Each year, millions of American adults are hospitalized with suspected sepsis. Current guidelines recommend prompt administration of antibiotics to reduce mortality among patients who are ultimately diagnosed with sepsis. However, diagnostic challenges impede safe delivery of rapid antibiotics to appropriate patients. Due to the high prevalence and significant associated harm of sepsis-related diagnostic errors, sepsis is one of the `Big Three' conditions prioritized for diagnostic error reduction efforts. Both sepsis under-treatment and over- treatment have significant population health impact due to misdiagnosis-related harm. However, health systems currently lack effective, contextually tailored strategies to improving and maintaining diagnostic quality in sepsis (i.e., correctly identifying patients most likely to benefit from treatment while avoiding adverse consequences of excessive treatment). The goal of this project is to improve the safety and quality of early sepsis diagnosis using principles and tools of safety (SaferDx) and implementation sciences (Consolidated Framework for Implementation Research). We aim to understand which sepsis diagnosis practices in which contexts are effective, necessary, or complementary for accurate early identification and treatment of sepsis patients presenting to the ED and develop a toolkit to facilitate translation of this evidence into practice. We will apply a mixed-methods evaluation approach to accomplish this by successfully completing the following aims: Characterize and describe variability in i) sepsis diagnosis practices (collected using stakeholder surveys), ii) contextual conditions that serve as barriers and facilitators to optimal sepsis diagnosis (captured from stakeholder interviews), and iii) risk-adjusted sepsis diagnosis outcomes (i.e., both under-treatment and over-treatment, measured using electronic health record data from hospitalizations during 2021-2023) across diverse hospitals (Aim 1); Apply coincidence analysis to identify combinations of sepsis diagnostic practices and the context in which they occur that are minimally sufficient and necessary for achieving sepsis diagnostic safety and quality (i.e., high-performance on both under- and over-treatment outcomes) (Aim 2); and Co-design, disseminate, and evaluate usability of an organizational toolkit to facilitate optimal sepsis diagnostic practices and evaluate the toolkit's utility using validated implementation outcome scales (Aim 3). The proposed project is innovative by leveraging the joint concepts of minimizing both under- and over-treatment in reframing sepsis diagnostic excellence and approaching sepsis diagnosis as a team, organizational, and institutional process, rather than an individual decision-making process alone. Our approach will produce an actionable toolkit that is conceptually grounded, informed by rigorous causal inference methods, and feasible to implement in diverse hospitals and contexts. Results will ...