PROJECT SUMMARY – PROJECT 5 An adage in acute care medicine is that while patients don’t want to die badly, they also don’t want to die. Increasingly, hospital-based physicians must negotiate this difficult balance. They must implement time- sensitive treatment algorithms for acute care conditions such as severe sepsis, stroke, and heart attack to mitigate mortality risk while simultaneously considering patients’ goals of care to respect their end-of-life (EOL) treatment wishes. Making this process even more challenging is that often the physician will be seeing the patient for the first time. To this end, hospitalists often have advance care planning (ACP) discussions with patients and their families, often regarding advance directives (ADs). Yet rates of treatment-limiting ADs implemented in the hospital (e.g., “do-not-resuscitate” (DNR) orders) vary widely from hospital to hospital for otherwise similar seriously ill older adults. Similarly, rates of cardiopulmonary resuscitation (CPR) and mechanical ventilation among ICU patients with pre-existing DNR orders vary from ICU to ICU, most likely reflecting both over-use and under-use of treatments. The goal of this project is to improve the quality of decision making for seriously-ill, hospitalized older adults. Based on our prior research involving participant observation, simulation, and cognitive interviewing, our overarching hypothesis is that a single physician judgment contributes to these variations, whether or not a patient is near the “end of life” (EOL). We hypothesize that this judgment is not simply a product of explicit, knowledge-based prognostication, but is influenced by physicians’ heuristics (implicit, unconscious cognition related to pattern recognition) and by local social norms. The specific aims of this project are: 1) To understand the cognitive processes that influence hospital-based physicians’ judgment that a patient is near the “EOL,” and therefore their likelihood of ACP discussions;; 2) To explore the association between acute care ACP and patient care outcomes;; and 3) To test the effect of interventions designed to influence physician cognitive processes, compared to usual care QI alone, on the likelihood of ACP discussions using a randomized trial. To achieve these aims, we will use a combination of observational and experimental research linking proprietary clinical data with claims in partnership with a national physician management group. This group represents 2,500 hospital-based physicians at 250 community hospitals across the U.S who manage approximately half a million admissions among patients 65 and older each year. This project uses resources from all 3 Cores, provides a data collection platform for Core C’s validation of network measures employed by all projects, and offers synergies with Project 1 and 2, including how labeling a patient with dementia may influence physicians’ decisions about acute care treatment...