The Discharge Severity Index: Early Research on ED Readmission Risk Assessment

From Triage to Discharge: As an emergency medicine clinician, you’ve likely become comfortable using the Emergency Severity Index (ESI), a critical tool helping triage patients entering the ED. But what happens when these patients leave your care? How can we anticipate who might need extra support to avoid readmission? Let’s discuss why ED discharge risk stratification matters, the landscape of existing tools, and introduce a new effort called the Discharge Severity Index (DSI), in the context of this evolving conversation. History of Emergency Severity Index (ESI) As emergency medicine clinicians, we’ve all become comfortable with using the ESI. It’s simple, intuitive, and has revolutionized triage since its introduction in the late 1990s. ESI stratifies our incoming patients quickly and reliably based on anticipated resource needs and hospitalization risks, making it easy to decide who gets seen first. Over the years, ESI has gone through multiple iterations to better reflect evolving clinical priorities, workflows, and patient populations [1–4]. It became a living tool that is as dynamic and adaptive as emergency care itself. However, as powerful as ESI is, it addresses only half the equation: what happens when patients arrive. But what about when they leave? Discharge: More than a Binary Decision Currently, ED discharge is largely treated as a binary decision—admit or discharge. But think about admissions: we never treat admissions as simple “yes/no” decisions. Patients can go to observation, a floor bed, step-down units, or the ICU. Each has varying resource needs and follow-up intensities. So why don’t we apply this nuanced thinking to discharge? ED discharges aren’t straightforward. Almost 14% of patients discharged from EDs return within 30 days, often due to issues that could be preventable with better follow-up [5]. Many face barriers like misunderstanding discharge instructions, inadequate social support, and difficulty accessing outpatient care. We have powerful new follow-up tools available (e.g., nursing callback programs, telehealth, remote patient monitoring) but we often lack a clear, systematic way of figuring out which patients truly need them. Existing Tools and Their Limitations Multiple scoring systems have attempted to predict post-discharge adverse outcomes. Some prominent examples include: LACE Score: Length of stay Acuity of admission Comorbidities Emergency visits HOSPITAL Score: Hemoglobin level Oncology diagnosis Sodium level Procedure during hospitalization Index admission type Admissions in previous year Length of stay Yet, many of these tools weren’t specifically designed for the ED population. Our recent scoping review highlighted significant variability, limited ED-specific validation, and complexity that can hinder practical use [6]. Introducing the Discharge Severity Index (DSI): An Early-Stage Tool Recognizing this gap, our team developed the DSI, an initial attempt at ED-specific discharge risk stratification. The idea behind DSI is to use straightforward, quickly accessible ED data points to identify patients who might benefit from enhanced follow-up. Our single-center retrospective study analyzed ED visits, dividing the data into the derivation (75%) and validation (25%) cohorts [7]. We attempted to stratify risk based on the DSI score and measuring their 7-day readmission rates. Our DSI score was calculated using 5 key clinical factors (0=lowest risk, 7=highest risk): Age > 65 years = 1 point Heart rate at discharge > 100 bpm = 1 point Oxygen saturation at discharge < 96% = 1 point Length of ED stay > 3 hours = 2 points Active medications > 5 during hospital stay = 2 points Here’s what we found: DSI LevelScore7-day Approximate Readmission Risk1 (highest risk)6-75%254%33–43%41–21%5 (lowest risk)0<0.5% A patient scoring a DSI 1 might benefit from immediate follow-up with telehealth, home health visits, and/or increased outpatient support. Conversely, a DSI 4 or 5 patient might safely manage standard outpatient care with minimal risk. How is DSI Different Existing Scoring Systems? Unlike the LACE or HOSPITAL scores, the DSI was built specifically for the ED context. It uses data readily available at discharge, allowing rapid identification of patients who may require more intensive post-discharge follow-up. It’s meant for nursing or automated tools to assign this to the patient, without requiring more provider resources. But, let’s be clear: the DSI is not perfect. We intentionally started simple (similar to how ESI began) to get people thinking about stratifying discharge risks. For instance: Length of Stay (LOS): Right now, LOS includes waiting room times, boarding delays, and other systems-level issues, making it an imperfect measure of medical complexity. Vital Signs at Discharge Only: Using only discharge vitals doesn’t account for patients who had unstable earlier vitals during their ED stay. Missing Comorbidities: The current DSI doesn’t explicitly factor in comorbidities or past medical history, which we know affect patient outcomes. Why This Matters to You It’s important to grasp the complexity behind discharge decisions just as clearly as they understand triage. Discharge isn’t simply sending patients home; it’s anticipating what happens next and appropriately preparing patients to succeed. Implementing structured discharge risk stratification not only supports better clinical outcomes but also helps teach clinicians to think about care beyond the ED walls. With more accurate identification of high-risk patients, residents can be better prepared to integrate innovative follow-up resources into patient care. Where do we go from here? The DSI represents an early, evolving concept. We don’t expect it to be adopted widely and imminently. Rather, we hope it sparks a broader conversation similar to the early years of the ESI. ESI began as a simple triage tool and matured through iterative development, field testing, and adaptation across varied ED environments. It became more robust, nuanced, and integrated into daily clinical operations over time. We envision a similar trajectory for the DSI. Future iterations of the DSI will undoubtedly incorporate additional clinical variables, operational data, and even social determinants of health. But before we get there, the next step is clear: we must operationalize the DSI and test it in multiple real-world settings. Its utility must be validated not just in theory or retrospective data, but in the dynamic, complex ecosystem of actual emergency departments. We encourage EM educators and residency programs to join us in refining the conversation about ED discharge stratification. Whether it’s integrating DSI into discharge planning discussions, piloting it during teaching rounds, or evaluating it in post-discharge follow-up workflows, there is now an opportunity to take this idea from concept to practice for the benefit of our patients. Let’s build upon this first step, creating tools that are practical, teachable, and clinically meaningful. Together, we can ensure that the decision to discharge is just as thoughtful, nuanced, and patient-focused as the decision to admit. References