Improved Estimates of the Comparative Treatment Effects of CABG and PCI

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): This application addresses Challenge Area 10 and specific Challenge Topic 10-HL- 101*,"Information Technology for Processing Healthcare Data for Research," specifically to the challenge topic 10-HL-101*, "Develop data sharing and analytic approaches to obtain from large- scale observational data, especially those derived from electronic health records, reliable estimates of the comparative treatment effects and outcomes of cardiovascular, lung, and blood diseases." The availability of electronic health data now permits researchers to compare the outcomes of patients who received different treatments for the same clinical condition, which would be advantageous by assessing treatment among typical patients seen in representative practice settings rather than relying solely upon formal randomized controlled trials with highly selected patients. The major limitation of observational comparisons is that many factors may affect either the choice of treatment, or clinical outcomes, so comparisons of treatment derived from observational data may not be reliable. The goal of this study is to evaluate several statistical approaches to derive reliable treatment comparisons from observational data using coronary artery bypass graft surgery and coronary angioplasty as the test case. Various analytic methods will be compared in a large observational database from the electronic records of Kaiser Permanente of Northern California, and the validity of these methods will subsequently be assessed using an independent database of 7,812 individual patients randomized to surgery or angioplasty in one of ten randomized clinical trials. A variety of observational methods, including propensity score approaches, instrumental variables approaches and emerging techniques, will be applied to determine both overall comparative effectiveness and variations in effectiveness according to baseline clinical characteristics. The most reliable method(s) will then be applied to compare surgery and angioplasty in patient groups that were poorly represented in the trials (e.g., patients more than 75 years old), as well as in other large datasets, such as Medicare files. The methods validated in this test case study of coronary disease will then be more general tools that may be applied broadly to evaluate other therapies and other patient populations. The current proposal would positively impact the economy by creating or retaining two new jobs at Stanford Medicine and elsewhere. (According to the California Biomedical Industry, for every one employee of a biomedical organization, another three to five will be employed in firms that service that industry.) Public Health Relevance: This project will develop methods to use information from electronic medical records to reliably compare treatments for heart disease. Better methods to use electronic data collected in routine practice will help compare the benefits and risks of treatments in routine medical care settings.