3Health Research Needs and Actions: Lessons from Recent Disasters

Why Research During a Response

Because funding and staffing situations can be fragile, questions on the need for research or activities not directly related to an emergency response are perpetually raised. Prezant elucidated an important example showing the need for timely research specifically related to the disaster at hand. As a result of the 2001 attacks in New York, two 110-story towers and several surrounding buildings were reduced to about four stories of pulverized rubble. What was not in the rubble pile was suspended in the air in an enormous dust cloud. Shortly after the collapse of the towers, news outlets were interviewing experts on television, some of whom said there would be no respiratory effects of exposure to the debris dust cloud. They based these statements, Prezant explained, on research that has demonstrated that particles must be less than 3 microns in size to enter the lower airways. At the very most, they predicted mild upper airway consequences (chronic sinusitis, rhinitis). These experts were wrong, Prezant said, because they were thinking about low-density, low-concentration inhalation experiments as would occur with tuberculosis or other infection agents or with metered-dose inhalers. This exposure lasted for several days, and about 16,000 FDNY rescue workers continued to be exposed to different levels of dust in encapsulated subterranean areas over the next 10 months during the intense rescue and recovery effort.

Prezant said firefighters came out of the rubble and told him that the air had a different smell and taste than anything they had experienced before in fighting fires. The firefighters at the WTC site were exposed to high concentrations of very alkaline dust (approximate pH of 10) containing a mixture of pulverized cement, gypsum, pulverized glass, asbestos, silica, fibrous glass, heavy metals, volatile organic compounds, and the organic combustion/pyrolysis products of all the building components and jet fuel (e.g., polycyclic aromatic hydrocarbons, dioxins, polychlorinated biphenyls). About 3 weeks into the recovery effort, a therapeutic, diagnostic bronchoscopic evaluation of a firefighter who was sent to a nearby hospital in respiratory distress and arrest revealed uncoated asbestos fibers, glass, and pulverized ash. In other cases, macrophages in sputum samples collected 10 months after 9/11 contained encapsulated dust similar to that at the WTC site (Fireman et al., 2004). Air and respiratory monitoring while the response and clean-up was going on might have allowed responders and clinicians to more quickly see the potential hazards involved and change actions accordingly.

The Importance of Early Baseline Data

In many disasters, baseline data may not exist or are hard to access for the same cohort looking to be studied. This lack of access can make it even more difficult to pinpoint causes or link health outcomes to disaster exposures. Prezant also stressed the importance of acting immediately to get monitoring and treatment systems in place, so that data are not lost and so that credibility as a partner is established. An FDNY WTC responder cohort was established after 9/11, and a well-organized medical monitoring program has minimized longitudinal dropout with consistently high retention rates (92–95 percent). Because FDNY has been capturing health data, including pulmonary function tests, systematically since 1996, this is the only cohort with pre-9/11 health data, Prezant noted. This cohort also has the highest exposure to respiratory irritants and combustion by-products of any WTC cohort. Monitoring and treatment protocols are in place for respiratory and mental health issues as well as for late emerging diseases (e.g., cancer). This can also foster trust with participants by allowing them to be cared for quickly, which will continue to encourage study involvement and high retention (see Figure 3-1).

FIGURE 3-1

The World Trade Center Health Program for the Fire Department of New York. NOTE: DIS = Diagnostic Interview Schedule; EKG = electrocardiogram; MSW = master of social work; PFT = Physical Fitness Test; RN = registered nurse. SOURCE: Prezant presentation, (more...)

Coordinating Logistics to Execute Rapid and Sustained Research in Disaster Response

Research is a slow, deliberative, thoughtful, and collaborative process. Disaster research, however, has to absorb the immediacy of the event, Prezant said. Without that information, any of the outcomes of interest will remain elusive, regardless of the techniques used. The immediate data are essential to determine what did or did not happen; who was exposed, to what, and to what extent; and what the immediate impact appears to be. The research questions are similar at every disaster, and preparedness requires prearranged studies that are ready to go, he added.

Partnering with Communities to Facilitate Pre-Event Assessment

Partnering with communities, Prezant said, including state and local government, is essential to facilitate access to pre-event data, collection of baseline data, and longitudinal assessment of the population (discussed further in Chapter 5). Participants need a reason to participate in research, Prezant continued; too often, participants view research as an experiment. They have already lived through one experiment (the disaster), and they are not inspired to volunteer for another. In addition, they are dealing with the aftermath of the disaster (e.g., the death of a loved one, loss of their home), and their energy and patience are limited. Therefore, participants need a reason to believe the research effort is worth their remaining energies and time. This involves developing trust, which requires credibility. IRB approval is not enough to garner trust and demonstrate credibility to people, Prezant said. They do not know what an IRB is. What is needed are partnerships and a history of service that preferably began before the disaster (as is seen with the FDNY responder cohort). There also needs to be a “stamp of excellence” from a major organization that the community trusts (e.g., the American Cancer Society, the American Heart Association, CDC) to help secure local buy-in and public labor and management support.

If we cannot monitor, assess, provide services, intervene, and reassess, we cannot provide what people need in a disaster, Prezant said. If we do not know what is going on in the communities, we cannot respond, or we respond with the wrong interventions, only making matters worse, he continued. The end point then is not the data, but the knowledge that can be used to guide recovery, including clinical services and systems improvements. Continued participation requires fulfillment of this promise, in the form of delivery on end points in the short term as well as in the longer term.

As an example of delivering on the promise of research, Prezant cited several published papers on declining pulmonary function in WTC responders in the short term and persistence of reduced function in the long term (Aldrich et al., 2010; Banauch et al., 2006; Prezant et al., 2002). FDNY was also the first to publish an early assessment of cancer outcomes, showing a cancer signal only 7 years after 9/11. This affected federal health care policy, Prezant said, because cancer was then added to WTC covered conditions (Zeig-Owens et al., 2011).

FDNY is also involved in basic research. For example, blood banked during the first 6 months after 9/11 has been used for studies linking pulmonary function test outcomes to biomarkers that could potentially predict both susceptibility and resistance to disease (Cho et al., 2013; Nolan et al., 2012). Firefighters are saying yes to participating in basic science research, as long as they have a sense that it is going to help them or another firefighter who is exposed.

Obtaining Rapid and Sustained Research Funding

Research funding after a disaster has several phases, and they are generally based on public perception, Prezant said. Initially, small amounts of funding flow from governments and philanthropies because it is “the right thing to do.” This is often followed by disbelief about whether people were actually exposed to something dangerous, whether it was just an acute issue that will resolve spontaneously, or whether the exposure actually caused disease. The next phase is apathy, when funders question whether this research is still important and how much is enough, especially in the face of other national priorities.

The way to overcome these barriers to funding is data-driven advocacy, which requires research, Prezant said. Emotional advocacy secures the initial funding to look at uncertain exposures or health effects, but without data to show importance, there is a loss of credibility. No study is ever perfect, and this is especially true for disaster research where there are numerous, simultaneous, and often unknown confounders. However, multiple studies showing similar effects lead to a more perfect understanding and a more perfect response (i.e., improved services and systems). Research requires educating everyone involved (researchers, participants, funding agencies) as to what is achievable given the situation and adjustment of expectations across all of the funding agencies.

Improving Research Systems and Data Collection

Although everyone thinks their idea is worthy of immediate study, and there are novel issues that can occur, we need to remain focused on the main questions, Prezant said. Affected systems and populations have limited time for testing and questionnaires, and overburdening people can lead to longitudinal dropout. In addition, trying to capture too much data can reduce focus and lead to paralysis. The questions, data collection, and analyses need to be part of a predisaster, agreed-upon, focused agenda, he said.

Data collection should allow for immediate collection and real-time access, Prezant said. For example, a computerized, online questionnaire was developed predisaster as part of the regular annual monitoring exams for firefighters, and within 4 weeks after 9/11, FDNY was able to change the entire monitoring exam to focus on WTC. Questions were designed so that the introductory phrase could be changed with a single-stroke command. In 2000, the phrase was “during your career of firefighting have you ever.…” or “in the last 12 months have you ever.…” Between 9/11/01 and 10/01/01, the phrases were changed to “after the World Trade Center.…” When FDNY was deployed to assist with the Hurricane Katrina response, the phrase was changed to “after Hurricane Katrina.…”

The use of validated questionnaires is essential, but they are not as validated as we think, suggested Prezant. Many of them have not been studied in different populations, and almost none have been studied in a disaster population. Questionnaires may have been validated for one disease, but disasters cause multiple diseases. Using multiple, different validated disease-set questionnaires can lead to additional concerns. Concerns may be attention deficit after answering too many questions, confusion with different time spans for answers to questions (e.g., “in the last 4 weeks have you.…” or “in the last 3 months….”), or similar questions in each disease set that reduce specificity of the answers.

Prezant described the World Trade Center Health Program and how it monitors and assesses firefighters. The same day that data are collected and assessed, firefighters are informed of recommended treatment or referral options. Prezant noted that FDNY cannot force firefighters to do the follow-up, but they make it free (through their health benefits program) and as easy as possible to get mental health treatment, physical health treatment, medications, benefits counseling, and so on.

Focusing on the Human Need to Help

Prezant stressed that the goal of research in response to a disaster is not just to record events, determine mechanisms and outcomes, and design interventions, although all of these are essential to response. The most important goal after a disaster, and the one required for all of the above to be possible, is to stimulate human decency, he said, by focusing on our natural impulse to reach out and help people and building a lasting relationship where at first there was only the desire and the urgency to help, and not the established scientific need to help.

Gulf Coast Child and Family Health Study

Abramson elaborated on one of the larger community-based studies in which he participated, the Gulf Coast Child and Family Health Study, begun in 2006 and funded through the Children's Health Fund and the Merck Foundation. The study focused on identifying health and social service needs among this displaced and heavily impacted population. Data were collected from 1,079 randomly sampled households over the span of 29 days for the start of a 5-year longitudinal study. The sample was intended to represent 60,000 to 100,000 displaced and/or heavily affected Katrina survivors.

The immediate challenges in operationalizing this study were enormous, Abramson said. Even though he and his colleagues started working on a protocol just a few weeks after the storm, the first barrier was access to sample frame data (or the set of information used to identify the population for research). Namely they were looking for information on where the displaced populations were, especially congregate settings to help frame their research protocol. FEMA was managing the congregate settings population and had meticulous data, he said, but they did not release any of it for his research. A second challenge, Abramson added, was that the sampling frame was a moving target after they got the initial access. For example, by the time the protocol was approved, FEMA had triggered its deadline for evacuees to leave the hotels, so identifying and locating participants had to be done all over again. Another challenge they anticipated was whether or not researchers would be able to gain access to these public and private sites once they found them. Time was the biggest issue, as it becomes more difficult to find people as time passes and recall bias can increase over time. Abramson said there was no reliable Internet access in many places they went, and to account for that anticipated challenge they developed field management databases for tablets that can batch data and upload later when an Internet connection is available. Researchers were also sure to have pens and hard copies of the survey they were administering.

To quicken the time line and mobilize the research infrastructure more rapidly, Abramson used the infrastructure and expertise he already had in place for an HIV longitudinal cohort study happening in the New York City metropolitan area. This existing infrastructure offered Abramson and his team rapid access to data systems, qualified interviewers, research staff, staff knowledge of working with vulnerable and elusive populations, and a common methodology for building and maintaining a longitudinal cohort using multistage, stratified sampling strategies. Building these elements from scratch would have delayed the time line and kept them from getting into the field as quickly as possible to capture the ephemeral data. In addition, Abramson was connected with more than 25 graduate students who were interested in coming to the Gulf Coast to conduct the research. Although this worked as a great model in this case, it may be difficult to fund this type of “attaching” to an existing research infrastructure as a sustainable future model for disaster research. Standardizing that process across institutions nationwide that are interested in disaster research may be a challenge as well, but in cases where relationships are already built, it may also be very efficient.

By the time they were ready for the IRB submission, it was approximately 4 months after the storm hit. The IRB understood the need for speed in this case and provided a rapid expedited review. The process involved a back-and-forth dialogue and helped to establish a relationship that is now in place to facilitate rapid IRB approval in future disasters. Upon initial submission, the head of Columbia University Medical Center's IRB informed Abramson that his protocol for the Gulf Coast Child and Family Health Study would need to consider additional elements: addressing the vulnerability of the subjects; systematic referral of subjects to care; ability to identify crises as they occur; endorsement from the community or government; the safety, training, and preparation of the field team; and compliance with local “Duty to Report” laws regarding abuse and neglect (which may differ by state). Agreeing that these items were important and needed to be examined, they were added to the protocol. The IRB director and chairs ultimately granted approval days before Abramson was to enter the field, and they now have an understanding of the time line importance for certain disaster-related projects. If an IRB is not as understanding and flexible as the one in this case, this could be another potentially vulnerable area for a researcher's time line, especially in the case of multiple institutions working together and needing approval by separate IRBs. Education and sharing of disaster-related research projects at IRB forums and net-working events could be a way to shift perception of the IRBs from a barrier to a partner.

A Transient Study Population

Key among the challenges to conducting oil spill research was working with the atypical study population of workers and volunteers. To identify potential study participants, NIOSH was able to provide a roster of 50,000 people who were involved, and the National Guard and the Coast Guard were able to provide lists of people involved in cleanup in some way. After many negotiations, British Petroleum (BP) also provided a list of more than 130,000 workers. NIH had hoped to recruit 50,000 people to assist in cleanup, but Birnbaum said it has been extremely challenging to recruit the 33,000 enrolled thus far. About 20 percent of the workers came from out of state and then dispersed, making it difficult to find them.

Timing Sensitivity

Although the intramural study development process was fairly rapid (the new study was developed, received the necessary approvals, and researchers were in field within 10 months after the event and 8 months from the time of funding), we need to be able to do research in a disaster situation from the start, Birnbaum said. Echoing Lurie's concerns, she said waiting 8 months to 1 year later means missing baseline and peak exposure data collection and studying mid-term rather than acute effects. Awarding NIH extramural grants takes even longer: usually 12 to 14 months elapse from the time of request for funding opportunities until the actual grants are received and funded. NIH was able to fund the extramural consortium within 8 months, but in most cases individual IRB and other necessary approvals added significant time, making it 18 to 20 months from the time of the spill for most of the extramural grantees to begin to recruit for their studies. Following a more nimble NSF model Abramson mentioned, where smaller amounts of money can be disbursed more quickly, could help to address this, as well as streamlining the IRB process so researchers do not have to gather approvals with different requirements.

True baseline preexposure data were available for only a small fraction of the cohort (e.g., those from the Coast Guard), and Birnbaum suggested the need for baseline data and biospecimen collection from people who are rostered to work in a disaster cleanup. In addition, while there was a great deal of exposure data collected by BP and multiple federal and state agencies, the databases are not integrated, and Birnbaum said it has taken significant time and resources to reconstruct exposures at different places and times during the cleanup.

Summarizing the lessons learned from the NIEHS experience with the Deepwater Horizon oil spill, Birnbaum said that in a disaster, in addition to the normal occupational cohort for the site, there may be National Guard and other servicemen and women, firefighters and police officers, volunteers, and other workers involved in the response, as well as local residents whose lives and livelihoods are impacted (e.g., Gulf fishermen). Birnbaum stressed the importance of rapid and ongoing communication with all of the stakeholders, including community groups, academic partners, and industrial partners. She also noted the need to develop better capabilities to rapidly evaluate the toxicity of the exposures.

Howard of NIOSH at CDC added that NIOSH has done extensive intramural research on the Deepwater Horizon disaster from the perspective of worker safety, and referred participants to recent articles for further details (Decker et al., 2013a,b; King and Gibbons, 2011; Kitt et al., 2012; Michaels and Howard, 2012; NIOSH, 2011).

Dissemination

NYU researchers are also considering mechanisms for rapid dissemination of experiential learning in a disaster (Laskowski et al., 2013). Goldfrank noted that the traditional peer-review process impeded speedy publication of Sandy research results. He raised several questions for discussion: Should there be “disaster standards” for publication with regard to rigor, methodology, response rate, control groups, etc.? Does the public good of disaster research demand open access, and if so, who funds this? Who is the author, the investigator, or the participants? Do disciplines not traditionally associated with health care delivery achieve larger importance in the face of disaster (e.g., materials science, water and sanitary engineering) and can they be cosupported?

The Impact of Disasters on Hospitals and Health Care

There were many questions about the role of hospitals, emergency departments, and ambulatory care units in meeting the needs of the community during the crisis, and Goldfrank said most decisions on these questions were made in “an evidence-free zone” where guidance from other disasters would have been helpful. While these issues were debated, health care clinical deficits continued or were exacerbated. Goldfrank raised concerns about the ability of a city or state to respond to a disaster without total integration of the health care system. He explained that in evacuating the hospitals, all private staff were sent to other private hospitals, public hospital personnel were sent to other public hospitals, and U.S. Department of Veterans Affairs (VA) staff were sent to other VA facilities. Because there was no universal credentialing, and it can take 2 or 3 days for a provider to be credentialed in another hospital, many would go and wait, unable to deliver care. Goldfrank suggested that staffing with providers from closed hospitals benefits the other hospitals because then they do not have to pay overtime to their staff to have the necessary coverage.

With the above issues in mind, NYU is working on a project funded by an ASPR Recovery Grant to study the impact of a major adverse climate event on health system care and development of disaster response- and resilience-based metrics. The study will examine the comparative effectiveness of adaptive options (e.g., setting up an ambulatory care center or freestanding emergency department) and the decision making necessary for the entire health community to function.

Removing Structural Barriers

A few participants discussed further the need to address critical structural barriers and build better systems for everyday use that can be scaled up in a disaster, rather than focusing on building systems just for use in disasters. Goldfrank stressed the importance of being able to deliver everyday health care to everyone and opined that we “will not achieve excellence, as we would consider acceptable, unless we accomplish universal health care.” Prezant added that before the disaster, there need to be opportunities to develop relationships that can be translated into an improved health care infrastructure when needed (discussed further on page 39).

Abramson raised the issue of silos and proprietary data systems as barriers to disaster research. Many entities (e.g., FEMA, U.S. Department of Homeland Security, other emergency management) create their own data systems to facilitate their work and become very protective of their data, systems, and projects, making interoperability before and during an incident a challenge. Some even seem unwilling to acknowledge that data they are collecting could be very helpful to the community at large. Abramson noted that this goes beyond academic researchers seeking access, and he has heard from local and state health officials who could not access government datasets to help facilitate their work. At the local level the silos of information can also trickle down, making it difficult for local health departments, hospitals, and emergency ma-nagement agencies to easily talk and share data across sectors. A part-icipant noted the potential for a symbiotic relationship between local health departments and researchers. The resources and people that local health departments can pool can be beneficial to researchers, and the information that researchers generate can be beneficial to local health departments.

Prezant highlighted the need for funding streams for preparedness research and called for data-driven advocacy to establish and maintain credibility of the field after the initial emotional funding response subsides. Disaster response research cannot compete in the process as it exists now; for example, there is not time to write extensive proposals, the proper control groups are not always possible, and there are numerous simultaneous, and often unknown, confounders. We need to change the expectations of federal funding agencies with regard to disaster response research, he said.

Building Coalitions and Goodwill in Advance

A discussion point throughout the case studies was the benefit that could be realized in getting different groups to work together to prepare and develop “prepositioned goodwill” that can help them to be ready to work together in a disaster. Abramson highlighted the role of health care coalitions in getting hospitals to begin to work together and noted the need to broaden those coalitions so they begin to engage other health care providers within a larger health system and then the community stakeholder groups. Prezant commented that each area has its own challenges for developing networks. In a real disaster, many barriers tend to disappear for the first few days, but then they recur. In some cases excellence can be a barrier, as expert entities do not want to make compromises. In other cases there may be little or no structure to begin with, making it difficult to develop any sort of network. What is needed is a solid public health infrastructure, because without a foundation for everyday medicine, there is no foundation for disaster medicine, he said.