Project Summary Almost 300,000 patients undergo cardiac surgery in the U.S. annually, and up to 30% develop cardiac surgery associated acute kidney injury (CSA-AKI). The complication is potentially preventable and results in 3-4 additional intensive care unit (ICU) days, $10-40K added health expenditures per case, and up to 10-fold increased odds of mortality. In severe cases requiring dialysis, consequences are even greater. While efforts to identify targets for CSA-AKI risk-reducing interventions have focused on underlying patient risk and surgical details, the intraoperative and early ICU periods remain understudied sources of outcome variation. Intraoperative and early ICU periods present unique opportunities for reducing CSA-AKI due to (i) unique renal insults due to altered hemodynamics of cardiopulmonary bypass; (ii) major physiologic shifts and clinical interventions detailed in granular intraoperative and ICU data, and (iii) amenability to process change. A paradigm shift in how CSA-AKI is understood during these periods remains possible through the addition of detailed minute-to-minute intraoperative and early ICU data to factors in traditional prediction models, but requires advanced analytical approaches to identify patterns within the 25,000 physiologic, fluid, medication, and intervention data points available for each patient. The potential value of such patterns is emerging in recent studies, yet remains unvalidated in large, contemporary cardiac surgery populations. Moreover, for CSA-AKI prediction models to be maximally informative, risk-reducing modifiable processes of care are likely not “one size fits all”, as are commonly applied in healthcare despite important heterogeneity of treatment effects across diverse pathologies and procedures. Finally, efforts to translate evidence to practice often fail, due to poor communication of evidence-based, patient-specific benchmarking data to individual clinicians. We propose a multicenter study leveraging the integration of two mature, unique research and quality improvement collaboratives built upon national, standardized registries: the Multicenter Perioperative Outcomes Group (MPOG) and the Society for Thoracic Surgeons (STS) Adult Cardiac Database. MPOG uses nurse- validated, detailed minute-to-minute intraoperative and early ICU data from the EHR for research and quality improvement. Headquartered at University of Michigan and guided by Associate Research Director PI Mathis, MPOG has integrated 16 million patient records across >40 health systems in 22 states and provides monthly automated performance improvement benchmarking reports to 5,000 frontline anesthesiology clinicians. In addition, MPOG has integrated each member hospital’s STS clinical registry to create a unique national “MPOG- STS” dataset of 80,000 cardiac surgeries. We will (i) identify high-impact, modifiable intraoperative and early ICU processes of care associated with reduced CSA-AKI; (ii) estimate the impact of...