Large Dataset Analysis
With a growing focus on outcomes assessment, the use of available data in hospital benchmarking and clinical research has grown significantly. Drawing strength from large numbers of observations, multiple data points and longitudinal follow-up, patient registries and claims databases have played a major role in assessing risk-adjusted outcomes in cardiothoracic surgery. With increased federal funding for research aimed at the development of clinical registries, clinical data networks, and electronic health care infrastructures, the role of this data in clinical research will continue to expand.
Our group has significant experience using "real-world" data to pursue clinical questions of interest. This includes the use of large patient registries and claims data, such as the United Network for Organ Sharing (UNOS) Scientific Transplant Registry, Healthcare Cost and Utilization Project (HCUP) Nationwide Inpatient Sample (NIS), New York State Cardiac Surgery Database, Thompson Reuters’ Marketscan Database, and Center for Medicare and Medicaid Services. Further, we are contributors to the Society of Thoracic Surgeon (STS) Database, International Registry of Acute Aortic Dissection (IRAD), and the Interagency Registry for Mechanically Assisted Circulatory Support (InterMACS).
Our group is also exploring the use of data mining for developing risk models. Data mining uses methods from statistical, artificial intelligence, and database management to discover new patterns.