The train was cut open like a can of tuna. We didn’t know who to treat first. There was a lot of blood, a lot of blood. —Enrique Sanchez, Madrid Emergency Medical Services, of simultaneous commuter train bombings, March 2004
Triage is the utilitarian sorting of patients into categories of priority to allocate limited resources rationally; it is, proverbially, to do “the greatest good for the greatest number.” Triage systems are used to determine the order in which patients will receive treatment and transport based on their condition, prognosis, and the availability of resources. Triage typically fails in one of two basic ways: undertriage and overtriage. Undertriage represents a failure to identify the casualties who could benefit from scarce medical resources, such as the serious casualty whose life would be saved with rapid evacuation and prompt emergency surgery. In terms of test characteristics, undertriage means poor sensitivity to those who would benefit from the medical resources available, whereas overtriage occurs when casualties who, relatively speaking, do not benefit from a scarce resource, nonetheless receive that resource. Overtriage signifies a sorting system with low specificity and overloads of scarce resources. Noncritical patients who receive immediate care even though they could safely wait (at the expense of more serious, savable casualties) constitute one form of overtriage. It can also occur when expectant patients, with little chance of survival, are provided with precious medical resources.
Triage following disaster is not a single processing step; rather, triage underlies all aspects of the response, including on-site rescue, evacuation, receiving hospital activities, decontamination, and so on. Because the resources available, the clinical conditions of casualties, and the information available all change throughout the time-course of the response, response priorities do as well; triage is a dynamic process.
It has been suggested that disaster responses should match normative practice as much as practically possible, , and that, of course, would apply to triage methods. However, normative practice could pose a liability for massive disasters, in which a set of reflexes could interfere with the truly rational allocation of resources. This trade-off is discussed throughout the chapter. As noted in the following section, many recent disaster responses have been characterized by a general lack of meaningful triage.
Historical perspective
The roots of disaster triage stretch back at least to eighteenth-century military casualty care. Baron Larré, a surgeon with Napoleon’s army, established a system under which wounded soldiers received initial treatment in the field before being transported to hospitals. In 1846, British naval physician John Wilson first proposed that treatment be deferred for casualties with either minor or likely fatal injuries, so that treatment could be provided to the severely injured, who were most likely to benefit. Military triage evolved sporadically, but, by World War II, a hierarchical structure for combat casualty care had been developed. During World War II, the average time from injury to definitive care was 12 to 18 hours. By the Vietnam War, improved triage and air-ambulance capabilities reduced this time to less than 2 hours. The early military history of triage was well reviewed by Kennedy.
In the 1980s, civilian prehospital systems became interested in trauma triage for the individual trauma casualty. West et al. showed that serious trauma casualties had superior outcomes when cared for at specialized trauma centers. There were investigations of prehospital criteria for differentiating between individual casualties that should be taken to major trauma centers and those that could receive care safely at community medical centers (to avoid overburdening trauma centers with nonserious casualties). Trauma registries enabled the development and validation of various field triage decision rules, although such criteria have proven problematic (see the Triage Scoring System section). Such civilian trauma scoring in turn influenced modern military triage. In the 1991 Persian Gulf War, Burkle et al. explored the use of Champion’s Revised Trauma Score for triage of combat casualties. Triage classification schemes for military and civilian mass casualties, to a large extent, have converged. Both military (e.g., NATO) and civilian (e.g., color coding) triage systems make use of comparable levels of acuity, and the Trauma Sieve and START, which are similar triage decision systems, have been used by both civilian emergency medical service(s) (EMS) as well as British Army soldiers.
Regarding civilian triage, the past several decades have seen a litany of tragic events in industrialized countries, including bombings, fires, shootings, and plane crashes. These events show a consistency of scale, with immediately surviving casualties numbering in the dozens or hundreds. Unfortunately, many retrospective reports continued to note unsatisfactory execution of triage, particularly prehospital triage, for these events. In the mid-1980s, Vayer et al. cited Butman’s analysis of 51 mass-casualty incidents (MCI) that identified a universal failure to execute proper triage. Inadequate prehospital triage continues to be reported following MCIs: an aircraft crash in Singapore in 2000, the Tokyo sarin attacks, and the Gothenburg Fire Disaster. Typically, documentation of prehospital triage during an MCI is quite poor; therefore, retrospective analysis of field triage is simply not feasible, as was the case for the Oklahoma City Bombing.
The urban community in a developed country may very well possess the resources necessary to treat dozens or even hundreds of casualties, provided the resources are mobilized and the patients in need of immediate care are identified in a timely fashion. Yet even in these settings, suboptimal use of available community resources is the rule rather than the exception; for instance, most casualties self-transport to the closest hospital and leave distant facilities underused. Community resources, such as urgent care centers and outpatient clinics, capable of treating the majority of casualties with relatively minor injuries, are almost always underused. Almogy et al., reporting on the Israeli response to the Jerusalem Sbarro Pizzeria Bombing in 2001, observed: “in these circumstances (e.g., a MCI such as a suicide bombing), ordinary hospital resources are heavily burdened, yet delivery of efficient medical treatment is possible by recruitment of all available personnel and resources.”
In contrast to the many reports of poor triage after disasters, exemplary responses to MCIs have stemmed from exemplary mobilization of resources. After the bombing at the Atlanta Centennial Olympics, 30 EMS units evacuated all 111 casualties to area hospitals within 32 minutes. The vast majority of serious casualties were taken to Grady Memorial Hospital, where, “at one point, there were more physicians than victims in the emergency care center.” Those casualties all had good or excellent outcomes. After the Jerusalem Sbarro Pizzeria Bombing, the prehospital response was largely “scoop and run.” The Ein Kerem Campus, receiving 132 surviving casualties, performed two emergency department (ED) thoracotomies, with no apparent shortage of resources available for those less-critical casualties. After the 2013 Boston Marathon Bombings, all 30 “red-tagged” patients, the most critically ill, were transported to area hospitals within minutes, and all survivied. Almogy’s dictum pertains to most recent disasters in developed countries, which have been limited to dozens or hundreds of casualties. Popular field triage systems (see the Triage Scoring Systems section) are most appropriate for this scale of events.
Such triage systems, however, are less applicable to disasters of enormous scale. The greater the scope of the disaster, both in terms of geographic area and number of casualties, the more challenging triage becomes. Including the 1918 Influenza Pandemic and Hurricane Katrina, in 2005, there have been only nine nonmilitary disasters in the entire history of the United States that resulted in more than 1000 deaths. Internationally, massive disasters have produced tens or hundreds of thousands of dead and injured, such as earthquakes in Tangshan (1976), Armenia (1988), Hanshin-Awaji (1995), and Iran (2003), as well as the Indian Ocean Tsunami (2004) and the Tohoku Earthquake and Tsunami in Japan (2011). Field triage systems tend to be sensitive and not specific, leading to overtriage, and they may be too unwieldy to triage massive numbers of casualties spread over a wide area. Additionally, they are tailored to traumatic pathology, but broad medical pathology (e.g., infectious disease and metabolic disarray) occurs in the aftermath of these massive disasters, and baseline medical emergencies (e.g., myocardial infarctions and ectopic pregnancies) continue to occur unabated.
To the extent that one seeks historical guidance when planning for the future, the emphasis that should be given to preparing for enormous-scale disasters in the United States is unknowable. Such events are rare, and we have little experience to guide us. Traditionally it has been difficult to find the resources-funding, time, and materials to develop and maintain preparedness. Some respected authorities have suggested that too much attention has been paid to the unprecedented scenario of weapons of mass destruction (WMD). Frykberg wrote: “We must resist our current tendency to become overly enamored with the ‘weapons of mass destruction’ of biologic, chemical, and radiologic attacks, in terms of funding priorities and resource allocations that are wholly disproportionate to the clear reality of the terrorist bombing threat.” Even though history has shown that smaller, more mundane MCIs are more likely, there remains concern for a nuclear attack on U.S. soil. One WMDs event could dwarf the sum of preceding MCIs in morbidity and mortality, and it seems prudent to prepare for this possibility and the unique challenges it would create for disaster triage. Natural disasters of enormous scale, even though uncommon in the United States, are also a threat. The New Madrid seismic zone, located in the Southern and Midwestern United States, produced several earthquakes of magnitude 7.0-8.0 in 1811 and 1812. Damage was not significant at the time because there were few settlements in the region, but a 2009 report published by the Mid-America Earthquake Center predicted that another earthquake of similar magnitude would result in approximately 86,000 casualties across eight states. The U.S. Geological Survey estimates a 7% to 10% probability of an earthquake of this magnitude within the next 50 years. Even though the extent to which preparedness can and should be maintained for massive disasters is unclear, it would be unwise to disregard the real possibility of such an event.
Current practice
This section offers a technical framework for triage following a disaster. The issues discussed here are important to consider for disaster planning and training (and real-time execution), but the reader must always bear in mind the sobering limitations of triage systems detailed in the previous section. All the same, triage should be carried out to the extent possible at the scene and at every facility and site providing disaster care. Box 54-1 lists considerations that should be made in advance of any disaster. Box 54-2 lists details that need to be determined in real time after a rapid survey of the disaster and the scope of casualties. Triage training is specifically discussed at the end of the chapter.
- •
What triage classification will be used, if any (e.g., four or five levels of severity)?
- •
Will a formal triage scoring system be used?
- •
What on-site/hospital documentation will be used?
- •
Who will be the triage officer(s)?
- •
Who will collect vital signs for the triage officer(s)?
- •
Physically, where will casualties from each triage category be treated (and who will staff each area)?
- •
What overtriage and undertriage rates are acceptable?
- •
What level of casualty gets “black-tagged”?
- •
Assuming 10 patients in 20 minutes per triage officer, are there enough officers?
- •
Are there enough other personnel to keep up with the triage team (e.g., litter bearers and care providers); if not, is the triage team too big?
- •
Are resources being used appropriately, including (1) on-site medical interventions, (2) evacuation resources, and (3) hospital-based resources (e.g., ED care and OR)?
- •
After initial triage, who reevaluates the casualties, and how often?
- •
Have all details of the triage plan been reevaluated on an ongoing basis?
- •
Has an updated triage plan been communicated to all active rescuers, including victims and participating bystanders, as is appropriate?
What Triage Classification Will Be Used?
There are several issues to consider for planning a disaster response. First, will a multiple-level triage classification system be used? As noted in the previous section, there are precedents for responses to MCIs with over 100 casualties that consist of prehospital “scoop and run.” Then, rather than a rigid triage scheme outside of normative practice, casualties were, in essence, sorted into one of two categories, using clinical judgment: OR or no OR , These events were notable for a lack of clinically significant bottlenecks in field evacuation and hospital capacity, thus reducing the importance of triage.
In planning for a disaster, it hardly seems prudent to assume such ample evacuative and hospital-based resources would be available, but it does speak to the importance of mobilizing maximum resources. This is the impetus for a formalized formal triage scheme, such as a four- or five-level classification system. In the common four-level system, such as the well-known civilian START system, the categories are: (1) those who will get immediate priority (color-coded red), (2) those casualties who must wait (color-coded yellow), (3) those with the least severe injuries (often referred to as “walking well,” color-coded green), and (4) those casualties whose prognosis is so poor that there is no justification for spending limited resources on them (color-coded black). Military triage hierarchies have a similar four- or five-level structure, although the nomenclature is different. For instance, color-code green is equivalent to NATO Level T3. People responding to international disasters should be aware of all of the triage classifications used by the host country and by the other responding agencies. A slightly more complicated five-level system makes a distinction between patients who will not survive (color-coded black) and those too gravely injured to receive limited resources (sometimes color-coded blue). Nevertheless, if enough resources become available, blue-coded casualties can then receive care. It has been argued that the option for an intermediate level such as blue may actually produce superior triage decision making. Given a stark choice between red or black for gravely injured patients who are not yet dead, it might be emotionally difficult for responders to apply a black tag, even though resources would be wasted on the casualty with a very poor prognosis.
When a tiered triage system is to be used, the most complicated issue is selecting the formal criteria to use for assigning each acuity level. This is discussed in detail in the sections “Will a Formal Triage Scoring System Be Used?” and “Are Normative Overtriage and Undertriage Rates Acceptable?” Most triage plans assume, and rightly so, that a casualty who can ambulate (i.e., the “walking wounded”) is truly low risk. In a review of nearly 30,000 routine civilian trauma casualties, Meredith found that the ability to follow commands (i.e., Glasgow Coma Scale [GCS] motor = 6) upon arrival to an ED is an outstanding positive predictor of patient survival. Still, these casualties do require medical attention at some point; head or extremity hemorrhage, open fractures, or penetrating abdominal trauma are injuries that might be present in ambulating patients, which are manageable conditions that become dangerous if not treated within an appropriate timeframe. In Meredith’s study, the patients with GCS motor of 6 and outstanding prognoses received normative medical care, not indefinite neglect.
Will a Formal Triage Scoring System Be Used?
A formal scoring system can help categorize casualties by objective criteria. Reliance on EMS subjective assessments of severity has been studied as an independent predictor of acuity in routine trauma patients (e.g., not disaster casualties). EMS judgment has been found to be better, equal, , and worse , than formal triage scoring. There is evidence that EMS judgment complements objective triage scoring. , , Prehospital triage systems are tailored to traumatic pathology, not squeezing chest pain or focal pain at McBurney’s point; EMS providers can easily recognize that patients with these signs and symptoms require urgent care. The GCS, published in 1974, offered a historic means of stratifying prognosis after head injury. Triage scores for individual trauma patients (see the Historical Perspective section) arose from the need to decide if individual trauma casualties should be transported to specialized trauma centers. The best known are Champion’s Trauma Score and Revised Trauma Score (RTS). The original Trauma Score used capillary refill and respiratory expansion, which were felt to be too unreliable. The RTS uses GCS, systolic blood pressure, and respiratory rate, and it yields a score between 0 and 12. , An RTS of 12 predicts mortality of less than 1% (when given routine clinical care). Mortality of roughly 50% is predicted by an RTS of 5. In general, the use of physiologic criteria tends to be specific but not as sensitive for predicting critical injury. Other notable triage scores include the contract research and manufacturing services (CRAMS) and Triage Index, which have also shown suboptimal test characteristics. The advantage of a scoring system is that no particular physiologic state preordains a specific triage category; rather, scores for “black” and “red” and “yellow” can be established in real time, based on the perceived balance of casualties and medical resources. Triage scoring offers a flexible, sophisticated approach to triage. However, in the chaos and uncertainty of a disaster—a high-acuity, low-probability event—“sophisticated” can become “complicated,” and “flexible” can become “uncertain.” If formal triage scoring is used, emergency responders must be extremely facile with its use. For triage of pediatric casualties, neurologic, cardiovascular, and respiratory triage criteria should be different from those for adults because of the differences in physiology and recuperative capabilities. The differences motivated the development of specialized pediatric triage criteria, including the pediatric trauma score , and pediatric triage tape. The former instrument did not prove superior to adult triage scoring systems, and the latter has not been extensively validated.
A far simpler alternative to triage scoring is a triage algorithm, such as the START system. This system, for example, assigns “red” based on a rigid set of criteria (if airway is compromised; if minute respiratory rate is over 30; if capillary refill is over 2 seconds; or if simple commands cannot be followed). The Triage Sieve is another similar example of an inflexible assignation algorithm; it includes an upper and lower limit for “red” respiratory rate, as well as a fixed upper limit for heart rate (unlike START, it does not use capillary refill). The advantage of such rigid systems is the simplicity to teach and learn, and the appropriateness for typical MCIs with dozens or hundreds of casualties. The disadvantage, the inflexibility, is discussed further in the Acceptable Overtriage and Undertriage Rate section. More complex triage systems have also been developed, such as the Sacco Triage Method (STM). The system uses mathematical modeling to predict patient survival and to prioritize triage and transport to maximize the number of survivors based on available resources. In a retrospective review of trauma registry patients, STM was shown to predict mortality more accurately compared with other triage systems, but whether this system identifies patients that would most benefit from expedited care is unclear. It should also be noted that STM requires proprietary software and assumes an operational incident command system.
In 2011, the Centers for Disease Control and Prevention (CDC) sponsored a project to review existing triage systems and develop a national standard for mass-casualty triage. The group concluded that there was insufficient evidence to compare systems rigorously or to identify a superior system. Combining features of known triage systems, they developed a new, algorithmic system, called SALT triage (Sort, Assess, Lifesaving Interventions, Treatment and/or Transport). The algorithm uses the ability to ambulate or follow commands to globally sort patients and prioritize assessment, and it emphasizes the rapid performance of simple, potentially lifesaving interventions such hemorrhage control and chest decompression. Ultimately, patients are assigned 1 of 5 categories and transported or treated accordingly: immediate, expectant, delayed, minimal, or dead. While SALT was designed in consideration of the best available evidence, it, like other triage systems, is largely consensus-based, and further research is required to determine if it is effective.
What On-Site/Hospital Documentation Will Be Used?
The triage tag, a minimal document that can be attached to each casualty, might be the only practical method of communicating findings, interventions, and so on, as countless casualties are passed through a chain of emergency care. However, it has been argued that triage tags are impractical to use, and geographic triage (see later) can obviate the need for tags. In a disaster, hundreds or thousands of tags for each triage category must be immediately available to the responders, who need to be exceptionally familiar with the tags to use them properly under trying circumstances, and frenzied casualties may not take proper care of the tags. After an enormous disaster (thousands of casualties), tags might be especially challenging to use properly, although they could also be especially useful.
Consideration of the use of triage tags requires some research on the part of the customer, since there are over 120 triage label systems in use internationally. Hogan and Burstein suggested the following criteria for the optimal triage tag: (1) It must attach securely to each casualty’s body, (2) it must be easy to write on, (3) it must be weather-proof, and (4) it should permit the documentation of the patient’s name, gender, injuries, interventions, care-provider IDs, casualty triage score, and an easily visible overall triage category. It must also permit changes to be made, because triage is always dynamic. One unfortunate potential limitation for such a tag is the presence of contamination that may limit the ability of the triage tag to persist through hospital-based decontamination efforts if the patient is not decontaminated prior to transport.
Who Will Be the Triage Officer(s)?
There must be a major first-pass decision to establish which casualties can wait, which are top priority, which are expectant, etc. This can occur in the field and/or receiving hospital, but the designated triage officer needs a deep understanding of emergency medical treatments, what outcomes are likely for various casualties, and what resources are necessary for treatments. It may be advantageous to have a physician in this role, but in all cases the person performing this role should be among the most-experienced clinicians available because of the vital importance of triage with respect to resource utilization. In Israel, the role is taken by the surgeon-in-charge, because the emergency care bottleneck following a suicide bombing MCI typically occurs in determining operative priority after rapid evacuation to a receiving hospital. Among multiple important traits, a triage officer needs the experience, disposition, and judgment to act as an able leader under incredible pressure. ,
Who Will Collect Vital Signs for the Triage Officer(s)?
The measurement of vital signs should be delegated to assistants to the triage officer. Unfortunately, even in routine clinical conditions, vital sign measurements are confounded by human error and poor technique. Retraining improves the accuracy of vital signs; this is important, because during the chaos and stress of an MCI, an accurate set of vital signs determines the fate of an individual (e.g., placement in a triage category).
Physically, Where Will Casualties from Each Triage Category Be Cared for (and Who Will Staff Each Area)?
Geographic triage means assigning casualties to different physical locations based on severity. Initially, casualties can be brought to a collection point. From there, the triage officer can determine the appropriate triage category (e.g., red, yellow, black, and possibly blue), and then the casualty can be moved to a triage category-specific collection point for further on-site treatment and/or transportation. The walking wounded (green) are readily separated from more seriously injured casualties through good crowd communication and control; most of these casualties could be taken care of in an urgent care center, clinic, or private physician offices, and having some mechanism for directing these casualties away from larger medical centers may be valuable. Briggs suggests the following are desirable characteristics for on-site triage and treatment locations: (1) proximity to the disaster site, (2) safety from hazards, (3) location upwind when contamination is an issue, (4) protection from climactic conditions, (5) easy visibility, and (6) convenient access for air and land evacuation. Vayer cited a number of MCIs, including the Hyatt Regency Skywalk Collapse and the IBM shooting incident, in which a collection point for triage was useful. However, in other situations, no ideal location may be available and an on-site collection point may be of mixed utility. Quarantelli, in a classic study of EMS in 29 U.S. disasters, noted that most casualties are not transported by a properly staffed ambulance, but by private cars, buses, taxis, and even on foot. Field triage and first aid stations are often bypassed, either because their location or existence is unknown or because they are considered an inferior level of care compared with hospitals. Following the Centennial Olympics Bombing, ample EMS resources allowed rapid, complete evacuation of casualties, so on-site geographic triage was unnecessary. In small or large disasters, it may be preferable for the triage officers and litter bearers to circulate and directly move casualties from the scene to appropriate treatment locations, eliminating any collection point. In extremely large disasters, several collection areas might be useful.
It is important to consider the manner and degree to which geographic separation will be enforced. The bigger the disaster, the more crowd control is an issue. For receiving hospitals, crowd control must be a priority; an unstructured influx of casualties should be anticipated, with most casualties ending up at the closest hospitals. Following the Tokyo Subway sarin attack, the St. Luke’s ED was deluged with 500 patients in the first hour alone. At the scene, crowd control is a more complicated issue. Einav described how Jerusalem EMS evacuates casualties so rapidly that only minimal medical interventions are performed, and no crowd control at all is attempted. Crowd control requires emergency resources and may eliminate the assistance by untrained local citizens, who often constitute the majority of the response resources after disasters. Regarding crowd control on scene, Vayer suggested that “if patients are medically and emotionally stable and choose to leave, they should be allowed to do so.”
What Overtriage and Undertriage Rates Are Acceptable, and What Level of Casualty Gets “Black-Tagged”?
Investigations of trauma triage rules (the prehospital decision of whether or not a patient’s severity warrants transport to a specialized trauma center) suggest that overtriage of 50% to 60% is necessary to achieve undertriage rates around 10%. , , , , , It has also been noted that the four-tiered START triage algorithm leads to substantial overtriage of disaster casualties. , This might be acceptable in a response to an MCI with a few hundred surviving casualties. Consider Table 54-1 : after a hypothetical terrorist bombing with 100 to 200 surviving casualties, conventional triage practice might overtriage 50% of the intermediate “true yellow,” increasing the number of total red. Would it be worth attempting a dramatic departure from normative triage to attempt to reduce the 10 or 20 “false reds”? Routine rates for trauma overtriage and undertriage may be tenable for incidents on this scale. It is worth examining other scenarios that suggest how overtriage and undertriage goals must be adapted to the scale of the disaster. An event with 1000 to 2000 surviving casualties would yield 100 to 200 “false reds” in addition to the 200 to 400 “true-reds” if triage criteria were not appropriately adjusted ( Table 54-1 ). Such numbers of red critical casualties would likely overwhelm even a large urban area’s facilities; therefore the rate of overtriage should be lowered even if that means increasing rates of undertriage (this is the classic sensitivity/specificity trade-off embodied by receiver-operator curves). There are many options of how to define “black” casualties, who are considered so sick (e.g., pulseless and apneic) that resources are wasted on them: an Israeli EMS protocol for MCIs is more specific, considering as dead those with amputated body parts who are not showing signs of movement, as well as those who are pulseless with dilated pupils. In published reports, there is mixed evidence that the overtriage of excessively sick casualties (e.g., the failure to apply a black tag to unsavable survivors) has been an issue. One retrospective analysis of seven separate terrorist bombings noted a correlation between high mortality rates for hospitalized casualties with critical injuries and the fraction of casualties hospitalized without critical injuries. One explanation would be that overtriage of noncritical casualties cost the truly critical casualties the medical attention they required. However, the paper did not describe other characteristics of the casualties, such as the age of victims, the distribution of injury severity, and the incidence of head injury. These factors might suggest alternative explanations. Historically, resuscitation of moribund casualties following MCIs offers dismal outcomes, , , and such heroic measures are generally discouraged. Failure to triage the unsavable as “black” (a form of overtriage) will produce “false reds,” who will pointlessly be given precious medical resources ( Table 54-1 ). Indeed, in designing a response to an earthquake with many thousands of casualties, it has been suggested that casualties with a likelihood of survival less than 50% may need to be “black-tagged.” If so, an RTS of 5 or less would be an appropriate criterion.
Related posts:
Role of Emergency Medical Services in Disaster Management and Preparedness
Psychological Impact of Disaster on Displaced Populations and Refugees of Multiple Traumas
Nongovernmental Organizations in Disaster Medicine
Worker Health and Safety in Disaster Response
Community Hazard Vulnerability Assessment
The Role of Social Media in Disasters
Full access? Get Clinical Tree
