Replication data for: Do physicians with electronic access to prior imaging actually order more tests? A reassessment of the evidence

McCormick et al. (2012) used 2008 National Ambulatory Medical Care Survey (NAMCS) data to argue that physicians with electronic access to prior imaging results ordered more diagnostic tests than physicians without this health information technology. Despite the inherent limitations of single cross-sectional analyses, we replicated the original study results, made several improvements to the statistical methods, and tested the original and improved models on NAMCS data from 2007 and 2009. We found the original results to be highly model dependent, suggesting the findings are tied to a particular statistical formulation rather than demonstrative of an underlying phenomenon. Governments should fund more rigorous evaluations of new health information technologies, such as time-series analyses or randomized controlled trials. In the absence of such research, health care costs may continue to grow without achieving corresponding improvements in health outcomes.