Understanding the Discharge Severity Index (DSI): A Tool for Risk Stratification and Post-ED Resource Allocation

Published in the American Journal of Emergency Medicine (2025), the Discharge Severity Index (DSI) represents a significant advancement in emergency medicine’s effort to optimize post-discharge care and reduce avoidable readmissions. Developed by researchers at Massachusetts General Hospital, Harvard Medical School, and international collaborators, DSI offers a validated, pragmatic framework for predicting short-term readmission risk among emergency department (ED) discharges.

Why the DSI Was Created

Emergency departments are increasingly discharging patients with moderate acuity illnesses, relying heavily on outpatient follow-up for recovery. However, adherence to follow-up plans remains alarmingly low—often due to patient confusion, lack of access, or unclear instructions. As a result, many patients return to the ED and are readmitted within days, placing pressure on already strained healthcare systems.

The DSI was conceived to meet a critical need: a simple, reliable, and generalizable tool that enables ED providers to prioritize patients for enhanced post-discharge care based on their risk of readmission.

How the DSI Works

The DSI is a five-level categorical score derived from retrospective analysis of over 229,000 ED discharges from the MIMIC-IV database. It assigns each discharged patient to one of five risk groups (DSI 1 = highest risk, DSI 5 = lowest) based on six key variables, all of which are routinely available at discharge:

Risk Factor	Points
Age > 65	1
Heart rate > 100 bpm	1
Oxygen saturation < 96%	1
>5 active medications prescribed	2
Length of ED stay > 3 hours	2

Each patient’s total score determines their DSI category:

DSI Category	Score Range	Odds Ratio for Readmission
DSI 5	0	1.00 (reference)
DSI 4	1–2	3.49
DSI 3	3–4	8.44
DSI 2	5–6	11.65
DSI 1	>6	14.63

In validation, patients in DSI 1 had a 4.67% 7-day readmission rate, compared to 0.39% for DSI 5.

Clinical Utility of the DSI

The DSI is designed to guide—not dictate—clinical decision-making. Its main value lies in:

• Stratified follow-up planning: Prioritize patients in DSI 1 and DSI 2 for early follow-up, home health monitoring, or telehealth support. 
• Resource optimization: Allocate limited post-discharge resources more efficiently based on patient risk. 
• Generalizability: Based solely on universally available discharge data (vital signs, medications, LOS), DSI is deployable across diverse settings without needing labs or complex EHR integration. 

Importantly, DSI is not prescriptive; rather, it supports local tailoring of follow-up protocols based on available resources.

DSI vs. Emergency Severity Index (ESI): What’s the Difference?

While both DSI and ESI use 5-level scales and support resource prioritization, their contexts, purposes, and timing are fundamentally different.

Feature	DSI	ESI
Purpose	Predict risk of readmission post-discharge	Prioritize urgency of care at ED triage
Timing	Applied at discharge	Applied at ED arrival
Input variables	Age, vitals at discharge, meds, LOS	Symptoms, vital signs, anticipated resources
Target application	Post-ED resource planning	Intra-ED flow and care prioritization
Outcome predicted	7-day ED return + readmission	Immediate acuity and need for timely care

In essence, ESI governs who needs care first; DSI governs who needs care next.

What Makes DSI Unique Compared to Other Readmission Scores?

Unlike tools such as the LACE or HOSPITAL scores, which focus on 30-day readmission from inpatient settings, DSI:

• Targets ED discharges only, 
• Focuses on short-term (7-day) risk, 
• Uses discharge-time vitals and observations  (e.g., LOS), 
• Avoids demographic variables like race or insurance, enhancing fairness and generalizability, 
• Prioritizes simplicity, making it usable even in lower-resource environments. 

It is particularly relevant in modern healthcare systems striving to expand outpatient care capacity while minimizing unplanned returns.

Limitations and Future Directions

While promising, the current version of the DSI has some limitations:

• Single-center development may affect generalizability. 
• High rate of exclusions due to missing or implausible vital signs (approx. 8% of the original dataset). 
• Lack of social determinants of health (SDOH) in the model—an important area for future enhancement. 
• The model does not yet prescribe specific interventions per DSI category. 

Future work may include:

• External validation across other hospitals and countries. 
• Integration into electronic health record (EHR) systems for real-time scoring. 
• Machine learning augmentation to include SDOH, lab data, and patient-reported outcomes. 
• Tailored intervention studies testing specific follow-up strategies by DSI level. 

Final Thoughts: A Step Toward Smarter Post-ED Care

The Discharge Severity Index (DSI) represents a conceptual and practical shift in how we approach post-ED discharge care. By enabling data-informed follow-up prioritization, the DSI helps bridge the gap between emergency stabilization and long-term health management.

As emergency medicine evolves, so must our tools—not only to save lives in the acute phase but also to prevent suffering and system strain afterward. DSI is one such tool—built not just for triage, but for transitions.

Author:
Ehsan Seif M.D.