Multiple-casualty disaster triage: START and SALT

A bus overturns on a road and the first ambulance arrives with two paramedics and a doctor. There are about thirty people scattered along the shoulder. Some walk and cry, a young woman screams with a large scalp cut that bleeds heavily, an older man sits silently, pale, holding his abdomen. The doctor knows he cannot attend to everyone at once and that the next minutes will define who receives help first. The scene is fictional and was written for teaching. This text uses that picture as a clinical reasoning scenario: why disaster triage changes the question from "what does this patient need" to "how to help the greatest number of people", how the START and SALT systems work, and where judgment usually errs.

NEXO is a daily medical diagnosis game: one clinical case a day, up to six progressive clues, and ICD-10 proximity feedback across five levels.

The text avoids being a protocol. Each service follows the triage system adopted in its region and trains the team to use it. The interest here is the reasoning: how to decide quickly with little information, how to keep the most striking wound from stealing attention, and how to live with decisions that carry moral weight. If you play the case of the day on NEXO, you will recognize the move: with few clues, the decision needs to be good enough and revised when new information arrives.

Key points

• In disaster triage, the goal is no longer the best care for each person but the greatest good for the greatest number of people.
• START and SALT are systems that organize this decision in a few steps and in a few seconds per victim.
• The available studies show that no system has high enough accuracy to forgo retriage.
• Overtriage overloads resources, and undertriage delays those who needed immediate help.
• The expectant category is the hardest decision in triage, and preparing the team for it is part of the training.

Why disaster triage changes the logic

Lerner and colleagues define triage in incidents with multiple victims as the process of prioritizing care when resources are not enough to treat all victims immediately (Lerner and colleagues, 2011). Jenkins and colleagues recall that triage arose as a necessity in wartime and became a civilian tool whose goal is to do the greatest good for the greatest number of people (Jenkins and colleagues, 2008).

In the ordinary emergency department, the doctor thinks about the patient in front of them and does everything within reach for that patient. In a disaster, this logic does not hold. If the doctor at the scene spends twenty minutes with the first victim they find, others may worsen without anyone noticing. The reasoning becomes population-based: who needs a simple action now to survive, who can wait without harm, and who would need so many resources that, at that moment, the help given to them would be missing for several others.

From the individual to the group

This change is uncomfortable. Health professionals are trained for individual care, and switching to deciding for the group requires a conscious effort. Dadoun and colleagues describe that, in a disaster, the goal of care becomes the survival of the population, and that this paradigm shift is difficult for those providing care (Dadoun and colleagues, 2017).

Reasoning done in seconds

Mass triage is not a diagnosis. It does not ask what the disease or injury is, but what the priority is. To do this, it replaces the detailed history and complete examination with a few quick observations. Clinical reasoning remains present but compressed: each victim receives a short assessment, a category and, when applicable, a simple intervention, and the rescuer moves on to the next. The post on clinical decision-making shows how to choose actions considering the risks of acting and not acting.

What changes in the opening scene

In the bus scene, the young woman with the scalp cut is the one who draws the most attention. The pale, silent man holding his abdomen draws less. Triage exists precisely so that the order of care is not defined by who shouts loudest or bleeds most visibly, but by criteria the team applies the same way to everyone.

START: a simple, fast algorithm

START, short for Simple Triage and Rapid Treatment, has been widely used in the United States since the 1980s (Kahn et al., 2009). Jenkins et al. list START alongside other mass triage systems, such as JumpSTART for children, Care Flight, Triage Sieve, Sacco and SAVE (Jenkins et al., 2008).

The idea behind the algorithm

Systems like START use a few questions, in sequence, to sort victims into priority groups. In START, the parameters assessed are the ability to walk, the presence or absence of spontaneous breathing, a respiratory rate above or below 30 per minute, perfusion, by palpable radial pulse or visible capillary refill, and mental status, by the ability to follow commands (REMM, START algorithm). Bazyar et al. list the same indicators: walking, breathing and its rate, capillary refill and the ability to follow commands (Bazyar et al., 2022). The method was created in 1983 by the Newport Beach fire department and Hoag Hospital, in California (REMM, START for adults), and in 1996 Benson et al. proposed using the radial pulse instead of capillary refill (Benson et al., 1996). The advantage is speed and reproducibility: different responders, applying the same steps, tend to reach the same category. The cutoffs used in each service are in its training materials.

Which signs carry the most information

Garner et al. analyzed 1144 trauma patients to test triage algorithms. The motor response of the Glasgow coma scale and systolic pressure were the variables most strongly associated with severe injury (Garner et al., 2001). In the same study, sensitivity for identifying severe injury was 82 in 100 for CareFlight, 85 for START and 84 for a modified version of START, with specificities of 96, 86 and 91 in 100, respectively. The authors themselves note that these results, obtained with ordinary trauma patients, may not be repeated in a real incident with multiple victims.

What START does well

Kahn et al. evaluated START after a 2003 train crash and concluded that the system was useful in ensuring that the most critical patients were transported first: the median time to hospital arrival for red patients was more than an hour shorter than for the others (Kahn et al., 2009). This result shows the practical value of a simple system: it organizes transport when the scene is chaotic.

SALT: sort, assess, lifesaving, treat

Lerner et al. reviewed the evidence on mass triage and found little basis for stating that any existing system had been validated. From the best available science and expert consensus, they proposed SALT, short for Sort, Assess, Lifesaving interventions, Treatment or transport (Lerner et al., 2008). SALT incorporates aspects of the systems that already existed.

The steps of SALT

The first step is a global sort, done by looking at the whole group, not one victim at a time. The second is the individual assessment, in which each victim is examined quickly. Between the two come the simple lifesaving interventions that can be done in seconds. Finally, each victim receives a treatment or transport category.

A common standard

Lerner and colleagues report that, in the United States, there was no agreement on categories or on the colors used in triage (Lerner and colleagues, 2008). Later, the same group described uniform core criteria, organized in four parts: general considerations, global sorting, lifesaving interventions and individual assessment (Lerner and colleagues, 2011). The authors point out that variation in labels and names between services can cause confusion when teams from different places work together.

Why standardization matters

In a major disaster, teams from several cities meet at the same scene. If one uses a color for "immediate" and another uses the same color for another category, the label stops communicating. Standardizing does not make the system more precise, but it reduces one type of error that has nothing to do with the patient: the error of communication between teams.

Categories and colors

The table below summarizes, in a teaching format, the categories most used in mass triage. Names and colors vary between systems and countries, and each service follows the standard it adopts.

Category	Most used color	Core idea	Reasoning question
Immediate	Red	Needs rapid action to survive and has a good chance with it	What to do now, in seconds?
Delayed	Yellow	Serious injury, but can wait a while without likely worsening	How long can they hold on?
Minor	Green	Mild injuries, usually able to walk	Who will reassess them later?
Expectant	Gray or another color, depending on the system	Low chance of survival and great need for resources	Has the situation changed?
Dead	Black	Death confirmed at the scene	Has the assessment been confirmed?

In the United States national triage criteria, each category has a color: immediate red, delayed yellow, minor green, expectant gray and dead black (FICEMS, 2011). In START, the expectant receives the black label, and the guidance is to offer palliative care and pain relief (REMM, START for adults).

The category as a hypothesis

It is worth thinking of each category as a working hypothesis, not as a definitive label. A green victim can become yellow, and a yellow one can become red. Labels with detachable parts exist precisely to allow the category to change. The post on differential diagnosis with clues shows how to keep hypotheses open and revise them with each new clue.

Retriage

Triage does not happen only once. It is repeated at the scene, at the treatment post, during transport and on arrival at the hospital. Each new assessment is a chance to correct the previous one. For clinical reasoning, retriage is the equivalent of reassessment in the individual patient: it recognizes that the first impression was made with little information.

Overtriage and undertriage

Triage errors go in two directions. Overtriage places a victim in a more severe category than they have. Undertriage does the opposite and places in the milder category someone who needed immediate help. The two have different consequences.

What happened in the train accident

In the study by Kahn and colleagues, with 148 records reviewed in 14 hospitals, field triage classified 22 victims as red, 68 as yellow and 58 as green. The later classification, based on outcome, found 2 red, 26 yellow and 120 green (Kahn and colleagues, 2009). Seventy-nine victims were overtriaged, 3 were undertriaged and 66 stayed in the correct category. The authors concluded that START kept undertriage at an acceptable level, but with a substantial amount of overtriage.

The weight of each error

Undertriage is the most serious error for the individual: someone who needed quick help waits. Overtriage seems safer, but it weighs on the group. Each victim wrongly placed as red occupies an ambulance, a team or a bed that could belong to someone else. In a disaster, resources spent on those who did not need them are missing for those who did.

The reasoning behind the tendency

The tendency to over-triage is understandable. Faced with uncertainty, labeling upward seems to protect. The post on diagnostic uncertainty discusses how to decide when erring in either direction has consequences, and triage is a clear example of that tension. Knowing which error the system tends to make helps the team compensate with re-triage.

What the evidence says about accuracy

Jenkins and colleagues concluded that the evidence on mass triage is limited and that the best-known algorithms were not built on evidence. They also noted the absence of studies in chemical, biological or radiological scenarios (Jenkins and colleagues, 2008).

START and SALT side by side

Bhalla and colleagues applied both systems to 100 medical records. With SALT, 65 victims were in the correct category, 5 were over-triaged and 30 were under-triaged. With START, 55 were correct, 12 were over-triaged and 33 were under-triaged. The authors concluded that neither system was sensitive or specific enough (Bhalla and colleagues, 2015).

A large study with trauma records

Cross and Cicero compared triage methods in 530695 records of pediatric, adult and elderly patients. The Sacco method performed best, and the Glasgow coma scale alone performed similarly to the dedicated methods (Cross and Cicero, 2013). CareFlight placed patients who could have been saved in the dead (black) category: 41 in every 100 of these patients survived, against about 10 in every 100 with the other methods.

The systematic review

Bazyar and colleagues gathered 13 studies on the accuracy of disaster triage systems. START, modified START, SALT and other systems fell below 90 in every 100 in accuracy, sensitivity and specificity, and only one system, Smart, exceeded that threshold (Bazyar and colleagues, 2022).

How to read these numbers

These studies use different methods, different outcomes and different populations, and so they do not allow electing a winning system. The common message is another one: no system replaces clinical judgment and re-triage. The post on pretest probability helps to understand why sensitivity and specificity depend on the context in which the test is applied.

The expectant category and the moral weight

The expectant category is the hardest decision in triage. Dadoun and colleagues describe that it applies to the victim with a low chance of survival and a great need for resources, and that invoking this category requires a paradigm shift that is hard for professionals (Dadoun and colleagues, 2017).

Expectant is not abandonment

Classifying someone as expectant does not mean ceasing to care. It means that, at that moment, with those resources, the priority for complex interventions goes to those with the best chance of benefiting from them. The expectant victim still deserves comfort, pain relief and reassessment, and can change category when more resources arrive or when the number of victims decreases.

A decision that depends on the context

The same victim can be expectant in a disaster with hundreds of injured people and immediate in an accident with few people and many resources. The category does not describe only the patient, but the relationship between the patient and what is available. That is why the decision must be reviewed when the context changes.

Preparing the team

In the work of Dadoun and colleagues, a teaching session on ethics improved the participants' confidence to use the expectant category (Dadoun and colleagues, 2017). Discussing these decisions before the disaster, in training, helps the team make them with more clarity when they really happen. The World Health Organization keeps material on the organization of mass casualty response systems, which deals with planning at the service level.

The bias of the wound that impresses

Croskerry described how cognitive errors lead to diagnostic errors and proposed strategies to reduce them (Croskerry, 2003). Graber and colleagues showed that cognitive factors, such as premature closure, are present in a good part of diagnostic errors in internal medicine (Graber and colleagues, 2005). In mass triage, one bias stands out: judging severity by the appearance of the wound.

Visible blood, hidden risk

In the opening scene, the young woman with the scalp cut bleeds heavily and screams. The wound is striking. The silent man, pale and holding his abdomen, has nothing visible. A structured triage system forces the team to look at the same signs in every victim, and not only at what draws attention. The post about heuristics in clinical reasoning shows how striking findings distort judgment.

Silence as a clue

A victim who does not speak, does not complain and does not move may be more seriously ill than one who calls for help out loud. In triage, the absence of a complaint is not a reassuring sign by itself. The post about signs, symptoms and history discusses how to interpret what the patient brings, including when they bring little.

Pause between victims

A brief pause before moving to the next victim helps check whether the category makes sense: was the victim assessed by the same criteria as the others? Was anything beyond the visible wound checked? The post about the diagnostic pause shows how a short moment of checking improves the decision.

Questions that protect

Some questions help keep the focus during triage:

1. Am I applying the same criteria to all victims?
2. Is the most visible wound in fact the most serious?
3. Who is silent and has not yet been assessed?
4. Who will do the retriage and when?
5. Does the expectant category still make sense with the resources available now?

How NEXO helps train this reasoning

In NEXO, the case of the day is fictional, with a real clinical presentation and up to six progressive clues. You propose a diagnosis by searching for the code in ICD-10 and receive proximity feedback at five levels, which shows whether the guess landed on the exact code, the same category, the same group, the same chapter or a different chapter. This format connects with the theme of this text: triage also decides with few clues, and the feedback shows how close the decision came to the target.

To use the case of the day as training, try three habits. Before each guess, ask which clue is the most striking and whether it is in fact the most important. When a new clue arrives, treat the previous guess as a category that may change. After the result, note whether you erred on the high side or the low side.

Three exercises with cases

Separate the striking from the important. In each case, list the clue that drew your attention most and the one that changed the probability most.

Practice retriage. With each new clue, revise your guess as if you were retriaging a victim.

Think about the group. Imagine the same case appears in ten patients at the same time and decide whom you would see first.

The post about prognostic reasoning helps you think about who tends to worsen and how soon.

The limits on using the game are described on the NEXO economy page. If you research the teaching of clinical reasoning, the page for researchers explains how to get in touch. Professionals who write or review cases find information in curation and in creators, and institutions can learn about the proposal on the partnerships page.

Frequently asked questions

What is disaster triage?

It is the process of prioritizing care when resources are not enough to treat all victims at the same time. The goal becomes doing the greatest good for the greatest number of people.

What is the difference between START and SALT?

START is a simple algorithm, widely used since the 1980s. SALT was proposed later, based on a review of the evidence and expert consensus, and begins with a global ordering of the group before individual assessment.

Which triage system is the most accurate?

The available studies show variable performance and, in general, below the ideal for all the most used systems. That is why retriage and clinical judgment remain essential, whatever system is adopted.

What are overtriage and undertriage?

Overtriage is placing a victim in a more serious category than they have. Undertriage is placing them in a lighter category than they needed. The first consumes resources, and the second delays care for those who needed it.

What does the expectant category mean?

It is the category for the victim with a low chance of survival and a great need for resources at that moment. They continue to receive comfort and reassessment, and may change category when the context changes.

Why does triage need to be repeated?

Because the condition of the victims changes and because the first assessment is made with little information. Repeating triage allows errors to be corrected and those who worsen to be tracked.

Final checklist

• I remembered that, in a disaster, the goal is the greatest good for the greatest number of people.
• I applied the same criteria to all victims.
• I looked for those who were silent, not only those who drew attention.
• I considered the risk of over-triage and under-triage.
• I treated each category as a guess that can change.
• I planned re-triage and reviewed the expectant category when the context changed.

Want to practice reasoning with progressive clues? Play the case of the day.

References

1. Bazyar J, Farrokhi M, Salari A, et al. Accuracy of triage systems in disasters and mass casualty incidents; a systematic review. Arch Acad Emerg Med. 2022. PMID 35573710.
2. Benson M, Koenig KL, Schultz CH. Disaster triage: START, then SAVE. A new method of dynamic triage for victims of a catastrophic earthquake. Prehosp Disaster Med. 1996. PMID 10159733.
3. Bhalla MC, Frey J, Rider C, et al. Simple Triage Algorithm and Rapid Treatment and Sort, Assess, Lifesaving, Interventions, Treatment, and Transportation mass casualty triage methods for sensitivity, specificity, and predictive values. Am J Emerg Med. 2015. PMID 26349777.
4. Croskerry P. The importance of cognitive errors in diagnosis and strategies to minimize them. Acad Med. 2003. PMID 12915363.
5. Cross KP, Cicero MX. Head-to-head comparison of disaster triage methods in pediatric, adult, and geriatric patients. Ann Emerg Med. 2013. PMID 23465555.
6. Dadoun A, Khalil E, Bank I. Invoking the "expectant" triage category: can we make the paradigm shift? Am J Disaster Med. 2017. PMID 29270959.
7. Federal Interagency Committee on Emergency Medical Services. National implementation of the Model Uniform Core Criteria for mass casualty incident triage. 2011. FICEMS document.
8. Garner A, Lee A, Harrison K, et al. Comparative analysis of multiple-casualty incident triage algorithms. Ann Emerg Med. 2001. PMID 11679866.
9. Graber ML, Franklin N, Gordon R. Diagnostic error in internal medicine. Arch Intern Med. 2005. PMID 16009864.
10. Jenkins JL, McCarthy ML, Sauer LM, et al. Mass-casualty triage: time for an evidence-based approach. Prehosp Disaster Med. 2008. PMID 18491654.
11. Kahn CA, Schultz CH, Miller KT, et al. Does START triage work? An outcomes assessment after a disaster. Ann Emerg Med. 2009. PMID 19195739.
12. Lerner EB, Cone DC, Weinstein ES, et al. Mass casualty triage: an evaluation of the science and refinement of a national guideline. Disaster Med Public Health Prep. 2011. PMID 21685309.
13. Lerner EB, Schwartz RB, Coule PL, et al. Mass casualty triage: an evaluation of the data and development of a proposed national guideline. Disaster Med Public Health Prep. 2008. PMID 18769263.
14. Radiation Emergency Medical Management (REMM), U.S. Department of Health and Human Services. START adult triage algorithm. REMM page and text version.
15. World Health Organization. Mass casualty management systems: strategies and guidelines for building health sector capacity. 2007. Publication page.