Applying a Group-based Trajectories Methodology to Measures of Patient Safety

In 2000, the Institute of Medicine released a report on patient safety that included an estimate that adverse events occur in 2.9 to 3.7 percent of hospitalizations (Institute of Medicine, 2000). Every year, 50,000 to 100,000 people die from preventable adverse events that occur during hospital stays in the United States. The estimated cost of these events is $17 billion (Institute of Medicine, 2000). Since the IOM report, measuring and reporting patient safety has become an important component of healthcare policy and research.

Despite the attention and funding that studies of patient safety have received, there are still gaps in our understanding of the scope of the problem and the causes (Leape and Berwick 2005; Zhan et al 2005). These gaps are due in part to the difficulty in identifying medical errors using existing data as well as the reliance on analytic and reporting methods that may be misleading. Government agencies report patient safety levels and trends in the U.S. (CDC 2004; AHRQ 2007c), and public and private organizations provide annual grades and rankings of hospitals in the nation based on their patient safety practices and records (The Leapfrog Group 2008; Hospital Compare 2009). These reports and rankings are either cross-sectional or present an average trend, both methods that can miss important information about the distribution of patient safety rates in the population of hospitals.