Thinking about Pandemic Influenza: Understanding Health Cognitions Helps Evaluate Response Options

May 2010
Topics: Emergency Preparedness and Response, Disasters, Disease Transmission
How we think about health emergencies like influenza affects our behavior and ultimately public safety.
woman in protective mask

When an emergency strikes, even the best-laid plans of government agencies to deal with crisis situations can go awry. During Hurricane Katrina, for instance, an inadequate understanding of how citizens would react to emergency directives hampered rescue efforts. "People were stranded on rooftops with their pets. Many refused to board rescue helicopters when ordered to leave the animals behind," says MITRE's Jill Egeth, a principal human factors engineer.

Had rescue workers understood how reluctant some hurricane victims would be to abandon their pets, government agencies could have tailored emergency response plans and messages to increase public cooperation with rescue efforts, she explains. Research now underway at MITRE aims to unravel these beliefs, or "cognitions," about public health emergencies—in particular, a potential outbreak of pandemic influenza. What we learn about how citizens are likely to behave during such a disaster will help the federal government and state and local agencies to develop more effective emergency plans and communications, Egeth says.

The idea isn't to help the government write disaster plans from scratch, since such plans exist already. But in the aftermath of Katrina and similar events, it's become clear that certain emergency response plans fail to take into consideration critical issues relating to public cognitions about public health emergencies, she explains.

"The federal government has national emergency plans for at least 15 different scenarios, such as earthquakes, bio-terror attacks, and pandemics," Egeth says. "Many of these plans address scenarios that our nation has not yet experienced, or that, if and when they do occur, may rapidly evolve and change. It's difficult, but critically important, to effectively plan for the unknown."

How People Think Affects Existing Plans

The government's existing pandemic flu disaster plans don't take into account a range of public beliefs. Examples include attitudes about the safety and reliability of pandemic influenza or seasonal influenza vaccines, whether it's advisable to stockpile food in the event of a flu outbreak, and which events would prompt people to stay home from work and school in mass numbers. "Cognitions help predict behaviors," Egeth says. "And research into public health cognitions has not yet been well applied to the field of emergency response."

MITRE's research into emergency preparedness and response cognitions applies a health psychology approach to the assessment of citizens' attitudes, beliefs, and judgments about pandemic influenza. Our researchers analyze data from MITRE's own surveys and from similar ones conducted by the Harvard School of Public Health (HSPH) and use it to develop and refine disease-spread simulation models. These models could help government agencies better understand the influence of citizens' attitudes towards vaccination programs on likely patterns of disease spread. Armed with the data implemented in simulation models used to explore possible outcomes, MITRE will develop recommendations for more targeted, effective vaccine administration policies and public health communications efforts.

Real-world Flu Outbreak Research

Although concerns about the H1N1 strain of influenza have been prevalent over the last year, MITRE's research in this area began in October 2008, predating the H1N1 outbreak. However, when the H1N1 outbreak began, it quickly became apparent that current events would force changes to the research program.

"Midway through our initial research process, in April 2009, the H1N1 situation arose," Egeth says. "We had to make some big changes to our research questions. We had been assuming that nobody knew anything about pandemic flu, but with so much on the news about H1N1, knowledge and attitudes were evolving. That summer, we were able to collect nationally representative survey data on citizens' attitudes toward flu vaccines and other issues relating to pandemic flu. But we knew that in order for research to remain current, we needed to get access to a constant stream of fresh survey data. In response to the outbreak, HSPH initiated a similar line of survey research, and we've been able to leverage some of their data and results."

For example, widespread public fears over vaccine safety in general appeared to be influencing people's plans to obtain the H1N1 vaccine for their children. A December 2009 survey conducted by HSPH showed that 60 percent of respondents cited safety fears as their primary reason for declining the vaccine for their children. Parents held these beliefs in spite of the Centers for Disease Control and Prevention's official determination that all children aged 6 months to 18 years constitute a high-priority group for the H1N1 vaccine.

"From the very beginning, we were interested in figuring out what people thought about vaccines," says Jennifer Mathieu, a MITRE lead multi-discipline systems engineer who also works on the project. "In addition, we studied social distancing [staying at home], antiviral use patterns, and food and water stockpiling. What would prompt people to prepare by stockpiling food and water or remaining at home during a pandemic? What if the government recommended people stay in their homes?

"We discovered that there were gaps in people's knowledge about what this would mean for everyday life. If you're told not to go to supermarkets, how do you get food?"

As the H1N1 outbreak progressed, it became clear that the main issue facing government agencies was public cognitions about vaccines. The MITRE team decided to focus all its research efforts on the new H1N1 vaccine, as opposed to related issues like school closure, Mathieu says. (However, insights the researchers gained into those related issues should be useful for MITRE's sponsors in the future.) The research team has since delved more deeply into the question of how public confusion over H1N1 vaccine availability and safety affected vaccination rates at the height of the recent outbreak.

Developing Models of Disease Spread and Response

Refinements to the team's hybrid model for evaluating local response to flu outbreaks are ongoing, says Mathieu. (The hybrid model consists of an agent-based model used to simulate how individuals or agents behave in a given situation, and a discrete event process model used to capture response processes. "Agents" stand in for individuals affected by the disease. So, for example, agents go through the flu clinic process to get the H1N1 vaccine.)

"How to use the survey data and have it make sense within the modeling framework is a challenge," she says. "We're trying to make our models represent enough of the reality of what people would actually do in these situations so that we can use the model to look at various response options. Once we have developed methods for using social behavioral science data in validated simulation models, government response organizations could use these methods to simulate possible outcomes, given certain decisions."

"For example, survey data was collected roughly monthly through the fall of 2009—how do you use this data in the hybrid model? We tried updating the agents in the model who got the vaccine when the new data was available, but in reality they got the vaccine over the course of the last month. Just using the data as it becomes available to update parameter values in the hybrid model is not necessarily the best way to use data in the model. Changing parameters at an earlier time and in a continuous fashion rather than in one large step might be more accurate." This type of feedback will help researches and sponsors know how best to use collected social behavioral science data in simulation models.

Mathieu also notes that when the researchers plug all the data into the hybrid model, it can determine how real-world disease rates would differ if certain factors had changed, including increased vaccination rates, faster vaccine production, and more effective public health communications plans. Relevant agencies, such as the Centers for Disease Control and Prevention, could use this approach to figure out which maladaptive cognitions would offer the most "bang for the buck" in terms of behavior change and disease spread. To be successful, the proposed changes must be targeted by a communication campaign aimed at changing attitudes.

"For example," Egeth adds, "we are currently working to figure out how changes in vaccination attitudes would influence vaccination behavior and disease spread. A government agency could use this information to run a cost-benefit analysis on the usefulness of a communication plan that educates parents about vaccination safety."

Real Data Provides Valuable Input

"Many researchers who work with these kinds of models plug in assumptions, while we use real data drawn from the social and behavioral sciences to make the outcome of the research more accurate," Egeth notes. "This is something that shows MITRE's value in a project like this—we have multi-disciplinary team members who can work together to come up with these results." The team includes specialists in health psychology, public health, and sociology, along with system engineers and computer scientists.

In the agent-based part of the disease spread model, MITRE's Epidemiological Model, first developed in 2004, was used. Each agent has a probability of contracting influenza. When the researchers plug the appropriate disease spread characteristics into the model, they can draw conclusions about which segments of the population will be hit hardest by the flu, depending on the lethality of the disease and the response measures implemented, Mathieu explains.

The MITRE team has expanded the discrete event processes that mimic responses to pandemic influenza that can be evaluated using the hybrid model. The model currently focuses on the area near Emerson Hospital in Concord, Mass., and includes 11 surrounding towns. Information, patient, and supply flows are represented in the model, enabling the evaluation of different outcomes for a variety of specific scenarios. The team fleshes out these scenarios using data about operational government procedures obtained from subject matter experts, government documents, and other available literature. The results of such analyses will provide valuable information to government decision-makers at all levels for disaster planning and infrastructure analysis.

"By modeling government operational procedures, we can help identify deficiencies and gaps in response plans—and perhaps help save lives," Mathieu says.

—by Maria S. Lee


Publication Search