Hierarchical models of prescription drug use in the elderly

描述(由申请人提供)：本提案中概述的五年指导研究和培训计划旨在将生物统计学家Brookhart博士培养成一名在老年学、生物统计学和药物流行病学的交叉点工作的独立调查员。这项建议的主要研究目标是开发和应用分级统计模型，以研究老年人使用处方药的复杂模式，这是一个具有重要临床和政策意义的领域。可用于了解用药情况的数据具有多层次结构，包括对聚集在医生、诊所和地理区域内的患者的重复观察。为了捕捉数据的复杂结构，并回答有关老年患者使用处方药的重要实质性问题，我们建议使用现代统计方法来分析分层数据。随着人们对处方药的适当使用和成本越来越关注，特别是在新的医疗保险药物福利的情况下，这种处方和遵守模式可以在改善药物使用方面发挥重要作用。通过了解处方药的常见使用模式，可以设计有针对性的、具有成本效益的质量改进干预措施。在该奖项的最后阶段，候选人将开发多层次统计方法来创建患者、医生、实践和地区概况，这些方法可用于针对性的质量改进干预措施，旨在促进更适当的处方和更高水平的患者对有效治疗的依从性。作为未来R01提案的一部分，将对不同的分析方法进行试点测试和比较。本提案中概述的指导性研究将得到正式培训计划的补充，该计划将涉及老年学、老年病学、慢性病流行病学、健康行为理论、处方药和医疗保健政策以及医疗服务研究的课程；定期与导师和其他高级教师会面；出席研讨会、研讨会和学术会议；以及系统地接触临床老年病学，包括在门诊、住院和长期护理环境中度过的时间。

DESCRIPTION (provided by applicant): The five year program of mentored research and training outlined in this proposal is designed to develop Dr. Brookhart, a biostatistician, into an independent investigator working at the intersection of geriatrics, biostatistics, and drug epidemiology. The primary research aim of this proposal is the development and application of hierarchical statistical models for the study of complex patterns of prescription drug use by the elderly, an area of great clinical and policy importance. The data that are available to understand medication use have a multilevel structure consisting of repeated observations on patients who cluster within physicians, practices, and geographic regions. To capture the complex structure of the data and answer important substantive questions about prescription drug use by older patients, we propose to employ modern statistical approaches for the analysis of hierarchical data. With increasing concern over the appropriate use and cost of prescription drugs, particularly with the new Medicare drug benefit, such models of prescribing and adherence can play an important role in improving the use of medications. By understanding common patterns of prescription medication use, targeted and cost-effective quality improvement intervention can be designed. In the final phase of this award the candidate will develop multilevel statistical approaches for creating patient, physician, practice, and region profiles that can be used to target quality improvement interventions that aim to promote more appropriate prescribing and higher levels of patient adherence to effective therapies. Different profiling approaches will be pilot tested and compared as part of a future R01 proposal. The mentored research outlined in this proposal will be complemented by a formal training program that will involve coursework in gerontology, geriatrics, chronic disease epidemiology, theories of health behavior, prescription drug and health care policy, and health services research; regular meetings with mentors and other senior faculty; attendance at seminars, workshops, and academic conferences; and systematic exposure to clinical geriatrics including time spent in outpatient, inpatient, and long-term care environments.