Abstract
Research has shown that the concurrent presence of Diabetes Mellitus (DM) and Thyroid Dysfunction (TD) can exacerbate diabetes-related complications and impose a significant economic burden on healthcare systems. Therefore, this study aimed to develop a nomogram model for predicting the risk of TD in patients with Type 2 Diabetes Mellitus (T2DM) and to validate its predictive performance. A total of 1853 patients with T2DM diagnosed at the First Hospital of Hebei Medical University from 2019 to 2024 were included in the study. The dataset was randomly divided into a training set (n = 1297) and a validation set (n = 556) at a 7:3 ratio using the R software. Univariate and multivariate logistic regression analyses were conducted to identify predictors of TD, and these predictors were used to construct the nomogram model. The model was evaluated using the receiver operating characteristic (ROC) curve and the area under the curve (AUC), calibration curve, the Hosmer-Lemeshow test, and decision curve analysis (DCA). HDL-C, BUN, gender, GLU, Hypertension, Hyperuricemia, Coronary Heart Disease, and Liver disease were identified as predictors of TD. A nomogram model was constructed based on these eight factors. The model demonstrated good discrimination in both the training and validation sets. The calibration curves indicated a good fit of the model in both datasets. The decision curve analysis showed that the model had good clinical applicability. The nomogram developed in this study can predict the risk of developing TD in patients with T2DM. It enables clinicians to identify T2DM patients at high risk of concurrent TD, which may help facilitate the development of effective preventive measures and improve patient prognosis.
Supplementary Information
The online version contains supplementary material available at 10.1038/s41598-026-36582-3.
Keywords: Type 2 diabetes mellitus, Thyroid disorders, Nomogram, Risk factor
Subject terms: Medical research, Diabetes, Risk factors
Introduction
Diabetes Mellitus (DM) and Thyroid dysfunction (TD) are among the most prevalent endocrine disorders globally. Diabetes is characterized by elevated blood glucose levels, while thyroid disease involves abnormal thyroid hormone secretion1. In clinical settings, it is common to observe the coexistence of DM and TD2–5. Furthermore, a growing body of evidence suggests a complex interplay between these two conditions. Thyroid hormones are essential for maintaining glucose homeostasis and can influence blood glucose levels by affecting insulin secretion6–8. Conversely, diabetes can impair thyroid function, leading to dysfunction9, which in turn affects insulin secretion and clearance, exacerbating glucose dysregulation and worsening diabetes symptoms10,11. Additionally, Type 2 Diabetes Mellitus (T2DM) patients with TD are more susceptible to complications and incur higher healthcare costs compared to those without thyroid dysfunction12–15. This interplay increases the complexity of disease management and necessitates more sophisticated clinical interventions.
Currently, the assessment of TD risk in DM relies heavily on clinical indicators and subjective judgment, lacking a comprehensive tool that integrates multiple risk factors to predict the risk of concurrent TD in DM patients. Therefore, developing a risk prediction model based on clinical data is imperative. This study aims to identify predictors of TD in patients with T2DM and to construct a nomogram-based risk prediction model using clinical data from T2DM patients. This model is intended to provide a practical tool for clinicians to assess the risk of TD in T2DM patients, enabling the early identification of T2DM patients at high risk of TD. This approach will help T2DM patients better control their blood glucose, reduce the risk of cardiovascular events, and improve their overall health status.
Methods
Study designs and participants
This is a retrospective study that included patients diagnosed with T2DM who were hospitalized at the First Hospital of Hebei Medical University from July 2019 to January 2024. The study was approved by the Ethics Committee of the First Hospital of Hebei Medical University (Approval Number: 20200615). Due to the single-center and retrospective design of the study, the review committee of the First Hospital of Hebei Medical University waived the requirement for written informed consent. The study was conducted in compliance with the Declaration of Helsinki. The study flowchart is presented in Fig. 1.
Fig. 1.
Research pathway diagram.
The inclusion criteria were as follows: (1) Age ≥ 18 years; (2) Patients diagnosed with T2DM according to the 1999 World Health Organization criteria; (3) Available thyroid function tests. The exclusion criteria were: (1) Other types of diabetes; (2) Cancer; (3) Infection; (4) Severe liver dysfunction; (5) Chronic renal insufficiency due to other causes; (6) Inaccurate urine albumin-to-creatinine ratio (UACR); (7) Patients with incomplete key data or missing target variable values.
TD was defined as: (1) Serum TSH < 0.35 mIU/L or > 4.94 mIU/L, or (2) Serum FT4 < 12 pmol/L or > 22 pmol/L, or (3) A previous diagnosis of hyper-/hypothyroidism or Hashimoto’s thyroiditis confirmed by an endocrinologist with current thyroid medication. The diagnostic criteria followed the 2022 European Thyroid Association guidelines.
Data collection
Following a comprehensive literature review, we identified 44 potential risk factors that may be associated with TD in patients with T2DM. These factors can be categorized as follows: (1) Demographic characteristics: age, gender, body mass index (BMI), blood pressure, smoking history, alcohol consumption, and education attainment; (2) Clinical information: Family history of diabetes, history of insulin use, and medical history including diabetic foot, hypertension, hyperuricemia, coronary heart disease, stroke, liver disease, gastrointestinal diseases, urinary diseases, and respiratory diseases; (3) Blood and urine biochemical indicators: White blood cells (WBC), platelets (PLT), hemoglobin (Hb), uric acid (UA), serum creatinine (SCr), fasting blood glucose (FBG), glycated hemoglobin (HbA1c), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), blood urea nitrogen (BUN), triglycerides (TG), alkaline phosphatase (ALP), albumin (ALB), total cholesterol (TC), hematocrit (HCT), mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean platelet volume (MPV), alanine aminotransferase (ALT), aspartate aminotransferase (AST), protein (PRO), glucose (GLU), leukocytes (LEU), erythrocytes (ERY), nitrites (NIT), and ketones (KET). All laboratory values are residual data generated during routine patient care by the Department of Clinical Laboratory (ISO 15189 accredited) and were exported from the LIS. No extra specimens or assays were performed specifically for this research.
Analytical platforms and quality control:
Biochemical parameters (BUN, Cr, glucose, lipid profile, etc.) were measured on a Beckman Coulter AU 5800 automatic analyser using original Beckman reagents; two-level internal controls were analysed daily and evaluated with Westgard rules.
Thyroid-function tests (TSH and FT4) were performed with a Siemens Centaur XP chemiluminescence immunoanalyser; reagent lots and calibration curves are traceable to WHO reference standards, with inter-assay CVs kept within ± 5%.
Urinalysis (PRO, GLU, LEU, ERY, NIT, KET) was carried out with a Siemens Clinitek Advantus analyser and Multistix 10SG dipsticks; negative and positive controls were run each day.
Data-extraction rule: for each patient only the most recent result obtained within 30 days before the index hospital or outpatient visit was retrieved, preventing bias from multiple repeated measurements. No analytical protocols were modified and no extra specimens or reagents were consumed for the study.
Statistical analysis
The data were analyzed using SPSS 26.0 software and R software (version 4.3.1). Quantitative data that met the criteria for normal distribution were presented as mean ± standard deviation (Mean ± SD), and the Student’s t-test was used for group comparisons. If the data did not meet the criteria for normal distribution, they were presented as median [M (P25, P75)] (P25 and P75 denote the 25th and 75th percentiles, respectively) and compared using the Mann-Whitney U test. Qualitative variables were reported as frequencies and percentages, and comparisons between groups were performed using Fisher’s exact, Pearson’s chi-square, or Freeman-Halton tests based on expected frequencies. Variables that were significant in the univariate analysis were included in the multivariate logistic regression analysis to identify statistically significant predictors for constructing the Nomogram. The rms package in R software was used to construct the Nomogram, and the rmda package was used for decision curve analysis (DCA) to assess the clinical effectiveness of the model. The area under the curve (AUC) of the receiver operating characteristic (ROC) curve was used to evaluate the discriminative ability of the model, and the Hosmer-Lemeshow test was used to assess the goodness of fit of the model. The level of statistical significance was defined as p < 0.05.
Results
Baseline characteristics
This study included a total of 1853 patients with T2DM, of whom 174 were diagnosed with TD, resulting in a prevalence rate of 9.4%. The mean age of the participants was 53.9 ± 13.6 years, and 60.9% of them were male. Compared to patients without TD, those with TD were more likely to be older females with higher education levels. Additionally, they exhibited lower levels of HbA1c, Hb, BUN, Cr, and HCT. Furthermore, TD patients showed a higher incidence of hypertension, hyperuricemia, coronary heart disease, liver disease, gastrointestinal diseases, and urinary diseases. Notably, TD patients had a lower positive rate of GLU (Table 1). The study sample was randomly divided into a training set and a validation set using the R package at a ratio of 7:3, with 1297 patients in the training set and 556 in the validation set. We analyzed the baseline characteristics between the training and validation sets to ensure the validity of the dataset division. The results indicated no significant statistical differences in the sample data between the two sets (p > 0.05) (Table S1).
Table 1.
Demographics and baseline data.
| Variables | Total | Non- Thyroid Disorders |
Thyroid Disorders |
p- value |
|---|---|---|---|---|
| N | 1853 | 1679 | 174 | |
| The demographic characteristics: | ||||
| Age (year) | 53.9 ± 13.6 | 53.6 ± 13.7 | 56.6 ± 11.9 | 0.003 |
| Gender/n (%) | < 0.001 | |||
| Male | 1128(60.9) | 1048(62.4) | 80(46.0) | |
| Female | 725(39.1) | 631(37.6) | 94(54.0) | |
| BMI (kg/m2) | 27.0 ± 4.0 | 27.0 ± 4.0 | 27.4 ± 3.9 | 0.226 |
| SBP (mmHg) | 133.2 ± 18.8 | 133.2 ± 19.0 | 133.4 ± 17.5 | 0.874 |
| DBP (mmHg) | 76.2 ± 11.5 | 76.4 ± 11.6 | 74.6 ± 10.5 | 0.049 |
| Smoking history/case (%) | 752(40.6) | 690(41.1) | 62(35.6) | 0.162 |
| Alcohol consumption/case (%) | 936(50.5) | 857(51) | 79(45.4) | 0.157 |
| Education attainment/n (%) | 0.010 | |||
| Below high school | 676(36.5) | 628(37.4) | 48(27.6) | |
| Above high school | 1177(63.5) | 1051(62.6) | 126(72.4) | |
| The clinical information: | ||||
| Family history of diabetes/n (%) | 1068(57.6) | 957(57.0) | 111(63.8) | 0.084 |
| Insulin use history/case (%) | 526(28.4) | 472(28.1) | 54(31.0) | 0.416 |
| Diabetic foot/case (%) | 95(5.1) | 88(5.2) | 7(4.0) | 0.488 |
| Hypertension/n (%) | 853(46.0) | 752(44.8) | 101(58.0) | 0.001 |
| Hyperuricemia/n (%) | 156(8.4) | 131(7.8) | 25(14.4) | 0.003 |
| CoronaryHeartDisease/n (%) | 219(11.8) | 181(10.8) | 38(21.8) | < 0.001 |
| Stroke/n (%) | 162(8.7) | 140(8.3) | 22(12.6) | 0.056 |
| Liver disease/n (%) | 187(10.1) | 144(8.6) | 43(24.7) | < 0.001 |
| Gastrointestinal diseases/n (%) | 86(4.6) | 70(4.2) | 16(9.2) | 0.003 |
| Urinary diseases/n (%) | 106(5.7) | 87(5.2) | 19(10.9) | 0.002 |
| RespiratorySystemDiseases/n (%) | 56(3.0) | 47(2.8) | 9(5.2) | 0.082 |
| Blood and urine biochemical indicators: | ||||
| HbA1c (%) | 9.4(7.6, 11.2) | 9.4(7.7, 11.2) | 8.6(7.3, 10.3) | 0.002 |
| PRO/case (%) | 337(18.3) | 315(18.8) | 22(13.3) | 0.085 |
| GLU/case (%) | 1141(61.9) | 1061(63.2) | 80(48.5) | < 0.001 |
| LEU/case (%) | 727(39.4) | 664(39.5) | 63(37.7) | 0.646 |
| ERY/case (%) | 335(18.2) | 303(18.0) | 32(19.4) | 0.668 |
| NIT/case (%) | 130(7.0) | 122(7.3) | 8(4.8) | 0.247 |
| KET/case (%) | 495(26.8) | 461(27.5) | 34(20.6) | 0.058 |
| FBG (mmol/L) | 9.2(7.5, 11.7) | 9.3(7.6, 11.8) | 8.6(7.4, 11.3) | 0.060 |
| HDL-C (mmol/L) | 1.0(0.9, 1.2) | 1.0(0.9, 1.2) | 1.0(0.9, 1.2) | 0.225 |
| LDL-C (mmol/L) | 3.1 ± 0.9 | 3.1 ± 0.9 | 3.0 ± 0.9 | 0.263 |
| WBC (×109/L) | 6.6 ± 2.0 | 6.6 ± 2.0 | 6.5 ± 1.7 | 0.520 |
| PLT (×109/L) | 219(182, 259.4) | 218(182, 259) | 222(180, 259) | 0.828 |
| Hb (g/L) | 141(130,153) | 141(130, 153) | 140(128, 148) | 0.021 |
| BUN (mmol/L) | 5.5(4.4, 6.7) | 5.5(4.5, 6.7) | 5.2(4.4, 6.4) | 0.046 |
| UA (µmol/L) | 327.9(271.2, 395.6) | 328.9(272.7, 395.2) | 334.7(270.7, 408.8) | 0.819 |
| sCr (µmol/L) | 60.7(50.8, 71.7) | 61(51.1, 71.8) | 58(46.6, 71.1) | 0.016 |
| ALP (U/L) | 78(64, 95) | 78(64, 95) | 76(64, 95) | 0.139 |
| ALB (g/L) | 41.2 ± 4.9 | 41.2 ± 5.0 | 41.1 ± 4.2 | 0.644 |
| HCT (%) | 0.42(0.39, 0.45) | 0.42(0.39, 0.45) | 0.42(0.38, 0.44) | 0.025 |
| MCV (fL) | 90.1(87.2, 93.1) | 90.1(87.2, 93.0) | 90.7(87.0, 93.4) | 0.528 |
| MCH (pg) | 30.3(29.2,31.3) | 30.3(29.2, 31.3) | 30.4(29.3, 31.3) | 0.639 |
| MPV (fL) | 8.9(8.2, 9.8) | 8.9(8.2, 9.8) | 9.0(8.2, 9.9) | 0.837 |
| ALT (U/L) | 20.4(14.3, 32) | 20.4(14.1, 31.7) | 20.1(14.2, 32.9) | 0.878 |
| AST (U/L) | 18.7(14.9, 25.2) | 18.6(14.8, 25.0) | 18.5(14.9, 25.5) | 0.676 |
| TG (mmol/L) | 1.6(1.1, 2.5) | 1.6(1.1, 2.5) | 1.6(1.1, 2.2) | 0.818 |
| TC (mmol/L) | 5.1 ± 1.4 | 5.1 ± 1.4 | 4.9 ± 1.3 | 0.177 |
PRO/GLU/LEU/ERY/NIT/KET: urine dipstick chemistry; negative = 0, positive = 1; data presented as frequency (case) and percentage (%).
BMI: Body mass index; DBP: Diastolic blood pressure; SBP:Systolic blood pressure; HbA1c: Hemoglobin A1c; PRO:Protein; GLU:Glucose; LEU:Leukocyte Esterase; ERY: Erythrocytes; NIT: Nitrites; KET:Ketones; FBG: Fasting Blood Glucose; HDL-C: High-density lipoproteincholesterol; LDL-C: Low-density lipoprotein cholesterol; WBC: White BloodCell; PLT: Platelet; Hb: Hemoglobin; UA: Uric Acid ; sCr: Serum Creatinine;BUN: Blood Urea Nitrogen; ALP: Alkaline Phosphatase; ALB: Albumin; HCT:Hematocrit; MCV: Mean Corpuscular Volume; MCH: Mean CorpuscularHemoglobin; MPV: Mean Platelet Volume; ALT: Alanine Aminotransferase;AST: Aspartate Aminotransferase; TC: Total cholesterol; TG: Triglycerides
Univariate variable screening and multivariate logistic regression analysis of TD occurrence in the T2DM
In the training sample, univariate regression analysis identified 12 statistically significant variables: Age, Gender, Family History of Diabetes, Hypertension, Hyperuricemia, Coronary Heart Disease, Liver disease, Gastrointestinal diseases, HbA1c, GLU, HDL-C, and BUN. We conducted a multivariate logistic regression analysis using the occurrence of TD as the dependent variable and the aforementioned 12 variables as independent variables. The analysis revealed that the following variables were statistically significant: HDL-C (OR = 0.404, 95% CI: 0.177–0.922, p = 0.032), BUN (OR = 0.876, 95% CI: 0.779–0.986, p = 0.028), Gender (OR = 2.090, 95% CI: 1.359–3.215, p = 0.001), GLU (OR = 0.644, 95% CI: 0.428–0.969, p = 0.035), Hypertension (OR = 1.609, 95% CI: 1.053–2.459, p = 0.028), Hyperuricemia (OR = 2.051, 95% CI: 1.147–3.667, p = 0.015), Coronary Heart Disease (OR = 1.831, 95% CI: 1.080–3.105, p = 0.025), and Liver disease (OR = 3.147, 95% CI: 1.897–5.221, p < 0.001), as shown in Table 2.
Table 2.
Variables by logistic regression analysis (training sample).
| Variables | Univariate analysis | Multivariate analysis | ||
|---|---|---|---|---|
| OR (95% CI) | p-value | OR (95% CI) | p-value | |
| Age (year) | 1.018(1.003–1.033) | 0.015 | 1.006(0.988–1.025) | 0.506 |
| Gender/ n (%) | ||||
| Male | Reference | |||
| Female | 1.994(1.374–2.894) | 0.000 | 2.090(1.359–3.215) | 0.001 |
| BMI | 1.029(0.984–1.076) | 0.212 | ||
| SBP | 0.999(0.989–1.009) | 0.836 | ||
| DBP | 0.989(0.973–1.006) | 0.195 | ||
| Smoking history/ n (%) | ||||
| No | Reference | |||
| Yes | 0.755(0.512–1.115) | 0.158 | ||
| Alcohol consumption/ n (%) | 0.900(0.621–1.303) | 0.576 | ||
| Education attainment/ n (%) | ||||
| Below high school | Reference | |||
| Above high school | 1.248(0.837–1.862) | 0.277 | ||
| Family history of diabetes/ n (%) | ||||
| No/Unknown | Reference | |||
| Yes | 1.667(1.124–2.471) | 0.011 | 1.308(0.854–2.003) | 0.217 |
| Insulin use history/ n (%) | ||||
| No/Unknown | Reference | |||
| Yes | 0.900(0.590–1.372) | 0.624 | ||
| Diabetic foot/ n (%) | ||||
| No/Unknown | Reference | |||
| Yes | 0.881(0.374–2.078) | 0.772 | ||
| Hypertension/ n (%) | ||||
| No/Unknown | Reference | |||
| Yes | 1.725(1.184–2.511) | 0.004 | 1.609(1.053–2.459) | 0.028 |
| Hyperuricemia/ n (%) | ||||
| No/Unknown | Reference | |||
| Yes | 2.133(1.266–3.594) | 0.004 | 2.051(1.147–3.667) | 0.015 |
| Coronary Heart Disease/ n (%) | ||||
| No/Unknown | Reference | |||
| Yes | 2.334(1.466–3.715) | 0.000 | 1.831(1.080–3.105) | 0.025 |
| Stroke/n (%) | ||||
| No/Unknown | Reference | |||
| Yes | 1.400(0.787–2.489) | 0.252 | ||
| Liver disease/ n (%) | ||||
| No/Unknown | Reference | |||
| Yes | 3.501(2.235–5.486) | 0.000 | 3.147(1.897–5.221) | 0.000 |
| Gastrointestinal diseases/ n (%) | ||||
| No/Unknown | Reference | |||
| Yes | 2.357(1.221–4.552) | 0.011 | 1.232(0.547–2.773) | 0.615 |
| Urinary diseases/ n (%) | ||||
| No/Unknown | Reference | |||
| Yes | 1.838(0.939-3.600) | 0.076 | ||
| Respiratory System Diseases/n (%) | ||||
| No/Unknown | Reference | |||
| Yes | 2.204(0.950–5.115) | 0.066 | ||
| HbA1c | 0.897(0.824–0.977) | 0.012 | 0.950(0.859–1.051) | 0.321 |
| PRO/ n (%) | ||||
| Negative | Reference | |||
| Positive | 0.562(0.310–1.020) | 0.058 | ||
| GLU/ n (%) | ||||
| Negative | Reference | |||
| Positive | 0.512(0.349–0.751) | 0.001 | 0.644(0.428–0.969) | 0.035 |
| LEU/ n (%) | ||||
| Negative | Reference | |||
| Positive | 0.858(0.579–1.270) | 0.443 | ||
| ERY/ n (%) | ||||
| Negative | Reference | |||
| Positive | 1.210(0.754–1.943) | 0.430 | ||
| NIT/ n (%) | ||||
| Negative | Reference | |||
| Positive | 0.495(0.178–1.376) | 0.177 | ||
| KET/ n (%) | ||||
| Negative | Reference | |||
| Positive | 0.654(0.404–1.060) | 0.085 | ||
| FBG | 0.944(0.886–1.006) | 0.076 | ||
| HDL-C | 0.437(0.207–0.925) | 0.031 | 0.404(0.177–0.922) | 0.032 |
| LDL-C | 0.955(0.767–1.190) | 0.682 | ||
| WBC | 0.963(0.868–1.069) | 0.480 | ||
| PLT | 1.000(0.997–1.003) | 0.903 | ||
| Hb | 0.992(0.982–1.002) | 0.108 | ||
| BUN | 0.897(0.808–0.997) | 0.044 | 0.876(0.779–0.986) | 0.028 |
| UA | 1.001(0.999–1.002) | 0.542 | ||
| Cr | 0.995(0.986–1.004) | 0.265 | ||
| ALP | 0.996(0.988–1.003) | 0.265 | ||
| ALB | 0.982(0.944–1.022) | 0.369 | ||
| HCT | 0.049(0.001–1.943) | 0.108 | ||
| MCV | 0.984(0.954–1.015) | 0.311 | ||
| MCH | 0.969(0.893–1.053) | 0.462 | ||
| MPV | 0.897(0.763–1.055) | 0.189 | ||
| ALT | 0.994(0.985–1.002) | 0.155 | ||
| AST | 0.993(0.980–1.006) | 0.304 | ||
| TG | 0.947(0.850–1.056) | 0.329 | ||
| TC | 0.935(0.810–1.079) | 0.354 | ||
Values of P < 0.05 were bolded. Gender: female = 0 (Ref), male = 1, Analogous notation is provided for all other categorical variables.
95% CI: 95% Confidence interval; OR, Odds ratio.
Creation of a nomogram for predicting the risk of TD occurrence in the T2DM
Based on the findings of the multivariate logistic regression analysis, we developed a nomogram to predict the probability of TD in patients with T2DM concurrent. The predictive factors: HDL-C, BUN, gender, GLU, Hypertension, Hyperuricemia, Coronary Heart Disease, and Liver disease are represented by specific values on the nomogram for scoring purposes. Each predictive factor is assigned a score on the upper scale, by summing the scores for each factor, we calculate the total score for each individual, which is displayed on the lower scale, ranging from 0.1 to 0.9. The total score corresponds to the diagnostic probability at the bottom of the figure, indicating the risk of TD occurrence. A higher total score suggests a greater likelihood of TD occurrence (Fig. 2).
Fig. 2.
Nomogram to estimate the risk of TD in T2DM.
Evaluation and validation of the nomogram model in the training and validation samples
By plotting the ROC curves for the training and validation sets and calculating the AUC, we assessed the discrimination ability and predictive value of the nomogram model. The AUC for the training set was 0.747 (95% CI: 0.680–0.815), while the AUC for the validation set was 0.709 (95% CI: 0.657–0.762). These results demonstrate that the nomogram model exhibits moderate discrimination and predictive value (Fig. 3). The calibration curves were used to evaluate the fit of the nomogram model in both the training and validation sets. The calibration curves aligned closely with the 45-degree diagonal line indicated that the model demonstrated good accuracy in its predictions. The Hosmer-Lemeshow goodness-of-fit test showed no statistically significant difference between the observed and predicted probabilities in both the training and validation sets (p = 0.952 and 0.574, respectively). The study findings indicated a strong agreement between the predicted and actual probabilities, suggesting that the nomogram model was well-calibrated and effectively predicted the risk of TD occurrence (Fig. 4). The DCA curve was employed to assess the clinical utility of the prediction model. The DCA curve revealed that the model provided a significant net benefit in predicting TD occurrence in both the training and validation sets, underscoring its clinical value (Fig. 5).
Fig. 3.
ROC curves for predicting TD among T2DM patients in thetraining sample (A) and the validation sample (B).
Fig. 4.
Calibration curves of the nomogram prediction in thetraining sample (A) and validation sample (B).
Fig. 5.
Decision curve analysis curve of the nomogram based onthe data of training sample (A) and validation sample (B).
Discussion
TD and DM are the most prevalent endocrine disorders in the general population and frequently coexist. Untreated TD can adversely affect blood glucose control and increase the risk of complications in diabetes patients. Consequently, thyroid screening in patients with T2DM is of significant clinical importance. However, there is currently no satisfactory predictive tool available to identify T2DM patients who may have TD. Therefore, this study retrospectively analyzed 1853 patients with T2DM from our hospital, using the presence or absence of TD as the outcome variable, and developed a clinical prediction model for TD in patients with T2DM. We employed univariate logistic regression for variable selection and further identified significant predictors using multivariate logistic regression analysis. Based on these predictors, we constructed a nomogram model. The nomogram model assigns a score to each predictor and sums these scores to calculate the probability of TD occurrence for each individual. Its primary advantage is that it transforms complex regression equations into a visual format, enabling more intuitive and personalized TD prediction16–18. The results revealed that HDL-C, BUN, gender, GLU, Hypertension, Hyperuricemia, Coronary Heart Disease, and Liver disease are predictors of TD. To evaluate the model’s reliability, we conducted internal validation, and the results indicated that the model demonstrated moderate discrimination ability in both the training and validation sets. The calibration curve showed the high accuracy of the model in both groups. Furthermore, the DCA indicated that the model had good clinical applicability in both the training and validation sets. Catma et al.19. and Tabari et al.20, who have also investigated the prevalence and risk factors of thyroid dysfunction (TD) in patients with type 2 diabetes mellitus (T2DM). Similar to our results, they found that gender, dyslipidemia (low HDL, high TC, high TG), and comorbidities (such as hypertension) are risk factors for TD in T2DM patients. However, our study distinguishes itself from these previous works in several ways. The Catma team primarily focused on the impact of traditional thyroid function indicators and demographic factors on TD in T2DM patients. Meanwhile, the Tabari team explored the risk factors for TD in T2DM patients from the aspects of demographic factors, lifestyle, and metabolic indicators. Our model, in contrast, comprehensively integrates a wide range of indicators, including demographic factors, lifestyle, biochemical indicators, hematology indicators, and urinalysis. This multi-faceted approach allows for a more thorough exploration of the risk factors for TD in T2DM patients. Furthermore, we have developed a nomogram that transforms these risk factors into a visual assessment tool. This innovation enables clinicians to more intuitively predict TD risk. In addition, while current clinical guidelines generally recommend screening for thyroid function at the time of diabetes diagnosis, they provide little specific guidance on subsequent screening frequency. Our model addresses this shortcoming by offering a quantitative risk assessment tool that can identify high-risk individuals requiring closer monitoring. This advancement has the potential to significantly improve the management of T2DM patients at risk for TD.
The results of this study indicated that female sex is an independent risk factor for TD, which was consistent with findings from previous studies21,22. In patients with T2DM, women were more susceptible to TD compared to men. This disparity may be attributed to several factors, including hormonal differences. Research has shown that sex hormones such as estrogens, progesterone, and androgens can influence the synthesis and metabolism of thyroid hormones. Women have higher levels of estrogens and progesterone, while men have higher levels of androgens such as testosterone. These hormones can affect the thyroid through various pathways, leading to differences in thyroid function. Additionally, the varying physiological stages and basal metabolic rates between men and women can also impact thyroid function.
Besides the female sex, low HDL-C levels were another independent risk factor for TD. Studies have found that low HDL-C levels are associated with an increased risk of TD, particularly in individuals with thyroid cancer23. Several studies24–28 have shown that lower HDL-C levels were linked to a higher risk of thyroid cancer, although some studies did not find significant differences. However, overall, the reduction in HDL-C levels may represent an early event in the development of TD. This finding can be explained at the molecular level by the impact of thyroid hormones on lipid metabolism. Thyroid hormones (TH) influence the metabolism of high-density lipoprotein (HDL) by regulating the activity of cholesterol ester transfer protein (CETP) and phospholipid transfer protein (PLTP)29. CETP was responsible for transferring cholesterol from HDL-C to low-density lipoprotein (LDL-C) and very low-density lipoprotein (VLDL), while PLTP was involved in the formation of mature HDL-C forms. These proteins played a crucial role in regulating the concentration, size, and composition of circulating HDL particles. Furthermore, TH affected HDL metabolism through other mechanisms. For example, TH could stimulate hepatic lipase (HL), leading to the hydrolysis of HDL2 into HDL3 and promoting the conversion of intermediate-density lipoprotein (IDL) to LDL30. TH also stimulated the activity of 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase, a key enzyme in the cholesterol biosynthesis pathway, and promoted the conversion of cholesterol to bile acids31. All these pathways could influence HDL-C levels.
BUN and GLU are key indicators for evaluating renal function and glucose metabolism in diabetes patients32–34. In our study, we found that these two factors are also independent risk factors for TD. The results indicated that BUN and GLU levels are negatively correlated with the occurrence of TD. Specifically, diabetic patients with lower BUN levels and negative GLU results were more susceptible to thyroid dysfunction. However, in the general population, higher BUN levels and positive GLU results are associated with a higher risk of TD. This discrepancy may be due to the bidirectional regulatory relationship between BUN, GLU, and thyroid function. Thyroid hormones can influence BUN and GLU levels, and conversely, changes in BUN and GLU levels can impact thyroid function. BUN is an important marker of renal function, reflecting the state of protein metabolism and kidney function. It can affect thyroid hormone metabolism and function by modulating the kidney’s clearance of thyroid hormones35. When kidney function is compromised, it affects the metabolism and clearance rate of thyroid hormones, leading to an imbalance in thyroid hormone levels and resulting in thyroid dysfunction. Conversely, TD can also affect renal function. In hyperthyroidism, protein breakdown accelerates, potentially leading to elevated BUN levels. In hypothyroidism, the metabolic rate decreases, which may result in lower BUN levels. GLU levels reflect the state of glucose metabolism in the body. Thyroid hormones have a significant impact on glucose metabolism. In hyperthyroidism, the metabolic rate increases, leading to faster absorption and utilization of glucose, which may result in elevated urinary glucose levels. Conversely, in hypothyroidism, the metabolic rate decreases, slowing down glucose metabolism and potentially leading to lower urinary glucose levels. Studies have shown that hypothyroidism and subclinical hypothyroidism are the most prevalent among T2DM patients2, and hypothyroidism may mask the clinical symptoms and signs of diabetes36,37. Therefore, in the context of T2DM, diabetic patients with lower BUN levels and negative GLU results may be more susceptible to hypothyroidism and subclinical hypothyroidism. However, this conclusion remains exploratory and may represent a physiological phenomenon reproducible only in the specific subpopulation of “early T2DM with preserved renal function,” potentially related to reduced protein catabolism and elevated renal glucose threshold during early-stage hypothyroidism. Its stability and causal direction require validation in prospective multicenter cohorts.
In addition to the previously mentioned indicators, our study found a positive correlation between liver disease, hypertension, hyperuricemia, and coronary heart disease with the occurrence of TD38–40. Patients with T2DM who suffered from these conditions were more susceptible to TD. Hypertension, hyperuricemia, liver disease, and coronary heart disease are common metabolic and cardiovascular conditions that may influence thyroid function through various mechanisms. Hypertension can affect thyroid function by altering the metabolism and regulation of thyroid hormones. Patients with hypertension often exhibit excessive activation of the sympathetic nervous system, which could lead to fluctuations in thyroid hormone levels. Additionally, hypertension may reduce renal blood flow, affecting the clearance and metabolism of thyroid hormones. Hyperuricemia is closely associated with metabolic syndrome, and elevated uric acid levels may induce oxidative stress and inflammatory responses, which can interfere with the synthesis and secretion of thyroid hormones, thereby affecting thyroid function. The liver is a key site for the metabolism of thyroid hormones. Liver diseases such as cirrhosis and fatty liver can lead to abnormal metabolism of thyroid hormones, affecting their synthesis, conversion, and clearance. Coronary heart disease is closely linked to thyroid function. Thyroid hormones have a direct impact on myocardial function and heart rate. Coronary heart disease may influence the demand and action of thyroid hormones by directly affecting myocardial function and heart rate.
Limitations
This study has several limitations. First, some data were collected through patient self-reports and interviews, which may affect the accuracy of the data and the objectivity of the results. Second, as a retrospective, single-center study lacking data from other centers or regions, this research may introduce selection bias and limit the generalizability of the results; therefore, these results should be regarded as exploratory and require validation in prospective, multicenter cohorts. Third, the nomogram was only internally validated and requires external validation to confirm its reliability. Furthermore, this study did not differentiate between different types of thyroid diseases, such as hyperthyroidism, hypothyroidism, and subclinical thyroid disease, which may also affect the reliability of the results.
Conclusion
This study confirmed that HDL-C, BUN, gender, GLU, Hypertension, Hyperuricemia, Coronary Heart Disease, and Liver disease are independent risk factors for the occurrence of TD in patients with T2DM. Based on these findings, a nomogram was developed to predict the occurrence of TD. This model has demonstrated good clinical predictive value after internal validation. Clinicians can utilize this tool to identify T2DM patients who are at high risk of occurrence of TD and to implement timely interventions, which may help improve patient prognosis.
Supplementary Information
Below is the link to the electronic supplementary material.
Author contributions
J.W. and D.S. conceived and supervised the project. Y.H.N. and Z.H.C. contributed to the design of the study. Y.T.L. L.L and X.W. contributed to the acquisition and analysis of the data. Y.H.N. wrote the main manuscript text. All authors reviewed and approved the submitted manuscript.
Funding
This study was sponsored by grants from the Medical Science Research Project of Hebei (20201143) and the Medical Science Research Project of Hebei (20241554).
Data availability
The data are available upon reasonable request from the corresponding author.
Declarations
Competing interests
The authors declare no competing interests.
Ethical approval
The Medical Ethics Committee of the First Hospital of Hebei Medical University approved this study. Informed consent was waived by the Medical Ethics Committee of the First Hospital of Hebei Medical University since it was a retrospective study.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
These authors contributed equally: Yinghao Niu and Zhihua Chen.
Change history
3/18/2026
The original online version of this Article was revised: The Funding section in the original version of this Article contained an error in a grand funders’ name. The Funding section now reads: “This study was sponsored by grants from the Medical Science Research Project of Hebei (20201143) and the Medical Science Research Project of Hebei (20241554).”
Contributor Information
Jun Wang, Email: 13730113570@163.com.
Dan Song, Email: 444711388@qq.com.
References
- 1.Rong, F. et al. Association between thyroid dysfunction and type 2 diabetes: A meta-analysis of prospective observational studies. BMC Med.19, 1–13 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Islam, M. R. & Sultana, N. Prevalence of thyroid dysfunction in type 2 diabetes mellitus. Mymensingh Med. J.33 (2), 341–349 (2024). [PubMed] [Google Scholar]
- 3.Li, Y. et al. Increased FT3/FT4 ratio in a certain range is associated with decreased glycemic variability in patients with type 2 diabetes. Sci. Rep.14 (1), 26556 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Vinod, V., Rajagambeeram, R. & Samal, R. Evaluation of thyroid function and its relation to glycemic status in pregnant women with gestational diabetes mellitus. Cureus16 (10), e72339 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Binjawhar, D. & Mohammedsaeed, W. The determinants of leptin, angiopoietin like 8, and thyroid hormones levels in Saudi females with type 2 diabetes mellitus: A retrospective study. Med. (Baltim).103 (36), e39339 (2024). [Google Scholar]
- 6.Eom, Y. S., Wilson, J. R. & Bernet, V. J. Links between thyroid disorders and glucose homeostasis. Diabetes Metab. J.46, 239 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Jang, J. et al. Association between thyroid hormones and the components of metabolic syndrome. BMC Endocr. Disord. 18, 29 10.1186/s12902-018-0256-0 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Chen, C., Xie, Z., Shen, Y. & Xia, S. F. The roles of thyroid and thyroid hormone in pancreas: Physiology and pathology. Int. J. Endocrinol.2018, 2861034 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Potenza, M., Via, M. A. & Yanagisawa, R. T. Excess thyroid hormone and carbohydrate metabolism. Endocr. Pract.15, 254–262 (2009). [DOI] [PubMed] [Google Scholar]
- 10.Mullur, R., Liu, Y. Y. & Brent, G. A. Thyroid hormone regulation of metabolism. Physiol. Rev.94 355 (2014).
- 11.Ward, R. J., Heald, A. H., Ogunmekan, S., Fryer, A. A. & Duff, C. J. Should we be screening for thyroid dysfunction in patients with type 2 diabetes mellitus? Br. J. Gen. Pract.68, 94–95 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Pavan Kumar, J. T. V. K. Comparison of thyroid function tests among type 2 diabetes patients with and without diabetic nephropathy and controls. Cureus16 (9), e70462 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Gao, J. & Liu, J. Correlation of serum Thyrotropin and thyroid hormone levels with diabetic kidney disease: a cross-sectional study. BMC Endocr. Disord. 24 (1), 170 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Sun, M. et al. Correlation between thyroid hormone sensitivity and diabetic peripheral neuropathy in euthyroid patients with type 2 diabetes mellitus. Sci. Rep.14 (1), 19603 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wang, Z. et al. Association between thyroid autoimmunity and bone mineral density in patients with type 2 diabetes mellitus: a cross-sectional study. Endocrine85 (2), 604–614 (2024). [DOI] [PubMed] [Google Scholar]
- 16.Li, Y. et al. Prognostic nomograms for young breast cancer: A retrospective study based on the SEER and METABRIC databases. Cancer Innov.3 (6), e152 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Schrader, A. et al. Prostate cancer risk assessment and avoidance of prostate biopsies using fully automatic deep learning in prostate MRI: comparison to PI-RADS and integration with clinical data in nomograms. Eur. Radiol.34 (12), 7909–7920 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Rojas-Garcia, P. et al. Using nomograms wisely: predicting Sentinel node positivity in melanoma. Ann. Surg. Oncol.31 (12), 8240–8244 (2024). [DOI] [PubMed] [Google Scholar]
- 19.Catma, Y. et al. Higher prevalence of thyroid dysfunction in type 2 diabetes mellitus: effects on glycemic Control, diabetic complications and comorbidities. Medicina61, 1427 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Moosazadeh, M. et al. The prevalence and determinants of diabetes mellitus and thyroid disorder comorbidity in Tabari cohort population. Sci. Rep., 14 17577 (2024).
- 21.Al-Geffari, M. et al. Risk factors for thyroid dysfunction among type 2 diabetic patients in a highly diabetes mellitus prevalent society. Int. J. Endocrinol. 2013 417920 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Khassawneh, A. H. et al. Prevalence and predictors of thyroid dysfunction among type 2 diabetic patients: A case-control study. Int. J. Gen. Med.13, 803–816 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Kim, J. et al. Repeated low High-Density lipoprotein cholesterol and the risk of thyroid cancer: A nationwide Population- based study in Korea. Endocrinol. Metab. (Seoul). 37 (2), 303–311 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Park, J. H., Cho, H. S. & Yoon, J. H. Thyroid cancer in patients with metabolic syndrome or its components: A NationwidePopulation-Based cohort study. Cancers14, 4106 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Park, J. H. et al. Metabolic syndrome and the risk of thyroid cancer: A nationwide Population-Based cohort study. Thyroid30, 1496–1504 (2020). [DOI] [PubMed] [Google Scholar]
- 26.Fussey, J. M. et al. Does obesity cause thyroid cancer? A Mendelian randomization study. J. Clin. Endocrinol. Metab.105, e2398–e2407 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Xu, H. et al. Monocyte to high-density lipoprotein cholesterol ratio as an independent risk factor for papillary thyroid carcinoma. J. Clin. Lab. Anal.35, e24014 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Zhao, J. et al. Hypercholesterolemia is an associated factor for risk of differentiated thyroid cancer in Chinese population. Front. Oncol.10, 508126 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Skoczyńska, A. et al. Serum lipid transfer proteins in hypothyreotic patients are inversely correlated with Thyroid-Stimulating hormone (TSH) levels. Med. Sci. Monit.22, 4661–4669 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Rizos, C. V., Elisaf, M. S. & Liberopoulos, E. N. Effects of thyroid dysfunction on lipid profile. Open. Cardiovasc. Med. J.5, 76–84 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Duntas, L. H. & Brenta, G. A renewed focus on the association between thyroid hormones and lipid metabolism. Front. Endocrinol. (Lausanne). 9, 511 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Yang, J. & Jiang, S. Development and validation of a model that predicts the risk of diabetic nephropathy in type 2 diabetes mellitus patients: A Cross-Sectional study. Int. J. Gen. Med.15, 5089–5101 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Yang, Q. P., Chen, Y. Y., Li, Z. & Xu, M. Major risk factors analysis of pruritus complicated by type 2 diabetes mellitus and the effect of comprehensive nursing intervention. Front. Surg.9, 842884 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Li, S. et al. An environment-wide association study for the identification of non-invasive factors for type 2 diabetes mellitus: analysis based on the Henan rural cohort study. Diabetes Res. Clin. Pract.204, 110917 (2023). [DOI] [PubMed] [Google Scholar]
- 35.An, Y. S., Lee, J., Kim, H. K., Lee, S. J. & Yoon, J. K. Effect of withdrawal of thyroid hormones versus administration of Recombinant human thyroid-stimulating hormone on renal function in thyroid cancer patients. Sci. Rep.13 (1), 206 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Ward, R. J., Heald, A. H., Ogunmekan, S., Fryer, A. A. & Duff, C. J. Should we be screening for thyroid dysfunction in patients with type 2 diabetes mellitus? Br. J. Gen. Pract.68 (667), 94–95 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Chauhan, A. & Patel, S. S. Thyroid Hormone and Diabetes Mellitus Interplay: Making Management of Comorbid Disorders Complicated. Horm. Metab. Res. (2024). . 10.1055/a-2374-8756
- 38.Sinha, R. A., Bruinstroop, E. & Yen, P. M. Actions of thyroid hormones and thyromimetics on the liver. Nat. Rev. Gastroenterol. Hepatol.17, 10 (2024). [Google Scholar]
- 39.Soetedjo, N. N. M., Agustini, D. & Permana, H. The impact of thyroid disorder on cardiovascular disease: unraveling the connection and implications for patient care. Int. J. Cardiol. Heart Vasc. 55, 101536 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Yang, Y. et al. Predicting thyroid involvement in primary Sjögren’s syndrome: development and validation of a predictive nomogram. Front. Immunol.15, 1445916 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data are available upon reasonable request from the corresponding author.