Objective 1: Developing Consensus on AO

For the purposes of this project inclusion criteria for potential AO must have the following characteristics: be potentially harmful; be available in the AMC, hospital, or associated research facilities; and have a practical means of dissemination. By definition, the Centers for Disease Control and Prevention (CDC) Select Agent list,Reference Rotz, Khan, Lillibridge, Ostroff and Hughes910 Health and Human Services, and the US Dairy Association Select Agents and Toxins list (7 CFR Part 331, 9 CFR Part 121, and 42 CFR Part 73),10 controlled pharmaceutical substances,11 and the Nuclear Regulatory Commission (NRC) Radionuclides of Concern12 are not considered AO because they are already highly regulated. However, if agents are present in amounts less than necessary to make them part of the above lists, such as agents listed on the NRC Radionuclides of Concern, which must be above specific amounts, then they were included as AO.

A multidisciplinary expert working group (EWG) was established from 1 AMC and 4 affiliated teaching hospitals to provide guidance and final consensus on potential AO. The EWG primarily comprised senior School of Medicine faculty with class-specific knowledge, and hazardous materials experts, hospital environmental health and safety personnel, and epidemiologists. Class-specific knowledge refers to faculty members with expertise in infectious diseases and control, radiation physics, medical toxicology, and clinical pharmacology. Panels of this type are used in the development of consensus for many national standards and guidelines. The biological panel also included an infectious disease/infection control physician and a microbiology laboratory director. The chemical and pharmaceutical panels included medical toxicologists and senior clinical pharmacists. The radiological expert panel included a radiation safety officer, a radiation physicist, and a nuclear medicine physician. The expert panel members were provided an AO attribute list (see Objective 2) in advance of the panel discussion and reviewed the attributes during deliberations.

To generate a broad list of potential AO, we e-mailed members of the Environment of Care committees of the 4 participating hospitals, the organizational unit mandated by The Joint Commission to oversee the operational environment of health care settings, and followed with e-mails to people recommended by this group.

The group completed an online survey that allowed each participant the opportunity to nominate up to 5 agents, along with information about the AO location and the participant's opinion of its toxicity (threat potential), availability/dissemination capability (vulnerability), and the impact a release could have on the capacity of AMC operations (consequences).

Independently, in-person interviews of members of the Environment of Care committees were conducted, focusing on the availability and potential use of AO in the AMC, including their location, entry, movement, storage, and removal/disposal. Interviewees (see Supplementary Appendix 1 at http://www.dmphp.org/misc/Online_Appendix_1.pdf) were chosen based on job function, work location, and perceived ability to provide AO information. Each interview included open-ended prompts concerning the chemical biological, radiological, and pharmaceutical agents within the interviewee's purview. In addition, insight into the circumstances under which the potential AO may produce harm to people or the infrastructure was also obtained. The findings were then presented to the EWG for final review and decision.

Table 1 is the final list of AOs determined by the expert panels (our EWGs) to be of the highest institutional risk. AO nominated by these surveys (see Supplementary Appendix 2 at http://www.dmphp.org/misc/Online_Appendix_2.pdf), interviews (see Supplementary Appendix 3 at http://www.dmphp.org/misc/Online_Appendix_3.pdf), and EWG (see Supplementary Appendix 4 at http://www.dmphp.org/misc/Online_Appendix_4.pdf) are comparable to recognized occupational hazards of health care providers, and include chemical, biological, radiological, and pharmaceutical substances.Reference Black and Hultquist13Reference Sullivan and Micale14Reference Wax15 Some of the nominated AOs, particularly disinfectants and sterilants such as ethylene oxide, formalin, glutaraldehyde, and picric acid, have been associated with hospital disaster activations.Reference Burgess16 These nominated agents were considered either not to be extremely harmful or difficult to obtain and disseminate, and some due to their familiarity were believed to possess a minimal psychological effect. Many other rejected AOs were not considered harmful, were not available at participating hospitals, or did not have an effective dissemination mechanism according to our EWG. Biological agents were included due to the recognized experiences of food contamination.Reference Török, Tauxe and Wise4Reference Kolavic, Kimura, Simons, Slutsker, Barth and Haley5Reference Wax, Becker and Curry717 Biological agents are accessible in the microbiology or virology laboratories at AMC due to identification from clinical specimens and storage of specimens for research. Salmonella species were kept on the list because of the prevalence of food contamination. Asbestos, found in many older buildings (likely near mechanical or engineering rooms) including AMCs, provides an easy opportunity for contamination of the air supply in most health care facilities due to accessible dissemination through the HVAC system. Radiological agents, found in the nuclear medicine and oncology suites for clinical testing and specific treatment regimens, are secured for safety purposes but not sufficiently to prevent theft. A release of iodine-125 or other radioactive materials may achieve a quantitative level of medical concern, but the psychological impact on the public could be consequential. Opioids, which are among the pharmaceutical agents of highest concern, are available on almost every hospital floor for the treatment of pain and anxiety and for sedation. Fentanyl was identified as AO because of its potency and toxicity, and if disseminated in a room or through the HVAC system may prove acutely life threatening.Reference Wax, Becker and Curry7 Fentanyl is not only an easily available source but it also has been used in a prior event.Reference Wax, Becker and Curry7 Other agents were excluded from the highest risk listing for similarly specific reasons: warfarin would require multiple contamination episodes for patients to develop grave clinical effects; therefore, the expert panel decided it should not be an agent of highest concern. Clostridium difficile, although hardy, is difficult to grow in a bacteriology laboratory and not likely to be present in adequate supply to be effectively disseminated. Propofol was removed because its mode of dissemination must be intravenous and its use would be limited to individual potential criminal acts and not the large-scale exposures contemplated by AO risk.

TABLE 1 List of Highest Concern to the Expert Panels

Objective 2: AO Risk Profile

AOs were categorized by identifying and quantifying specific agent attributes to describe the risk of a particular AO release by characterizing its threat, understand the AMC's vulnerability, and determine the appropriate management of the consequences, medically (to humans) and operationally (to the building infrastructure).18

Attributes for each AO were compiled into a document known as the agent profile. This profile is designed to provide AMC with rapid access to extensive, detailed, and pertinent information in the event of an AO release, including management of the consequences. We performed a structured literature search in the major biomedical databases (MEDLINE, Embase, Web of Knowledge, Google Scholar) using different variations and combinations of the following search terms: universities, academic medical centers, medical schools, hospitals, health facilities, hazardous substances, agents (chemical, biological, radiological, pharmaceutical), dual-use research, security, risk management, risk analysis, biosecurity, biosafety, access control, bioterrorism, terrorism, disaster, preparedness, warfare, weapons, mass casualties, food contamination, and water pollution. We also reviewed the gray literature (including the DHS, Health and Human Services, CDC, Department of Defense, Food and Drug Administration, RAND Corporation, and major news publications. The search was limited to articles written in English (Table 2 lists the databases used).

TABLE 2 Online Databases Used to Complete Agent Profiles

We identified 3 documents that provided useful strategies based on the characteristics permitting a ranking or classification of AO of a specific class as hazardous, select agents, or “of concern.” These are presented in Table 3.Reference Rotz, Khan, Lillibridge, Ostroff and Hughes9Reference Hauschild and Bratt19Reference Dun, Wood and Martin20 No source document assesses the risk of all of the agent classes, and none evaluated pharmaceuticals. For example, the US Army Center for Health Promotion and Preventive Medicine only assessed and ranked chemicalsReference Hauschild and Bratt19 and the CDC Biodefense Preparedness and Response Working Group on Civilian Biodefense focused only on biological agents, identifying the “select agents” for increased regulations.Reference Rotz, Khan, Lillibridge, Ostroff and Hughes910 The Environmental Protection Agency National Homeland Security Research Center and Decontamination and Consequence Management Division classification scheme addressed both biological and chemical agents and was designed to facilitate the process of decontaminating an area.Reference Dun, Wood and Martin20

TABLE 3 Attribute Comparison From Source Documents

Where the lethality of an agent has been well established, heightened security requirements are required. For example, Select Agent controls for highly lethal biological agents include maximized security for laboratories using such agents. For radiologicals, the NRC guidelines for increased controls for Radionuclides of Concern identify threshold quantities for regulation of security, storage, and transit.12 Similarly, a DHS Chemicals of Interest list indicates threshold amounts and addresses particular security concerns, including theft, as a means to regulate chemicals.21

AMCs often store amounts of chemicals, biologicals, and radiologicals below the CDC, NRC and DHS thresholds, highlighting a potential gap between regulation and reporting standards and institutional risk. The AO risk framework attempted to establish informed risk by evaluating this gap, giving particular attention to the psychological consequences of potentially dangerous agents.

In addition to many agent characteristics, the National Homeland Security Research Center and Decontamination and Consequence Management Division assessed building-specific ranking factors, including access into the building and HVAC systems and size of the rooms. In conjunction with Federal Emergency Management Agency and Department of Defense sources and existing hazard vulnerability analysis frameworks such as the Haddon matrix for pre-event, event, and postevent analysis,Reference Chipley, Kaminskas, Lyon, Beshlin and Hester222Reference Barnett, Balicer, Blodgett, Fews, Parker and Links23 validation was established for the environmental variables (the ability to contaminate via air, food, and/or water supplies or surface exposures) included as a fundamental basis of AO risk and the agent profile.

The agent profile does not rank the risk levels of agents, but rather establishes risk by the harm potential, availability, and vulnerability to an agent. Moreover, the AO framework simultaneously addressed the continuity of operations and the health risks of vulnerable populations in addition to the common decontamination constraints.

The classification strategies and specific attributes of each source document were used to determine the attributes that are essential to the analysis of identified AO regarding threat, vulnerability, consequence, and management, based on the DHS Security Risk Assessment Framework of the National Infrastructure Protection Plan.18 The final attributes list can be identified as the 16 boldface items in Table 4 (see Supplementary Appendix 5 at http://www.dmphp.org/misc/Online_Appendix_5.pdf for a blank example of an agent profile).

TABLE 4 Agent Profile AO Attributes (Bold)

The agent profile was grouped into 3 sections (threat, vulnerability, consequence) consistent with the National Infrastructure Protection Plan risk management framework.18 Threat requires that the AO is known to be present in the institution. Threat is heightened if the AO has resulted in an event, either intentionally or unintentionally in the past (historical/prior event). Vulnerability includes the information necessary to describe how the AO may be accessed, the required route(s) of their dissemination, and the means by which exposure and transmission to others can occur. Consequence, indicates the AO's morbidity and mortality, toxicity, and psychological impact (public perception). A fourth section unique to the agent profile, consequence management, provides information about the detection, identification, means of decontamination of the exposed person, and treatment available to the patient. Included in this section of the agent profile is also the type of precautions (standard, droplet, airborne, lead aprons) that health care providers must use to prevent exposure to themselves so that they can continue to care for patients. Other attributes that are present in the agent profile enhance the description of the AO and include physical state, persistence, and latency.

Risk assessment and subsequent management requirements need to be integrated across multiple agent classes. By including all of the agent classes in a single framework, it may become possible to assess AO risk across class types and better understand how to communicate AO risk, mitigation, and response in disaster planning to a variety of occupations and professions. The determination that the AO is a chemical, biological, radiological, or pharmaceutical allows emergency personnel and hospital administration to determine which experts need to be contacted if information is not available in the agent profile.

Merely recognizing an agent as “toxic” is not sufficiently robust to understand and mitigate risk or to manage consequences. The agent profile will provide hospital administrators and first responders with a broad overview of a particular agent and serve as an instrument to identify specific approaches to AO management, information important to have immediately available should an AO event occur.

The inability to determine a rank order of severity or risk results from lack of information in several areas. Because the quantity of an AO and the institutional characteristics (eg, security measures) will differ by institution, producing quantitative designations was not practical. Other obstacles to ranking included an inability to determine what combination of attributes makes 1 agent more harmful than another; the uncertain utility of specific doses such as LD₅₀ commonly used to convey chemical toxicity; the diverse ways to deliver or disseminate an AO that may alter potential morbidity and mortality; lack of information on final concentrations of AO after dissemination due to lack of prior events, and lack of predictive models; diverse occupational exposure levels for chemicals and pharmaceuticals; and the exceptional uncertainty for contamination by biologicals and radiologicals (eg, minimal bacilli for Mycobacterium tuberculosis expressionReference Riley and Mills24). Most substances have not been modeled for human exposure after dissemination during an inadvertent or purposeful release. Thus, a threshold exposure over time producing a toxicity threshold could not be defined.

The potential psychological impact of an exposure may be 1 of the most important predictors of ultimate institutional impact.Reference Butler, Panzer and Goldfrank25 Agents with limited toxicity are categorized as AO due to their “toxic connotation” to the public and health care employees who have inadequate health literacy or substantial uncertainty with regard to a complex situation. The agent profile incorporates the psychological and behavioral impact for 2 reasons: psychosocial impact will be greater with an uncertain toxic connotation and terrorists may exploit such uncertainty and use it to their advantage. The significant contribution of the agent profiles is the integration of chemical, biological, radiological, and pharmaceutical substances as well as pre-event, event, and postevent management recommendations.

Objective 3: Implications for Public Health Preparedness

Our EWG was able to derive a consensus-based framework and risk profile for AO in AMC. However, this should be viewed as only the beginning of this effort. Our research also uncovered a dearth of practical information regarding the physicochemical, pharmacokinetic, and environmental parameters of many of the nominated AO, even after an extended literature review and discussions with national AOs experts (nonpanel members). This was generally due to the lack of prior AO dissemination events or adequate modeling to allow prediction. This is an important finding, strongly suggesting that specific information on AO toxicity may not be available (publicly or through government or military data sources). Because of the need to make assumptions regarding issues such as quantities released, exposure doses, and concentrations after dissemination, a rank order could not be determined with any level of certainty. Furthermore, local experts may find it difficult to react to a situation with an indeterminate risk level, particularly those with limited information. Due to lack of prior experience and precedent and of adequate mathematical or computer modeling, the experts were unable to provide information regarding the level of risk for human occupants or the buildings. In addition, federal and local agencies (eg, Federal Bureau of Investigation, CDC, Department of Health) will likely assess the AMC as a crime scene and assess the level of building contamination to determine whether closure or decontamination is warranted. Given that unintentional local spills or contamination can result in a section of an AMC building being closed and evacuated,Reference Burgess16 purposeful dissemination of AO through a large area of an AMC building would likely result in a similar outcome on a larger scale.26

The findings of this project may provide a foundation for a discussion and consensus efforts to determine a nationally accepted risk profile framework for AO. This would additionally lead to the implementation of appropriate regulatory policies to improve public health preparedness.

As an additional tool, a modified Haddon matrix was developed to identify ways to prevent or mitigate the impact of a potential AO release (Table 5; see Supplementary Appendixes 6 and 7 at http://www.dmphp.org/misc/Online_Appendix_6.pdf and http://www.dmphp.org/misc/Online_Appendix_7.pdf for examples). The Haddon matrix is used in injury prevention to assist in the identification, organization, and classification of factors that contribute to an event. The factors are placed in categories determined by their relation to the host (human victim or hospital building), agent (AO, or damaging mechanism), physical environment, and social environment surrounding an event, and sorted by the time in relation to an event—pre-event, event, and postevent. By analyzing each factor (eg, agent) at a particular time (eg, pre-event), the Haddon matrix has been used to analyze terrorist events and identify best-practices strategies to prevent or mitigate the severity of the event.Reference Varney, Hirshon, Dischinger and Mackenzie27Reference Arnold28

TABLE 5 Sample Haddon Matrix for Aerosolized Infectious Agent Placed in Air Distribution System

Interventions to reduce the overall risk in AMC would primarily address the availability and accessibility of AO on site and means of dissemination, implying a need for more substantial readiness and security measures. The accompanying article in this issue offers further details with regard to risk mitigation involving people, engineering, and security efficacy.Reference Graham, Tunik and Farmer29

As part of this project a series of 4 pilot tabletop exercise workshops were held to provide insight into the state of preparedness and the potential responses of individuals and administrators working in AMC to a threat scenario. Agents of opportunity in all 4 categories were chosen (chemical, biological, radiological, pharmacological) and various dissemination mechanisms were used in the threat-scenario workshops. Although there were strong differences of opinion expressed by the participants, it was clear after the workshops that although a particular AO may not be highly dangerous, there was uncertainty about the actual risk; notably, it could not be stated that there was no risk. The fear generated by the AO release in patients, their families and the public—the “toxic connotation” of an AO—was viewed as highly important and would result in significant consequences, such as AMC closure. This psychosocial impact strongly influenced the discussions and many of the decisions made during the workshops. The following were the themes of the threat scenario workshops:

• 1. The AO is sufficiently toxic to humans, and contamination of the building is of concern.

• 2. The AO and the dissemination mechanism are inadequately secured, creating a vulnerable situation.

• 3. Confirmation of the theft and confirmation of the dissemination of the AO are important facts that are necessary to determine actions.

• 4. Dissemination of AO creates a contamination risk to the building, sufficient to require decontamination or evacuation.

• 5. The purposeful AO dissemination would cause the hospital to be declared a crime scene, placing decision making as to how to manage the hospital in the hands of federal agencies, such as the Federal Bureau of Investigation.

• 6. After dissemination the AO exist in sufficient concentrations to be toxic to humans.

• 7. The fear generated by knowledge of the disseminated AO is sufficient to cause disruption of hospital function, decrease patient use of the hospital, or closure of the hospital.

• 8. The hospital's reputation would suffer as a result of widespread knowledge of the occurrence of an AO dissemination event

• 9. Security measures exist that can prevent or mitigate the severity of an AO dissemination event.

• 10. The process of better securing AOs and dissemination mechanisms is costly, and hospital budgets are restricted.

• 11. An attack on a hospital is a low-risk event, and spending time and funds to prevent the event is unwarranted.

• 12. Increased security decreases the ability to care for patients and/or the ability of researchers to use AO and perform research.

Next Steps

The EWG seeks opportunities to vet this framework through an august national body such as the Institute of Medicine. The EWG is also going to use and pilot a computerized data collection and reporting instrument for AO, which may provide greater data of use and misuse. This could also aid in modeling and simulation scenarios.