更新的Diamond-Forrester法和Duke临床评分预测模型对可疑冠心病患者的诊断价值

摘要

目的：比较更新的 Diamond - Forrester 法（UDFM）和 Duke 临床评分（DCS）两种预测模型对我国可疑冠心病患者评估的准确性，并进一步分析两者在不同性别之间准确性的差异。方法选取2010年1月—2015年5月因胸痛在中南大学湘雅三医院心内科行冠状动脉造影（CAG）的患者1311例。分别利用 UDFM、DCS 估算患者的验前概率（PTP），并分为低（<30%）、中（30%~70%）、高（>70%） PTP 组，高 PTP 组即认为患有冠心病。以 CAG为金标准，分析 UDFM、DCS 对我国可疑冠心病患者诊断的准确性，并进一步分析两者在不同性别之间准确性的差异。结果 CAG 结果显示，739例（56.37%）患者确诊为冠心病。UDFM 结果显示，294（22.43%）、673（51.33%）、344（26.24%）例患者分别纳入低、中、高 PTP 组； DCS 结果显示，165（12.59%）、403（30.74%）、743（56.67%）例患者分别纳入低、中、高 PTP 组。两者 PTP 分布比较，差异有统计学意义（χ2=379.00，P <0.001）。在确诊的739例患者中，UDFM 将125（16.91%）、372（50.34%）、242（32.75%）例患者分别纳入低、中、高 PTP组，DCS 将64（8.66%）、189（25.58%）、486（65.76%）例患者分别纳入低、中、高 PTP 组，两者确诊患者 PTP分布比较，差异有统计学意义（χ2=257.00，P <0.001）。以 CAG 为金标准，UDFM 诊断冠心病的灵敏度为32.8%，特异度为82.2%，正确率为54.3%，受试者工作特征（ROC）曲线下面积为0.64〔95% CI（0.61，0.67）〕；DCS 诊断冠心病的灵敏度为65.8%，特异度为55.1%，正确率为61.1%，ROC 曲线下面积为0.63〔95% CI（0.60，0.66）〕。两者 ROC 曲线下面积比较，差异无统计学意义（Z =0.33，P >0.05）。UDFM 对男性典型胸痛患者、不典型胸痛患者、非心绞痛型胸痛患者计算的 PTP 与实际阳性率较接近，而对女性患者，除50~59岁典型心绞痛患者存在高估现象外，余均存在低估现象；DCS 对男性患者存在高估现象，对女性患者，除典型心绞痛患者存在高估现象外，余均存在低估现象。结论 UDFM、DCS 对于我国可疑冠心病患者的评估准确性不高，且存在性别差异。与欧美国家明显不同的是两者均对女性存在低估现象。因此，对我国可疑冠心病患者进行 PTP 的估算需要依据不同性别建立更为准确的模型。%Objective To compare the evaluation accuracy of the two prediction models———updated Diamond -Forrester method(UDFM)and Duke clinical score(DCS)for patients with suspected coronary artery disease in China,and further analyze their accuracy differences between different genders. Methods 1 311 patients who had underwent CAG surgery because of chest pain in the Department of Cardiology of the Third Xiangya Hospital of Central South University from January 2010 to May 2015 were enrolled in this study. UDFM and DCS were employed to estimate the pretest probability(PTP)of the patients respectively,and PTP was divided into three groups,which were low PTP( < 30% ),medium PTP(30% - 70% )and high PTP( > 70% )groups;patients in the high PTP group were taken as having coronary artery disease. Taking CAG as the golden standard,we analyzed the diagnostic accuracy of UDFM and DCS in patients with suspected coronary artery disease in China, and further analyzed the accuracy differences of these two models between different genders. Results Of the 1 311 patients, CAG result displayed that 739 (56. 37% ) cases were confirmed as coronary artery disease. UDFM result showed that 294 (22. 43% ),673(51. 33% )and 344(26. 24% )cases were grouped into low PTP,medium PTP and high PTP respectively;DCS result demonstrated that 165(12. 59% ),403(30. 74% ) and 743(56. 67% ) cases fell into the groups of low PTP, medium PTP and high PTP. The two PTP distribution comparisons showed significant difference(χ2 = 379. 00,P < 0. 001). Of the confirmed 739 cases,UDFM brought 125(16. 91% ),372(50. 34% ) and 242(32. 75% )cases into the low PTP, medium PTP and high PTP groups,while 64(8. 66% ),189(25. 58% )and 486(65. 76% )were classified into the groups of low PTP,medium PTP and high PTP by DCS,and the PTP distribution comparison of the confirmed patients of the two showed significant difference(χ2 = 257. 00,P < 0. 001). Under the CAG golden standard,the diagnostic sensitivity,specificity and the accuracy rate of UDFM in coronary artery disease were 32. 8% ,82. 2% and 54. 3% respectively,and the areas under the curve of ROC was 0. 64〔95% CI(0. 61,0. 67)〕;the diagnostic sensitivity,specificity and the accuracy rate of DCS in coronary artery disease were 65. 8% ,55. 1% and 61. 1% respectively,and the area under the curve of ROC was 0. 63〔95% CI (0. 60,0. 66)〕. There was no significant difference in the areas under the curve of their ROC(Z = 0. 33,P > 0. 05). The calculated PTP of UDFM for male patients with typical angina,patients with atypical angina,and patients with non - anginal chest pain were closer to the actual positive rate,while for female patients,besides the overestimation in the typical angina patients under the age of 50 - 59,the rest were all underestimated;DCS overestimated the male patients,for the female patients,besides the overestimated phenomenon in typical angina patients, the other were all in underestimation state. Conclusion The diagnostic accuracy of UDFM and DCS for patients with suspected coronary artery disease in China is not high, moreover there are gender differences. The two both have the underestimation phenomenon in female patients,which is different from European and American. Therefore,a more accurate model based on different genders is needed to be established for the PTP estimation of patients with suspected coronary artery disease in China.