Report Summary

With funding from the Agency for Healthcare Research and Quality (AHRQ), we developed the Hospital Surge Model, which estimates the amount of hospital resources needed to treat casualties of major disasters. The Hospital Surge Model is available at http://hospitalsurgemodel.ahrq.gov.

The Hospital Surge Model includes ten different scenarios:

• Biological (anthrax, smallpox, and pandemic flu).
• Chemical (chlorine, sulfur mustard, or sarin).
• Nuclear (1 KT or 10 KT explosion).
• Radiological (dispersion device or point source).

Three additional scenarios—plague, food contamination, and a conventional explosive attack—are under development and will be included in the next version of the Hospital Surge Model.

When the Hospital Surge Model is run, the user selects one of the above scenarios and specifies the number of casualties that their hospital(s) will need to treat. Casualties are treated, as necessary, in the emergency department (ED), in the intensive care unit (ICU), or on a general medical/surgical bed ward ("the floor"). Hospitals are assumed to have unlimited capacity and provide a standard level of care to all casualties—that is, the Hospital Surge Model assumes that care is not degraded by the surge in patients or by resource constraints. Eventually, casualties in the model are either discharged or die in the hospital(s). While patients are in the hospital(s), the Hospital Surge Model estimates the amount of resources (e.g., personnel, equipment, supplies) they require.

For the selected scenario, the Hospital Surge Model estimates:

• The number of casualties in the hospital(s) by hospital unit (ED, ICU, or floor) and day.
• The cumulative number of dead or discharged casualties by day.
• The required hospital resources (personnel, equipment, and supplies) to treat casualties by hospital unit and day.

This document is a description of the Version 1.2 of the Hospital Surge Model and its underlying assumptions. A companion report—the Hospital Surge Model User Manual—provides instructions on how to run the Hospital Surge Model. The remainder of this report is organized as follows:

• An overall description of the Hospital Surge Model and the Hospital Surge Model Web site.
• Chapters 1 through 9 discuss the assumptions for the individual scenarios.
• References are cited in the individual sections.

The Hospital Surge Model is a modified version of the original Surge Model, developed by the project team in 2005. A project steering committee guided the original Surge Model project—go to the list of members in the appendix.

Model and Web Site Overview

This section provides an overview of the Hospital Surge Model and how it has been implemented on the Web site http://hospitalsurgemodel.ahrq.gov. Following this overview are nine chapters (Chapters 1 through 9) that discuss the details of each of the ten scenarios.¹

Exhibit 1 shows the overall structure of the Hospital Surge Model. On the Hospital Surge Model Web site, the user specifies a number of different inputs, including a scenario (e.g., anthrax or smallpox) and the number and/or type of casualties assumed to present at their hospital(s).

Exhibit 1: Hospital Surge Model Structure

Users specify the number and/or type of casualties arriving at their hospital(s). The model then determines when these casualties present at the hospital(s) and estimates the hospital resources required to treat the casualties. The Hospital Module section provides an overview of the Hospital Module, which tracks the movement, treatment, and outcomes of patients in the hospital. The Hospital Module incorporates numerous assumptions, including:

• The length of stay by hospital unit (ED, ICU, or the floor).
• The probability that a given patient will be transferred between hospital units (i.e., "transition probabilities").
• The overall outcome probabilities (i.e., probability of discharge and probability of death).
• The assumed level of resource consumption per patient per day.

Go to Chapters 1 through 9 for specific assumptions for each scenario. References are cited for those assumptions that are based on empirical evidence. In the absence of empirical evidence, project staff with clinical experience have made their own estimates, which clearly require future validation. Project staff are continually updating each assumptions as new empirical data or expert opinion are obtained.

The Hospital Surge Model outputs, including details about the casualties, resources, and requirements, are summarized in the Hospital Surge Model Outputs section.

1. The 1 KT and 10 KT nuclear scenarios have been combined into one section.

User Inputs

When the Hospital Surge Model is run, the user selects one of ten scenarios; they include biological, chemical, nuclear, and radiological scenarios. After selecting one of the scenarios, the user must specify the number and (optionally) the type of casualties that will present at their hospital(s).

Scenarios

Biological

• The covert release of aerosol Bacillus anthracis through a city.
• The covert release of smallpox virus in a large indoor theater.
• A pandemic flu outbreak that sweeps across the country.

These three scenarios include both communicable and non-communicable diseases, a disease that is treatable with antibiotics and two that are not, and indoor and outdoor attacks.

Chemical

• The overt explosion of a liquid chlorine tank in a densely populated suburb.
• The overt release of the blister agent sulfur mustard at a large outdoor event.
• The semi-covert release of the nerve agent sarin in a crowded arena.

These three scenarios include both indoor and outdoor scenarios and include toxic industrial chemicals, blister agents, and nerve agents. Blood agents are not included because of the perceived difficulty in using blood agents to inflict mass casualties in a non-confined space; choking agents are not included because they are only lethal in extremely large quantities.

Radiological/Nuclear

• The overt detonation of 1KT of nuclear material (improvised nuclear device) in a city center.
• The overt detonation of 10KT of nuclear material (improvised nuclear device) in a city center.
• The overt release of cesium-137 dust (commonly referred to as a "dirty bomb" or radiological dispersal device) in a city center.
• The covert placement of a cesium-137 source in a public place.

These four scenarios cover the threat space posed by radionuclides, which can be dispersed in an extremely violent reaction (nuclear), more gently (radiological dispersion device, or RDD), or not at all (a radioactive point source).

Exhibit 2 shows how the user selects a scenario on the Hospital Surge Model Web site—http://hospitalsurgemodel.ahrq.gov.

Exhibit 2: Scenario Selection Screen

When the "Next" button is clicked, the user is asked to specify the number and (optionally) type of casualties that will present at their hospital(s).

For each of the scenarios, users have two options for specifying casualties:

• Specify the number and type of casualties. For example, users could specify that their hospital(s) receives 100 mild, 20 moderate, and 10 severe cases of sarin; or.
• Specify the number of casualties. For example, users could simply specify that their hospital(s) receives 130 cases of sarin. In this case, the casualties are randomly selected from among all casualties from the attack.

The table below lists the type of casualties users can specify for each scenario:

Scenario	Types of Casualties
Anthrax	Mild: When victims become ill they are more than 3 days away from death, assuming no treatment. Severe: When victims become ill they are 3 or fewer days away from death, assuming no treatment.
Smallpox	Onset: mild, generalized symptoms.
Pandemic Flu	Moderate: Patients with moderate flu symptoms go directly to the floor for treatment. Severe: Patients with severe flu symptoms go directly to the ICU for treatment.
Chlorine	Irritated: burning in eyes or respiratory system, exposed to a detectable odor. Incapacitated: intolerable irritation of respiratory system and lungs.
Mustard	Irritated: hoarseness or burning in throat and lungs, irritation in eyes. Severe: temporary blindness, permanent eye damage, bronchopneumonia and skin damage.
Sarin	Mild: miosis, ocular pain, tearing, rhinorrhea, bronchospasm, slight dyspnea, respiratory secretions, salivation, diaphoresis. Moderate: moderate dyspnea, nausea, vomiting, diarrhea. Severe: loss of consciousness, convulsions, paralysis, copious secretions, apnea.
Nuclear device	Burns/moderate: 2nd degree burns on hands and face. Burns/severe: 3rd degree burns on hands and face (and 1st degree over the rest of the body). Trauma/people in collapsed skyscrapers . Trauma/people in collapsed houses and other light buildings. Trauma/people who receive multiple lacerations from flying glass . Trauma/people outside who receive blunt trauma . Radiation/mild: nausea, vomiting, anorexia, fever, infections . Radiation/moderate: mild symptoms as above, plus bleeding, fatigue and weakness. Radiation/severe: moderate symptoms as above, plus headache, prostration, dizziness and disorientation . Fallout/mild: 1Gy equivalent dose for blood effects, no other equivalent dose (some bleeding and infection issues). Fallout/severe: 4Gy for blood effects (problems with bleeding and infection) and about 0.75 Gy for lethality and GI effects (some small fraction of people will die, and others get nausea, vomiting, etc.).
Radiological dispersion device (RDD)	Mild: 1Gy equivalent dose for blood effects, no other equivalent dose (some bleeding and infection issues). Severe: 4Gy for blood effects (problems with bleeding and infection) and about 0.75 Gy for lethality and GI effects (some small fraction of people will die, and others get nausea, vomiting, etc.).
Radiological point source	Mild: nausea, vomiting, anorexia, fever, infections. Moderate: mild symptoms as above, plus bleeding, fatigue and weakness. Severe: moderate symptoms as above, plus headache, prostration, dizziness and disorientation.

Hospital Module

The Hospital Module calculates resource requirements in different units (ED, floor, and ICU) of a hospital for each scenario, given the number and type (i.e., severity of injury) of patients arriving at a hospital.

Exhibit 3 below shows the different ways in the Hospital Module that patients can move between different units in a hospital. Patients arrive at the ED of a hospital. The model groups them according to severity of their injury. The model assumes they spend a certain amount of time in the ED (dependent on severity). After the end of this period of stay, either they are discharged or they enter the floor or the ICU. On the floor, if their conditions worsen, they could be sent to the ICU. Otherwise, after a predetermined length of stay (depending on severity), they are discharged. Patients in the ICU recover after a predetermined length of stay (again, depending on the severity) and are transferred to the floor. These patients are assumed to not enter the ICU again. Patients can die in the ICU or on the floor. Lengths of stay, transition probabilities, and death rates are assumed to be fixed (i.e., the Hospital Module is a deterministic, rather than random, model). However, different cases corresponding to different values of lengths of stay and death rates are considered. Assumptions regarding these data are in Chapters 1 through 9.

Exhibit 3: Hospital Module System Diagram

Methods

The hospital module sends all arriving patients through the system, transferring them from one hospital unit (i.e., the ED, the ICU, or the floor) to another according to the probabilities and assumed lengths of stay, and assuming certain mortality rates (see sections 3 through 11). All patients will eventually exit a hospital either due to death or by being discharged. Probabilities of death per hospital unit are compounded daily. Once we have a count of how many patients are present in each hospital unit in each period, we can determine overall resource requirements by looking up a database of resource requirements per patient.

Data

A database drives the hospital module. It contains the following information for each scenario:

• A list of resources.
• Daily requirements for each resource per patient in each hospital unit (i.e., the ED, the ICU, or the floor).
• Length of stay in each hospital unit.
• Mortality rates in each hospital unit—these are expressed as the probability that a patient in that hospital unit will die in the next period, given that he has survived until that period.
• Transition probabilities between hospital units (e.g., the probability that a patient of a certain severity will move from the ED to the ICU, as opposed to the ED to the floor).

While a patient is hospitalized, some resources are used at a constant rate; for other resources, a decreasing amount is used each day. For example, while a patient needs full use of a bed every day s/he is hospitalized, the amount of time physicians and other health care staff must devote to a single patient is likely to decrease each day that a patient is in the hospital. The daily "decay" rate for resource utilization is captured in a model variable called lambda. The following table defines "time to < 50% of use" (i.e., time to use of half of listed amount of resource) and "time to <10% of use" for different values of lambda.

<50%	Lambda	<10%
	0.1	Day 3
Day 2	0.3
	0.5	Day 4
	0.6	Day 5
Day 3	0.7	Day 6-8
Day 4	0.8
Day 5	0.85
Day 6	0.875
Day 7	0.9

In this initial version of the Hospital Surge Model, we have included those resources that are viewed as the most critical in a hospital setting for treating casualties of major disasters. We have not attempted to include all possible resources, although future versions of the Hospital Surge Model will include additional resources. The specific resources we have included for each scenario are listed in Chapters 1 through 9.

General References

Scenario-specific references used to estimate parameters in the Hospital Module are noted in Chapters 1 through 9. In the absence of specific references, estimates for lengths of stay, transition probabilities, path routing through a hospital, and resource utilization were obtained from a combination of general references (see below), general references for each overall scenario type (noted in Chapters 4 through 9 for chemical and radiological scenarios), and scenario-specific references.

1. Braun BI, Wineman NV, Finn NL, et al. Integrating hospitals into community emergency preparedness planning. Ann Intern Med 2006;144(11):799-811.

2. Brilli RJ, Spevetz A, Branson RD, et al. Critical care delivery in the intensive care unit: defining clinical roles and the best practice model. Crit Care Med 2001;29(10):2007-19.

3. Connelly LG, Bair AE. Discrete event simulation of emergency department activity: a platform for system-level operations research. Acad Emerg Med 2004;11(11):1177-85.

4. Fone D, Hollinghurst S, Temple M, et al. Systematic review of the use and value of computer simulation modeling in population health and health care delivery. J Public Health Med 2003;25(4):325-35.

5. Halpern NA, Pastores SM, Greenstein RJ. Critical care medicine in the United States 1985-2000: an analysis of bed numbers, use, and costs. Crit Care Med 2004;32(6):1254-9.

6. Halpern NA, Pastores SM, Thaler HT, et al. Changes in critical care beds and occupancy in the United States 1985-2000: differences attributable to hospital size. Crit Care Med 2006;34(8):2105-12.

7. Hick JL, O'Laughlin DT. Concept of operations for triage of mechanical ventilation in an epidemic. Acad Emerg Med 2006;13(2):223-9.

8. OSHA. OSHA Best Practices for Hospital-Based Receivers of Victims from Mass Casualty Incidents Involving the Release of Hazardous Substances. Occupational Safety and Health Administration, U.S.; 2004.

9. Rubinson L, Nuzzo JB, Talmor DS, et al. Augmentation of hospital critical care capacity after bioterrorist attacks or epidemics: recommendations of the Working Group on Emergency Mass Critical Care. Crit Care Med 2005;33(10):2393-403.

10. Shapiro DS. Surge capacity for response to bioterrorism in hospital clinical microbiology laboratories. J Clin Microbiol 2003;41(12):5372-6.

Hospital Surge Model Outputs

On the Hospital Surge Model Web site, the output from the Hospital Surge Model is organized into five sections: summary results, casualty arrival pattern at the hospital(s), number of patients in the hospital(s), number of dead and discharged patients, and resource requirements.

Summary results

Two summary measures from the daily results are displayed:

• The day that the most patients arrive at the hospital(s), and the number of patients that arrive on that day.
• The day that the most patients are in the hospital(s), and the number of patients that are in the hospital(s) on that day.

Casualty arrival pattern at the hospital(s)

The casualty arrival pattern at the hospital(s) is shown in graphical and tabular form. Casualties are assumed to arrive at your hospital(s) when symptoms present. For the nuclear and chemical attacks, casualties arrive at your hospital(s) immediately after the attack (i.e., on Day 1). For the biological and radiological attacks, there is a delay between exposure and when symptoms present. The Hospital Surge Model computes the delay for these attacks and generates an arrival pattern for the scenario. Exhibit 4 shows an illustrative arrival pattern for the anthrax scenario.

Exhibit 4: Illustrative Hospital Arrival Pattern

Number of patients in the hospital(s) by day

The number of patients in the hospital(s), by hospital unit and by day, is displayed in a table and in a graph (see Exhibit 5). The assumptions regarding lengths of stay, transition probabilities between hospital units, and mortality rates for the different scenarios are listed in Chapters 1 through 9.

Exhibit 5: Illustrative Number of Patients in the Hospital(s)

Number of dead and discharged patients

The Web site shows, in both graphical and tabular form, the number of patients in the hospital(s), the cumulative number of deaths to date, and the cumulative number of patients discharged from the hospital(s) to date (see Exhibit 6).

Exhibit 6: Number of Dead and Discharged Patients by Day

Resource requirements

While patients are in the hospital(s), they require resources (e.g., medical personnel and equipment) and consume resources (e.g., medical supplies). (As noted earlier, the Hospital Surge Model assumes the hospital(s) has unlimited resources.) The Web site displays the daily resource requirements for patients by resource, day, and hospital unit.

The following information is displayed for each resource:

• Resource name.
• Resource unit. The units of the resource requirements are: FTEs, machine time, and unit of use.
  • Machine time refers to the amount of time needed per patient per day on a diagnostic machine, such as a CT scanner.
  • Unit of use is a generic term for a daily dose or other definable amount of a consumable resource such as a medication or a set of laboratory reagents. For example, the unit of use for antibiotics for anthrax would be two 400mg doses of intravenous ciprofloxacin or two 100mg doses of intravenous doxycycline plus one or two additional antibiotics.
• Resource category and subcategory.
• Consumable (yes/no).
• Day of peak use. The day that the greatest amount of the resource is required.
• Amount of use on peak day. The amount of the resource that is required on the peak day.
• Daily requirements, as computed by the model.

Daily requirements are displayed in a table (see Exhibit 7). Requirements for individual resources can be graphed (see Exhibit 8).

Exhibit 7: Resource Requirements (Tabular Display)

Exhibit 8: Requirements for a Specific Resource (Graph Form)

Return to Contents
Proceed to Next Section