The Use of High-Dimensional Visualizations in Explaining Hospital Pricing Patterns

Abstract

The Centers for Medicare and Medicaid Services (CMS) has made public a data set showing what hospitals charged and what Medicare paid for the one hundred most common inpatient stays. By law payments approximate a hospital's cost of providing a service. However, the data shows a wide dollar gap between Medicare payments and what hospitals actually charged. We explore the origins of hospital pricing using the technique of Independent Component Analysis, a form of blind source separation. Discovered source signals were interpreted by putting them into context with conditions, including characteristics of individual hospitals and the marketplaces in which they operate. This high-dimensional data consisting of over 100 variables was explored using Weave, a web-based analysis and visualization environment. Four underlying processes that exert influence on hospital pricing were identified, including one that revealed distinguishing features of hospitals at the extreme high and low ends of the charge distribution. Perhaps surprisingly, hospitals that lie on opposite ends of the price scale were found to have many attributes in common as well.