A systems-based approach to revealing patterns of care with the strongest relationship to patient outcomes and costs, with the ultimate goal of generating and disseminating information on how to improve hospital care.
Hospitals exert strong influence on patient outcomes. Performance is quite variable and some institutions achieve markedly better outcomes at lower costs than others, even after controlling for differences in their patients’ characteristics. We have a poor understanding of what accounts for these variations in performance. Lacking tools to characterize the complexities of hospital care for specific patient groups, we remain ignorant of which variations in care are consequential and which are wasteful.
Our collaboration with the Premier network of hospitals gave us access to one of the nation’s most comprehensive data warehouses which contained patient-level clinical and billing data from more than 400 hospitals. Using this database, we created hospital-specific profiles of care or “fingerprints” for a given medical condition. We then sought to identify among these profiles the patterns that were most strongly associated with the best clinical outcomes and the lowest cost, laying the groundwork for possible interventions to improve performance. The steps involved an investigation of systems dynamics and networks, use of data mining and modeling tools, as well as an accurate measurement of outcomes and costs.
Hospitalomics, funded by a 4-year Donaghue Foundation grant, was a multi-disciplinary data-driven approach that combined systems thinking, outcomes sciences, and information sciences to generate information on variation in hospital care and to identify those patterns of care that are strongly associated with the best performance as measured by clinical outcomes (mortality and readmission rates) and hospital costs. Our ultimate goal was to disseminate this information among the hospitals to guide improvement strategies.
The future of the healthcare system depends on our ability to focus our resources where there is a return for patients, promoting value and identifying areas where spending more is not associated with better outcomes. The infrastructure created by this effort provided the ability for ongoing research into comparative effectiveness of organizational strategies to deliver care more safely, effectively, equitably, and efficiently.
May 2012 - The Hospitalomics team assembled for a fourth retreat in New Haven, CT including representatives from the Donaghue Foundation, and from Premier, Inc. (data source). The half day event provided an update on progress-to-date and agreed upon project milestones to be achieved during the second half of the year. Analyses reviewed included temporal trends in the use of Nesiritide, utilization and cost, hospital variation in ICU admission rates for heart failure (HF) patients, and treatment of acute coexisting conditions in patients hospitalized with HF. Methodologies were presented for hospital cost profiling and joint analysis of practice patterns and multi-level health outcomes.
May 2012 - Serene Chen, a second-year medical student and a member of the Hospitalomics team, was nominated as a finalist for the Young Investigator Abstract Award for her presentation at the American Heart Association’s Quality of Care and Outcomes Research (QCOR) meeting in May.)