Assessing and Improving Patient Understanding of Publicly Reported Healthcare-Associated Infection-Related Hospital Quality Measures

Abstract

Background Public reporting of hospital quality data is a key element of health care reform in the United States, with the goals of improving quality of care while reducing costs by encouraging reductions in preventable adverse events (PAE). Healthcare-associated infections (HAIs) are a common PAE that cause substantial morbidity and mortality. HAI rates for hospitals are widely available online, including via http://medicare.gov/ hospitalcompare. Publishing these data requires considerable effort and expense for hospitals and the government. However, there has been little research on the ability of the general public to understand published HAI quality measures. Methods Aim 1: We assessed understanding of HAI data as presented on CMS Hospital Compare among a random sample of University of Maryland Medical Center (UMMC) patients. Participants compared HAI data for two hospitals, and the accuracy of their comparisons was assessed. Aim 2: We analyzed nationwide HAI data to determine their utility in distinguishing among hospitals, and assessed characteristics of this dataset (e.g. geographic areas with hospitals that have substantially different HAI denominators or risk-adjustment profiles) that inform how HAI data are presented. Aim 3: We developed a new method for presenting HAI data to the public. We then conducted a randomized controlled trial comparing this new method to the method from CMS Hospital Compare among a random sample of UMMC patients. Results Aim 1: Participants were able to correctly assess hospital performance 38% of the time (most complex data) to 72% (least complex) of the time. Aim 2: In many geographic areas, HAI-related quality data are diverse enough to distinguish among hospitals. The methods for presenting HAI data on CMS Hospital Compare were suboptimal for displaying characteristics observed in the underlying data. Aim 3: Participants in the experimental arm (with the new data presentation method) got 55.8% of questions correct on average, compared to 31.5% correct in the control arm (p=0.0002). Conclusions The current tabular methods for presenting hospital-level HAI data to the general public on CMS Hospital Compare are understood by one third of patients but can be improved through user centered design.