基于随机森林模型的社区糖尿病患者合并高血压的影响因素分析

李文涛; 高文娟; 刘新颖; 王玥; 吴浩

doi:10.16766/j.cnki.issn.1674-4152.003991

基于随机森林模型的社区糖尿病患者合并高血压的影响因素分析

doi: 10.16766/j.cnki.issn.1674-4152.003991

1.
首都医科大学全科医学与继续教育学院，北京 100069
2.
北京市丰台区方庄社区卫生服务中心，北京 100078

基金项目:

智慧家医监管平台关键技术的构建和效果评价(首都卫生发展科研专项)项目首2022-2-7052

北京市卫生系统高层次卫生技术人才队伍建设专项 2022-1-005

详细信息

通讯作者:
吴浩，E-mail：wushunzhe@ccmu.edu.cn

中图分类号: R544.1 R587.1
计量
- 文章访问数: 8
- HTML全文浏览量: 4
- PDF下载量: 1
- 被引次数: 0
出版历程
- 收稿日期: 2024-09-24
- 网络出版日期: 2025-08-14

Factors influencing hypertension comorbidity in community-dwelling patients with diabetes based on random forest model

1.
School of General Practice and Continuing Education, Capital Medical University, Beijing 100069, China

摘要

摘要: 目的探讨社区糖尿病人群合并高血压的现状及其影响因素，指导糖尿病患者个性化健康管理。方法 2023年1月—2024年1月，采用横断面调查方法纳入北京市丰台区方庄社区卫生服务中心签约的糖尿病患者的健康档案2 591份，基于随机森林模型与LASSO回归法探讨糖尿病合并高血压的影响因素。结果社区糖尿病患者合并高血压的发生率为87.42%(2 265人)。随机森林算法分析显示，当lambda(λ)值为0.003 9时误差最小，对应的影响因素数目为8个，重要性排序为TG/HDL-C、估算的肾小球滤过率(eGFR)、BMI、空腹血糖、年龄、糖尿病患病年限、收缩压、学历。多元逐步回归分析显示，eGFR(OR=0.980, 95% CI：0.969~0.990)、TG/HDL-C(OR=1.083, 95% CI：1.022~1.147)、BMI(24~28组OR=1.469, 95% CI：1.140~1.893, >28组OR=2.340, 95% CI：1.561~3.509)、年龄>75岁组OR=1.844, 95% CI：1.125~3.021)、收缩压(OR=1.053, 95% CI：1.031~1.076)均为糖尿病患者共病高血压的影响因素(P < 0.05)。结论社区糖尿病患者合并高血压的患病率较高，TG/HDL-C指数越高、eGFR越低、年龄越大、肥胖的糖尿病患者高血压发病风险越高。在该类人群的管理过程中，应定期监测TG/HDL-C指数、评估肾功能，根据患者的年龄及体重制定个性化的控糖、减重方案及健康教育计划，降低糖尿病患者合并高血压的发生风险，减少远期心血管疾病并发症，实现糖尿病患者管理的关口前移。
- 糖尿病 /
- 高血压 /
- 共病 /
- 随机森林模型 /
- 影响因素
Abstract: Objective To investigate the current situation of hypertension in community-dwelling patients with diabetes and to analyze its influencing factors, guide personalized health management for patients with diabetes. Methods A total of 2 591 health records of diabetes patients registered at the Fangzhuang Community Health Service Center in Fengtai District, Beijing, from January 2023 to January 2024, were included in a cross-sectional survey. Based on random forest model and LASSO regression, the influencing factors associated with hypertension in patients with diabetes were explored. Results The prevalence of comorbid hypertension in community-dwelling patients with diabetes was 87.42% (2 265 people). The results of random forest algorithm showed that the lowest error was observed when the lambda (λ) value was 0.003 9, corresponding to eight key influencing factors. The were TG/HDL-C, estimated glomerular filtration rate (eGFR), BMI, fasting blood glucose, age, duration of diabetes, systolic blood pressure, and education background. The results of the multivariate stepwise regression analysis indicated that eGFR (OR=0.980, 95% CI: 0.969-0.990), TG/HDL-C (OR=1.083, 95% CI: 1.022-1.147), BMI (24-28 group: OR=1.469, 95% CI: 1.140-1.893; >28 group: OR=2.340, 95% CI: 1.561-3.509), age (>75 years group: OR=1.844, 95% CI: 1.125-3.021), and systolic blood pressure (OR=1.053, 95% CI: 1.031-1.076) were influencing factors of comorbid hypertension in diabetic patients (P < 0.05). Conclusion The prevalence of hypertension was notably high among community-dwelling patients with diabetes. Elevated TG/HDL-C ratio, reduced eGFR, older age, and obesity were significantly associated with increased risk of hypertension. Effective management should include regular monitoring of TG/HDL-C levels, assessment of renal function, and personalized interventions focused on glycemic control, weight management, and health education tailored to the patient' s age and body mass index. This approach can help reduce the risk of hypertension in diabetic patients, decrease their long-term cardiovascular disease burden, and facilitate an upstream shift diabetes management.
- Diabetes /
- Hypertension /
- Comorbidity /
- Random forest model /
- Influence factors

HTML全文

图 1 基于RF模型的影响因素重要性排序

Figure 1. Importance ranking of influencing factors based on RF model

下载: 全尺寸图片幻灯片

图 2 基于LASSO回归的变量筛选

Figure 2. Variable screening based on LASSO regression

下载: 全尺寸图片幻灯片

表 1 社区糖尿病患者共病高血压的单因素分析

Table 1. Univariate analysis of hypertension comorbidity in community-dwelling diabetic patients

项目	人数(n=2 591)	未患高血压(n=326)	患高血压(n=2 265)	统计量	P值
年龄[例(%)]				21.972^a	< 0.001
< 60岁	419(16.17)	75(23.01)	344(15.19)
60~74岁	1 693(65.34)	215(65.95)	1 478(65.25)
≥75岁	479(18.49)	36(11.04)	443(19.56)
学历[例(%)]				19.015^a	< 0.001
小学及以下	132(5.09)	6(1.84)	126(5.56)
初中	523(20.19)	67(20.55)	456(20.14)
高中	753(29.06)	75(23.01)	678(29.93)
大学及以上	1 183(45.66)	178(54.60)	1 005(44.37)
职业[例(%)]				6.585^a	0.010
在职	2 382(91.93)	312(95.71)	2 070(91.39)
退休	209(8.07)	14(4.29)	195(8.61)
医保类型[例(%)]				9.226^a	0.002
城镇居民医保	2 437(94.06)	294(90.18)	2 143(94.61)
城镇职工医保	154(5.94)	32(9.82)	122(5.39)
饮酒[例(%)]				5.347^a	0.021
从不	1 987(76.69)	233(71.47)	1 754(77.44)
既往/正在	604(23.31)	93(28.53)	511(22.56)
运动情况[例(%)]				14.775^a	0.002
从不	102(3.94)	12(3.68)	90(3.97)
偶尔	462(17.83)	34(10.43)	428(18.90)
每周>1次	517(19.95)	67(20.55)	450(19.87)
每天	1 510(58.28)	213(65.34)	1 297(57.26)
BMI[例(%)]				23.427^a	< 0.001
< 24	1 021(39.41)	165(50.61)	856(37.79)
24~28	1 147(44.27)	129(39.57)	1 018(44.95)
>28	423(16.33)	32(9.82)	391(17.26)
腰高比[M(P₂₅, P₇₅)]	0.51(0.48, 0.54)	0.51(0.48, 0.53)	0.51(0.49, 0.54)	-2.950^b	0.003
收缩压[M(P₂₅, P₇₅), mmHg]	125.34(119.52, 131.16)	123.81(118.36, 129.25)	125.56(119.72, 131.40)	-4.900^b	< 0.001
TyG[M(P₂₅, P₇₅)]	8.99(8.62, 9.41)	8.89(8.56, 9.33)	9.01(8.63, 9.42)	-2.178^b	0.029
TG/HDL-C[M(P₂₅, P₇₅)]	2.52(1.74, 4.00)	2.19(1.58, 3.40)	2.59(1.78, 4.03)	-4.091^b	< 0.001
METS-IR[M(P₂₅, P₇₅)]	38.11(34.12, 42.55)	36.9(32.55, 40.66)	38.28(34.38, 42.73)	-4.012^b	< 0.001
空腹血糖[M(P₂₅, P₇₅), mmol/L]	6.77(5.75, 8.16)	7.00(6.01, 8.34)	6.70(5.72, 8.10)	-2.613^b	0.009
甘油三酯[M(P₂₅, P₇₅), mmol/L]	1.44(1.08, 2.03)	1.32(0.95, 1.86)	1.46(1.10, 2.06)	-3.676^b	< 0.001
高密度脂蛋白[M(P₂₅, P₇₅), mmol/L]	1.31(1.11, 1.52)	1.35(1.16, 1.55)	1.30(1.11, 1.51)	-3.087^b	0.002
血清肌酐[M(P₂₅, P₇₅), mg/dL]	0.71(0.60, 0.85)	0.69(0.59, 0.80)	0.72(0.60, 0.86)	-2.944^b	0.003
eGFR [M(P₂₅, P₇₅), mL/(min·1.73 m²)]	92.32(84.15, 99.06)	94.68(88.25, 101.69)	92.04(83.25, 98.61)	-5.369^b	< 0.001
尿素氮[M(P₂₅, P₇₅), mmol/L])	5.88(4.86, 7.02)	5.66(4.67, 6.74)	5.90(4.88, 7.06)	-2.427^b	0.015
糖尿病患病年限[M(P₂₅, P₇₅), 年]	12.34(5.33, 17.35)	11.36(4.22, 18.50)	12.49(5.51, 19.47)	-2.941^b	0.003
高脂血症[例(%)]				45.242^a	< 0.001
是	325(12.54)	79(24.23)	246(10.86)
否	2 266(87.46)	247(75.77)	2 019(89.14)
注：^a为χ²值，^b为Z值。1 mmHg=0.133 kPa。eGFR为估算的肾小球滤过率(estimated glomerular filtration rate)。本表仅列出差异有统计学意义的指标。

下载: 导出CSV

表 2 社区糖尿病患者高血压共病影响因素多元logistic回归分析

Table 2. Multivariate logistic regression analysis for influence factors for comorbidity hypertension in diabetes patients

变量	B	SE	Waldχ²	P值	OR(95% CI)
TG/HDL-C	0.079	0.029	7.319	0.007	1.083(1.022~1.147)
eGFR	-0.020	0.006	13.730	< 0.001	0.980(0.969~0.990)
BMI
24~28	0.385	0.129	8.856	0.003	1.469(1.140~1.893)
>28	0.850	0.207	16.948	< 0.001	2.340(1.561~3.509)
年龄
60~75岁	0.240	0.163	2.159	0.142	1.272(0.923~1.751)
>75岁	0.612	0.252	5.894	0.015	1.844(1.125~3.021)
收缩压	0.052	0.011	22.464	< 0.001	1.053(1.031~1.076)
注：以BMI < 24、年龄 < 60岁为参照。

下载: 导出CSV

参考文献(22)

[1]	PENG W, CHEN S, CHEN X, et al. Trends in major non-communicable diseases and related risk factors in China 2002-2019: an analysis of nationally representative survey data[J]. Lancet Reg Health West Pac, 2023, 43: 100809. DOI: 10.1016/j.lanwpc.2023.100809.
[2]	WANG L, PENG W, ZHAO Z, et al. Prevalence and treatment of diabetes in China, 2013-2018[J]. JAMA, 2021, 326(24): 2498-2506.
[3]	KIM H J, KIM K I. Blood pressure target in type 2 diabetes mellitus[J]. Diabetes Metab J, 2022, 46(5): 667-674.
[4]	KAMALUMPUNDI V, SHAMS E, TUCKER C, et al. Mechanisms and pharmacotherapy of hypertension associated with type 2 diabetes[J]. Biochem Pharmacol, 2022, 206: 115304. DOI: 10.1016/j.bcp.2022.115304.
[5]	王烁珊, 吴敬国, 荆小莉, 等. 甘油三酯葡萄糖指数与冠状动脉粥样硬化性心脏病关系的研究进展[J]. 中华全科医学, 2021, 19(6): 1008-1012. doi: 10.16766/j.cnki.issn.1674-4152.001972 WANG S S, WU J G, JING X L, et al. Research progress of triglyceride glucose index in coronary atherosclerotic heart disease[J]. Chinese Journal of General Practice, 2021, 19(6): 1008-1012. doi: 10.16766/j.cnki.issn.1674-4152.001972
[6]	WANG Z, YANG T, FU H. Prevalence of diabetes and hypertension and their interaction effects on cardio-cerebrovascular diseases: a cross-sectional study[J]. BMC Public Health, 2021, 21(1): 1224.
[7]	LI Y, SUN X, WEI J, et al. Time trends in comorbidity and management of hypertension and self-reported diabetes: a 15-year nationwide longitudinal study in China[J]. Am J Hypertens, 2021, 34(8): 810-820.
[8]	中华医学会糖尿病学分会. 中国2型糖尿病防治指南(2020年版)(上)[J]. 中国实用内科杂志, 2021, 41(8): 668-695. Diabetes Society of Chinese Medical Association. Guideline for the prevention and treatment of type 2 diabetes mellitus in China(2020 edition)(Part 1)[J]. Chinese Journal of Practical Internal Medicine, 2021, 41(8): 668-695.
[9]	UNGER T, BORGHI C, CHARCHAR F, et al. 2020 International society of hypertension global hypertension practice guidelines[J]. Hypertension, 2020, 75(6): 1334-1357.
[10]	ZOU L X, SUN L, NICHOLAS S B, et al. Comparison of bias and accuracy using cystatin C and creatinine in CKD-EPI equations for GFR estimation[J]. Eur J Intern Med, 2020, 80: 29-34. DOI: 10.1016/j.ejim.2020.04.044.
[11]	LIU H, YAN S, CHEN G, et al. Association of the ratio of triglycerides to high-density lipoprotein cholesterol levels with the risk of type 2 dabetes: a retrospective cohort study in Beijing[J]. J Diabetes Res, 2021, 2021: 552472 8. DOI: 10.1155/2021/5524728.
[12]	QIAN T, SHENG X, SHEN P, et al. Mets-IR as a predictor of cardiovascular events in the middle-aged and elderly population and mediator role of blood lipids[J]. Front Endocrinol(Lausanne), 2023, 14: 1224967. DOI: 10.3389/fendo.2023.1224967.
[13]	BAE J H, HAN K D, KO S H, et al. Diabetes fact sheet in Korea 2021[J]. Diabetes Metab J, 2022, 46(3): 417-426.
[14]	MUNTNER P, WHELTON P K, WOODWARD M, et al. A comparison of the 2017 American College of Cardiology/American Heart Association Blood Pressure Guideline and the 2017 American Diabetes Association Diabetes and hpertension position satement for U.S. A dults with diabetes[J]. Diabetes Care, 2018, 41(11): 2322-2329.
[15]	ZHANG W, HUO W, HU H, et al. Dose-response associations of triglyceride to high-density lipoprotein cholesterol ratio and triglyceride-glucose index with arterial stiffness risk[J]. Lipids Health Dis, 2024, 23(1): 115.
[16]	CHE B, ZHONG C, ZHANG R, et al. Triglyceride-glucose index and triglyceride to high-density lipoprotein cholesterol ratio as potential cardiovascular disease risk factors: an analysis of UK biobank data[J]. Cardiovasc Diabetol, 2023, 22(1): 34.
[17]	DEMIR S, NAWROTH PP, HERZIG S, EKIM VSTVNEL B. Emerging targets in type 2 diabetes and diabetic complications[J]. Adv Sci(Weinh), 2021, 8(18): e2100275. DOI: 10.1002/advs.202100275.
[18]	WANG T, LU J, SHI L, et al. China Cardiometabolic Disease and Cancer Cohort Study Group. Association of insulin resistance and β-cell dysfunction with incident diabetes among adults in China: a nationwide, population-based, prospective cohort study[J]. Lancet Diabetes Endocrinol, 2020, 8(2): 115-124.
[19]	YANG T, LIU Y, LI L, et al. Correlation between the triglyceride-to-high-density lipoprotein cholesterol ratio and other unconventional lipid parameters with the risk of prediabetes and Type 2 diabetes in patients with coronary heart disease: a RCSCD-TCM study in China[J]. Cardiovasc Diabetol, 2022;21(1): 93.
[20]	SHAHWAN M J, GACEM S A, ZAIDI S K. Prevalence of diabetic nephropathy and associated risk factors among type 2 diabetes mellitus patients in Ramallah, Palestine[J]. Diabetes Metab Syndr, 2019, 13(2): 1491-1496.
[21]	USUI I. Common metabolic features of hypertension and type 2 diabetes[J]. Hypertens Res, 2023, 46(5): 1227-1233.
[22]	YANG J, WANG X, JIANG S. Development and validation of a nomogram model for individualized prediction of hypertension risk in patients with type 2 diabetes mellitus[J]. Sci Rep, 2023, 13(1): 1298.