DOI: https://doi.org/10.21203/rs.3.rs-2530709/v1
To analyze the risk factors of adult patients with type 2 diabetes mellitus complicated with hypertension in Jiangsu province, and establish an individualized risk prediction model of hypertension.
We analyzed 485 newly diagnosed patients with type 2 diabetes mellitus from 2020 to 2021 in Lianshui County People's Hospital, Huai'an City, Jiangsu province, China. According to the ratio of 7:3, they were randomly divided into training group and validation group. After primary screening by univariate analysis, the predicted variables were determined by multivariate analysis. The nomogram model of hypertension was constructed and evaluated by receiver operating characteristic curve (ROC curve), calibration curve and decision curve analysis(DCA).
This study investigated 485 people, and the prevalence of hypertension was 56.90%. Residential area, systolic blood pressure, family history of diabetes and uric acid are independent risk factors of hypertension among adults in Jiangsu province (P < 0.05). The area under ROC curve (AUC) of hypertension risk prediction model in training group and validation group were 0.7401 and 0.7392, respectively, and the calibration curves showed excellent consistency. DCA shows that the training group shows a great positive rate of return in the risk range of 13%~57% and 58%~100%.
The prediction model based on the related risk factors of hypertension among adult residents in Jiangsu province has excellent accuracy and clinical application value. It can provide a more intuitive way to assess the risk of diabetic patients with hypertension, and has guiding significance for the prevention and treatment of hypertension.
Diabetes is an endocrine disease, which can't produce enough insulin, or their bodies can't make normal use of the insulin they produce[1]. Hypertension is one of the most common chronic noninfectious diseases and the most important risk factor for cardiovascular and cerebrovascular diseases [2]. The World Health Organization defines hypertension as systolic blood pressure (SBP) and/or diastolic blood pressure (DBP) ≥ 140/90 mmHg. Noninfectious diseases such as hypertension and diabetes are among the most deadly diseases, which are prevalent among adults all over the world, including China, and pose a threat to public health [3]. Hypertension is a common complication of diabetes. Several studies have reported that the prevalence of hypertension in diabetic patients is high[4 –6]; and other studies also mentioned the high prevalence of diabetes in patients with hypertension[7–8]. Therefore, diabetes is closely related to hypertension. Diabetes complicated with hypertension is far more harmful to cardiovascular system than a single disease. It may effectively reduce the risk of diabetes complicated with hypertension by exploring the related risk factors and taking effective preventive measures.
The disease risk prediction model can predict the disease risk more accurately by quantitative research, and display the prediction results intuitively. This study analyzed the baseline data of 485 hypertensive patients in Jiangsu province, China, to explore the risk factors of diabetes combined with hypertension and establish a nomogram prediction model, so as to provide medical workers with early identification of high-risk population of diabetes combined with hypertension among the physical examination population, and formulate prevention and treatment strategies.
After filtering according to various exclusion criteria, we selected 485 subjects. The detailed sample selection procedure is shown in Fig. 1.Materials were collected from 485 newly diagnosed type 2 diabetes mellitus patients in Lianshui County People's Hospital, Huai 'an City, Jiangsu province, China from 2020 to 2021. Inclusion criteria: ① Age ≥ 18 years old; ② It conforms to the relevant diagnostic criteria of "Guidelines for Prevention and Treatment of Type 2 Diabetes in China (2017 Edition)" [9]; ③ It meets the diagnostic criteria of "2018 Chinese Guidelines for Prevention and Treatment of Hypertension"[10]; ④ Patients and their families informed consent to this study. Exclusion criteria: ① Other types of diabetes and secondary hypertension; ② Complicated with severe cardiovascular and cerebrovascular complications, diabetic ketoacidosis, diabetic hyperosmolar coma, acute and chronic nephritis, secondary kidney disease and malignant tumor; ③ Pregnant and lactating women. ④ Incomplete information.
Collect general information such as the name, gender, age, residential area, living habits (smoking or not) of the subjects. Collect subjects' family history and personal history: family history of thyroid disease, history of thyroid disease, family history of diabetes, hyperuricemia and gout. Physical examination: height, weight, body mass index (BMI,kg/m2), systolic blood pressure (SBP, mm Hg) and resting heart rate (RHR, beats/minute) were measured. Biochemical examination: peripheral venous blood was collected to detect fasting blood glucose (FBG, mmol/L), triglyceride (TG, mmol/L), total cholesterol (TC, mmol/L), low-density lipoprotein cholesterol (LDL-C, mmol/L), high-density lipoprotein cholesterol (HDL-C, mmol/L), uric acid (UA) ,glycosylated hemoglobin, type a1c(HbA1c,%). Thyroid hormone examination: TSH (uIu/ml), free triiodothyronine (FT3, pmol/L) and free tetraiodothyronine (FT4, pmol/L).
The software R 4.0.4 was used for statistical analysis, Kolmogorov-Smimovz normal test was used for continuous variables, (x̄ ± s) was used to describe the measurement data conforming to normal distribution, and t test was used for comparison between the two groups. The data were analyzed by χ2 test; by strictly data filtering and preprocessing, eligible patients (n = 485) were randomly divided into training group (n = 340) and validation group (n = 145) according to the ratio of 7∶3. Taking hypertension in the training group data as the outcome variable, the independent predictors were screened out by univariate analysis, on this basis, the nomogram prediction model was further discussed and established by multivariate Logistic regression analysis, and the risk nomogram model of type 2 diabetes mellitus complicated with hypertension was established by R( R4.0.4) software and rms software package, and the prediction model was verified by the validation group data. In order to evaluate the effectiveness of the prediction model, the area under the ROC curve was used to verify the prediction effect of the prediction model, and the goodness of fit of the model was judged by Hosmer-Lemeshow goodness of fit test, and draw calibration curve. At the same time, the risk of hypertension was predicted by decision curve analysis (DCA). P < 0.05, the difference was statistically significant.
485 patients with type 2 diabetes mellitus include 209 non-hypertensive patients and 276 hypertensive patients. There were no statistically significant difference between non-hypertensive patients and hypertensive patients in sex, smoking, height, RHR, family history of thyroid disease, history of thyroid disease, hyperuricemia, gout, FBG, TG, TC, LDL-C, HDL-C, HbA1c, TSH and FT3 (P > 0.05). There were statistically significant differences between non-hypertensive patients and hypertensive patients in urban and rural categories, weight, BMI, SBP, family history of diabetes, UA and FT4 (P < 0.05), the results are shown in Table 1.
| Variable | Total Number (N = 485) | Non-hypertension Group (N = 209) | Hypertension Group (N = 276) | t/χ2 value | P value |
|---|---|---|---|---|---|
| Gender /n(%) | 0.718 | 0.397 | |||
| Man | 255 (52.6) | 115 (55.0) | 140 (50.7) | ||
| Woman | 230 (47.4) | 94 (45.0) | 136 (49.3) | ||
| Residential area /n(%) | 16.223 | < 0.001 | |||
| City | 158 (32.6) | 47 (22.5) | 111 (40.2) | ||
| Rural | 327 (67.4) | 162 (77.5) | 165 (59.8) | ||
| Smoking /n(%) | 0 | 1.000 | |||
| No smoking | 484 (99.8) | 209 (100.0) | 275 (99.6) | ||
| Occasionally smoking | 1 (0.2) | 0 (0.0) | 1 (0.4) | ||
| Height (cm) /(mean ± SD) | 164.12 ± 7.65 | 164.53 ± 7.83 | 163.80 ± 7.52 | 1.036 | 0.301 |
| Weight (kg) /(mean ± SD) | 69.06 ± 11.37 | 67.28 ± 11.52 | 70.42 ± 11.09 | -3.038 | 0.003 |
| BMI(kg/m2) /(mean ± SD) | 25.57 ± 3.38 | 24.75 ± 3.23 | 26.19 ± 3.36 | -4.732 | < 0.001 |
| SBP(mm Hg) /(mean ± SD) | 129.46 ± 13.01 | 125.22 ± 7.97 | 132.67 ± 15.03 | -6.500 | < 0.001 |
| Resting heart rate (times/minute) /(mean ± SD) | 77.09 ± 11.91 | 76.86 ± 11.64 | 77.27 ± 12.13 | -0.376 | 0.707 |
| Has family thyroid diseases /n(%) | 0 | 1.000 | |||
| Yes | 2 (0.4) | 1 (0.5) | 1 (0.4) | ||
| No | 483 (99.6) | 208 (99.5) | 275 (99.6) | ||
| Has history of thyroid diseases /n(%) | 0 | 1.000 | |||
| Yes | 3 (0.6) | 1 (0.5) | 2 (0.7) | ||
| No | 482 (99.4) | 208 (99.5) | 274 (99.3) | ||
| Has family history of diabetes /n(%) | 16.271 | < 0.001 | |||
| Yes | 460 (94.8) | 188 (90.0) | 272 (98.6) | ||
| No | 25 (5.2) | 21 (10.0) | 4 (1.4) | ||
| Has hyperuricemia /n(%) | 0 | 1.000 | |||
| Yes | 10 (2.1) | 4 (1.9) | 6 (2.2) | ||
| No | 475 (97.9) | 205 (98.1) | 270 (97.8) | ||
| Has gout /n(%) | 0.022 | 0.882 | |||
| Yes | 11 (2.3) | 4 (1.9) | 7 (2.5) | ||
| No | 474 (97.7) | 205 (98.1) | 269 (97.5) | ||
| FBG(mmol/L) /(mean ± SD) | 8.95 ± 3.41 | 9.13 ± 3.35 | 8.82 ± 3.46 | 0.999 | 0.318 |
| TG(mmol/L) /(mean ± SD) | 2.02 ± 1.62 | 1.98 ± 1.69 | 2.05 ± 1.56 | -0.461 | 0.645 |
| TC(mmol/L) /(mean ± SD) | 4.42 ± 0.96 | 4.47 ± 0.99 | 4.38 ± 0.95 | 0.967 | 0.334 |
| LDL-C(mmol/L) /(mean ± SD) | 2.40 ± 0.75 | 2.45 ± 0.81 | 2.36 ± 0.70 | 1.409 | 0.160 |
| HDL-C(mmol/L) /(mean ± SD) | 1.19 ± 0.39 | 1.20 ± 0.41 | 1.18 ± 0.37 | 0.577 | 0.564 |
| UA(mmol/L) /(mean ± SD) | 333.33 ± 112.52 | 310.89 ± 108.48 | 350.32 ± 112.74 | -3.876 | < 0.001 |
| HbA1c(%) /(mean ± SD) | 9.30 ± 2.24 | 9.52 ± 2.33 | 9.14 ± 2.16 | 1.837 | 0.067 |
| TSH(uIu/ml) /(mean ± SD) | 2.99 ± 2.55 | 2.95 ± 2.76 | 3.02 ± 2.38 | -0.275 | 0.783 |
| FT4(pmol/L) /(mean ± SD) | 1.27 ± 0.21 | 1.25 ± 0.20 | 1.29 ± 0.22 | -2.088 | 0.037 |
| FT3(pmol/L) /(mean ± SD) | 2.70 ± 0.54 | 2.68 ± 0.58 | 2.71 ± 0.50 | -0.723 | 0.470 |
The dependent variable is whether hypertension occurs in the training group (assigned value: yes = 1, no = 0), and the above factors were taken as independent variables for univariate analysis. Univariate analysis showed that the statistically significant differences between urban and rural areas, weight, BMI, SBP, family history of diabetes, and uric acid were included in the Logistic regression equation, and the independent variables were screened by multivariate Logistic regression. The results showed that residential area, SBP, family history of diabetes and uric acid were the influencing factors of hypertension in type 2 diabetic patients. The results are shown in Table 2.
| Variable | β value | SE | Wald value | P value | OR | 95% CI |
|---|---|---|---|---|---|---|
| Residential area | ||||||
| Rural | 1.000 | |||||
| Urban | 0.799 | 0.269 | 8.840 | 0.003 | 2.222 | 1.321,3.799 |
| Family history of diabetes | ||||||
| No | 1.000 | |||||
| Yes | 1.440 | 0.603 | 5.696 | 0.017 | 4.220 | 1.398,15.775 |
| SBP | 0.049 | 0.013 | 14.365 | < 0.001 | 1.051 | 1.026,1.079 |
| UA | 0.004 | 0.001 | 9.636 | 0.002 | 1.004 | 1.001,1.006 |
The variables selected from the above multivariate Logistic regression analysis results are included in the nomogram prediction model, and the prevalence risk of hypertension is selected as the outcome index, and draw a nomogram(Fig. 2.). According to the scale above the nomogram corresponding to each risk factor, the single score of that factor is obtained, and the total score of all risk factors is added to get the incidence of hypertension of the corresponding patient. The higher the total score, the greater the possibility of hypertension risk.
Draw the ROC curve of nomogram prediction accuracy. The area under the ROC curve of the nomogram prediction model in the training group is 0.7401 [95% CI (0.6866, 0.7935)] (Fig. 3.A), the cut-off value is 0.335, the sensitivity is 74.5% and the specificity is 66.2%; the area under the ROC curve of the predicted nomogram is 0.7392 [95% CI (0.6587, 0.8197)] (Fig. 3.B), the cut-off value is 0.273, the sensitivity is 75.0%, and the specificity is 67.9%. In addition, the area under the ROC curve of the validation group is only 0.009 lower than that of the training group, which indicates that the prediction model has good prediction discrimination in both the training group and the validation group.
Draw the calibration curves of hypertension in the training group and the validation group. The calibration curves of the training group show that the predicted results are in good agreement with the observation. (Fig. 4.A). Hosmer-Lemeshow goodness-of-fit test shows that the model has excellent calibration (χ2 = 13.316, P = 0.1014), and the model is not significant (P > 0.05), which indicates that the model is in good agreement with the observed data. The validation of the validation group data set also shows that the prediction is in good agreement with the observation(Fig. 4.B). The Hosmer-Lemeshow goodness-of-fit test shows that the model has excellent calibration (χ2 = 9.8847, P = 0.2732), and the model is not significant (P > 0.05), which indicates that the model is in good agreement with the observation data.
On the basis of the nomogram prediction model, the selected variables were analyzed by the DCA (Fig. 5.A). The results showed that when the threshold probability of patients was 13%~57% and 58%~100%, the net benefit of using the nomogram to predict the risk of hypertension in type 2 diabetes mellitus was higher, which was also confirmed in the validation group (Fig. 5.B). Therefore, the extensive alternative threshold probability showed that the model had excellent evaluation ability.
At present, research has found that type 2 diabetes mellitus is closely related to hypertension. Both diseases can cause chronic damage to blood vessels, often called homologous diseases.Type 2 diabetes mellitus combined with hypertension may cause serious cardiovascular diseases, even life-threatening[11]. Hypertension is an important public health problem in the world. Patients with type 2 diabetes mellitus complicated with hypertension have a high prevalence[12–13].According to the 2020 Diabetes Survey Report of Korea, 13.8% of adults over the age of 30 have diabetes, and 61.3% of diabetic patients have hypertension[14]. In addition, the report shows that among the elderly over 65 years old, 27.6% suffer from diabetes, and 74.3% of diabetic patients also suffer from hypertension. In the study of prevention of cardiovascular risk factors in Hong Kong, 58% of diabetic patients have hypertension[15]. Sabuncu et al.[16]pointed out that in Turkey, 67.5% of adult diabetic patients suffer from hypertension. Petrie et al. [17]pointed out that diabetic patients are twice as likely to suffer from hypertension as non-diabetic patients. The above research shows that subjects with hypertension have a greater risk of diabetes than subjects with normal blood pressure. Hypertension is a common complication of diabetes, and the risk of diabetic patients with hypertension is higher than that of the general population. In this study, the prevalence of hypertension in 485 patients with type 2 diabetes mellitus was 56.90%, which was close to the prevalence of the above study, suggesting that type 2 diabetes mellitus patients were prone to hypertension.
Logistic regression is used in this study. Logistic regression is a nonlinear probabilistic prediction model that can study the relationship between some covariates and classified observation results. Logistic regression is often used in clinical analysis of high-risk factors causing diseases. Logistic regression analysis showed that urban, SBP increased, family history of diabetes and elevated UA were risk factors for type 2 diabetes mellitus complicated with hypertension.
In this study, urban factors are the risk factors of hypertension in diabetic patients. Some studies show that the prevalence of hypertension in urban areas is significantly higher than that in rural areas[18–20]. Due to the different levels of economic development between urban and rural areas, limited health care facilities and different lifestyles of residents, the prevalence of hypertension and diabetes varies from region to region. The prevalence of hypertension in urban areas is significantly higher than that in rural areas[21–22]. According to the above factors, the prevalence of hypertension in urban areas is higher than that in rural areas, which is the same as the conclusion of this study.
This study shows that SBP is an important and unchangeable risk factor for type 2 diabetes mellitus complicated with hypertension. With the increase of SBP, the risk of hypertension increases greatly, which is consistent with the results of previous studies. The OR value of type 2 diabetes mellitus patients with hypertension with elevated SBP was 1.06 (95% CI: 1.03 ~ 1.08). In FHS and Mesa areas, the increase of SBP level is considered as the main risk factor of hypertension[23]. Studies have shown that from 1980 to 2008, the average SBP increase was observed in patients with hypertension in South Asia and Southeast Asia[24]. Previous studies have shown that baseline SBP and DBP are the main determinants of hypertension. Every unit of increase in blood pressure exceeds the optimal value of 119/79 mm Hg, which will also increase the risk of hypertension[25–27]. In addition, it may be necessary to actively intervene the individuals with elevated SBP, which should be the focus of hypertension prevention.
Diabetes has strong family aggregation, and it is often used to assess the risk of diabetes in clinical practice[28]. Studies have shown that parents influence the metabolism of offspring through different mechanisms and ways, and the similar genetic background and common living environment among family members are the key factors of disease risk[29]. In this study, the risk of hypertension in diabetic patients with family history of diabetes is 4.22 times higher than that without family history. Therefore, people with family history should pay more attention, strictly control other risk factors and do self-prevention.
It is reported that the increase of serum UA is related to hypertension[30–32]. In recent years, the prevalence of hyperuricemia and hypertension is on the rise, with 5.46–19.30% and 23.2% respectively[33]. Studies have shown that the prevalence of hypertension increases with the increase of UA level[34–37]. Recent meta-analysis shows that there is a significant correlation between UA and hypertension, and UA level in patients with hypertension is significantly increased, independent of traditional risk factors[38–41]. The above analysis shows that the increase of UA is positively correlated with the development of type 2 diabetes mellitus patients with hypertension, and it is a risk factor for type 2 diabetes mellitus patients with hypertension.
There are several limitations to the study. This study is an observational study. The sample size is limited and the source of cases is single, so it has some limitations. Although there is external validation, but the sample belongs to a medical center, its model validation ability is still insufficient. In the follow-up study, it still needs to be verified by multi-center research, and the research conclusion needs to be further explored and confirmed by large-scale prospective clinical research.
To sum up, this study has constructed a risk prediction model of type 2 diabetes mellitus complicated with hypertension with excellent discrimination ability, which includes four variables: urban, SBP increase, family history of diabetes and UA increase. By referring to the nomogram model, community medical workers can identify the high-risk group of type 2 diabetes mellitus complicated with hypertension in the physical examination population at an early stage, strengthen the health publicity and education of residents in multiple ways and develop targeted intervention strategies, which has important guiding significance for reducing the risk of type 2 diabetes mellitus with hypertension, improving the quality of life of residents, and standardizing health management.
Acknowledgements
The authors thank the patients and their caregivers in addition to the investigators and their teams who contributed to this study.
Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Funding
No funding was acquired for this work.
Competing interests
The authors declare that they have no competing interests.
Author contributions
Conception and design:Guanzhong Tian,Wei Wang, Tuerxunyiming Muhadasi
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Wei Wang, Tuerxunyiming Muhadasi
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Guanzhong Tian, Shan Li
Writing, review, and/or revision of the manuscript:Guanzhong Tian, Shan Li
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Tuerxunyiming Muhadasi,Wei Wang
Study supervision: Tuerxunyiming Muhadasi,Wei Wang
Ethical approval
This is an observational study. The Lianshui County People's Hospital Research Ethics Committee has confirmed that no ethical approval is required.
Author sorting
Guanzhong Tian, The first author
Shan Li, The second author
Tuerxunyiming Muhadasi, The first corresponding author
Wei Wang, The second corresponding author
Guanzhong Tian, Shihezi University School of Medicine, Shihezi, Xinjiang province, 832003 P.R. China.
email:[email protected].
Shan Li, Shihezi University School of Medicine, Shihezi, Xinjiang province, 832003 P.R. China.
email:[email protected].
Tuerxunyiming Muhadasi, Zhejiang University City College, Hangzhou, Zhejiang province, 310015 P.R. China.
email:[email protected].
Wei Wang, Department of Clinical Laboratory, Lianshui People's Hospital of kangda college affiliated to Nanjing Medical University, Huai'an, Jiangsu province, 223400 P.R. China.
email:[email protected].