Shaw L J, Miller D D, Gillespie K N, Younis L T, Chaitman B R, Romeis J C
Duke University Medical Center, Durham, NC 27705, USA.
Clin Perform Qual Health Care. 1995 Oct-Dec;3(4):209-17.
An outpatient-based scoring system was developed for at-risk patients with coronary artery disease based on data derived from the clinical history and noninvasive testing results for the prediction of an adverse event, the development of risk subsets, and the evaluation of the appropriateness of utilization patterns in an ambulatory care patient population.
This was a hospital-based cohort study. From a population of 3,795 consecutively tested patients, 872 with suspected coronary artery disease were enrolled from a midwestern university tertiary medical center from 1988 to 1989.
Multivariable Cox modeling was used to develop scoring weights with scores ranging from -1.6 to 8.5 points. Significant multivariable disease predictors of cardiac death or myocardial infarction were use of nitroglycerin or insulin, ST-T wave changes, female gender, left ventricular hypertrophy, and a reversible thallium 201 defect. Receiver operating characteristics curves by use of the hazard score were comparable by gender. A probability threshold of .30 for cardiac death or myocardial infarction yielded a cut point of acceptable sensitivity and specificity for prompting medical management decisions. Below this threshold, the rate of follow-up diagnostic testing was 16.9% for women and 57.8% for men (p=.00001). Above this threshold, the rate of follow-up diagnostic testing was 40.6% for women and 64.3% for men (p= .04).
Use of cardiac diagnostic services and cardiac event-free survival varies by gender in patients screened by noninvasive testing. For men at low risk of cardiac death or myocardial infarction, a statistically greater use of follow-up diagnostic testing was reported, thus reflecting more aggressive treatment and overuse of services for men as compared with women.
基于临床病史和无创检测结果数据，为冠心病高危患者开发一种门诊评分系统，用于预测不良事件、划分风险亚组以及评估门诊患者人群利用模式的适宜性。
这是一项基于医院的队列研究。1988年至1989年，从一所中西部大学三级医疗中心连续检测的3795名患者中，纳入了872名疑似冠心病患者。
采用多变量Cox模型制定评分权重，分数范围为-1.6至8.5分。心脏死亡或心肌梗死的显著多变量疾病预测因素包括硝酸甘油或胰岛素的使用、ST-T波改变、女性性别、左心室肥厚以及可逆性铊201缺损。使用风险评分绘制的受试者工作特征曲线在性别间具有可比性。心脏死亡或心肌梗死的概率阈值为0.30时，产生了一个具有可接受敏感性和特异性的切点，以促使做出医疗管理决策。低于该阈值，女性的随访诊断检测率为16.9%，男性为57.8%（p = 0.00001）。高于该阈值，女性的随访诊断检测率为40.6%，男性为64.3%（p = 0.04）。
在通过无创检测筛查的患者中，心脏诊断服务的使用和无心脏事件生存率因性别而异。对于心脏死亡或心肌梗死低风险的男性，报告显示其随访诊断检测的使用在统计学上更多，因此与女性相比，反映出男性接受了更积极的治疗以及服务的过度使用。
