以代謝症候群為基礎的心血管病風險預測模型

Abstract

背景 闡明代謝症候群的概況與其對心血管病的衝擊對公衛政策制定與臨床指南建立至為重要。 方法 本論文所研究的資料來自參與基隆闔家歡篩檢(Keelung Community-based Integrated Screening, KCIS)計劃的民眾。首先，描述代謝症候群的標準化盛行率與發生率。在貝式架構下，依序置入人口學特徵、代謝症候群因子和傳統危險因子來建立心血管病預測模型。最後，利用六階段馬爾可夫模式描述代謝症候群的自然病史、心血管疾病的發生率與死亡率。 結果 代謝症候群盛行率、發生率在男性、女性分別為19.0％、5.4％和13.4％、4.6％。代謝症候群患者相對於未罹患代謝症候群者其死於心血管病死亡相對危險性為1.5。針對某47歲男子，利用建立的貝式模型納入其代謝分數(概似比1.488)後，可將其五年心血管病風險從基本的11.2％更新為15.8％。若加上抽菸(概似比1.62)、心血管病家族史(概似比1.47)等資訊，風險升至30.9％。我們估計代謝異常與心血管病自然病史各階段的轉移率。每年從輕度代謝異常變成代謝症候群的機會6.15％，反之，回復正常的機會為29.11%。代謝症候群患者每年心血管病發生率為20.2%，且風險與代謝異常嚴重度同步。 結論 本文描述代謝症候群與心血管病的流行病學、建立貝式模型以容納不同程度證據來預測心血管病風險、並探討心血管病與代謝症候群的動態變化。這些風險預測模型不論對高危險病人的臨床鑑別或衛生政策的評估均極為有用。

Background Elucidating the epidemic profiles and the impact of Mets on CVD in general population is important in developing public health policy and clinical guideline. Methods Subjects from Keelung Community-based Integrated Screening (KCIS) Program were enrolled. We reported standardizes prevalence and incidence rate of Mets, and constructed Bayesian models to estimate the CVD risk by sequentially incorporating demographic features, metabolic syndrome components and conventional risk factors. Finally, a six-state Markov process was used to describe the natural history of Mets, and incidence and mortality of CVD. Results The age-adjusted prevalence and incidence in men and women were 19.0%, 5.4% and 13.4% 4.6%, respectively. The adjusted RR of Mets on CVD was 1.5. By applying the constructed Bayesian model, the predicted five-year CVD risk for a 47-year-old man could be updated to 15.8% from basic 11.2% after incorporating his metabolic score (Likelihood ratio (LR)=1.488), and further 30.9% after considering smoking status (LR=1.62) and family history(LR=1.47). We estimated transition rates between different states of Mets by the Markov model. In subjects of MMD, about 6.15% progressed to Mets annually and 29.11% regressed to FMD. The annual CVD incidence of Mets cohort was 20.2% and the risk increased with the severity of metabolic status. Conclusions In this thesis, we described the epidemiological profiles for Mets and CVD, demonstrated a sequential Bayesian approach for individual risk prediction, and studied the dynamic changes regarding Mets and CVD. These models are very useful either from personal viewpoints of high risk group identification or population viewpoints of police policies evaluation.