冠狀動脈心臟病危險因子及診療指引之探討：卡方自動互動偵測法之應用; Identifying the risk factors and establishing a diagnostic guideline of coronary artery disease---An application of Chi-Square automatic Interaction Detector method

CMUR > College of Public Health > Department and Graduate of Health Services Administration > Theses & dissertations > Item 310903500/24172

Please use this identifier to cite or link to this item: http://ir.cmu.edu.tw/ir/handle/310903500/24172

Title:	冠狀動脈心臟病危險因子及診療指引之探討：卡方自動互動偵測法之應用;Identifying the risk factors and establishing a diagnostic guideline of coronary artery disease---An application of Chi-Square automatic Interaction Detector method text/html
Authors:	鍾國屏
Contributors:	中國醫藥學院醫務管理研究所
Keywords:	冠狀動脈心臟病;危險因子;卡方自動互動偵測法;診療指引;Coronary artery disease;Risk factor;Chi-Square Automatic Interaction Detector method;Diagnostic guideline
Date:	1991
Issue Date:	2009-12-03 09:46:45 (UTC+8)
Abstract:	為了定義出適合台灣地區之冠狀動脈心臟病之危險因子，及利用所定義出的危險因子以決策樹之模型、定義出具有不同程度及類別危險因子之病患其罹患冠狀動脈心臟病之機率，而建立一套可協助醫師決定是否進行心導管的診斷指引，我們搜集2001年中部地區為了檢查是否罹患冠狀動脈心臟病病所進行之心導管檢查個案1001筆，以適當的統計方法﹝卡方自動互動偵測法﹞進行分析。在1001位樣本中，有305位(30.47%)心導管檢查結果為假陽性。在所有11個變項中，「性別」是最能夠預測罹患冠狀動脈心臟疾病與否的變項，在614位男性中，471位(76.71%)被證實至少有一條冠狀動脈50% 狡窄，而在387位女性中，只有225位(58.14%)有此結果。在男性中，高血壓為下一個最顯著的預測變項，在患有高血壓的男性中，82.1%的機率會罹患冠狀動脈心臟疾病，在無高血壓的男性中，此機率只有63.01%。而在患有高血壓的男性中，下一個能夠顯著預測的變數為糖尿病，患有糖尿病的高血壓男性，其罹病之機率為90.44%，而未患有糖尿病的高血壓男性，其罹病的機率為78.36%。在同時患有高血壓及糖尿病的男性之中，若再加入是否缺乏運動，可發現最高罹病群組其罹病機率為95.40%，亦即是患有高血壓、糖尿病及缺乏運動之男性，其罹病機率最高。而在女性部份，因變數的交互作用，本研究發現與男性不同的結果。在女性部份，下一個最顯著預測變項為缺乏運動，缺乏運動的女性，接受心導管檢查出現假陽性的機率高達69.39%，也就是說此組接受心導管檢查真正罹患冠狀動脈心臟疾病的機率只有30.61%。有無糖尿病對冠狀動脈心臟疾病之罹病率也有顯著之影響、分別為82.09%及63.01%。在無糖尿病、缺乏運動之女性，肥胖或過重對冠狀動脈心臟疾病之罹病率也有顯著之影響，其罹病率高達90.0%，而正常體重為71.05%。急躁性格為最後ㄧ預測變項。本研究之敏感度經計算為76.4%；特異度為78.8%；陽性預測值為91.95%。本研究最後結果將1001個樣本依其在個危險因子上之數值，以及對依變項之結果，總計分割為12組相似之群組。12組中，冠狀動脈心臟疾病之罹病率從最高95.4 %至最低28.57%。; To identifying the risk factor of coronary artery disease and establishing a diagnostic guideline of cardiac catheterization in Mid-Taiwan, one thousand and one patients were included for evaluation of the risk factor of coronary artery disease after cardiac catheterization from a medical center and regional hospital using CHAID. Of the 1001 suspected patients, 69.53% were abnormal and 30.47% were normal and near normal. Sex, hypertension, diabetes, lack of exercise, age, obesity, and type A personality were chosen as predictors of developing CAD in the exhaustive CHAID analysis. The strongest predictor of outcomes on the basis of the exhaustive CHAID analysis is sex. Of all suspected patients, males have more chance of developing CAD than females. The probability of developing CAD is 76.71% for males and 58.14% for females. For the male group, hypertension is the most important factor in further segmentation. Those with hypertension have a 0.821% probability of developing CAD, while those without hypertension have only a 63.01% probability. Diabetes is the next most significant variable in this subgroup. For patients with hypertension, the probability of developing CAD increases from 78.36% without diabetes, to 90.44% with diabetes. Lack of exercise also plays an important role in males with hypertension, but only affecting those with diabetes. Male patients with suspected hypertension and diabetes who don’t exercise regularly have a 95.4% probability of developing CAD, as compared to a lower 81.36% for those taking regular exercise. Focusing on the female group, lack of exercise is the next most significant variable. Suspected female patients who take regular exercise have a 46% chance of developing CAD, as compared to a 69.39% chance for those who lack regular exercise. For the latter group, diabetes is the most important factor in further segmentation. Those with diabetes have an 82.09% chance of developing CAD and those without diabetes have only a 63.01% chance. For the female group who do not take regular exercise and do not suffer from diabetes, suspected patients who are overweight and obese are more likely to develop CAD. The probability decreases from 90% for those who are overweight and obese to 71.05% for those with normal weight. Finally, type A personality is the next significant factor, but at a lower level of the tree. The sensitivity, specificity and predictive accuracy of the exhaustive CHAID model are 76.40%, 78.8% and 91.95% respectively. In summary, 12 homogeneous groups have been obtained from the analysis of exhaustive CHAID, with the probabilities of developing CAD ranging from as high as 95.4% to as low as 28.57%.
Appears in Collections:	[Department and Graduate of Health Services Administration] Theses & dissertations

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	221	View/Open

Loading...