以混合性馬可夫共病廻歸模型探討代謝症候群與心血管疾病及中風之相關性

Mixture Markov-based co-morbidity regression model for Metabolic Syndrome Associated with Cardiovascular Disease and Stroke

王珮樺(Pei-Hua Wang)

指導教授 ： 陳秀熙

國立臺灣大學/公共衛生學院/流行病學與預防醫學研究所/碩士(2017年)

https://doi.org/10.6342/NTU201703557

摘要

研究背景
　　儘管代謝症候群由五個生物標誌所組合而成已被確立，然而對於疾病如何由一個組成進展至多於兩個的組成，意即共病發展的危險性，或是兩個組成由哪一個先發生、哪一個後發生的問題，是值得研究的問題。此外，是否發生第一個疾病的風險和發展成後續疾病的風險，可被其他代謝症候群組成或受其他因素(如憂鬱症、腺瘤及尿酸)所影響，這些不同類型的疾病組合及其他因素，可能較僅探討代謝症候群，對推定心血管疾病和腦血管疾病發生之危險性有所助益。

研究目的
本論文以高血壓和糖尿病為例，探討單一疾病和兩種疾病共病的疾病路徑，研究目的包含：
(1) 估計族群中會發展高血壓和糖尿病共病所佔的比例。
(2) 估計由第一個(或單一)疾病的發病危險和轉變到第二個疾病的速率
(3) 比較第一個疾病具共病潛力與單一疾病無共病潛力之不同風險。
(4) 比較在給定第一個疾病情形下，第二個續發疾病之不同風險。
(5) 評估代謝症候群組成及其他因子對第一個(或單一)疾病及第二個續發疾病發病風險的效應。
(6) 考量人口學、憂鬱和腺瘤等因子下，估算不同共病路徑發病機率下之心血管疾病和中風疾病風險。

材料及方法
本研究用於估計慢性病共病路徑相關資料是取自1999年至2015年基隆社區整合式篩檢計畫。除了人口特徵和生活型態的問卷調查，總共有104,898位代代謝症候群篩檢參加者納入分析。腺瘤資料是透過免疫法大腸癌糞便潛血檢查篩檢取得。本研究追蹤參加者後續就醫，以了解包含憂鬱症、心血管疾病和中風之罹病情形。
　　本研究利用混合性馬可夫共同廻歸模型評估高血壓和糖尿病之共病可能性，以及評估兩個疾病間發生的先後關係。並利用比例風險回歸模式評估其他代謝症候群組成及其他因素對共病路徑的影響。Cox比例風險回歸模型用以評估上述不同共病路徑發病風險對心血管疾病和中風罹病風險之影響。。

結果
　　研究結果顯示族群中有23.1％(95％CI：22.6％-23.6％)具有潛力形成高血壓和糖尿病之共病。這23％的高血壓或糖尿病患者中，有86％(95％CI：85％-87％)第一個疾病是先有高血壓而之後發展為糖尿病共病。具共病可能性的患者，其先發病為糖尿病是無共病者的36.2倍，先發病為高血壓則是無共病者的11.9倍。相較於第一次發病為高血壓者其發展第二個續發疾病的風險，是第一次發病為糖尿病者的69.7倍。
　　關於代謝症候群的三種其他組成對第一次(單一)疾病(對糖尿病的OR 從2.49倍到8.17倍和對高血壓的OR從1.25倍到1.54倍)，和同時有高血壓和糖尿病的風險(OR 倍1.47到1.63倍)的效應，皆具統計上顯著意義。尿酸對血壓(OR=1.62, 95%CI: 1.50-1.76)，糖尿病(OR=5.43, 95%CI: 4.08-7.23)，和同時有高血壓及糖尿病(OR=1.35, 95%CI: 1.15-1.59)的風險皆顯著效應。然而，腺瘤和憂鬱症對糖尿病與高血壓發展及共同效應均不具統計上顯著意義。
　　調整人口特徵和其他變項後，共病路徑發病危險分數對於心血管疾病的風險（aOR=1.03, 95%CI:1.02-1.05）和中風（aOR=1.05, 95%CI:1.02-1.08）據統計上顯著意義。憂鬱症為心血冠疾病（aOR = 2.19,95％CI：1.81-2.65及中風（aOR = 2.52,95％CI：1.79-3.56）發生之獨立因子），腺瘤也有一致性的效應，但接近統計上顯著意義。

結論
　　透過高血壓及糖尿病的先後發病時序，和兩種疾病的共同可能性評估，研究結果顯示與代謝症候群其三個組成(高密度脂蛋白、三酸甘油酯和肥胖)及尿酸有相關。這種共病路徑對心血管疾病和中風罹病風險的影響不能予以忽略。本研究提供一種嶄新的量化模式，並考量其他共病因子對共病潛在性及不同的疾病發病時序影響，有助於心血管和中風等疾病預防。

關鍵字

代謝症候群 ； 心血管疾病 ； 中風 ； 共病 ； 混合性馬可夫廻歸模型

並列摘要

Introduction
Despite the well-established metabolic syndrome (MetS) with the clustering of five components of biomarkers, it is still elusive whether and how the risk of developing one component (disease) of MetS is affected by the potential of being co-morbidity (more than two components), a latent variable, and also the temporal sequence between two diseases. Moreover, whether the risk of onset of first disease and the risk of developing subsequent disease is affected by the other components of MetS and extraneous variable (such as depression, adenoma, and uric acid) that are putative factors for the risk of cardiovascular disease and cerebrovascular disease (e.g. stroke) other than the components of MetS.
Aims
This thesis is intended to elucidate the pathway, taking hypertension and diabetes mellitus (DM) for example, leading to single disease and comorbidity with both of diseases
(1) to estimate the proportion of being susceptible to hypertension and DM;
(2) to estimate the force of being the first (single) disease and the force of transition to the second subsequent disease;
(3) to compare the risk of being first disease with the potential of having co-morbidity and that of single disease without potential of having co-morbidity based on (1) and (2);
(4) to compare the risk of being second subsequent disease given the first disease by two reverse orders of both hypertension and DM based on (1) and (2);
(5) to assess the effect size of other components of MetS and that of extraneous variables on the risk of first (single) disease and the risk of second subsequent disease;
(6) to relate the probability of risk score based on the pathway of (1) and (2) to the risk of CVD and stroke with adjustment for demographic features and other extraneous variables.
Material and Methods
Data sources: The empirical data used for estimating parameters related to the bivariate pathway on two components of MetS are derived from the Keelung community-based integrated screening (KCIS) since 1999 until 2015. In addition to questionnaire on demographic features and life-style factors, annual health check-up for five components of MetS have been offered for approximately 104, 898 residents. Information on adenoma has been collected through one of colorectal cancer screening with fecal immunochemical test (FIT). Moreover, depression and the outcomes on CVD and stroke during has been obtained by linking the entire cohort with claimed data on national health insurance during follow-up.
Mixture Markov-based co-morbidity model: The previously proposed bivariate co-morbidity pathway was applied to the model the potential of being both hypertension and DM and also the temporal sequence of onset of disease between hypertension and DM. The effects of other components of MetS and extraneous variables such as adenoma, depression, and uric acid were evaluated by using proportional hazards regression form. The probability of being each mode built in bivariate pathway of hypertension and DM was further incorporated into the Cox proportional hazards regression model to assess the influence of bivariate comorbidity pathway on the risk for CVD and stroke with adjustment for the variables that were associated with the risk for CVD and stroke.
Results
The estimated results on bivariate co-morbidity pathway show 23.1% (95% CI: 22.6 %-23.6%) of the underlying population had potential of developing both hypertension and DM. Approximately 86% (95% CI: 85%-87%) out of these 23% with the potential of having both diseases developed first hypertension and then DM later. It is very interesting not that those taking the pathway of co-morbidity were at great risk of being first disease by 36.2-fold for DM and by 11.9-foldfor hypertension compared with those taking the pathway of no potential of co-morbidity. The risk of developing the second subsequent disease was greater for the first disease with DM than that with hypertension by 69.7-fold.
The statistical significance was noted regarding the effects of all three other components of MetS on the first (single) disease (ORs from 2.49 to 8.17 for DM and ORs from 1.25 to 1.54 for hypertension) and also on the risk for having both hypertension and DM (ORs from 1.47 to 1.63). The significant effects of uric acid on hypertension (OR=1.62, 95% CI: 1.50-1.76), on DM (OR=5.43, 95% CI: 4.08, 7.23), and also on the risk of having both hypertension and DM (OR: 1.35, 95%CI: 1.15-1.59) were also noted. However, there was lacking of statistical significance regarding the effect of adenoma and depression for each of two risks.
After adjustment for demographic features and other significant variables, the effect of the probability of being each mode of bivariate co-morbidity was statistically significant for the risk for CVD (aOR=1.03, 95%CI:1.02-1.05) and stroke (aOR=1.05, 95%CI:1.02-1.08). It is very interesting to note that the independent effects of depression on the risk for CVD (aOR=2.19, 95%CI: 1.81-2.65) and stroke (aOR=2.52, 95%CI: 1.79-3.56) were found whereas the corresponding effect of adenoma was of borderline statistical significance.
Conclusions
Bivariate co-morbidity pathway for hypertension and DM in relation to three other components of Mets (HDL, Triglyceride, and Obesity) and also extraneous variables such as adenoma and depression has been quantitatively assessed by providing a clear picture of temporal sequence and the potential of having both disease. The influence of such a bivariate co-morbidity pathway on the risk for CVD and stroke has been noted making allowance for the independent effects demographic features and depression. The findings provide a new quantitative insight into prevention of CVD and stroke by considering the potential of being co-morbidity and also the temporal sequence of first hypertension and then second DM.

並列關鍵字

Metabolic Syndrome ； Cardiovascular Disease ； Stroke ； Co-morbidity ； Mixture Markov-based Regression Model

參考文獻

Ahmed RL, Schmitz KH, Anderson KE, et al. The metabolic syndrome and risk of incident colorectal cancer. Cancer. 2006;107:28–36.. Bowers K, Albanes D, Limburg P, et al. A prospective study of anthropometric and clinical measurements associated with insulin resistance syndrome and colorectal cancer in male smokers. Am J Epidemiol. 2006;164:652–664.

Ahto M, Isoaho R, Puolijoki H, et al. Stronger symptoms of depression predict high coronary heart disease mortality in older men and women. Int J Geriatr Psychiatry. 2007; 22: 757–763.

Aleksandrova K, Boeing H, Jenab M, et al. Metabolic syndrome and risks of colon and rectal cancer: the European Prospective Investigation into Cancer and Nutrition Study.Cancer Prev Res (Phila). 2011;4:1873–1883.

Al-Mahroos, F., Al-Roomi, K., McKeigue, P.M., 2000. Relation of high blood pressure to glucose intolerance, plasma lipids and educational status in an Arabian Gulf population. International Journal of Epidemiology 29, 71–76.

Anda R, Williamson D, Jones D, Macera C, Eaker E, Glassman A, Marks J. Depressed affect, hopelessness, and the risk of ischaemic heart disease in a cohort of US adults. Epidemiology. 1993; 4: 285–294.