Hospitals prioritize patients primarily by how urgently they need care, using structured triage systems that sort people into acuity levels within minutes of arrival. In most emergency departments worldwide, a trained triage nurse evaluates every patient who walks in and assigns them a score on a five-level scale, with resuscitation-level emergencies seen immediately and minor complaints waiting longest. But that emergency department process is only one layer of a much broader system. How a hospital decides who gets an ICU bed, whose surgery happens first, or who receives a donated organ involves different tools, different trade-offs, and sometimes different ethical frameworks entirely.
The Five-Level Triage System
When you arrive at an emergency department, you are typically assessed within a few minutes by a triage nurse who assigns you a score from 1 (most urgent) to 5 (least urgent). The most widely used version in the United States is the Emergency Severity Index, or ESI. Level 1 means you need immediate, life-saving intervention. Level 2 is a high-risk situation where delays could be dangerous. Levels 3 through 5 reflect decreasing urgency, with Level 5 covering conditions that could reasonably be handled in a clinic rather than an emergency room.
Other countries use similar five-level systems with different names. Canada uses the Canadian Triage and Acuity Scale (CTAS), the United Kingdom and much of Europe use the Manchester Triage System (MTS), and Australia has the Australasian Triage Scale (ATS). All of them share the same core logic: assess the patient quickly, assign a priority level, and use that level to determine who gets seen in what order. Research consistently shows that these five-level systems meaningfully predict patient outcomes. A study using the ESI found a strong statistical relationship between the triage level a nurse assigned and the patient’s eventual clinical outcome.1PubMed Central. Accuracy of the Emergency Department Triage System using the Emergency Severity Index for Predicting Patient Outcome; A Single Center Experience
The concept has surprisingly old roots. The word “triage” comes from the French verb trier, meaning to sort. It was first applied in a medical context around 1792 by Baron Dominique Jean Larrey, Napoleon’s chief military surgeon, who sorted wounded soldiers on the battlefield by the severity of their injuries rather than by rank.2PubMed Central. Evolution of triage systems That principle, treating the sickest first regardless of who they are, still anchors modern triage, though the systems for implementing it have become far more sophisticated.
Time-Critical Conditions Get Their Own Lanes
For certain diagnoses, the standard triage-and-wait process is too slow. Strokes and heart attacks are the most prominent examples. Brain tissue dies with every passing minute during a stroke, so hospitals have developed “fast-track” protocols that bypass normal emergency department flow entirely. When paramedics radio ahead with a suspected stroke, the receiving hospital pre-activates a care team, and the patient may go straight from the ambulance to a CT scanner.
These protocols make a measurable difference. One hospital implementing a multidisciplinary stroke fast-track cut its median door-to-needle time, the interval from arrival to receiving clot-busting medication, from about 82 minutes to 46 minutes. The share of stroke patients treated within the critical 60-minute window jumped from 28% to 81%, and functional independence at 90 days improved substantially.3CME Journal Geriatric Medicine. Reducing Door-to-Needle Time in Acute Ischemic Stroke: Identifying Emergency Department Barriers and Evaluating the Impact of a Multidisciplinary Stroke Fast-Track Protocol A separate study of a digital fast-track workflow found that median door-to-needle time dropped from about 58 minutes to 42 minutes, with every patient in the study group treated within 60 minutes.4PubMed. The Smart Stroke Fast Track: A Digital Workflow Innovation to Reduce Door-to-Needle Time in Acute Ischemic Stroke
Digital communication tools are accelerating this further. A study of an app that allowed ambulance crews to send real-time patient data to the receiving hospital found that stroke patients got their CT scan about 44 minutes faster when the app was used, and time from arrival to first doctor review was cut dramatically.5BMJ Open. Real-world, feasibility study to investigate the use of a multidisciplinary app (Pulsara) to improve prehospital communication and timelines for acute stroke/STEMI care The pattern here is clear: for conditions where minutes matter, hospitals create dedicated pathways that override the general queue.
How ICU Beds Get Allocated
Intensive care units run at high occupancy most of the time, and who gets a bed when demand outstrips supply is one of the more consequential decisions a hospital makes. Under normal conditions, the decision is largely clinical: the patient who most needs continuous monitoring, ventilation, or hemodynamic support gets the ICU bed. Intensivists and admitting physicians make these calls in real time, often multiple times per shift.
When units fill up, trade-offs emerge. Research shows that as ICU occupancy climbs, hospitals start discharging current patients sooner to make room for sicker arrivals. One study of over 65,000 ICU discharges found that going from low to high unit strain was associated with patients being discharged about six hours earlier and having a modest increase in ICU readmissions.6PubMed Central. Outcomes Among Patients Discharged From Busy Intensive Care Units A separate analysis found that once ICU occupancy exceeds roughly 80%, the rate of premature discharges climbs steeply, and above 90% occupancy the problem becomes especially pronounced.7PubMed. Intensive care unit occupancy and premature discharge rates: A cohort study assessing the reporting of quality indicators
This strain does not just inconvenience patients. A large Australian study found that being in an ICU during periods of high strain was associated with roughly 50% higher odds of in-hospital death, increased odds of readmission, and nearly doubled odds of being transferred to another hospital.8PubMed Central. Measuring the Impact of ICU Strain on Mortality, After-Hours Discharge, Discharge Delay, Interhospital Transfer, and Readmission in Australia With the Activity Index The takeaway: ICU prioritization works reasonably well under normal conditions, but the system degrades when the unit is full, and that degradation has real consequences for patients already receiving care.
Elective Surgery Waiting Lists
For surgeries that are necessary but not immediately life-threatening, hospitals use priority scoring systems to decide who goes first. These systems typically combine two factors: how urgently the patient needs the operation and how long they have already been waiting. A patient assigned to the most urgent elective category might have a maximum recommended wait time of 30 days, while a lower-urgency case might be allowed to wait up to a year.
One approach, studied across multiple health systems, uses a dynamic priority formula where a patient’s score increases over time. The longer you wait relative to your recommended maximum wait, the higher your score climbs, so a patient who has been waiting beyond their target window eventually overtakes a newer referral with similar clinical need.9PubMed Central. Managing surgical waiting lists through dynamic priority scoring Clinical factors like pain severity, functional limitation, and risk of progression are also weighted, so the formula does not simply reward patience.10PubMed Central. A model to prioritize access to elective surgery on the basis of clinical urgency and waiting time
The practical result is that two patients with the same diagnosis might wait very different lengths of time depending on symptom severity, comorbidities, and how many other patients are already in the queue for the same specialty. Emergency cases always bump elective ones, which is why a scheduled knee replacement can get postponed repeatedly during periods of high emergency admissions.
Organ Transplant Allocation
Transplant prioritization is the most formalized and regulated version of hospital triage. Each organ type has its own allocation system, and the rules are set at a national level by organizations like the United Network for Organ Sharing (UNOS) in the United States.
For liver transplants, the primary tool since 2002 has been the Model for End-Stage Liver Disease, or MELD, score. It uses a handful of lab values to estimate how sick a patient is, and organs are offered first to the patient with the highest score, following a “sickest first” principle.11PubMed. Improving liver allocation: MELD and PELD This replaced an older system that relied on broader status categories and was seen as less precise.12PubMed. A Critical Review of MELD as a Reliable Tool for Transplant Prioritization
Kidney allocation works differently. The current system, updated in 2014, introduced a concept called longevity matching: the highest-quality donor organs are preferentially offered to candidates predicted to survive the longest after transplant, rather than simply going to whoever has waited the most years. Wait time still matters, but it now starts from the date dialysis began rather than the date of listing, a change designed to help patients who faced delays getting onto the list in the first place. The system also allows candidates with blood type B to receive organs from certain donors with blood type A, reducing what were historically longer waits for that group.13PubMed Central. Changing organ allocation policy for kidney transplantation in the United States
Mass Casualty and Disaster Triage
When a hospital faces a mass casualty incident, say a large-scale accident, a natural disaster, or a terrorist attack, the normal triage system is replaced with a simpler, faster protocol designed to sort dozens or hundreds of patients in minutes. The most widely used is START (Simple Triage and Rapid Treatment), which categorizes patients into four color-coded groups: red (immediate, life-threatening), yellow (delayed but serious), green (walking wounded), and black (deceased or not survivable).
START sacrifices nuance for speed. Research shows it has limitations in precisely matching patients to the right category, but it is highly sensitive for identifying the most critical patients and predicting who will need surgery, ICU admission, or who will die in the emergency department.14PubMed. Simple triage and rapid treatment protocol for emergency department mass casualty incident victim triage The fundamental ethical shift in disaster triage is that the goal moves from doing the most for each individual patient to doing the most good for the greatest number. A patient who would consume enormous resources with a low chance of survival might be categorized differently than they would on a normal day.
Crisis Standards of Care
The COVID-19 pandemic forced hospitals across the world to formalize what had previously been mostly theoretical: crisis standards of care (CSC), the rules for rationing when resources like ventilators or ICU beds genuinely run out. These frameworks attempt to balance saving the most lives, respecting fairness, and avoiding discrimination.
In practice, most CSC protocols rely on clinical scoring tools, such as the Sequential Organ Failure Assessment (SOFA) score, to rank patients by predicted survival. The idea is straightforward: prioritize patients most likely to benefit from the scarce resource. But the implementation is ethically fraught. A simulation study of ventilator allocation protocols found that while different protocols varied in how many lives or life-years they saved, the real tension was between maximizing total lives saved and ensuring equal survival rates across racial groups.15PubMed Central. Investigating ethical tradeoffs in crisis standards of care through simulation of ventilator allocation protocols
Minnesota’s experience during COVID illustrated this tension. The state’s ethical guidance called for saving the most lives while respecting individual rights and fairness, but no guidance specified which duty should take precedence when the two conflicted.16Critical Care Explorations. Operationalizing Ethical Guidance for Ventilator Allocation in Minnesota: Saving the Most Lives or Exacerbating Health Disparities? A National Academy of Medicine review emphasized that no triage process is value-neutral and that all systems may have effects on underlying disparities.17NAM Perspectives. Crisis Standards of Care and COVID-19: What Did We Learn? How Do We Ensure Equity? What Should We Do?
Racial and Socioeconomic Disparities in Triage
Even well-designed prioritization systems are operated by humans, and research consistently finds that race affects how patients are triaged. A large study found that Black and Hispanic patients presenting to the emergency department were significantly less likely to be triaged to high-acuity beds than white patients. The gap was especially pronounced for complaints that depend on patient self-report, like chest pain and difficulty breathing: Black patients with chest pain had about 24% lower odds of being placed in a high-acuity area compared to white patients with the same complaint.18PubMed Central. Racial Differences in Triage for Emergency Department Patients with Subjective Chief Complaints
A separate study of a rapid-triage system found that even after matching patients by clinical characteristics, Black patients were about 28% more likely to be triaged to a lower-acuity track and about 27% less likely to receive a high-acuity score. Among those who ultimately proved to be high-acuity cases, Black patients were 40% more likely to have been initially sent to a fast-track (lower-priority) area.19PubMed. Investigating racial disparities within an emergency department rapid-triage system
These disparities extend into ICU-level decisions. A study examining the SOFA score, the same tool used in crisis standards of care, found that Black patients had lower mortality than white patients at the same SOFA score. That sounds like a neutral finding, but in practice it means SOFA-based rationing systematically underestimates how much Black patients stand to benefit from treatment. Correcting for this would have upgraded between 2% and 16% of Black patients to higher priority, depending on the severity of the shortage.20JAMA Network Open. Accuracy of the Sequential Organ Failure Assessment Score for In-Hospital Mortality by Race and Relevance to Crisis Standards of Care
Insurance status creates a separate layer of disadvantage. A study of surgical transfers found that uninsured patients were more than twice as likely as commercially insured patients to be designated as emergency medical condition transfers by referring hospitals, even after accounting for clinical factors. The likely mechanism is that referring hospitals classify uninsured patients as emergencies to trigger legal obligations under the federal EMTALA law, which requires hospitals to stabilize emergency patients regardless of ability to pay.21PubMed Central. Insurance Status Influences Emergent Designation in Surgical Transfers EMTALA itself, passed in 1986, was created specifically to prevent hospitals from “dumping” uninsured patients onto public hospitals without stabilizing them first.22PubMed Central. The Emergency Medical Treatment and Active Labor Act (EMTALA): what it is and what it means for physicians
When the System Breaks Down
Emergency department crowding is the single biggest threat to orderly prioritization. When an ED is overcrowded, wait times spike, patients assigned to lower triage levels wait longer than intended, and the system’s ability to detect deterioration in waiting patients degrades. A systematic review found that every retrospective study examining the relationship between ED crowding and inpatient mortality reached the same conclusion: mortality went up as crowding worsened.23PLoS ONE. Emergency department crowding: A systematic review of causes, consequences and solutions One study estimated that crowding contributed to roughly 13 excess deaths per year at a single hospital.24PubMed Central. Overcrowding in Emergency Department: Causes, Consequences, and Solutions—A Narrative Review
A study of non-critical patients found that when emergency department occupancy exceeded about 89%, ten-day mortality rose by roughly 40% compared to less crowded periods.25PubMed Central. Emergency department crowding increases 10-day mortality for non-critical patients: a retrospective observational study The patients most harmed are often those in the middle triage levels: sick enough to need treatment, but not so sick that they jump to the front of the line.
Crowding also drives people to leave before being seen at all. In one Canadian study, every five additional patients in the waiting area raised the odds of someone leaving by about 17%.26PubMed Central. Factors Associated with Patients Leaving Without Being Seen in a Canadian Emergency Department Nationally in the United States, leave-without-being-seen rates are higher at urban hospitals, government-owned hospitals, and safety-net hospitals that serve disproportionately low-income populations.27JAMA Network Open. Trends and Hospital Factors in Emergency Department Patients Leaving Without Being Seen, 2015-2024 A Swiss hospital found that patients who left without being seen had typically already waited a median of about 102 minutes.28PubMed Central. Missed Opportunities: Evolution of Patients Leaving without Being Seen or against Medical Advice during a Six-Year Period in a Swiss Tertiary Hospital Emergency Department
How AI Is Entering the Picture
Hospitals are increasingly experimenting with machine learning tools that predict which patients in the emergency department are about to deteriorate. The idea is to catch patients who look stable at triage but are actually on the verge of cardiac arrest, respiratory failure, or other emergencies. One model, trained on over 237,000 ED visits, used vitals, lab results, and imaging data to predict events like in-hospital cardiac arrest and ICU admission. Its predictions based solely on triage-time information outperformed traditional statistical models, and accuracy improved as more data became available during the visit.29Scientific Reports. A novel deep learning algorithm for real-time prediction of clinical deterioration in the emergency department for a multimodal clinical decision support system
But algorithmic triage carries risks. A widely cited study in Science examined a commercial algorithm used by health systems to decide which patients needed extra care. The algorithm predicted health care costs as a proxy for illness, but because Black patients historically had less money spent on their care due to unequal access, the algorithm systematically rated them as healthier than equally sick white patients. Fixing that single design choice would have nearly tripled the percentage of Black patients flagged for additional help, from about 18% to 47%.30PubMed. Dissecting racial bias in an algorithm used to manage the health of populations The lesson applies broadly: any algorithm trained on historical data risks encoding the biases present in that data, and using cost as a stand-in for need is particularly dangerous in a health system where access is unevenly distributed.
Triaging Children and Psychiatric Patients
Pediatric triage is widely acknowledged to be harder than adult triage. Children have different vital sign ranges at different ages, cannot always describe their symptoms, and their physiological reserves can mask serious illness until they crash suddenly. The main adult triage scales have been adapted for pediatric use, with varying success. The pediatric version of the Canadian system (PaedCTAS), the Manchester system, and ESI version 4 all have evidence supporting their reliability in children, but studies consistently note that triage accuracy is lower for younger age groups.31PubMed Central. Pediatric emergency triage systems Supplementary tools like the Pediatric Early Warning Score and the Pediatric Assessment Triangle are increasingly used alongside these standard scales to catch deterioration that the primary triage score might miss.32PubMed Central. Validation of different pediatric triage systems in the emergency department
Psychiatric emergencies present a different challenge. Standard triage systems were designed around medical and traumatic complaints, and they perform poorly at distinguishing between a patient having a panic attack and one on the verge of a psychotic break. An expert Delphi study developed a dedicated five-level mental health triage tool where the highest priority goes to patients who are extremely agitated, at immediate risk of self-harm or harming others, or require physical restraint. At lower priority levels, factors like psychiatric history and collateral information from family or caregivers become more important.33Emergency Medicine Journal. An Expert Delphi study to derive a tool for mental health triage in emergency departments Structured mental health triage systems, particularly those led by specially trained nurses using validated scales, have shown promise in reducing unnecessary psychiatric admissions and better identifying patients who need immediate intervention.34PubMed. Effectiveness of mental health triage systems in reducing psychiatric emergency department admissions
Virtual and Pre-Hospital Triage
An emerging layer of the prioritization process happens before the patient ever reaches the hospital. Emergency medical services in some regions now conduct secondary triage by phone for lower-acuity calls, assessing whether a patient really needs an ambulance ride to the ED or could be redirected elsewhere. One Australian program embedded a virtual emergency department pathway into EMS secondary triage for elderly residents of aged-care facilities, aiming to reduce unnecessary ED visits by connecting patients with remote physicians who could manage their conditions without a hospital transfer.35PubMed. Embedding a Virtual Emergency Department Pathway Within Emergency Medical Services Secondary Triage for People Living in Residential Aged Care
This kind of upstream triage serves two purposes: it gets patients to the right level of care faster, and it keeps lower-acuity cases out of emergency departments that are already struggling with crowding. As telehealth infrastructure matures, these virtual triage models are likely to become a more routine part of how hospitals manage the flow of patients before they ever arrive at the front door.