Improving Decision-Making in Health Care Quality Through Biostatistics

• Activity Number: 266
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Quality and Productivity Section
• Author(s): Henry Domenico*
• Companies: Vanderbilt University School of Medicine
• Keywords: Healthcare ; Quality Improvement ; Study Design ; Predictive Modeling

Abstract

The field of quality improvement in healthcare has not always applied the most rigorous, robust research methods. Pre/Post designs, regression to the mean, and a lack of randomization have made it difficult to determine the effectiveness of many quality improvement interventions. In this poster we demonstrate how biostatisticians working in quality improvement can address these issues and give examples of projects that allow healthcare leaders make data driven decisions. One of the projects we will highlight describes a predictive modeling system for hospital adverse events currently being run as a randomized trial. We will discuss model building with electronic health record data, model validation, how to use an automated predictive model to improve outcomes in a healthcare setting, and the importance of randomization in decision making. We will also describe a unit level crossover randomized trial that can be done at a low cost and is much stronger than a pre/post design. In addition we will show how graphics and advanced methods, such as the bootstrap, have helped answer questions related to healthcare quality.