Gastrointestinal bleeding (GIB) is a major cause of inpatient hospitalizations in the United States, and comprises a heterogeneous population with appreciable variability in patient outcomes amongst hospitals. Nationally, 30-day mortality rates are estimated at 7%; however, mortality and readmission rates vary widely across hospitals. Such variability suggests the potential to improve patient care and save lives by identifying modifiable factors that influence outcomes after hospitalization with GIB. Patient-related factors, hospital structure and processes of care play major roles in determining outcomes for many other diseases, but the impact of these factors on outcomes related to GIB has not been adequately defined. The objective of this proposal is to identify and understand sources of variation in care for patients with GIB, including patient-level factors, organization (i.e. structure) and processes of care (e.g.. endoscopic and interventional radiology utilization) amongst hospitals. Our central hypothesis is that modifiable factors impact patient outcomes during hospitalizations for GIB, which include both structural characteristics and processes of care. Our proposal is to build a multi-level model of patient outcomes standardized across hospitals, accounting for patient factors and assess the impact of hospital factors, and then qualitatively explore additional processes of care, through the following interrelated specific aims:
Aim 1 - To develop a risk-standardized model for GIB outcomes across hospitals using Medicare inpatient, outpatient, and pharmaceutical claims data. To ensure generalizability and clinical applicability, model validation will be performed in a multipayer, multicenter database and using UPenn health system data which contains clinical variables.
Aim 2 - to identify hospital factors (structures and processes of care) associated with GIB outcomes, using a combined dataset of the aforementioned Medicare data with American Hospital Association Annual Survey data.
Aim 3 ? to explore barriers and facilitators of processes of care for inpatient GIB among hospitals with better and worse patient outcomes, using qualitative methods. This mixed methods approach will allow for a comprehensive evaluation of drivers of variation in care for gastrointestinal bleeding. The expected outcome for the proposed research is that it will identify important modifiable factors in care associated with patient outcomes for those hospitalized with GIB. This can lead to the development and validation of quality metrics to improve patient care. To facilitate completion of the research and further the academic development of the applicant, the PI will enroll in coursework that builds on her Master of Science by focusing on statistical methods using clustered and longitudinal data, data analytics in confounding, and qualitative methodologies. This proposal has unequivocal and outstanding divisional and institutional support and exceptional mentorship of experienced faculty with a strong track record in epidemiology, health services research, and mentoring prior K-award grant recipients. These activities will allow the PI to develop the skills necessary to become a successful independent NIH-funded investigator and a leader in GI-related health services research.
There is substantial variability in patient outcomes after hospitalization for gastrointestinal bleeding (GIB) across hospitals. Patient-related factors, hospital structure and processes of care play major roles in determining outcomes for many other diseases, but the impact of these factors on outcomes related to GIB has not been adequately defined. This novel proposal will evaluate the effect of patient-level factors to generate risk-standardized rates of GIB outcomes across hospitals and to identify modifiable hospital-level factors using a mixed methods approach to study comprehensively how modifiable factors are associated with better outcomes.
