Critical illness is a significant public health problem affecting millions of Americans each year and associated with considerable morbidity and mortality. Many of these deaths are likely preventable, as evidenced by considerable variation in risk-adjusted mortality across ICUs and persistent gaps between clinical evidence and practice. Therefore, system-wide strategies to improve ICU outcomes and speed translation of evidence into practice are needed. Public reporting of healthcare outcome measures is one such strategy. In theory, public reporting works by providing consumers with the necessary information to select high-quality healthcare providers and by motivating physicians and healthcare systems to compete on quality. Although public reporting programs are expanding, there is limited evidence that these programs improve healthcare outcomes along with concerns that they may cause unintended negative consequences including the avoidance of sick patients by healthcare providers since these patients may negatively impact provider's publicly reported healthcare performance measures. Given the potential benefits and risks associated with public reporting, it is essential to better understand its impact in the ICU prior to implementing it on large scale. This proposal will address these knowledge gaps by taking advantage of a natural experiment in which California has been publicly reporting ICU in-hospital mortality rates since 2007, while other states have not. The overall goal of this proposal is to determine the effect of public reporting in the ICU by comparing changes in ICU case-mix and outcomes in California to changes in other parts of the country before and after the implementation of public reporting. First, we will determine the effect of ICU public reporting on selection of patients for ICU admission, examining whether the type of patients admitted to ICUs changed differentially in California compared to other regions. Second, we will determine the effect of ICU public reporting on risk-adjusted ICU outcomes, examining whether in-hospital mortality, 30-day mortality, and post-acute care utilization changed differentially in California compared to other regions. Our project will use national, patient-level data from the Medicare Provider Analysis and Review (MedPAR) files, state-of-the-art Bayesian statistical models, and a "difference-in-differences" approach to help control for variation in case-mix and temporal trends across regions. Overall our results will provide important knowledge that can be used to guide clinical decision making and healthcare policy regarding system-wide performance improvement initiatives in the ICU. Additionally, this project will provide the applicant the opportunity to expand her research skills in the area of critical care outcomes and advanced statistical modeling. Through the applicant's training plan which includes project- based learning, mentoring, coursework, and conferences, the candidate will acquire training in outcomes research, biostatistics, and healthcare policy. This training plan has been designed to assure the candidate's successful development to an independent researcher.
Over 5 million Americans are admitted to an intensive care unit each year, and critical illness syndromes such as severe sepsis and acute lung injury claim hundreds of thousands of lives annually. Public reporting of ICU performance data is a potentially important strategy to improve outcomes for critically ill patients, but its overall impct is unknown. By determining the impact of public reporting in the ICU, this project will provide clinicians, researchers, and policy makers with information on how to best design system-level strategies to improve ICU performance and outcomes nationwide.