority (96.01%) were classified as either benign or probably benign: 2,812 (63.78%) as TI-RADS 2 (benign) and 1,421 (32.23%) as TI-RADS 3 (probably benign). A smaller proportion of nodules exhibited features warranting higher clinical suspicion, with 156 (3.54%) categorized as TI-RADS 4A (mildly suspicious) and 20 (0.45%) as TI-RADS 4B–5 (highly suspicious). Collectively, nodules considered suspicious for malignancy (defined as TI-RADS 4A or higher) accounted for 3.99% of all detected nodules. This distribution underscores that while thyroid nodules are a common sonographic finding in this population, the overwhelming majority display imaging features associated with a low risk of malignancy.

In the univariate analysis, multiple demographic, metabolic, and lifestyle factors demonstrated significant associations with the presence of thyroid nodules. Specifically, female sex, advanced age, insufficient or excessive iodine intake, obesity (BMI ≥ 28 kg/m²), abnormal TSH levels, positive TPOAb status, history of thyroiditis, smoking, alcohol consumption, physical inactivity, family history of thyroid disease, and history of non-neck radiation exposure were all significantly associated with a higher prevalence of thyroid nodules (all p < 0.05). In contrast, neither hypertension nor diabetes showed a statistically significant association with nodule presence in this initial analysis (both p > 0.05; see Table 2). These findings not only align with previously reported risk profiles but also support the inclusion of these significant variables in the subsequent multivariable regression model to identify independent correlates.

Multivariable logistic regression, adjusted for key demographic, metabolic, and immunological confounders, identified several factors independently associated with thyroid nodule presence. Notably, the stability of the coefficient estimates and the absence of inflated standard errors in the final model indicate that collinearity—including the anticipated biological correlation between variables such as TSH and TPOAb status—did not materially threaten the reliability or interpretability of the odds ratios presented below.

Female sex was significantly associated with increased odds of nodules (OR = 1.65, 95% CI: 1.51–1.80), as was advancing age, with each 10-year increase corresponding to an OR of 1.55 (95% CI, 1.47–1.64). Modifiable metabolic and endocrine factors also demonstrated significant associations: insufficient iodine intake (OR = 1.38, 95% CI: 1.23–1.55), obesity (OR = 1.35, 95% CI: 1.21–1.50), abnormal TSH levels (OR = 1.32, 95% CI: 1.19–1.46), and TPOAb positivity (OR = 1.28, 95% CI: 1.13–1.45) were each independently correlated with higher nodule prevalence. A strong association was observed among participants with a family history of thyroid disease (OR = 2.31, 95% CI: 2.04–2.61), while a history of non-neck radiation exposure was associated with more modestly elevated odds (OR = 1.17, 95% CI: 1.05–1.31). In contrast, regular moderate-to-vigorous physical activity emerged as a significant protective factor (OR = 0.73, 95% CI: 0.66–0.81). Several factors that were significant in univariate analysis—including history of thyroiditis, smoking, alcohol consumption, educational attainment, and occupation—did not retain independent significance in the fully adjusted model (all p > 0.05), suggesting that their effects may be mediated or confounded by other variables. These results highlight a distinct profile of both non-modifiable and potentially modifiable factors independently associated with thyroid nodules in this adult health examination population (see Table 3).

Among the 4,409 participants with sonographically detected thyroid nodules, 176 (4.0%) were classified as having suspicious malignant nodules (TI-RADS 4A or higher). Descriptive analyses revealed notable differences in the profile of these participants compared to those with benign or probably benign nodules. Individuals with suspicious nodules were more frequently aged 50 years or older (58.52% vs. 46.89%), reported a family history of thyroid disease in 28.41% of cases (vs. 8.92%), and exhibited higher rates of thyroid autoimmunity, with TPOAb positivity observed in 26.14% (vs. 11.65%). Furthermore, abnormal TSH levels were more prevalent in this subgroup (29.55% vs. 19.87%). These patterns align with established risk profiles for thyroid malignancy reported in the literature. However, due to the limited number of cases in this category (n = 176), a formal multivariable analysis was not performed to control for potential confounders. Therefore, the observed associations should be interpreted as exploratory and hypothesis-generating, highlighting factors that warrant investigation in larger, prospective studies with pathological confirmation. The full comparative profile is presented in Table 4.