B-blocker Compliance Post-MI: Costs, Causes, Disparities

心脏病是65岁及以上美国人死亡的主要原因，也是45-64岁年龄组的第二大死因。这些死亡中很大一部分是由于急性心肌梗死。AMI后长期-受体阻滞剂治疗是一种已被证实的降低死亡率、心血管死亡率和重复AMI的方法。-受体阻滞剂治疗的短期依从性很好，而且正在改善，但最近的研究表明，在AMI后的一年内，-受体阻滞剂治疗的依从性急剧下降：在AMI后一年内，只有一半或更少的患者依从。此外，人们对这种下降的原因和后果知之甚少。鉴于心脏病的治疗和结果普遍存在年龄、性别和种族差异，依从性差异可能特别突出。因此，在使用大型健康保险公司索赔数据的观察性、回顾性设置中，我们的目标是：1)测量ami后长期β受体阻滞剂治疗依从性和非依从性患者之间的医疗费用差异。2)衡量基于年龄、性别、种族、收入和保险状况的依从性差异。3)测量依从性如何随患者的医生特征和医患互动特征而变化。4)将依从性变化分解为医生特异性和患者特异性成分。5)测量依从性和非依从性患者在死亡率、再梗死和再入院方面的差异，以评估我们数据中的选择偏倚。

Heart disease is the leading cause of death among Americans 65 and older and the second leading cause of death in the 45-64 age group. A substantial fraction of these deaths are due to acute myocardial infarction. Long-term beta blocker therapy after AMI is a proven method for reducing mortality, cardiovascular mortality, and repeat AMI. Short-term compliance with beta-blocker therapy is good and improving, but recent studies show that in the year after AMI compliance with beta-blocker therapy drops sharply: half or fewer of patients are compliant at one year post-AMI. Furthermore, little is known about the causes or consequences of this decline. Given the widespread age, sex, and racial disparities in treatment and outcome for heart disease, disparities in compliance may be particularly salient. In an observational, retrospective setting using claims data from a large health insurer, therefore, we aim to: 1) measure the difference in healthcare costs between patients compliant and non-compliant with long-term beta-blocker therapy post-AMI. 2) measure disparities in compliance based upon age, sex, race, income, and insurance status. 3) measure how compliance varies with the characteristics of a patient's physician and with the characteristics of the physician-patient interaction. 4) decompose the variation in compliance into physician-specific and patient-specific components. 5) measure differences in mortality, reinfarction, and readmission between compliant and non-compliant patients, in order to assess selection bias in our data.