NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Timbie JW, Ringel JS, Fox DS, et al. Allocation of Scarce Resources During Mass Casualty Events. Rockville (MD): Agency for Healthcare Research and Quality (US); 2012 Jun. (Evidence Reports/Technology Assessments, No. 207.)

Cover of Allocation of Scarce Resources During Mass Casualty Events

Allocation of Scarce Resources During Mass Casualty Events.

Show details


Literature Search

The peer-reviewed literature searches identified a total of 5,146 potentially relevant citations. A search of the grey literature yielded 297 citations, and our technical expert panel (TEP) suggested an additional 56 titles. Reference mining contributed an additional 217 citations. All 5,716 citations were imported into EndNote and then into DistillerSR, a web-based application designed specifically for the screening and data extraction phases of a systematic review. Reviewers selected 2,395 relevant and unduplicated titles for abstract review. During the review, they excluded 995 articles either because the abstract did not appear to answer a Key Question (664 articles) or because the abstract did not indicate a quantitative or qualitative data analysis (331 articles). After the abstracts had been reviewed, 1,400 full-text articles were available for further review.

Screening these articles with the aid of a short form led to the exclusion of 1,000 additional articles. Articles were excluded for at least one of the following reasons: (1) The article did not answer a Key Question (692 articles), (2) the article described a training program but did not report outcomes using performance measures (14 articles), or (3) the article was a proposed strategy but was not based on adequate consensus (277 articles for Key Questions 1 and 2; 17 articles for Key Question 4).

For Key Question 1, we considered 57 articles for data abstraction. We included nineteen articles that described tested strategies. We included seven additional articles in a separate group because they lacked a comparison population. One additional article was included that described a proposed strategy with a level of consensus that met our criteria. The major reasons for excluding articles at the data abstraction stage for Key Question 1 were insufficient evidence or inadequate consensus.

For Key Question 2, we considered 295 articles for data abstraction and ultimately included 55 articles that described tested strategies. We included an additional 47 articles in a separate group because they lacked a comparison population, and seventeen articles that described a proposed strategy with adequate consensus in a third group. Reasons for excluding articles included either insufficient evidence or inadequate consensus.

For Key Question 3, we identified 37 articles, ten of which we included in the review. Reasons for exclusion included either failure to address a resource allocation context or failure to assess the public’s opinions directly.

For Key Question 4, we identified 14 articles and included all of them.

In summary, we considered 400 articles for data abstraction. Ultimately, 170 met our selection criteria, including 27 studies that focused on policymakers (Key Question 1), 119 that addressed the decisions of providers (Key Question 2), 10 that considered the perspectives of the public (Key Question 3), and 14 that addressed engagement of providers in developing resource allocation strategies (Key Question 4). Five articles were written in languages other than English (4 German and 1 Portuguese). No articles were excluded due to lack of translational resources.

Reviewers used data abstraction tools as shown in Appendix B. We provide the evidence tables containing key data from the included studies in Appendix C. Citations of articles that we excluded and the reason for exclusion appear in Appendix D. Figure 3 depicts the literature flow, indicating the number of studies included and excluded at each screening level and the reasons for exclusion.

Figure 3 depicts our literature flow, including the number of titles identified for review, the number of articles excluded at each stage (and the reason for their exclusion), and the total number of articles accepted for each Key Question. For Key Question 1, 19 articles were included that described tested strategies, seven articles were included in a separate group because they lacked a comparison population, and one additional article was included that described a proposed strategy with adequate consensus. For Key Question 2, 55 articles describing tested strategies were included, 47 articles were included in a separate group because they lacked a comparison population, and 17 articles that described a proposed strategy with adequate consensus were included in a third group. For Key Question 3, 10 articles were included in the review, and for Key Question 4, 14 articles were included.

Figure 3

Literature flow. KQ = Key Question; TEP = Technical Expert Panel

Key Question 1. What Strategies Are Available to Policymakers To Optimize Allocation of Scarce Resources During MCEs?

What current or proposed strategies are available to policymakers to optimize the allocation and management of scarce resources during mass casualty events (MCEs)? What outcomes are associated with these strategies? What factors act as facilitators or barriers to their implementation or effectiveness?

Key Points

  • The small number of studies that met inclusion criteria (n = 19), and the marked variability in design, focus and content for this Key Question provide a relatively weak evidence base to inform policymakers. The 19 studies included more computer simulations (10) than intervention studies (9). Only a few studies examined similar resource allocation strategies using similar endpoints.
  • Each computer simulation was distinctly different from the others. Thus, their results cannot be meaningfully compared across studies. The computer simulations were often of lower quality than the intervention studies.
  • Three intervention studies examined the throughput achieved (or simulated) using different approaches to mass dispensing of medical countermeasures against anthrax. The standard “centralized” model for point of dispensing was efficient, but a decision-support software tool tested in Georgia further enhanced its efficiency.29 A “push” strategy using U.S. mail carriers produced even higher throughput than administration through fixed sites.30
  • We could not meaningfully compare results from the three studies that examined different approaches to augmenting health care resources following a major hurricane. Each employed a vastly different strategy and examined effectiveness using different end points. Nonetheless, each describes an empirically tested strategy deemed successful by the authors, ranging from opening alternate care sites to a mobile field hospital to more efficient distribution of patients via a regional medical operations center.
  • None of the included studies examined the implementation of crisis standards of care.

Description of Included Studies—Tested Strategies

The 19 papers included in this review address tested strategies for policymakers to reduce or manage less urgent demand for health care services (15 studies2943), optimize use of existing resources (two studies44, 45), or augment existing resources (four studies 32, 41, 46, 47); two studies included strategies that were classified in multiple categories.32, 41 No studies examining the implementation of crisis standards of care met our inclusion criteria. To meaningfully synthesize the available evidence we further classified strategies into subcategories (Table 1).

Table 1. Summary of strategies addressing Key Question 1, by category.

Table 1

Summary of strategies addressing Key Question 1, by category.

The 19 studies comprised three main types of analyses. Nine studies were intervention studies, including four drills and five analyses involving actual MCEs. Eight of the intervention studies occurred in the United States, and one study took place in Canada. The remaining ten studies were computer simulations.

Fifteen studies addressed biological threats, including anthrax (6), pandemic influenza (7), smallpox (1), and SARS (1). Three addressed natural disasters (hurricanes in each case, including Hurricane Katrina), and one addressed an explosive event (one of the September 11 attacks). All ten computer simulations addressed biological threats, including pandemic influenza, anthrax, and smallpox.

Among the five studies examining actual MCEs, three used a pre-post design, and two included only post-test assessments; none used a randomized controlled trial design. Studies assessing drills included one pre-post design and three post-only designs. Eight of the nine intervention studies had moderately high quality (50 percent or more of the total possible points across the quality domains) compared to six of the ten computer simulations.

Detailed Synthesis of Tested Strategies

Strategies To Reduce or Manage Less-Urgent Demand for Health Care Services

Twelve of the 15 studies reviewed under the broad category of strategies to reduce or manage less-urgent demand for health care services involved biological countermeasures. The specific strategies included modeling stockpile allocation, exercising stockpile dispensing, and mass distribution of antibiotics using mail carriers. The other three studies assessed the effectiveness of nonbiological countermeasures. These studies included a simulation of the impact of physical barriers to disease transmission, an exercise to raise awareness of legally acceptable intervention measures to stop the spread of pandemic flu, and implementing restrictions on elective surgery.

Biological Countermeasures

The 12 studies in this group included three intervention studies and nine computer simulations. The three intervention studies, all judged to be of relatively high quality, addressed point of dispensing (POD) operations for medical countermeasures against anthrax (presumably ciprofloxacillin). Two of the three studies provided quantitative end points that suggested they could be compared across studies (Table 2). One provided evidence that a traditional “centralized” POD system—where persons come to a fixed site to receive a medical countermeasure—provided slightly faster and more accurate processing than a hybrid model that combined both the centralized “pull” approach and a “push” approach in which countermeasures are delivered to some persons at their work site.31

Table 2. Comparison of different point of dispensing strategies.

Table 2

Comparison of different point of dispensing strategies.

The second study compared the standard centralized “pull” model to a different “push” model—one that used U.S. Postal Service mail carriers to deliver the medical countermeasure. The push approach in that study served more people per hour per provider than the fixed dispensing sites.30 When we converted the findings of one study into the units measured in the other, the “push” strategy using mail carriers appeared to produce the highest throughput. If the figures are indeed comparable, which is not entirely clear, then the centralized POD operations reported in the first study31 were more efficient than those in the second,30 and the “push” dispensing via mail carriers was the most efficient method of all.

The third study documented that POD operations supported by a specific decision-support software tool were demonstrably more efficient on several dimensions than traditional dispensing systems using no or existing software support. However, the quantitative endpoints were not comparable, and most comparisons between the one county using the tool and the seven counties not using it were mostly qualitative.29

The nine computer simulations were more varied in focus. Five addressed pandemic influenza, three addressed anthrax, and one addressed smallpox. Most of the influenza simulations examined different questions and thus were not comparable to one another. One study examined the use of the same or different drugs for treatment and prophylaxis,36 and one looked at allocation of the single stockpiled antiviral drug, including its use for treatment or prophylaxis.32 In the former, the authors found that a two-drug strategy for pandemic influenza (one drug for prophylaxis and a different drug for treatment) is more effective in delaying the propagation of disease and the emergence of drug resistance (including multi-drug resistance) than the use of a single drug for both prophylaxis and treatment.36 However, the simulation also indicated that the two-drug model is more likely to result in multidrug resistance than resistance to a single drug, which is a significant drawback. The other simulation provided useful, albeit somewhat less compelling, evidence. It noted that allocation of an antiviral stockpile should not be determined in advance; instead, it should be based on population attack rates and, potentially, age. It also indicates that when supplies of effective antiviral drugs are limited, they should be used for treatment rather than prophylaxis.32

Three studies assessed optimal vaccination-targeting strategies; two focused on the general population38, 35 and one focused on health care workers specifically.37 The first simulation, which we rated as high quality, reflects the importance of young children in influenza transmission and concluded that vaccinating children aged 5 to19 and their parents (ages 30 to 39) is a particularly effective vaccine targeting strategy, since these children are often vectors of transmission to others.38 In the second simulation, prioritizing prophylaxis to health care workers was shown to be an effective use of an antiviral stockpile, and this strategy did not have a deleterious effect on disease control in the population.37 Another simulation indicated that the most effective targeting strategy may depend on a policymaker’s objective.35 Specifically, to minimize population morbidity, the results suggested that children, adolescents, and young adults should be targeted; in contrast, to minimize mortality, infants, young adults, and older adults should be targeted.

One of the three anthrax simulations examined rapid mass distribution of prophylactic drugs versus treatment only of symptomatic persons. As expected, the simulation found that the former strategy prevents significantly more deaths than the latter.41 That study also showed the significant impact of adequate hospital surge capacity on reducing patient deaths. A second simulation found that local dispensing capacity was a critical factor in determining the cost-effectiveness of other strategies, such as increasing the size of stockpiles and improving surveillance.33 The other anthrax simulation was of poor quality and thus does not provide persuasive evidence to support its rather general findings.42

The smallpox simulation provided evidence to suggest that a combination of mass vaccination and targeted vaccination of contacts is needed to limit disease transmission. It also noted that school closures would further enhance the impact of such interventions.43

Several of the studies that tested strategies for implementing PODs involved relatively large-scale exercises that were conducted in different geographic regions. Evidence from these studies appears to be generalizable across locations and settings. The applicability of the evidence generated from computer simulations is exceedingly hard to assess. These studies may not provide highly applicable evidence if their conclusions rely heavily on assumptions or model parameters that are contextually inappropriate. Outcomes from tabletop exercises (e.g., increases in participants’ knowledge and confidence) may not be the most relevant outcomes for policymakers, who might be more interested in health outcomes or public perceptions of fairness. But taken together, the studies of biological countermeasures provide reasonably applicable evidence.

Nonbiological Countermeasures

Three studies assessed nonbiological countermeasures. Two studies involved a pandemic influenza context, and one study was based on the 2003 SARS epidemic. Among the influenza-related studies, one was an intervention study and one was a computer simulation; because they addressed entirely different issues, they were not comparable. The intervention study was a tabletop exercise addressing measures that policymakers could legally take during an infectious disease event affecting a community. Compared to pre-exercise measurements, post-exercise measurements reflected significant increases in knowledge and confidence regarding deployment of such measures.39 The computer simulation indicated that N95 respirators provide better protection against influenza infection than do surgical masks for both droplet and airborne virus transmission, but only if compliance with their use is nearly universal.34

The third study in this category assessed the effectiveness of imposing restrictions on ambulatory and inpatient medical and surgical care for nonurgent cases across all 32 hospitals in the greater Toronto area during the 2003 SARS epidemic.40 The authors showed that, while nonurgent admissions decreased significantly, high-acuity emergency department (ED) visits and interhospital transfers also decreased, suggesting that some patients may not have received needed care.

Strategies To Optimize Use of Existing Resources

We identified two studies in this category. One was of poor quality despite its providing highly applicable evidence from an actual MCE45; it is therefore not a robust source of evidence for this review. It described response strategies following the September 11, 2001, terrorist attacks in New York City. The study found that the absence of an enforced patient distribution system led to uneven load in three trauma centers, and attack damage to the Office of Emergency Management and disruption of cell phone and radio communications exacerbated problems with coordination and communication.

The second study documented significant reduction in patient transfer times once a coordinated regional trauma system was introduced for routine, small-scale trauma events.44 A comparably designed system based on a regional medical operations center was able to efficiently transfer and manage evacuation patients following Hurricane Katrina and transfer at-risk patients prior to Hurricane Rita.

Strategies To Augment Existing Resources

We reviewed four studies of strategies to augment existing resources. Three of the four were intervention studies evaluating measures taken after a major hurricane. The fourth was an influenza computer simulation discussed above under strategies to reduce or manage less-urgent demand for health care services.

The hurricane-related intervention studies did not report comparable end points; therefore we cannot make valid comparisons across their different strategies. One study documented the extra patient load cared for by a mobile field hospital deployed to care for evacuees from Hurricane Katrina.47 A second reported that an alternate care site in Dallas provided so much medical surge capacity following Hurricane Katrina that the emergency departments and trauma centers in the city saw no significant rise in patient visit rates during the two weeks postevent.46 The third study, which was a computer simulation, concluded that “mobile servers” (augmented hospital capacity provided by Federal health care providers) reduced predicted patient mortality.41

One study examining the impact of mutual aid agreements allowing transshipment of antivirals during an influenza epidemic found that the policy mainly favors less densely populated counties and is only cost-effective when there is geographic variability in the epidemic.32

The strategies we identified for augmenting capacity during MCEs relied on data from two real events and two computer simulations. All of the strategies were tested at a single site or within a single region; however, most strategies appear to be broadly applicable across settings. Several studies within this category reported process outcomes—mainly the number of patients served—while more relevant outcomes for policymakers might involve health outcomes. While the mobile field hospital appears to be particularly useful for a broad range of MCEs, the alternate care site that was established during Hurricane Katrina may only be useful for MCEs in which victims suffer less severe injuries.

Table 3 outlines the strength of evidence for Key Question 1.

Table 3. Strength of evidence for Key Question 1.

Table 3

Strength of evidence for Key Question 1.

Tested Strategies Lacking Comparison Groups

Seven studies were included in this section of the review. One study presented the results of an exercise that tested a disaster response protocol for Super Bowl XXXVIII.48 A second study, conducted as a simulation in Hawaii, demonstrated that prophylactic medication can be efficiently dispensed with minimal human-to-human contact using a drive-through clinic model.49 Another simulation study, conducted in the Netherlands, examined laboratory capacity during MCEs. It found that a national diagnostic laboratory network could handle diagnostic requests from hospitals during an MCE, but it would have insufficient capacity to manage the surge of tests that could be generated by the nonhospitalized population.50 A second study related to laboratory capacity described a customized laboratory information system to support Centers for Disease Control and Prevention (CDC) activities for rapid sample analysis and data reporting.51

Two studies assessed resource allocation strategies during hurricanes. Irwin et al. reported details about the successful use of a multidisciplinary treatment center in Houston to treat large numbers of evacuees for non-emergent medical concerns in the aftermath of Hurricane Katrina.52 During the time this large facility was in operation, it substantially reduced use of local emergency departments for non-emergent problems. A related study indicated that deployable military hospitals can effectively supplement surviving local health care capabilities after disasters.53

One study reported outcomes of an information technology applications deployed during the height of the 2009 H1N1 pandemic—an interactive, Web-based decision-support tool to help adults with influenza-like illness self-assess their need for ED care.54 The tool closely adhered to a diagnostic algorithm the group developed in collaboration with the CDC and subsequently validated using electronic health information collected from a large HMO in Colorado. The interactive, Web-based version of this algorithm was offered to the public via and a free Web site operated by Microsoft ( Users accessed it approximately 800,000 times before the end of the pandemic, with no reported adverse events. Although the report suggests that the concept of a web-based self-triage for influenza-like illnesses is feasible, it could not quantify the impact of the decision support tool on surge capacity. Similar Web sites exist, including one developed by a collaboration led by the American Medical Association (

Finally, in a study examining resource allocation under crisis standards of care, Etienne et al. described how a Multidisciplinary Healthcare Ethics Committee determined allocation of resources during the Haiti earthquake.55 The authors found that this process enabled ethical decision making in a timely manner.

Proposed Strategies

Our systematic review identified one study that described a proposed strategy for use by policymakers to allocate resources during MCEs. In 2008, a Federal interagency working group developed the current national plan for guiding the allocation of influenza vaccines during pandemics. The guidance is intended for use by Federal, State, local, and tribal governments; communities; and the private sector.56 Prioritizing the allocation of vaccine was accomplished by defining four categories in order of importance: (1) homeland and national security; (2) health care and community support services; (3) critical infrastructures; and (4) the general population. These target groups are further prioritized into tiers within each category, and, within tiers by the severity of the pandemic. The rationale behind the prioritization scheme is clearly elaborated in the report. For example, the highest-tier target group within homeland and national security comprises deployed and mission-critical personnel, recognizing that “these individuals are critical to protect national security” and have “a potential greater risk of infection due to geographic location and crowded living or working conditions.”

Key Question 2. What Strategies Are Available to Providers To Optimize Allocation of Scarce Resources During MCEs?

Key Points

  • A wide range of provider-oriented strategies has been tested in various contexts, including actual MCEs, exercises, drills, and computer simulations. However, with the exception of pre-hospital or “field” triage during MCEs, the body of high-quality evidence addressing any individual strategy is small, usually with no more than one or two studies providing evidence in each area. There is insufficient evidence to support the use of any one strategy over another.
  • Various triage systems and triage acuity scales have been used in emergency department operations for many years and have been extensively studied. But triage in the setting of MCEs is quite different, particularly triage practiced in pre-hospital settings where first responders may be required to assess large numbers of victims in a very short timeframe. Many of the studies on this topic raised significant concerns about current triage systems when used during actual MCEs. Other studies tested triage systems during exercises or drills and provided evidence with limited applicability.
  • The evidence base available to assess the effectiveness of the remaining strategies identified under this Key Question is thin. Few studies that met our inclusion criteria were based on data that were collected during one or more actual MCEs. The quality of these studies was substantially lower than drill-based studies. Few studies employed a randomized design. The computer simulations we identified provided low-quality evidence.
  • The majority of identified studies reported process measures (e.g., improved throughput times or triage accuracy) rather than outcomes. Studies that reported outcome data used less rigorous designs, such as comparing outcomes against historical control groups or a benchmarked performance rate, rather than a contemporaneous comparison group.
  • Few of the articles we identified examined specific barriers and facilitators to the implementation of provider strategies. Those that did reported this information inconsistently.
  • Evidence derived from drills and exercises did not report data on outcomes that are particularly relevant to patients and providers. The applicability of the findings beyond the immediate exercise setting is questionable.
  • With few exceptions, strategies proposed by national provider organizations were vague. Many did not propose actionable steps to help health care providers make difficult decisions regarding allocation of scarce resources under crisis standards of care.

Description of Included Studies—Tested Strategies

The 55 studies included in this part of the review address tested strategies available to providers to reduce or manage less-urgent demand for health care services (3 studies5759), optimize use of existing resources (48 studies60107), augment existing resources (1 study108), and implement crisis standards of care (5 studies77, 79, 109111). Two studies included strategies that were classified in multiple categories.77, 79 To meaningfully synthesize the available evidence we further classified strategies into subcategories (Table 4).

Table 4. Summary of strategies addressing Key Question 2, by category.

Table 4

Summary of strategies addressing Key Question 2, by category.

The 55 studies comprised a diverse set of analyses. Thirty-nine studies were intervention studies, including 19 studies evaluating the outcomes of drills and 20 analyses involving actual MCEs. Of the remaining 16 studies, 7 were computer simulations, 2 were systematic reviews, 5 were validation analyses, and 2 were laboratory analyses. Seventeen of the 39 intervention studies took place in the United States, while the remaining 22 represented a range of international contexts, including Europe (8), Israel (6), Asia (3), Canada (1), Australia (1), Mexico (1), Rwanda (1), and Haiti (1).

The studies addressed a wide range of MCEs, including explosive events (9), pandemic influenza (6), natural disasters (6, all of which involved earthquakes), nuclear/radiological events (3), transportation accidents (3), chemical events (3), multiple hazards (10), other MCEs (5), and unspecified events (10). The quality ratings were at least moderately high (50 percent or more of the total possible points across the quality domains) for 41 of the 55 studies.

Among the 20 studies examining actual MCEs, 6 used a pre-post design, 13 included only post-test assessments, and only a single study used a randomized controlled trial design. Studies assessing drills included 4 randomized designs, 5 pre-post designs, and 10 post-only designs.

Eighteen of the 19 studies involving strategies tested in drills had moderately high quality, compared with 11 of the 20 analyses of strategies tested during actual MCEs. Both meta-analyses were high quality, but only 3 of the 7 computer simulations were rated as having at least moderately high quality.

Detailed Synthesis of Tested Strategies

Strategies To Reduce or Manage Less-Urgent Demand for Health Care Services

Three studies described strategies to reduce less-urgent demand for health care services. Two studies examined techniques to rapidly dispense prophylactic medication, while the third study assessed the impact of a centralized information distribution system to support the information needs of the public. The strength of evidence provided by these studies was insufficient.

Among the two studies involving delivery of mass prophylaxis, one study demonstrated that communities can implement existing CDC mass vaccination protocols during a real MCE and achieve benchmark levels of throughput.58 The second study used a computer simulation to demonstrate that the design of PODs may require better command and control structures to address variability in patient flow.59 The third study showed that implementing an automated, centralized information distribution system in Israel prevented overloading of a hospital’s communication lines.57

Although each of these studies cleared the threshold for evidence, the two simulations were of low quality. Moreover, the incident command system proposed as a solution to address bottlenecks in the operation of PODs has not been tested in an actual MCE. The study of the mass vaccination clinic used data from an actual event (an outbreak of Hepatitis A) in a community with apparently average levels of preparedness. The results may be generalizable to similar communities but may not be generalizable to other types of MCEs. Although the outcomes of the centralized public information system were assessed at a single hospital, the findings are likely to be applicable to other Israeli hospitals. However, the requirements for implementing such a system in the United States are unclear.

Strategies To Optimize Use of Existing Resources

A total of 48 studies included a test of a strategy for optimizing existing resources during an MCE. Because of the large number of studies reporting the development or implementation of triage systems, we synthesized the evidence from these studies separately from the remaining studies in this category. The strength of evidence for both the triage studies and the nontriage studies is low.

Triage Systems

The 24 studies that examined triage systems can be classified in two main groups: (1) those that examined the validity of new or existing systems, and (2) those that assessed the degree to which these systems accurately triaged patients during drills or actual MCEs. One recent systematic review of the validity of triage systems comprising 11 articles (8 triage systems) concluded that limited evidence supported their validity.62 Among existing systems, the reviewers considered the Sacco Triage Method the most promising because it was the only system that combined estimates of patients’ survival probabilities with data on available capacity at receiving hospitals. A later validation study that was not included in the review showed that the Field Triage Score predicted patient mortality comparably to the Revised Trauma Score but was easier to calculate at the scene of an MCE.94 Collectively, these validation studies have low methodological quality. Most rely on small sample sizes, and few studies assessed the validity of the tool using prospectively collected health outcomes data from real events. In addition, few triage tools are applicable to pediatric disaster victims.

Several studies examined the implementation of triage systems during real or simulated events (Table 5). The vast majority assessed the accuracy of classifying patients into triage categories using the system’s specific criteria compared to a gold standard (e.g., medical record review or “true” triage categories determined prior to a drill). Only three studies reported data on the accuracy of a specific triage system used during an MCE. The reported accuracy of triage ranged from 62 percent to 100 percent across systems.

Table 5. Accuracy of triage for individual triage tools reported in 10 included studies.

Table 5

Accuracy of triage for individual triage tools reported in 10 included studies.

A few studies described implementation problems associated with triage systems. For example, in a commuter rail accident, implementation of Simple Triage and Rapid Treatment (START) led to poor allocation of patients between trauma centers and community hospitals, mainly because of confusion about the meaning of each triage category.78 Another study demonstrated that START triage categories were not sensitive to patients experiencing myocardial infarction or an asthma attack and may lead to under-triage of individuals with these conditions.100 Some studies reported triage performance using time-based outcomes, but these outcomes had limited comparability across studies due to differences in design.

Other studies provided evidence to inform triage approaches beyond the use of specific tools. For example, one hospital-based triage approach that was found to be superior in a computer simulation used a two-stage process in which mild cases were first separated from more severely ill or injured patients, after which the critically ill patients were distinguished from urgent cases.66 During the Sichuan earthquake of 2008, adding a resuscitation category to the standard START protocol enabled higher survival rates for a subset of victims who would have otherwise been categorized as “expectant” and not vigorously resuscitated.67 Other promising triage protocols included modified dosimetry methods, such as using fewer metaphase spreads for dicentric chromosome assays.60 One study demonstrated the effectiveness of a quality improvement program that was initiated in response to triage failures during a 2005 train crash and reported improvements in performance during a similar crash three years later.95

Although MCE triage has been examined more extensively than any other strategy, many of the studies we reviewed neither included a contemporaneous comparison group nor reported patient outcomes associated with the triage protocol. Studies tended to report throughput times or triage accuracy relative to an existing benchmark. Established standards for what constitutes acceptable triage performance are lacking, complicating efforts to infer the effectiveness of specific tools. Few studies tested triage protocols during MCEs. In general, triage accuracy rates that are measured during drills or exercises may provide evidence with limited applicability, because few drills are likely to capture the unique decision-making context imposed by a real MCE, and because results may be confounded by training that is part of the exercise.65, 69

Nontriage Studies

A total of 24 studies addressed resource optimization strategies other than the use of triage systems. We describe these results by subcategory and then assess the strength of evidence and applicability across all studies.

Case Managers

The use of case managers in an Israeli hospital was found to significantly expedite the delivery of critical tests and procedures and to lower the duration of hospital stays for critically injured patients.72


A randomized trial of alternative showering strategies suggested that providing washcloths with instructions to exposed victims of a radiological MCE enhanced the effectiveness of decontamination compared to the other methods.63

Health Care Worker Prophylaxis

One randomized trial conducted during the 2009 H1N1 influenza epidemic demonstrated that surgical masks were as effective as more costly and less readily available N95 respirators at preventing health care workers from contracting influenza.91

Health Information Technology

A computer simulation showed that a regional telemedicine system could potentially reduce mortality by limiting needless ED bed use and specialty care, thereby increasing surge capacity following a simulated earthquake.61 A second study demonstrated that triage accuracy can be enhanced through the use of electronic triage tags that monitor vital signs and permit reclassification of patients as their status evolves.87


Four studies evaluated strategies involving the use of imaging to optimize triage or ED throughput. In two studies, use of imaging improved initial assessment of large numbers of trauma patients. In the first, Focused Assessment by Sonography in Trauma (FAST) exams were found to have comparable diagnostic accuracy to CT and other diagnostic techniques.93 A second study showed that sonography was sufficiently accurate to be used as a primary triage tool during a major earthquake.86 In two drills, ED throughput was increased through the use of accelerated multislice CT protocols80, 104


Four studies provided evidence that load-sharing strategies can optimize allocation of patients to trauma centers and avoid the need to adopt crisis standards of care. In one study, the use of an incident command system successfully allocated victims of a terrorist bombing to avoid overwhelming the nearest hospital.75 A second study—also describing a terrorist event—demonstrated that centralized allocation of patients to hospitals, based on available capacity, achieved balanced allocation of patients to hospitals.97 A third study used a computer simulation to show that a regional surge distribution strategy reduced mortality among pediatric mass casualty victims.77 Finally, a load-sharing protocol developed in Germany for disaster situations involving mass gatherings was found to meet national standards.70

Medical Interventions

Two studies evaluated specific medical interventions for disaster victims and both reported favorable results. One demonstrated that many disaster victims with rhabdomyolysis from crush syndrome can avoid renal failure through vigorous fluid resuscitation.103 In a chemical exposure drill, a novel infusion device proved to be effective at delivering antidote, which enhanced throughput and increased predicted survival rates.82

Space Optimization

Three studies examined space optimization strategies. Two studies examined “reverse triage” protocols. One was implemented during a major transportation accident and the other during the 2009 H1N1 influenza epidemic. In the first study, the authors report that the protocol successfully created additional surge capacity without worsening the prognosis of patients who were discharged early.64 In the second study, the protocol failed to increase surge capacity—presumably because hospital management never formally implemented the protocol.107 The third study demonstrated that re-appropriating hospital lobbies, subspecialty clinics, and short-stay units (in conjunction with other strategies) increased surge capacity and reduced waiting times during the 2009 H1N1 epidemic.105


Each of the six studies of preparedness training for MCEs reported that the strategy is effective. One systematic review on training found that disaster drills were effective in improving response to MCEs, whereas evidence from computer simulations and tabletop exercises was inadequate to draw conclusions.81 Among studies that were not included in that review, one found that a computer game-based triage exercise was more effective in improving triage accuracy than tabletop exercises.88 A virtual reality method of teaching mass casualty triage skills reportedly improved accuracy,73 whereas a second study indicated that it did not improve provider performance using the START protocol.90 Another study that used podcasts and multi-manikin simulations improved triage performance by medical students.71 A typical “JumpSTART” training session improved triage performance in a subsequent drill.99

Although a clear majority of studies assessing resource optimization strategies indicate that these methods are effective, the limited level of evidence within each category does not allow us to draw definitive conclusions. Only three studies used randomized designs, and many studies failed to include a robust comparison group. Rather, many studies relied on performance benchmarks from prior events—a potentially subjective standard. For example, it is unclear what an “acceptable” false negative rate might be for an accelerated imaging protocol. Outcomes of load-sharing strategies that demonstrated balanced allocation are difficult to interpret: the few studies published on this topic did not report health outcomes or adverse events associated with these strategies, three occurred outside the United States, and the remaining study was a computer simulation. Nearly all studies reported positive results, suggesting that publication bias may be a threat to the validity of these findings. Although many of these studies drew on data collected during actual MCEs, they were often limited to a single setting and relied on small sample sizes, undermining both the validity and applicability of the results.

Despite providing outcomes data with published sources or comparison groups, many of these strategies can be regarded more accurately as promising pilot tests. For example, strategies involving electronic triage tags, and technology-enhanced triage training have not been taken to scale. As a result, important details of these strategies may not yet be fully understood. Load-sharing examples developed in Israel, a compact country where emergency care utilizes a national incident command system, may not work as well in other settings. Likewise, because the effectiveness of the telemedicine system was based on simulated data only, an unknown number of implementation issues may arise when applying the strategy in practice.

Strategies To Augment Existing Resources

A single study assessed different strategies for augmenting scarce resources during an MCE. Researchers demonstrated the feasibility of augmenting supplies of nerve agent antidote by converting a more widely available intramuscular formulation of pralidoxime to enable intravenous administration—a route more suitable for treating critically ill victims of a mass nerve agent attack.108 The strength of evidence in this category is insufficient.

Strategies for Use Under Crisis Standards of Care

Five studies evaluated outcomes of strategies for use under crisis standards of care during actual or simulated MCEs. These studies assessed a wide range of non-comparable outcomes that may have limited relevance to most providers. The strength of evidence from these studies is insufficient to support firm conclusions.

One article described the use of very early discharge from the intensive care unit (ICU) in a field hospital during the 2010 Haiti earthquake. The authors reported that this strategy enabled the hospital to treat a greater number of patients than would have otherwise been possible.109 Two studies assessed outcomes associated with a limited approach to trauma surgery under crisis standards. The first evaluated impact of “damage control” surgery to treat the initial influx of complex trauma victims from the London transit bombings. The authors report that this strategy resulted in lower than expected mortality rates.79 In the second study, hospitals that implemented damage control surgery in the aftermath of a major earthquake improved their operating room throughput with limited impact on patient outcomes.110 Another study examined the impact of crisis standards of care for orthopedic surgery under battlefield conditions. It reported faster throughput, but at the cost of higher complication rates, particularly surgical infections.111 Finally, a computer simulation study found that implementing crisis standards of care for pediatric disaster victims could reduce mortality, particularly if preceded by strategies to improve allocation of patients under surge conditions.77 However, this study has limited use because the specific approach used to implement crisis standards of care was not defined.

Collectively, these studies present encouraging findings but not definitive evidence. Most studies were of low quality because they used study designs that did not adequately control for potential confounders. Moreover, in the studies of actual events, data collection was typically nonsystematic, and measures of effectiveness were often compared to historical benchmarks that are open to interpretation. Several studies did not measure health outcomes or even the most relevant process outcomes. Instead, most of the studies focused on measures of throughput.

Reports based on actual MCEs were generally less rigorous but provided more applicable evidence. Computer simulations and exercises provided low-quality evidence, and their findings have limited applicability to real MCEs or to other settings. Crisis standards in the studies we reviewed were implemented in very specific contexts, including an earthquake and a terrorist bombing, and likely involved different types of injuries and different protocols. Crisis standards were typically implemented on a small scale and occasionally at a single site, limiting the generalizability of those studies.

Table 6 outlines the strength of evidence for Key Question 2.

Table 6. Strength of evidence for Key Question 2.

Table 6

Strength of evidence for Key Question 2.

Tested Strategies Lacking Comparison Groups

We identified 47 additional articles that presented evidence relevant to Key Question 2 but that did not meet the level of evidence required to be formally included in our review because they lacked a comparison group (Appendix Table C-5). Although the impact of these strategies on patient outcomes has not been conclusively demonstrated, many used novel techniques to optimize use of existing resources, augment existing capacity, and implement crisis standards of care. None of these articles addressed reducing less-urgent demand for health care services. Some strategies are sufficiently promising to warrant consideration for future research to advance the field.

Optimizing Resource Use

Seventeen studies sought to optimize resource use through improved approaches to triage or the use of imaging to support triage decisions. Two reports from actual disaster events112, 113 described the use of ultrasound, particularly the FAST exam, as a screening tool to support triage decisions. A third study, based on a simulation, assessed the feasibility of implementing ultrasound screening in the context of an MCE.114 Another simulation study examined the use of a modified approach to CT scanning as an adjunctive tool for clinicians evaluating large numbers of patients with complex injuries.115 A retrospective study described the use of three levels of triage—at disaster sites, primary health care centers, and tertiary referral centers.116 Okumura et al.117 described an approach that uses colored clothespins to perform color-coded triage for MCEs that require decontamination. Another drill-based study showed that a care team comprising both ambulance and hospital staff allowed timely triage for simulated disaster victims.118

Several exercises and simulations tested the effectiveness of information technology applications to facilitate MCE triage in the pre-hospital setting. One study employed a portable data collection tool for first responders. The authors claim that it reduced triage collection time and improved data collection accuracy in two field simulations.119 Another simulation exercise demonstrated that it was feasible to use a prototype Radio Frequency Identification (RFID) technology in the field as part of an online triage system for MCEs.120 A simple navigation device designed to guide walking wounded to a target destination was successfully tested in a third study.121 A pilot test of a “Scalable Medical Alert Response Technology,” or SMART, to monitor unattended patients showed promise in several emergency departments and scenes of actual MCEs.122 Electronic patient tracking through bar codes123 and web-based triage tools124 have also been shown to be promising techniques for optimizing resources.

Information technology has also been used to improve resource use inside health care facilities during MCEs. One article described the use of an electronic health information system—including patient medical records, picture archiving, and communications—that facilitated patient care in a field hospital established after the 2010 earthquake in Haiti.125 In another study, Roth and colleagues described a web-based all-hazards electronic disaster management system designed to optimize resource use by integrating health care data from multiple sources.126 A test of an automated call-down system demonstrated that roughly a third of personnel contacted were able to report to the facility in less than 60 minutes.127 Another pilot study tested an educational tool that linked participants’ resource allocation decisions to patient outcomes.128

Augmenting Existing Resources

Twenty-nine studies in this group focused on augmenting resources by repurposing drugs or devices; opening ancillary facilities; providing additional training to providers; or modifying existing equipment, such as ventilators, to serve multiple patients simultaneously. Two studies involved simulations to test whether a single ventilator could be modified to sustain up to four individuals.129, 130 One of the studies, conducted with four sheep, concluded that it may be possible to use this strategy during an MCE, such as a pandemic, when ventilators are in critically short supply.130 However, the other study, based on a simulation, suggested that such an approach would sustain only four adults for a very limited period of time.129 Another study of mechanical ventilation devised a prototype that could be quickly manufactured during an emergency.131 Automatic gas-powered resuscitators have been proposed to augment the supply of ventilators, but questions about their capacity and usefulness remain.132, 133 Other studies of respiratory support focused on enhancing capacity to deliver oxygen via an improvised system,134 testing the feasibility of just-in-time training for medical students to provide bag-valve-mask ventilation,135 and assessing the feasibility of cross-training non-respiratory therapists to assist in mechanical ventilation.136 Both of the cross-training studies demonstrated successful competency of trainees.

Several studies examined load-sharing strategies. A descriptive study, based on an actual MCE, reported successful use of an alternate care site as a temporary burn center coupled with successful long-distance transfer of some patients.137 Another described the implementation of a fully equipped mobile surgical hospital (MED-1) during Hurricane Katrina that succeeded in providing services to approximately 350 patients per day.138 During the 2009 H1N1 pandemic, an alternate care site effectively expanded ED capacity by 42 percent without any adverse events.139 Other studies reported the successful conversion of a charter plane to transport a large number of injured and ill tsunami victims back to their country of origin140 and a successful trans-provincial mass transfer of patients following a major earthquake in China.141 One study, conducted in a non-disaster situation, demonstrated that it is possible to implement load-sharing by transferring pediatric patients, including critically ill children, without adverse outcomes.142 Lessons learned during the mass interstate transfer of pediatric patients during Hurricane Katrina highlight the need for improved regionalization of pediatric services prior to an MCE.143 Trauma system structures have been tested as a mechanism for distributing victims of an MCE. For example, the Medical Alert Center in Los Angeles County has demonstrated its ability to coordinate distribution of critical casualties among area hospitals and trauma centers.144

Several articles in this group pointed to the role that information technology can play in augmenting health care resources. One team used a web-based application to assess surge capacity and other resources in a State disaster exercise.145 Another used a mass-casualty tracking system to improve coordination and reduce confusion during a simulated MCE.146 A wireless handheld device for recording and transmitting patient information between first responders and incident command has also been successfully field tested.147 A system that uses bar-coded identifiers to represent patients, injuries, facilities, and locations has been shown to facilitate information transfer and minimize errors during a simulated MCE.148 Two separate pilot tests demonstrated that electronic medical information tags can increase patient care capacity in the field and facilitate successful transfer of information to receiving facilities.149 Another study described the use of “pervasive computing technology” for MCEs, using a device that would capture contextual information from individuals in a non-intrusive manner to facilitate response. However, a prototype has not been built or tested.150

A few studies examined other approaches to augmenting resources. One study tested a tool designed to rapidly mobilize anesthesiology staff;151 another used a tool to estimate manpower reserve and service capacity for radiology staff.152 Two studies focused on lab capacity and scalability, particularly for chemical and radiological disasters. One of the studies described a customized laboratory information system developed at the CDC to support emergency response laboratory activities that would be required for the rapid analysis of samples such as chemical warfare agents.51 In another study, the Biodosimetry Laboratory in the State of Connecticut identified 30 willing and qualified labs that could perform initial biodosimetry processing should a radiological disaster occur.153 One study demonstrated the use of a unilateral external fixation device for stabilizing musculoskeletal injuries prior to major surgery.154 Two studies examined infectious disease control strategies within health care facilities. The first explored the feasibility of repurposing existing space to serve highly infectious patients and described the conversion of existing space within a health care facility into a temporary negative-pressure room through use of portable, HEPA-filtered forced air.155 The second tested a cost-effective method of establishing an airborne infection isolation area using a commercially available portable filtration unit and basic hardware supplies.156

Crisis Standards of Care

A single study focused on crisis standards of care met our criteria for inclusion in the review. The authors applied a decision support tool previously developed for ventilator allocation during an influenza pandemic to evaluate ventilator allocation decisions during the Haitian earthquake of 2010. The study used a case study design and assessed the allocation decisions made for five pediatric victims of the earthquake.157

Proposed Strategies

We identified 17 additional articles that proposed strategies to help providers allocate scarce resources during MCEs. These strategies have not been tested in the context of a real event, exercise, drill, or simulation, but represent the consensus opinion of one or more national professional organizations or task forces convened by the Federal government. The proposed strategies reviewed here addressed two major activities: performance of pre-hospital (field) triage and allocation of scarce resources in the hospital setting.

Prehospital Triage
National Association of EMS Physicians Workgroup

A national workgroup convened by multiple professional societies, provider organizations, public health organizations, the CDC, and the National Highway Traffic Safety Administration (NHTSA) reviewed nine existing mass casualty triage systems with the goal of recommending a single, national standard.158, 159 The work group used elements from existing systems to develop a new triage method known as SALT (Sort-Assess-Lifesaving Interventions-Treatment and/or Transport) that could serve as an initial all-hazards triage method. Although this work group ultimately endorsed the SALT triage system, it viewed it as “a beginning rather than final product.”

This workgroup subsequently developed the Model Uniform Core Criteria for Mass Casualty Triage to serve as a national guideline for mass casualty triage while enabling local flexibility in implementation.160 The Core Criteria consist of four categories: general considerations, global sorting, lifesaving interventions, and individual assessment of triage categories. Examples of recommendations include withholding lifesaving interventions if the intervention is not within the provider’s scope of practice, cannot be performed quickly (i.e., in less than 1 minute), or requires the provider to stay with the patient. Criteria for individual assessment include using the “dead” triage category for any patient not breathing after one attempt to open the patient’s airway and to refrain from counting or timing vital signs during the initial assessment.

Scarce Resource Allocation in the Hospital Setting
IOM Committee on Guidance for Establishing Standards of Care for Use in Disaster Situations

The 2009 IOM Letter Report called on health care providers, organizations, government officials, and the public to approach the challenge of allocating scarce resources in MCEs in a proactive and thoughtful way.13 The committee declared that such an effort should be grounded in the principles of fairness; equitable processes; community and provider engagement, education, and communication; and the rule of law. The committee called for the development of “consistent crisis standard of care protocols within each State,” and expressed the hope that their guidance could produce “a single, national framework for responding to crises in a fair, equitable, and transparent matter.” The Letter Report outlined a comprehensive framework for developing appropriate guidelines, based on an inclusive process and the best available medical evidence. However, it did not offer concrete recommendations to policymakers or providers about how they should make difficult resource allocation decisions under crisis standards of care. Our review identified no additional consensus recommendations on crisis standards of care in response to the Letter Report.

Task Force for Mass Critical Care

The task force developed a series of recommendations during the course of a summit meeting on definitive care for the critically ill during disasters. We have included three papers containing detailed recommendations. In the first paper, the Task Force developed recommendations on the use of equipment and space for creating surge capacity during MCEs.161 It recommended the use of one mechanical ventilator per patient (rather than the use of a multiple-limb ventilator circuits)—numerous examples of which were reported in the previous section. It also produced a list of ideal characteristics for stockpiled surge mechanical ventilators, recommended equipment for surge PPV, and recommended non-respiratory medical equipment. In the event that ICUs, post-anesthesia care units, and emergency departments have reached capacity, the Task Force recommended the following treatment spaces (in order): (1) intermediate care units, step-down units, and large procedure suites; (2) telemetry units; and (3) hospital wards. The Task Force strongly discouraged the use of nonmedical facilities to serve as alternate care sites. Finally, the Task Force endorsed a collaborative team model for staffing during critical care surge.

In the second paper, the Task Force proposed a bundle of seven services that comprise emergency mass critical care (EMCC).162 Each of these services requires inexpensive equipment and can be implemented without consuming extensive staff or hospital resources. The Task Force also developed a framework for optimizing surge capacity that includes various activities along a continuum from minimal patient need to overwhelming patient need and consists of 5 major types of activities: substitution, adaptation, conservation, reuse, and reallocation. The Task Force also adopted a multi-tiered critical care surge capacity framework that delineated specific triggers for escalation to higher tiers.

In the third paper, the Task Force presented a framework for resource allocation during MCEs that included specific inclusion criteria for the receipt of medical or palliative care.163 The inclusion criteria recommended by the Task Force are based on those developed by Christian et al.164 Recommended exclusion criteria take into account both the Sequential Organ Failure Assessment (SOFA) score and a patient’s chronic illnesses. The Task Force proposed a SOFA score cutoff corresponding to an 80% risk of mortality, and it also enumerated the specific chronic illnesses that should be used as exclusion criteria. The Task Force recommended prioritizing patients in the order of their latest SOFA score and daily SOFA trend. Finally, the Task Force described the recommended responsibilities of the triage officer and the recommended composition of the triage team—a critical care nurse, respiratory therapist, and/or clinical pharmacist.

Pediatric Mass Critical Care Task Force

The Task Force proposed minimum resource requirements for pediatric emergency mass critical care 165 that are largely consistent with those developed by the Adult Task Force on Emergency Mass Critical Care.161163 The Task Force also developed specific recommendations for non-pediatric hospitals, including a recommendation that adult ICUs keep adolescent patients without consultation (and patients aged 5–8 years after consulting with pediatricians). The Task Force was unable to recommend a pediatric prognostic scoring system to guide the triage of pediatric MCE victims due to the poor performance of existing systems. Moreover, the Task Force declined to endorse exclusion criteria for the use of life support based on pre-existing conditions despite the fact that other groups have proposed such criteria. The Task Force was also unable to develop recommendations on criteria for withdrawing life support for pediatric patients during MCEs. Finally, the Task Force called for the development of a triage protocol that not only took into account a patient’s likelihood of survival but also the likelihood that a patient would require a prolonged ICU stay. (This latter point is a notable difference from the adult recommendations that did not consider prolonged use of ICU resources).

Working Group on Emergency Mass Critical Care

This working group was convened by the Society of Critical Care Medicine and the Center for Biosecurity at the University of Pittsburgh Medical Center. The work group recommended that minimal requirements during crisis standards of care include: basic modes of mechanical ventilation, hemodynamic support, antibiotic or other disease-specific countermeasure therapy, and a minimum set of prophylactic interventions that can reduce the serious adverse consequences of critical illness.166 The work group also emphasized that the goal of crisis standards was to help the greatest number of people survive the crisis, and favored the use of triage protocols rather than a first come first served model. Additional recommendations included the personnel that should be involved with emergency mass critical care, the location where care should be provided, and specific infection control practices.

Society of Critical Care Medicine Ethics Committee

The SCCM Ethics Committee recommended that resource allocation decisions for patients with otherwise equivalent prognoses should be made on a “first come, first served” basis.167 Although the SCCM listed factors that should be considered when allocating ICU beds, such as the likelihood of a successful outcome, the patient’s remaining life expectancy, and the patient’s anticipated quality of life, it did not provide specific inclusion/exclusion criteria for these decisions. Ultimately, the SCCM Committee argued that “institutions should establish an explicit mechanism for implementing policies to allocate ICU resources.”

American Thoracic Society Bioethics Task Force

The Task Force reached similar conclusions to those of the SCCM Ethics Committee168 It emphasized that patients who continue to meet criteria for medical need and benefit should continue to receive ICU care, even if new candidates for ICU admission have an even greater potential for benefit. This task force went further and applied these same principles to all ICU services, not simply the allocation of ventilators or ICU beds.

Other Recommendations

Other recommendations, such as those by the European Society of Intensive Care Medicine, offer illustrative inclusion/exclusion ICU admission criteria but stop short of providing recommendations.169 The Australasian Surge Strategy Working Group enumerated strategies involving the use of space, staffing, supplies and equipment, and flow to optimize the ED response to mass casualty events, but it did not specifically address crisis standards of care, noting that this effort was “beyond the scope of [their] paper.”170 Similarly, other articles specified objectives for disaster preparedness and response, but not a path to achieving them. For example, the CDC convened an interdisciplinary panel of experts to develop strategies to assure surge capacity for sudden MCEs, particularly terrorist bombings.171 The effort culminated in the development of “surge action templates” tailored to ten distinct disciplines to address known challenges. The EMS template, for example, calls on local EMS organizations to “describe in a plan how alternative transport for 200 ambulatory patients will be initiated in the first 10 minutes after an explosion.” But it does not offer guidance on how to accomplish these objectives.

Another study focused specifically on appropriate use of immunization and postexposure prophylaxis for tetanus and occupational and non-occupational exposures to bloodborne pathogens during mass casualty events.172 However, the recommendations did not directly address altered standards of care when vaccines are in short supply. The European Society of Intensive Care Medicine’s Task Force for Intensive Care Unit Triage also provided recommendations and standard operating procedures for patient and staff prophylaxis during a pandemic.173. Finally, In 2007, the American Medical Association and American Public Health Association jointly released a set of eight goals for expanding health system surge capacity.174

Key Question 3. What Are the Public’s Concerns Regarding Strategies to Allocate Scarce Resources?

What are the public’s key perceptions and concerns (e.g., values, equity, transparency, communication, and public input) regarding the development and implementation of strategies to allocate and manage scarce resources during both actual and potential MCEs?

Key Points

The evidence across studies is relatively consistent in supporting the following concepts:

  • A successful allocation system should balance the goals of ensuring the functioning of society, saving the greatest number of people, protecting at-risk populations, reducing deaths and hospitalizations, and treating people fairly and equitably.
  • Multiple criteria are used to prioritize recipients of resources during an MCE. Health care professionals, health care workers, and first responders were among the highest priority groups; politicians were among the lowest.
  • High priority should be given to children and young adults for receipt of care.
  • Prioritization criteria should not be based on ability to pay, “first come, first served,” or random selection (lottery system).
  • The public has a high degree of faith and trust in medical professionals to make appropriate allocation decisions based on their expert opinions.
  • Resource allocation guidelines should be generally consistent, but should allow health care institutions some degree of flexibility to make allocation decisions based on their specific demand and supply situation.

Description of Included Studies

Our search identified ten studies that addressed this Key Question.175181, 182184b Six studies 177180 182 were conducted in seven different U.S. States (Georgia, Massachusetts, Minnesota, Nebraska, Oregon, Washington, and Louisiana); two studies were conducted in Australia,175, 176 and one each in Canada181 and Brazil183. Seven studies reported public opinions related to pandemic influenza, while three177, 183, 184 did not involve a specific MCE context. Two basic approaches were used to solicit public opinions: (1) public engagement activities in various forms, such as deliberative meetings, community forums, and small group discussions; and (2) surveys, including web-based questionnaires, telephone surveys, and solicitation of written comments.

The number of citizens participating in the studies ranged from fewer than 10 to more than 5,000; public engagement forums (sample size 9–441) involved fewer participants in general but generated substantially more in-depth discussions among participants. As a result, public engagement activities provided substantially more detailed information than surveys, although the latter were more broad-based (sample sizes 1,030–5,220).

Detailed Synthesis

A wide range of issues were discussed regarding public opinions on policies and strategies to allocate and manage scarce medical resources during an MCE. The ten papers all addressed at least one of two main themes: development of resource allocation policy and criteria for who should receive treatment under crisis standards of care. Resource allocation policy covered the public’s perceptions about allocation systems in general such as whether or not resource allocation guidelines were needed; what goals the allocation system should achieve; who should make allocation decisions; and what role the Federal and State governments should play in developing, managing, and implementing such a system. Priority criteria reflected the public’s views of which groups should be considered high versus low priority for receiving scarce medical resources during an MCE.

We rated the overall strength of evidence for these studies as medium (Table 7). Because of the limited number of studies addressing the question, and because four were from outside the United States, we rated the risk of bias for the set of results as medium. The evidence from the seven forums and three surveys was remarkably consistent, and, by construction, the evidence was derived directly from the public (indirect reports of public opinion were excluded). Because much of the evidence comprised rankings and consensus opinions, we could not meaningfully evaluate the precision of the results. Key themes arising from public engagement activities are summarized below.

Table 7. Strength of evidence for Key Question 3.

Table 7

Strength of evidence for Key Question 3.

Allocation Guidelines

The public agreed that MCEs are highly unusual situations that require decision-making processes and protocols different from those used in normal clinical circumstances. They stressed the need to proactively establish allocation standards or guidelines that will be followed by health care facilities and other providers. Participants generally felt that it will be important to take into consideration the different capacities that each region or facility might have, as well as different service demands they might face. Thus, although they widely agreed that guidelines for crisis standards of care should be generally consistent across health care facilities, they believed that institutions should have some degree of flexibility to make allocation decisions based on their specific demand and supply situation. Participants also agreed that guidelines should be relatively simple so that they could be successfully implemented.180

Goals of Allocation Systems

Participants in these forums listed several goals for a successful resource allocation system: ensuring the functioning of society, saving the greatest number of people, protecting at-risk populations, reducing deaths and hospitalizations, and treating people fairly and equitably.

Some participants preferred one goal over another, but one study found that many participants showed some degree of internal conflict when weighing different goals.181 Other participants suggested a balance of objectives.179 When forced to choose only one goal, participants explicitly stated that they would choose ensuring the function of society in the long run.176 To achieve the goals, most participants agreed that certain compromises might have to be made. For example, seeking to save the greatest number of people might result in lowered standards of care.180

Allocation Decisionmakers and the Role of Government

Across most studies, the public showed a high degree of faith and trust in medical professionals to make appropriate allocation decisions based on their expert opinions. They believed that health care professionals and experts were essential to ensure a fair and effective allocation system. Some participants preferred a joint committee consisting of a variety of experts and policymakers (but not politicians) elected by their peers.175 The public expressed a lack of trust in elected or appointed representatives and politicians without public health qualifications to make health resource allocation decisions.

Participants in the Public Engagement Pilot Project on Pandemic Influenza study suggested that the role of the Federal government should be to provide broad guidance, while responsibilities for interpreting and implementing the guidance should remain at the State and local level.178

Prioritization Criteria

Although the underlying rationale of prioritization has always been to ensure the best use of limited resources without capricious discrimination, participants used mixed criteria to prioritize recipients of resources during an MCE. Given different situations, participants expressed their preferences for a range of criteria, including the individual’s role in society (e.g., occupation), equity, survivability (the number of years a person would live if they are treated and survive), vulnerability, risk of exposure, and likelihood of recovery. Below, we summarize the key considerations raised by the public regarding each criterion.

“Role in Society” Criterion

A majority of participants across studies seemed to accept the criterion of ranking people based on their role in maintaining a properly functioning society. Professionals and health workers were always among the groups given highest priority to ensure an adequate workforce for providing continuous services to all people. For the same reason, first responders, essential services (e.g., power, water, electricity, gas), and military personnel were also listed as priority groups by many participants. This prioritization seemed to reflect the public’s perception that a successful allocation system should assure the functioning of society. However, one problem with this criterion, as pointed out by some other participants, was that it was not always easy to assess an individual’s “value” to society because individuals contribute to society in different ways.

“Equity” Criterion

Equity was a somewhat expected criterion, given America’s egalitarian nature and the role of equity concerns in public health in general. All participants in all studies unanimously agreed that decisions based on race, gender, culture, legal status, nationality, language, or income were unacceptable; prioritization based on age seemed to favor children and young people over the elderly. The elderly were not generally perceived as a priority group, although a small proportion of participants expressed the belief that all age groups should be equally valued and valuable.179 Together with chronically ill and disabled people, the elderly were perceived by some participants as “not contributors to a future society” and therefore were accorded lower priority for receipt of scarce health care resources.175 In fact, some participants in one study supported a policy that would “de-prioritize” persons more than 85 years of age.179

In contrast, many participants listed children and young adults as priority groups. For example, in a study from Australia, priority was given to children and young people aged 2–30, because “they are the future.”176 In the United States, children and pregnant women were prioritized, although to a lesser degree than health care professionals and health workers.180 Findings from a nationwide telephone survey conducted by the American Academy of Pediatrics highlighted the significant lack of medications for children during disasters.177 A majority of respondents in the studies we reviewed supported giving higher priority to children who need life-saving treatment.

“Survivability” Criterion

Many participants expressed the belief that patients’ survivability should be considered and that health care providers should be the ones to make allocation decisions accordingly. They argued that allocation of significant resources to an individual with low probability of survival is a suboptimal use of limited resources, regardless of the importance of that individual’s role in society.180

Other Findings Related to Prioritization

Political decision makers were generally among the groups accorded the lowest priority, mainly due to lack of public trust and public suspicion that they would misuse their authority. Participants raised the issue that improving transparency of decision-making processes and funding streams and providing more information to the public could be important tools to gain the public’s trust.

A few prioritization methods were rejected by most participants. These methods included decisions based on ability to pay, “first come, first served,” and random selection via a lottery system.

Another interesting finding was that some participants changed their priority decisions when those choices were reassessed in follow-up surveys, implying that their opinions could be influenced by the process of group deliberation, as well as by exposure to public briefings by experts. Data from the King County post-forum survey showed that many participants had shifted their opinions during the time between the forum and the post-forum survey.180 For example, the percentage of participants who considered children and pregnant women to be a high-priority group dropped from 71 percent during the forum to 40 percent after the forum.

Special Concerns of At-Risk Participants

Few studies separately reported public opinions on resource allocation regarding at-risk populations (e.g., minority groups, frail elderly). In most instances, members of these groups were actively recruited and included in the discussions. The only notable finding was from a public engagement forum in Seattle and King County, Washington, where Hispanic participants voiced much stronger opinions about prioritizing children and pregnant women than did non-Hispanic participants (70 percent indicating that children and pregnant women should be a priority vs. 27 percent of non-Hispanics). They also emphasized the needs of minorities and immigrant populations.

Other Relevant Findings

The public’s perceptions and concerns about medical resource allocation during an MCE did not always agree with those of policymakers, public health experts, or other stakeholders. Some doubted how much their concerns and perceptions would be taken into account in establishing a disaster plan. But in other cases, the public and health policymakers shared the same opinions. For example, in Australia, the priority groups selected by the public (health care workers and other functioning groups) based on the criterion of “the need to maintain functioning of critical infrastructure” corresponded to those outlined in the national pandemic plan.176 In Minnesota, a majority of the participants agreed on the three resource rationing objectives proposed by expert panels (reduce deaths, treat people fairly, and protect public health and infrastructure).179 However, other studies showed some nuanced differences in perspectives between the general public and experts or other stakeholders. For example, the King County study found that while the goals of prioritization were similar, experts tended to focus on maximizing resources by assessing survivability and saving the greatest number of people, and the public appeared to focus more on response capabilities by prioritizing health care workers and first responders.

It was notable that participants generally did not choose prioritization strategies that specifically favored themselves or their families. For example, the study in Canada found that participants who had children themselves did not necessarily give priority to children: Only 9.7 percent of participants who had children preferred the child-focused priority plans.181 Similarly, in the Minnesota public engagement project, which focused on prioritization for socially vulnerable groups, members of these groups seldom chose to prioritize themselves, but rather were more likely to prioritize groups associated with critical infrastructures.182 Fear of stigma following the implementation of such a policy was one of the main reasons cited by these representatives.

Participants acknowledged that an MCE is a difficult situation that would affect everybody. Some suggested that the number of pharmaceutical manufacturers should be increased to produce more supplies to meet the needs of an influenza pandemic. Others urged that in an MCE when medical resources were scarce and difficult allocation decisions must be made, more communication, information, education, and training would be needed to prepare the public.178 Some participants reported that they would be willing to accept some increase in their income taxes now as a form of insurance against an inadequate response to a future disaster.176

Key Question 4. What Methods Are Available To Engage Providers in Developing Strategies To Allocate Scarce Resources During MCEs?

What current and proposed methods are available to engage providers in discussions regarding the development and implementation of strategies to allocate and manage scarce resources both in planning for and during an MCE? What outcomes are associated with these strategies? What factors are identified as facilitators or barriers to engaging providers in these discussions?

Key Points

  • Nearly all studies described successful engagement strategies that involved multiple stakeholders and employed an inclusive, systematic, and often iterative process for reaching decisions or crafting a final plan. The articles we reviewed did not clearly identify one approach as superior to the others.
  • Engagement strategies varied by type of policymaker, provider, and range and mix of participants. Engagement strategies addressed planning for scarce resource allocation at different jurisdictional levels, ranging from local to regional, State, and even interstate levels.
  • Most engagement strategies were not specific to a particular type of disaster or to any single broad category of adaptive strategy for scarce resource allocation. However, only 5 of 14 studies addressed the development of strategies for implementing crisis standards of care.
  • Only 2 of 14 studies described an engagement process that included the public.
  • Provider engagement was led both by providers and by local or State government officials. The latter often did so in partnership with other institutions, including academic institutions.
  • Technical (e.g., clinical) experts and health leaders both led and participated in provider engagement strategies, adding credibility to the engagement process and the resulting plan, protocol, framework, or strategy.

Description of Included Studies

The 14 studies included in this part of the review address a wide range of planning activities and exercises with the goal of developing resource allocation strategies for MCEs. Many engagement activities involved a combination of adaptive strategies for resource allocation, but fewer than half of the studies (5) addressed the implementation of crisis standards of care. Six studies reported the results of engagement activities led by providers, while seven studies reported on those led or co-led by policymakers. One study reviewed planning models that included both provider-led and public health department-led engagement models.

All 14 studies took place in the United States but reflected broad geographic diversity: 11 studies described local-, regional-, or State-level planning in urban or rural settings in 16 different, specified States. Two studies were carried out in multiple, unspecified locations. One study drew experts from across the country.

Nearly half of the studies (6) did not specify the type of MCE to which planning activities or exercises were oriented. Among the remaining 8 studies, 4 addressed pandemic influenza preparedness, 2 addressed all-hazards preparedness, 1 addressed biological threats of various types, and 1 addressed radiological or nuclear threats. Of the 14 included studies, 11 were largely descriptive, while 3 were intervention studies with at least one post-test measurement.

All engagement strategies involved multiple stakeholders and systematic, often iterative, consensus building to undertake planning or multi-party exercises. Different studies described planning at the local, intrastate regional or county, State, or interstate level. Nearly all studies described engagement of hospitals—often by other hospitals. State and/or local public health departments were also included in most, though not all, studies. Leaders of engagement processes, commonly in partnership, included hospitals, State or local public health departments, academic institutions, intrastate or interstate regional entities, and de novo planning entities. The range of providers who were targeted by engagement strategies included professional staff in general or specialty hospitals, clinics, community health centers, pharmacy departments, laboratories, and front-line health care workers (e.g., emergency medical technicians). Although most of the studies described well established engagement strategies, some described more novel strategies. Of note, only 2 of 14 studies included public representation as part of the engagement process.179, 185 A summary of strategies addressing Key Question 4 is located in Table 8.

Table 8. Summary of strategies addressing Key Question 4, by category.

Table 8

Summary of strategies addressing Key Question 4, by category.

Detailed Synthesis

Nearly all studies described a successful engagement process that led to one or more desirable outcomes, including the development of resource allocation plans, training, or a commitment of resources. Synthesizing the evidence for Key Question 4 was challenging because of the nature of this question (related to provider engagement methods, rather than testing of the resource allocation strategies developed as a result of the engagement process) and the variability in study focus. However, several facilitators and barriers emerged as general themes across multiple studies.

Common facilitators of provider engagement strategies included the personal relationships established, the willing commitment of actors to participate in cooperative planning, the iterative and broadly inclusive engagement of key stakeholders, and the technical excellence and credibility of partner institutions or experts. Some papers referred to barriers stemming from the differences in the organizational cultures of collaborating partners, such as public health and hospitals186 or public health and business.187 Other barriers related to the long time required to build critical relationships,188 government regulations,187, 189 the complexity of interstate agreements,188 and the variability across facilities or other differences that impede a “one size fits all” approach.179, 190

We rated the overall strength of evidence for Key Question 4 as medium (Table 9). The risk of bias was medium, given the high likelihood for publication bias (unfavorable engagement strategies may be significantly less likely to be published). While the evidence on the effectiveness of the engagement models was consistently positive, it was indirect because the studies did not report how implementation of the strategies developed from the engagement process affected population health outcomes. We could not assess precision, given the qualitative nature of the evidence.

Table 9. Strength of evidence for Key Question 4.

Table 9

Strength of evidence for Key Question 4.

No study appeared to be highly unique to the site where it was carried out; however, the applicability of the evidence may be somewhat limited to the contexts described in each study. Most of the studies were at least moderately dependent on the scale of the MCE, such as the public health–business partnership to dispense medical countermeasures and the different approaches to optimize or augment resources through the use of existing personnel, health centers, laboratories, or pharmacy departments to provide surge medical resources. All strategies related to crisis standards of care were very dependent on scale of the MCE.

Below we summarize the key results according to whether providers or policymakers led or co-led the engagement process.

Engagement Strategies Led by Providers

Individual providers tended to engage other providers to develop highly technical or clinically oriented resource allocation strategies. For example, one study described how academic medical leaders engaged clinician and non-clinician experts to develop a 5-category classification system for “reverse triage” of hospital inpatients, based on their agreement about varying levels of risk tolerance for major medical consequences.191 In another study, hospital pediatric leaders engaged other acute care pediatricians from across the country to develop pediatric crisis standards of care.189

Two studies described more novel engagement approaches. In these instances, the providers who initiated the engagement represented ancillary clinical services, such as the laboratory and pharmacy department. In one study, the State biodosimetry laboratory engaged all public and commercial laboratories in the State to assess and support development of additional capacity to prepare laboratory specimens for diagnosis of radiation exposure following a major nuclear or radiological event.153 In the other study, the pharmacy department of a hospital helped lead development of a regional mass casualty “pharmaceutical preparedness” plan, including pharmaceutical resource sharing among regional providers.192

Institutional providers such as hospitals engaged other institutional providers in medical surge planning. In one study, an entirely new planning institute was created: Four unaffiliated hospitals in Brooklyn engaged the New York City Department of Health to organize the “New York Institute of All-Hazards Preparedness,” which in turn engaged individual hospitals to work together to identify enough surge beds to meet national standards across the region as a whole.193 Another study presented extant U.S. models for medical surge planning. Florida and Louisiana reflect decentralized planning models in which hospitals and the State hospital association engage other hospitals in surge planning.186 The same study described the decentralized rural surge planning process in Oregon, in which a regional medical center engaged other hospitals in surge planning. This study also described hospital-directed tiered regional planning models in Illinois, Louisiana, and Missouri. In these States, a designated regional hospital engaged other hospitals in surge planning.

A particularly interesting model is that of the Veterans Health Administration (VHA), because it is both a very large provider (the largest integrated health care delivery system in the United States) and a Federal policymaker. One study described a series of pilot tabletop exercises for the local, regional, and national levels of the VHA system, in which the VHA engaged other local and regional providers, as well as local and State public health departments and first responders.194

Engagement Strategies Led or Co-Led by Policymakers

With the exception of the VHA study just noted, government-led engagement strategies were largely at the State and local government level. In most instances, State or local public health departments partnered with other institutions, such as academic medical centers, to engage other providers in planning for scarce resource allocation. Some studies described engagement strategies involving the traditional and typically large range of partners, while others described more novel partnerships. For example, the case study compilation of planning models describes the top-down county planning model with master (State-level) mutual aid agreement exemplified by California and Illinois, and the third-party-directed planning model of Missouri, where the State’s health department and a designated hospital engaged other hospitals in surge planning.186

Another traditional example is Boston’s public health department and the State primary care association. Working together, they engaged hospitals, community health centers (CHCs), and the emergency medical system in planning that added CHCs to the city’s medical surge plan; the city health department then engaged the Harvard School of Public Health to provide training and exercises for CHCs. This plan was subsequently tested in three actual events: preparation for the Democratic National Convention and the public health investigation of two disease outbreaks.190 In another study from Massachusetts, the State’s public health department and a partner academic institution engaged a wide range of institutional health care providers, other health agencies, and the general public in developing consensus State-level guidelines and a decision-making protocol for crisis standards of care.185 In 2004, RAND Corporation, in conjunction with local public health departments used tabletop exercise templates that could be locally customized to assess the strength of relationships between local public health agencies and local delivery systems when faced with a hypothetical pandemic flu emergency.195

Another study described a similarly inclusive planning process in Utah, in which the State health department and university medical center engaged multiple hospital and non-hospital facilities, professional associations, local public health departments, transit, EMS, and church groups in an iterative process to develop a regional medical surge plan.196 Yet another study described the initiative of two State health departments and the regional public health preparedness center in engaging pediatric hospitals, major pediatric clinics, State public health departments, and emergency responders into a five-State voluntary pediatric surge network; in doing so, they created a network, an operational handbook, and a formal memorandum of understanding.188

Examples of less traditional approaches include the partnership of a State government (Minnesota), a State university, and a health care ethics center to engage local governments, experts, the general public, and a few hospitals and clinics in developing proposed ethical frameworks and procedures for rationing scarce medical resources within the State during an influenza pandemic.179 Another study described a public health–business partnership in Georgia that engaged providers from the public and business side to refine approaches to, and expand sites for, mass dispensing of medical countermeasures.187

Analysis of State Reports

The IOM Letter Report13 called for development of “consistent crisis standards of care protocols” within each State, with neighboring States, and in collaboration with public and private sector partners. The Letter Report went on to recommend that each crisis standards of care protocol address five key elements:

  1. A strong ethical grounding
  2. Integrated and ongoing community and provider engagement, education, and communication
  3. Assurances regarding legal authority and environment
  4. Clear indicators, triggers, and lines of responsibility
  5. Evidence-based clinical processes and operations.

We reviewed a set of existing State plans to identify and describe the strategies developed by States to allocate scarce resources during MCEs. The majority of these State plans, plus Guam, (N=23 States) were compiled as part of the research that contributed to the IOM Letter Report and were forwarded to us by AHRQ. However, several of the documents we received did not qualify as a formal State plan or did not directly address the issue of scarce resources. We identified two additional State plans—New York and Wisconsin—through a search of the references in the plans we received. When States had multiple plans for different MCE contexts, we synthesized their content to give the reader a sense of the totality of the State’s strategies. Ultimately, we reviewed plans from 11 States and one U.S. territory. Collectively, these plans provide an important window into the current status of State planning and the specific resource allocation strategies that will be used in response to an MCE.

In general, the strategies outlined in the State plans fit into the same four categories of adaptive strategies used to guide our CER. These include (1) early actions to reduce or divert less-urgent demand for health care services; (2) steps to optimize use of existing resources; (3) efforts to augment existing resources; and finally, if and when these measures prove to be inadequate to meet demonstrable need, (4) the ability to shift rapidly from strategies designed to deliver optimal care to each patient to a modified approach calculated to do the most good for the most people with the resources at hand. In cases where strategies might be classified in multiple categories, we explain the rationale for our choice.

In the sections that follow, we qualitatively summarize how these recurring strategies and themes were addressed across States with plans, plus Guam. Table 10 displays specific elements of the various plans on a State-by-State basis.

Table 10. Key elements of State Plans.

Table 10

Key elements of State Plans.

Reduce Less-Urgent Demand for Medical Resources

The State plans we reviewed described several proposed strategies to reduce demand on the health care system during MCEs. Their strategies followed two basic approaches: keep non-critical patients out of the hospital, and, in the case of an infectious disease outbreak, urge non-ill members of the public to self-quarantine through social distancing.

Keep Noncritical Patients Out of the Hospital

The State of California, in particular, has devoted considerable attention to strategies to reduce demand for services that could be provided outside of hospital settings. The State plans indicate that all elective surgeries should be canceled so medical staff can refocus their energies and other key resources on patients who require urgent care, and to keep healthy patients away from those who may be contaminated.197, 198 Although the cancellation of elective surgeries might alleviate demand to a limited degree, a substantial MCE will likely necessitate further measures to ensure that sufficient supplies, staff, and facilities are available to treat critically ill or injured patients. Therefore the plan argues that non-critical care (e.g., first aid, primary care) could be safely and efficiently provided in off-premises facilities, such as community clinics or temporary health care facilities to reduce the demand on hospital resources.199202

Encourage the Public to Self-Quarantine (Social Distancing)

For certain infectious disease outbreaks, such as an influenza pandemic, a few States201, 202 discussed measures to impede or delay disease transmission by encouraging the public to self-quarantine. Specific strategies included encouraging employers to allow their employees to telecommute, closing schools, and educating the public regarding easily implementable non-pharmaceutical interventions, such as wearing a facemask.

Optimize Existing Resources

Nine of the 12 plans we reviewed recommended strategies to leverage the most benefit from existing health care resources, including staff, stuff (i.e., supplies) and structure.

Repurpose Existing Resources

Several States incorporated a range of approaches to increase bed capacity in their plans, including repurposing nonpatient care space for patient care, establishing temporary health care facilities such as tent hospitals, and “freeing up” space through early discharge of stable patients. Three of 12 States suggested repurposing space by converting overflow space and non-patient care areas (e.g., waiting rooms) into patient care areas or using outpatient areas for inpatient care.197, 199 One of the plans recommended enhancing capacity by converting single-occupancy rooms to accommodate two or three patients.199 Another option described in one of the California State plans is to triage ventilator-dependent patients directly to step-down units.199 Lastly, preserving bed capacity might be accomplished by canceling elective surgeries and limiting those that are done to “life or limb” situations in order to facilitate discharge.203

Optimize Use of Space

Several State plans recommend optimizing the use of space by establishing temporary health care facilities in non–health care settings.198, 204 Alternatively, California and Guam plan to expand bed capacity through strategies such as “reverse triage” that either allow for early discharge of stable patients from the emergency room or the hospital or that persuade outside facilities, such as long-term care units, to accept lower-acuity patients in transfer.197, 205 Load balancing by distributing care across a region (e.g., mutual aid) is another common approach to optimize the use of space within individual facilities. Plans in several States recognized that morgue capacity could be exceeded and call for the establishment of temporary morgues in certain scenarios.197, 198, 205

Use Health Care Providers and Nonmedical Staff More Efficiently

During MCEs, medical and nursing staff are likely to quickly become limited resources. State plans described five strategies involving staffing, including the shifting of duties and priorities, to accommodate potentially large and rapidly growing patient populations. Several State plans recommend increasing nursing shift duration (from 8 to 12 hours or from 12 to 16 hours) as well as increasing provider-to-patient ratios to extend the reach of available personnel.197, 198, 205 Cross-training staff through “just-in-time” training might allow for more staffing flexibility.205, 206 Examples of potential uses of this strategy during an influenza pandemic include training health care professionals who are not respiratory therapists to provide basic respiratory care, including ventilator management (Project XTREME), or teaching emergency medical services (EMS) personnel to administer vaccines.198 In addition, non-health care personnel could be deputized to carry out essential non-clinical functions and free up nursing staff.197 During a pandemic, cohorting patients having similar ailments in a single ward or facility may allow specially trained staff to provide care more efficiently and effectively.197 Finally, relaxing the requirements for medical documentation may enable staff to focus on patient care or other higher-priority duties.197


Florida’s prehospital triage strategy indicates that the State’s hospitals are using or implementing standard triage strategies, including Simple Triage and Rapid Treatment (START).207 JumpSTART extends the concept of a standardized triage to children. Florida’s plan also mentions an alternative triage system called the START2Finish® Surge Capacity Response Model for Healthcare. This model focuses on optimizing allocation of labor, supplies, and space during an MCE.207 In a similar vein, Utah has devised State-level Pandemic Influenza Hospital and ICU Triage Guidelines to systematically match patients to appropriate levels of resources based on their need in order to preserve bed capacity and oxygen capacity, limit or stop elective surgeries, and maximize available personnel to care for victims of a future flu pandemic.203

Substitute Effective Alternatives

Plans in two States, Wisconsin and Minnesota, focus on reuse or substitution methods to optimize available resources. Wisconsin, in its “Oxygen Conservation Strategies in Resource-Limited Situations” plan, recommends several detailed methods for conserving medical oxygen: (1) Discontinue high-flow applications, such as restricting the use of Simple Mask and partial rebreather to 10 Ipm; (2) decrease the number of inhalation medication applications or restrict continuous nebulization therapy; (3) maximize reuse of expendable oxygen appliances, including disinfecting via high-level procedures (bleach concentrations of 1:10; high-level chemical disinfection or irradiation if available); and (4) terminally sterilize ventilator circuits, as well as low- and high-bore tubing.206, 208 Minnesota’s State plan includes similar strategies but also recommends substituting oral or nasogastric hydration for intravenous hydration or substituting epinephrine for vasopressor if the need arises.

Strategies To Augment Existing Resources

Increase Reserves and Stockpiles

Several State plans incorporated strategies to draw on equipment, supplies, drugs, and personnel held in reserve or stockpiled for such contingencies or to secure these resources from other States or institutions that are not experiencing surge conditions. These strategies included the use of mutual aid agreements, and coordination with outside agencies, such as the American Red Cross and the Medical Reserve Corps.

One of California’s plans recommends stockpiling supplies at 20 to 25 percent above conventional levels to last for at least the first 72 hours (ideally, 96 hours) of an MCE.197 Other plans recommend inventories or plans to increase critical supplies to assess considerations for stockpiling, such as ventilators and critical medications.199, 205 Several State plans call for accessing either drug caches (antibiotics, antivirals) or supplies from the Strategic National Stockpiles (SNS).197, 206, 209, 210

Mutual Aid Agreements

Mutual aid agreements are key elements of several California plans as well as one from Washington.197, 200, 210212 Other partnerships that can augment personnel include volunteer clinical staff, such as the California Medical Assistance Team (CalMAT), federal Disaster Medical Assistance Teams (DMATs), Emergency System for Advance Registration of Volunteer Health Professionals (ESAR VHP), Colorado’s Volunteer Mobilizer (CVM) for Medical and Public Health (CDPHE), the American Red Cross, and the federal Medical Reserve Corps (MRCs).197, 198, 200, 205

Adopt Crisis Standards of Care

All of the State plans we reviewed addressed general parameters for the shift to crisis standards of care. Most commented on the following elements:

Define Priority Groups

The first step in defining crisis standards of care is to identify priority groups for certain types of resources. For example, several States, including Nevada, California, and North Dakota, discuss the protocol for allocating antiviral agents during a pandemic flu outbreak.204, 209, 210 The priority groups include those at the highest risk for infection, such as medical personnel, young children, pregnant women, and the elderly.

Be Prepared To Provide Comfort Care

In the event that lifesaving resources cannot be allocated to patients who need them, either because they are unavailable or because the patient has a low probability of survival, experts agree that protocols should be put in place to ensure that these patients are made as comfortable as possible. Only a single State mentioned comfort care in the plans we reviewed. California has noted the importance of this issue in their Enhancing Surge Capacity and Partnership Effort (ESCAPE) Crisis Care Guidelines plan, developed by the University of California, Davis, Health Systems.204

Allocate Resources Under Crisis Standards of Care

Some State plans offer guidance on how to allocate critical resources under crisis standards of care. For example, Minnesota and New York have plans to allocate certain medical equipment and supplies by patient prognosis, using triage methods such as the Sequential Organ Failure Assessment score (SOFA) and a tool based on the recommendations of the Ontario Health Plan for an Influenza Pandemic (OHPIP).206, 213 Many of these strategies focus on the distribution of mechanical ventilators, advocating that assignment (and in some cases reallocation) of ventilators should be directed toward those patients who are most likely to benefit. New York’s draft plan for ventilator allocation was cited by several other State plans when they convened a working group to study this issue.206 Nevertheless, although all of the State plans reference the need for crisis standards of care, few have articulated guidelines or cited published evidence to support provider decisions.



Doctor, 2011175. Docter SP, Street J, Braunack-Mayer AJ, et al. Public perceptions of pandemic influenza resource allocation: A deliberative forum using Grid/Group analysis. J Public Health Policy. 2011 Jan 13PMID 21228887. and Braunack-Mayer, 2010176. Braunack-Mayer AJ, Street JM, Rogers WA, et al. Including the public in pandemic planning: a deliberative approach. BMC Public Health. 2010;10:501. PMID 20718996. reported data from the same public engagement activities. Since they had slightly different focuses of the data reported, we included them both.