Introduction
The public health impact of accidental or deliberate release of chemical, biological, radiological, or nuclear agents (CBRN) can be considerable.Reference Falkenrath1 The Union Carbide disaster in Bhopal in 1984 and Sarin gas attacks in Japan in 1995 illustrated not just the potential for large numbers of casualties in the acute phase, but also the burden of adverse health effects among exposed persons in the longer term.Reference Yanagisawa, Morita and Nakajima2, Reference Dhara and Dhara3
Public health agencies have a responsibility to plan for, prepare, and respond to incidents causing widespread population exposure to harmful agents.Reference Hyams, Murphy and Wessely4 In order to systematically study the short- and long-term health impacts and outcomes following exposure to CBRN agents, a health register of all potentially exposed persons is essential. Indeed, health registers have been used to study health outcomes of individuals exposed in recent major incidents.Reference Brackbill, Thorpe and DiGrande5-Reference Stather9
The rationale for deciding when and whether to establish a health register is not straightforward. Such a decision needs to be made within the first few hours to days of a major incident, since retrospective collection of registry data is more difficult and could lead to substantial loss of information.Reference Brackbill, Thorpe and DiGrande5 However, the resource implications for initiating and maintaining a health register may be considerable, and the potential benefits may not always be clear at the outset. Importantly, not all major incidents require that a register to be set up. While there are a number of drivers for initiating a health register (eg, scientific uncertainty or political pressures), no objective criteria or decision frameworks are currently available to enable rapid decisions in a scientific and transparent manner. This study aimed to develop a decision framework that could be used by key UK Health Protection Agency (HPA) experts and other stakeholders in considering the need for a health register in a major incident involving CBRN agents.
Methods
The study comprised three stages. In the first stage, the study team (authors of this paper) broadly considered all potential criteria relevant for setting up a health register of individuals exposed in major incidents. A list of statements was prepared on the basis of a literature review, supplemented by discussions and consultations among the study team, and piloted among a small number of staff working in the HPA.
In the second stage, the list of criteria statements was evaluated in two Delphi rounds involving a larger group of 16 experts from the HPA and academic organizations. The Delphi process is a well-established technique to support consensus building in difficult areas of public health decision making.Reference Jones and Hunter10 The process is conducted using a group of experts to provide feedback on statements by mail in two to three rounds and by making multiple iterations to the statements based on the response in each round. Key stakeholders were chosen on the basis of two principles: (1) they were recognised experts in chemical, radiological, or infectious incidents; or (2) they were senior decision makers with a position of authority and influence in emergency preparedness and response.Reference Rauch11 The involvement of senior decision makers was aimed at familiarizing them with this aspect of emergency planning, as well as enabling easier implementation of the agreed upon criteria if put to use at a later date. Twenty-four people were approached, of whom 16 agreed to participate. The remaining eight declined, did not reply, or were unable to be contacted. Study participants included expert or senior staff from the HPA, partner organizations, and academic institutions within the UK.
In the first Delphi round, participants were asked to express their level of agreement with each statement using a 9-point Likert scaleReference Likert12 (Table 1). To assist participants in evaluating the relevance of each statement, seven fictional scenarios of major incidents were provided as examples. Participants were encouraged to provide comments on the proposed criteria and to suggest other new criteria that may be of relevance. Consensus for a statement was defined by the study group as the achievement of a mean Likert score of 6 and above; a score below 6 indicated neutral or negative relevance to the decision process. Data were collated and descriptively analysed using Microsoft Excel (2007 version, Redmond, Washington, United States).
Table 1 Example of Delphi Statement in Round 1
In the second Delphi round, the original statements were presented to study participants together with a summary of the rest of the group's scores (Table 2). The individual participant's original score in Round 1 was highlighted in bold and underlined as shown. All 11 statements were accompanied by additional comments to improve clarity, provided on the basis of feedback from the first round. Participants were advised that these new comments would be incorporated into each statement at the end of the second round. Participants were given the option of changing their scores after considering the opinions of the rest of the group.
Table 2 Example of Delphi Statement in Round 2
At the end of the second Delphi round, the study team reviewed the scores and comments provided by participants. Statements that achieved consensus were used to develop a structured decision-making framework.
In the final stage of the study, the decision framework was circulated to all participants, and comments were invited. In addition, participants were asked to pilot test the framework using four fictional scenarios. For each fictional scenario, participants were asked to consider each statement in the decision framework for relevance and to make a conclusive decision on the need for setting up a health register. Finally, the responses were considered by the study team if they necessitated further changes to individual statements or the overall decision-framework.
Results
Stage One: Development of Criteria
Based on literature review and internal consultation, the study group prepared a list of key attributes of a major incident that are crucial in the decision to set up a health register. As published literature on this topic was sparse, the study group predominantly was guided by personal experiences in previous major incidents and the lessons learned from non-register based studies on significant major incidents involving CBRN agents across the world. The attributes critical in the decision to set up a health register are discussed below.
Novel Exposure and Limited Knowledge of Health Effects
Health registers traditionally have been set up following exposure to novel or unknown agents during major incidents. Recent examples include health registers following the World Trade Center attacks in New York, the London bombings, the Buncefield (UK) fire, and the Prestige oil spill in Spain.Reference Brackbill, Thorpe and DiGrande5-8 In all these instances, the main objective was to understand exposure-related health effects that were complex and difficult to predict at the outset.
Defining Target Population
In general, a major limitation of registers is that any association between exposure and health can be documented only to the extent that the exposure was considered and assessed when setting up the register in the first place.Reference Bongers, Janssen and Reiss13 As a consequence, the ability to define the population at risk in terms of likely exposure of interest is a key criterion for setting up a register.
Effects on Vulnerable Groups
The effects of novel or unknown agents on the exposed population may not be immediately predictable due to gaps in current knowledge, as is frequently the case for pregnant women and children.Reference Bhandari, Syal and Kambo14-Reference Pesatori, Consonni and Rubagotti15
Health Benefits
Systematic follow-up studies of exposed persons should be undertaken in circumstances and ways that will improve our understanding of immediate and long-term health effects due to novel exposures, which may not be practical or ethical to study in laboratory investigations or clinical trials.Reference Yanagisawa, Morita and Nakajima2, Reference Lo, Chan and Chen16-Reference Bromet and Havenaar19 Such knowledge may benefit the clinical management of the exposed persons and increase the ability to reduce harm by limiting exposure to a much wider population in the future.
Context
Scientific reasons are paramount in the decision to set up a health register following a major incident. However, it is acknowledged that political or public concerns must also be taken in to account.Reference Ackermann-Liebrich, Braun and Rapp20-Reference Traupe, Menge and Kandt22 The need to provide reassurance to the general public that the particular exposure does not have significant long-term health effects may necessitate the setting up of a health register in some instances.
Feasibility
In addition to all the above, it is important to take into account practical considerations in setting up a health register, such as the ability to recruit all exposed in a health register, to collect adequate information on exposed persons, and to address uncertainties regarding the assessment of exposure by geographical or clinical information.7, Reference Bongers, Janssen and Reiss13
The key themes that emerged from the review of literature and personal experiences were put together in the form of 11 statements. After piloting with a small number of HPA staff, the statements were distributed to the wider expert panel in the second stage of the study.
Stage Two: Agreeing Upon the Criteria
Participant scores in the two Delphi rounds are summarized in Table 3. At the end of the first Delphi round with 16 participants, eight statements attained a mean Likert score of 6 and above. The remaining three statements had mean scores of 5.6, 5.7, and 5.9. Based on the comments made by study participants, additional clarity was provided for each statement. The original statements were not amended at this stage to avoid confusion and to help study participants reassess their first-round scores in light of summary group scores. All 11 statements achieved a mean score of 6 and above in the second Delphi round completed by thirteen participants, thus confirming consensus. Of note, except for statement 5, the mean scores of statements were higher in the second Delphi round compared with the first round.
Table 3 Summary of Scores of Statements in Delphi Round One and Round Two.
aNote that some of the statements were significantly changed between Round One and Round Two.
Stage Three: Testing the Criteria
The 11 agreed-upon statements were converted into a decision framework (Table 4) by the study team. Thirteen participants applied the decision framework to the following four fictional scenarios and decided whether a health register should be set up.
Table 4 Decision Framework for Setting Up Health Registers After a Major Incident
aThe availability of complete information is not essential at the decision stage as most of the relevant data can be gathered at a later date based on preliminary information on exposure.
Scenario 1
It is a Saturday in August on the first day of the summer sales. An unknown terrorist group call Scotland Yard and the media groups claiming that they have released an Agent X in three locations along Oxford Street. There are an estimated 750,000 people in or around the area shopping and working, including many people from overseas (25%) and outside London (21%). The news spreads quickly and panic ensues, with some shoppers complaining of health symptoms, including difficulty breathing, nausea, dizziness, blurred vision and headaches. It is also a poor air quality day, with very high temperatures. People start self-presenting to hospitals in central London and the suburbs, and within a few hours turning up in emergency departments in Manchester, Southampton and Hemel Hempstead.
Scenario 2
Following an accident at a nuclear reactor in a neighbouring country, a radioactive cloud is blown over much of South and East of England, including Greater London and the surrounding area, before moving out over the North Sea. Environmental measurements identify the radionuclides released from the reactor and the level of deposition at various locations across south-east England. Foodstuffs and drinking water are also monitored. These measurements all indicate that radiation exposures to the population in the affected areas would equate to a small percentage of the exposure received annually from natural radiation. In the days following the incident, several thousand people ring NHS Direct [the UK National Health Service], expressing concern and reporting a multitude of symptoms.
Scenario 3
Between September and November, reports are received about an illness affecting a large town (population 100,000) in the south of England. About 100 individuals, mostly young adults and adolescents, have presented at the local hospital complaining of shortness of breath, chest pain and dizziness. The Cardiologist in the hospital thinks this may be myocarditis (inflammation of the heart muscle), which can cause long-term damage to the heart. However, some of his colleagues dispute this diagnosis. To date, all microbiological, virological and toxicological investigations have proven inconclusive. Further cases have recently been reported from neighbouring villages. There has been one sudden death in a young male. Autopsy results are not yet available.
Scenario 4
It is a Sunday in March at 11:00, and a large explosion and fire(s) is reported having occurred at a large chemical site near Beckton in East London. The fire continues for 6 days, releasing a cocktail of chemicals into the atmosphere, as well as dirty fire water, which the Environment Agency has not managed to contain. It is thought that as well as contamination reaching the Thames, it may have affected the nearby sewage works. The following week the Local Authority identifies that a nearby storage unit containing approximately 200 tonnes of endrin [an organic solid insecticide], awaiting shipment to a disposal facility in Nigeria, had been damaged by the fire. It is thought that some of the firewater [highly polluted water remaining after the fire has been extinguished] escaped to a section of a canal connecting the Barking Creek with the Grand Union Canal. The amount of endrin that contaminated the canal sediment is currently unknown. There have been several calls to the Local Authority from local residents complaining of headache and dizziness. Given the persistence of endrin, eventual seepage to the groundwater table is likely.
Among the thirteen participants who tested the criteria using four fictional scenarios, one provided feedback on one scenario only, as this participant felt the other scenarios were beyond the participant's area of expertise. In four instances, participants were unable to decide on the need for a health register citing lack of critical information. As can be seen from the results summarized in Table 5, the decision framework enabled good consensus in three scenarios (Scenarios 2, 3 and 4). The uncertainty regarding the exposure (Agent X) in Scenario 1 was the probable reason for absence of agreement, as evidenced by the comments made by participants.
Table 5 Summary of the Decisions of Participants on Whether to Set Up a Health Register in Four Fictional Scenarios
Summary of Comments from Participants
While overall consensus was achieved for all 11 statements, some points of debate raised by participants merit further discussion.
Some participants commented that the feasibility of setting up a health register was not a significant factor in the decision process. They felt that if there were overwhelming scientific reasons, all efforts should be made to set up a health register regardless of the potential difficulties and limitations. However, others suggested that feasibility and need were interrelated and closely linked to the value of a health register; therefore, both were important considerations in the decision process. Others stated that financial costs and resource implications were important considerations, as setting up and maintaining a health register might involve commitment of significant resources.
Opinion was divided on the need to accurately measure or estimate exposure levels for persons included in a health register, with some arguing that availability of detailed exposure data was essential and others suggesting that such information could be obtained at a later period by clinical history or biomonitoring methods. The latter group pointed out that in some situations, the occurrence of unique acute or long-term effects in a person included in the health register may suggest a degree of exposure that was not measured or identified previously.
Some participants observed that when a health register was set up on political grounds due to public or media pressure and not on scientific rationale, the decision should be made transparently and explained to the public. Others argued that pressure from the public and media usually indicate a significant incident, necessitating appropriate investigation by the relevant authorities with the responsibility to provide expert advice to the public on health outcomes. Scientific uncertainty in itself may exacerbate illness in the exposed persons and a health register may be a crucial tool to study outcomes systematically and provide reassurance to the public.
Another participant highlighted that in major incidents, the potential for psychological effects should not be underestimated, and that the likelihood of long-term psychological distress itself may constitute sufficient grounds for setting up a health register. It was noted that the long-term rehabilitation of the exposed persons should be planned as part of the acute response, and a health register would enable appropriate implementation of follow-up care and management. One participant remarked that a health register was justified in all incidents with novel agents or novel pathways of exposure, to ensure complete documentation of health effects for legal and financial requirements.
The potential for long-term health benefits to the persons included in a health register and to the wider population was another debated issue. Some participants contended that the prime consideration in the decision process was direct health benefits to the persons included in the health register and that possible benefits to the wider population were not relevant. Others countered this view, saying that the societal benefits from studying exposed persons may be more significant by enabling interventions to limit such exposures in the population. One participant commented that it was possible to overstate the benefits of a health register in creating new knowledge of potential benefit to public health.
Another participant asked whether explicit prior consent from the exposed persons included in a health register was necessary so registrants could be contacted in any future studies, since such studies may be conducted by another organization or by academic institutions. However, obtaining prior consent may be impractical in the acute phase when the focus would be directed largely toward emergency health response during a time of uncertainty and anxiety. One participant rejected the notion that decision frameworks such as the one agreed upon in this study were appropriate, as individual circumstances in a major incident dictate the need for a health register, and those circumstances cannot be pre-judged.
Discussion
In this study, a decision framework with a set of criteria for assisting decision makers on setting up a health register after a major incident was agreed upon and tested for practical use. The use of the Delphi consensus-building process applied to a difficult area of health protection resulted in good consensus among the experts and key stakeholders.
Following a disaster, the immediate period of emergency health response overlaps with the need to initiate a health register, so it is necessary to have protocols in place prior to the events. The Camelford water pollution incident demonstrated that a failure to rapidly establish a register of potentially exposed individuals might attract criticism at a later date. One of the recommendations of the Committee on Toxicity Report was that “it is vital to identify populations which may need to be monitored in any later epidemiology studies as early as possible after the incident. If identification of these populations is delayed exposed individuals may move out of the area and be lost to follow up.”23 The Committee also recommended that “if the exposed population includes a large number of transient residents, such as holiday makers who are in the area temporarily, consideration must be given as to how to identify this population in any future monitoring programme.” Similarly, among the lessons identified from the overall health response to the terrorist events of July 7, 2005, the London Regional Resilience Forum recommended that “there should be a requirement to record details of those in close proximity to the scene of an incident (an exposure Register) for subsequent monitoring.”24 These UK experiences have led to the conclusion that health monitoring and assessment for future disasters could benefit from additional centrally coordinated planning before the event occurs. In order to design a successful epidemiological follow-up to such an event, it is important to recruit the population of interest in a health register as soon as possible after the event.25, 26
Recognizing this need from previous experiences, the Agency for Toxic Substances and Disease Registry (ATSDR) in the United States has established a Rapid Response Registry (RRR) service to give timely information to public health officials and to enroll people exposed to, or potentially exposed to, an event.27 Similarly, following a series of major incidents, the Dutch government set up the Center for Health Impact Assessment of Disasters (CGOR), to improve national preparedness and, if necessary, facilitate rapid recruitment of exposed persons in a health register.Reference Ruijten28
Gaps in information are likely during the early phase of the response. One scientifically important reason may constitute adequate grounds to set up a health register and therefore a scoring system with threshold values was considered inappropriate for the decision framework. The final decision is ultimately based on a careful consideration of the feasibility and potential usefulness of a health register in light of current knowledge and circumstances by a core group of scientific experts and professionals who have particular experience in dealing with major incidents.
A crucial aspect in designing a population register is the need to identify the exposed persons and to undertake adequate and rapid exposure assessment at the earliest stage, as otherwise retrospective data collection may not be possible.Reference Brackbill, Thorpe and DiGrande5 In other words, the exposure distribution within the defined population needs to be considered and included at the design phase of a health register. In addition, it is important to develop and test tools for epidemiological responses that could be employed during the acute phase. Further work is necessary to develop and test operational guidance that outlines advisory and management structures and addresses the capacity issues involved in setting up population registers.
Limitations
This study has a number of limitations. First, the number of study participants in the second and third stages of this study was relatively low as a significant number of invited experts were unable to participate due to time constraints or other reasons. Second, a modified Delphi approach was used in this study with significant iterations in the statements between the first and second round. It could be argued that these additional comments provided in the second round changed the clarity of the statements and therefore, the participants’ scores for the first and second round statements were not directly comparable. This method was chosen as a pragmatic method to elicit consensus and at the same time limit the time required from participants to complete the study questionnaire. Third, the study participants were asked to pilot test the decision framework on four different incident types, some of which might not relate to their areas of personal expertise (ie, chemical or nuclear or infectious events). Finally, it is possible that having four different decision frameworks specific to CBRN events might have better value in the decision process.
Despite the above limitations, this study is a first attempt to develop an objective set of decision criteria for setting up a health register following a major incident. There is growing public expectation that public health agencies will provide specific services, such as health follow-up for those exposed to hazards in major incidents. Public health agencies need to be able to clearly state and defend why, how, and when a study will be undertaken. It is hoped that the development of an objective decision framework for setting up a health register would help public health authorities in using limited resources in a transparent and effective manner.
Conclusions
A decision framework to assess the need for setting up a health register after a major incident was agreed using the Delphi process and pilot tested using fictional incidents. Further areas of work for practical implementation of the criteria and related planning for systems and protocols have been identified.
Acknowledgments
The authors acknowledge the work done by Dr. Oliver Morgan in preparing the first draft of the protocol. They are very grateful to the study participants for taking the time to participate in the Delphi rounds.
Abbreviations
- CBRN:
chemical, biological, radiological and nuclear
- HPA:
Health Protection Agency (UK)
