Biosurveillance begins at the local level. Data gathered by health care providers, public health officials, emergency responders, and others are the foundation on which biosurveillance is built. These data then must be analyzed to generate the information and knowledge that drive specific actions.
Three speakers at the workshop—representing state and city perspectives—discussed how people at these levels prepare for and respond to biothreats and events, both from the planning and surveillance perspective and from the coordination and response perspective, highlighting the importance of developing effective collaboration early. Analysis of previous episodes and planning for possible future events are both necessary to uncover and fill gaps.
Under normal circumstances, within and between organizations and sectors, information sharing is considered maladaptive, said Joel Ackelsberg, Bureau of Communicable Diseases, New York City Department of Health and Mental Hygiene. Information is power, and sharing of information may lead to loss of control or autonomy. Unidirectional information flows are preferred, especially if that direction is pointed to
1This section is based on the presentation by Joel Ackelsberg, Bureau of Communicable Diseases, New York City Department of Health and Mental Hygiene.
oneself and one’s organization. As Ackelsberg quipped, a common perception is that “sharing information is for chumps.”
In emergencies, however, these generalizations no longer apply. Sharing information supports common missions, goals, and objectives. Multidirectional information flows are preferred to arrive at a common operating picture.
By examining experiences before and after 9/11 from this perspective, questions that were asked before the terrorist attacks can be reframed in ways that are still relevant today.
Responses to an Event
Public health responses to a covert biological release fall into six categories, Ackelsberg observed:
• Notification of key partners and the public
• Rapid investigation to confirm diagnoses, identify hazards and risk factors, and track impacts
• Risk communication and safety recommendations
• Coordinated interventions, including mass treatments and prophylaxis
Biosurveillance is typically interpreted to include the first three of these items. However, most biosurveillance today, Ackelsberg posits, focuses on detection and notification, but not characterization. These latter two tasks are “complicated and messy,” said Ackelsberg. “It takes people who are experienced to go through information, to share it, to analyze it, to interpret it. It involves instruments, but it goes well beyond gizmos. It’s heavy on people and heavy on the skills that they bring to problem solving.”
Before and After 9/11
The New York City Department of Health and Mental Hygiene was working on biological preparedness even before 9/11. It had instituted an internal incident management structure and had established interagency coordination with the mayor’s Office of Emergency Management, law enforcement, New York City hospitals, and regional public health agen-
cies. It had enhanced surveillance systems and had prepared for the mass distribution of antibiotics. It provided biological threat agent training for providers, worked on emergency communications, and conducted frequent tabletop exercises and drills.
“Exercises are one thing; experience, of course, is another,” said Ackelsberg. Though the department was primed for scenarios like the 9/11 attacks and the anthrax incidents that followed, it could never be ready for such events. The agency was nine blocks away from the World Trade Center and had to relocate to its laboratory. “Our assistant commissioner physically carried a couple of computers into the van to bring with her to the lab because it was on those computers that we had our software to broadcast faxes to the medical community.” The city had lost its emergency operations center, and communications were almost nonexistent. “We were operating in the dark. We had lost colleagues. We had lost phone service. It was very difficult for the public and our partners to communicate with us.”
The department characterized the initial problems after the attack, many of which involved the environment and the need for shelter, and established how best to address them. Large numbers of people who were exposed to harmful substances were evaluated, and the department still maintains a registry that tracks those impacts. The department also had the benefit of Epidemic Intelligence Service officers, Centers for Disease Control and Prevention (CDC) employees in a 2-year training program who respond to requests for epidemiological assistance, who were posted in 15 hospital emergency departments collecting information around the clock.
After the first anthrax inhalation case was identified in Florida, the department started active surveillance with intensive care units, engaged microbiology laboratories, and worked with infectious disease and infection control personnel. When anthrax cases began to occur in New York City, joint public health and law enforcement teams conducted multiple and simultaneous investigations of thousands of suspected cases. Public health liaisons within the criminal investigation made it possible to share ideas, generate hypotheses, and share analyses. “All of this was extremely personnel heavy,” said Ackelsberg. “It’s people who have to collect the data—or at least interpret the data—go through and clean up the data, and figure out what’s going on.”
The anthrax investigations revealed a number of biosurveillance-related gaps in the way information is collected and shared, according to Ackelsberg, including the following:
• Outreach is needed to clinical communities to aid in physician reporting and dissemination of information.
• Rapid mobilization capacity is needed across agencies to handle surges in reported cases, including case management and laboratories.
• Enhanced data collection and data management tools are needed, including integration with laboratory systems.
• It is advisable for laboratory and law enforcement personnel to develop relationships prior to an incident.
• It is potentially hazardous to deploy new surveillance systems during emergencies; “alarms” will occur frequently.
• Improved information sharing is needed between investigations located in other jurisdictions.
The underlying conclusion that can be drawn from these gaps, said Ackelsberg, is that reliable communication underlies all effective responses. As a result, communication has been a focus of change since 9/11 in the public health system in New York City. For example, the Department of Health has a protocol—“well exercised, unfortunately”—with the New York Police Department and the Federal Bureau of Investigation. “This is a good example of the way in which information needs to be shared in order for common missions and objectives to be successfully reached.” It is leveraging social networks for communications and public health surveillance.
The public health system in New York City has made different amounts of progress in different areas, Ackelsberg concluded. But information sharing will continue to be a major emphasis as the Department of Health prepares for future events. “The more that we can find ways to convince our colleagues, both in our agencies and in others, that information sharing is not for chumps, and that information sharing can actually bring us the knowledge that we need when responding to an incident, the more successful that will be in the long term.”
In response to a question about data integration centers, Ackelsberg observed that local and state health departments should be seen as customers and not just providers of information. The best people to interpret data are the ones who routinely use the data and are familiar with the patterns that exist at the local and state levels.
He also pointed to the importance of using words carefully. For example, “aberration” is a better term than “alarm.” “When we instituted drop-in surveillance in the fall of 2001, we had alarms going off all over the place. We had no baseline, so we decided the first thing we needed to do was to start calling it something else.” The terms used should convey the uncertainty of a situation while accurately describing and interpreting the available information. This becomes especially hard when people in different agencies are doing different interpretations or analyses, requiring not only communication but iteration. “What we thought we understood on day one is going to be very different from the way that we understand the situation days into the incident.”
Finally, Ackelsberg observed that all public health is local, but it may not be local to the place where a person lives. A place like New York City has millions of people coming in all the time by plane, by bus, and by boat, in which case local is much more expansive than the five boroughs. Meanwhile, the ability to monitor diseases around the world has huge vulnerabilities, despite the potential influence on localities like New York City.
Joseph Gibson, from the Marion County Public Health Department—which includes the city of Indianapolis—described his department’s experiences in gaining access to three kinds of surveillance data: clinical data, school absenteeism data, and data from the state health department. In doing so, Gibson highlighted issues of sharing and trust, and how to overcome such barriers.
The Health Insurance Portability and Accountability Act has a clause that allows health care providers to give clinical data to public health
2This section is based on the presentation by Joseph Gibson, Marion County Public Health Department, Indiana.
agencies. But the clause does not require the data to be shared, so hospitals can be reluctant to do so. “They aren’t covered for the risks that they open up themselves to if they do give us the data and something goes wrong,” Gibson said. In addition, state laws address how clinical data may be used in research, but omit public health uses, which puts such uses into a gray area.
Similarly, schools are not required to give the public health department absenteeism data. They did so during the H1N1 epidemic out of a sense of civic duty, but schools are strapped just trying to cover their core missions, said Gibson, so they do not have much time or resources for public health activities. Furthermore, when they were willing to share these data, the format and content of the data often varied. As a result, the public health department had to write custom computer programs and use manual processes to make the data useable.
Finally, the relationship between the Marion County Public Health Department and the state health department has varied from close to distant. Generally the state health department has been cooperative, but it needs to be careful not to violate state law, and these laws are not consistent across diseases. Furthermore, state officials have worried about establishing a precedent with Marion County, which is the largest health department in the state and has greater resources, that then would apply to counties with fewer human resources and less ability to protect the confidentiality of the data.
What Does Not Work
Gibson described several approaches that have not worked in requesting data to protect public health at the county level. First, broad requests for data are almost always rejected. “I had to provide specific uses to which I was going to put the data, and then I could start to get the data. And slowly, we’ve been able to develop more trust and expand those uses, but still, I have to be very specific in terms of what I’m going to use this data for.”
Relying on authority, power plays, or legal debate also has not worked. The law generally does not require sharing and is often gray, and recourse to authority often generates resistance.
Not understanding the restrictions faced by senders is a barrier to sharing data, Gibson said. “As I understand what their processes are and understand what their greatest concerns are about this data, I can much
more carefully draft my requests and my processes to fit theirs, so I’m more successful in getting the data.”
Finally, it does not work to request data that senders are not already gathering. Requests for data that are not already in the system generally are not successful. “Anything where I’m trying to change their work processes has almost always failed.”
What Does Work
The most important element in making data sharing occur, said Gibson, is trust. Understand a provider’s data protection rules can help build that trust. Also, finding opportunities to interact with data providers is important. “Whenever I have an excuse to go meet with my counterparts at the state, I go meet with them. Whenever there’s some event going on that they might be interested in, I try to get them there, so that we just have more interaction. That interaction builds understanding, and that understanding makes them much more comfortable in sharing the data with me.” Finally, it is important to be incremental and start with narrow requests, said Gibson. As the sender recognizes that data are handled carefully, it will be easier to get more data in the future.
The issue of trust is one reason why Gibson can be reluctant to share data with the federal government. His sources of data need to know that he is managing their data carefully. If he violates that trust—for example, by sending data to the federal government for one purpose that are then used for another purpose, such as law enforcement—the data provider might stop providing Gibson with the data.
Another approach that works is to minimize the burden and maximize the value for the sender. For example, Marion County and then the state of Indiana made a substantial effort to get syndromic surveillance data back to hospitals so they could see how the data were being used and use these data themselves. Similarly, schools were included in situation report distribution so they could see how their absenteeism data were being used and could understand the value to communities. For school absenteeism data, every school district received a summary of its absentee data and a comparison to the aggregate for all districts. Hospitals also received ways to compare their information to the information for all hospitals.
Along the same lines, during the H1N1 epidemic the public health department distributed swabs to clinics and then collected them every day to take to the laboratory. “It was really work intensive, and eventual-
ly it was not sustainable. But we tried to set up a system that was going to minimize the amount of work [the clinics] would have to do.” Similarly, with the schools, the department accepted absenteeism data in many different formats and then converted the data to a more useful form.
A legal mandate or top executive support is often important, said Gibson. Even institutions that want to provide data need legal coverage to do so. School absenteeism data did not become available until superintendents gave their blessing.
Finally, finding the right person to deliver the data can be critical. In schools, the right person was sometimes an information technology specialist and sometimes a nurse. “We had school nurses doing yeoman’s labor to give us the information . . . because it was important to them.”
A workshop participant asked about the efforts of the Office of the National Coordinator in the Department of Health and Human Services (HHS) to promote sharing among health care organizations through common standards and funding incentives and to build an interoperability framework to house information, suggesting that biosurveillance efforts should be coordinated with such health information systems efforts. Gibson remarked that the states are working hard to ensure that they receive this information as well, but they have very little funding to do so. “At the local and state level, there’s a lot of activity around that.”
Ackelsberg asked about sharing information while protecting the interests of an agency and its partners, and Gibson responded that “the more you share information, the stronger your community gets.” He said that he has a sign on his wall that says, “Information is power, so spread it around.” Information is not lost by sharing it—it just builds. However, when data come from somewhere else, it is advisable to direct people to the source, “because we don’t want to be sharing somebody else’s data; we’d rather have it come directly from them.”
This is also an issue when law enforcement becomes involved, because hospitals are unlikely to keep sharing information with a public health agency if that information is then passed on to law enforcement agencies, said Gibson. “They’re giving me that information for a very specific use, and if I’m sharing that information in a way that goes beyond that specific use, then they’re going to stop giving me the information.”
Moderator Lisa Gordon-Hagerty noted that an unwillingness to share means that information will be stovepiped. “The whole idea here is for information sharing, recognizing we’re talking about two very different cultures—the public health community and the law enforcement community.” How can the wealth and power of information be spread around if barriers exist between agencies? Gibson said that he has to maintain the trust of the people who are providing him with information. “That’s the tension that I have to try to work through.”
North Carolina relies on two main systems to detect disease outbreaks, said Jean-Marie Maillard, North Carolina Department of Health and Human Services (NC DHHS). One is the North Carolina Electronic Disease Surveillance System, which provides case reports and laboratory results. The other is the North Carolina Disease Event Tracking and Epidemiologic Collection Tool (NC-DETECT), which gathers information from emergency departments, poison centers, and ambulance runs and is used for syndromic surveillance and situation analysis.
NC-DETECT was started in the 1990s by a group of researchers at the University of North Carolina who were trying to gather information from emergency department visits. The NC DHHS partnered with this group and, after 2001, received increased funding to improve disease surveillance and epidemiologic capacity. The department also worked with the hospital trade organization in North Carolina to ensure that emergency department data would be reportable. In addition, it drew on data from the Carolina Poison Center, which receives about 120,000 calls per year, about 20 percent of which are from physicians and emergency departments.
NC-DETECT provides data in the form of customizable tables, graphs, and maps and is available 24 hours a day. Reports from emergency departments are received twice a day from every hospital reporting to the system, and for the poison center the update is every hour. Users have rights defined by their role, so a local person working with the hospital system could see local data, a regionally based public health professional
3This section is based on the presentation by Jean-Marie Maillard, NC DHHS.
could see regional data, and users at the state level could see the statewide picture.
As an example of the type of event seen by NC-DETECT, Maillard cited a 2007 outbreak of salmonellosis caused by contaminated peanut butter. Hospital visits immediately rose, along with a sharp and quick increase in calls to the poison center. The system “allows us to look at different aspects of a health event in the community,” Maillard said.
Surveillance networks in North Carolina include local health departments, including communicable disease nurses, hospital-based epidemiologists, laboratory directors, and epidemiologists at the state level who can provide support remotely and on site as needed. For instance, the state has used funding received since 2002 to set up a network of hospital-based public health epidemiologists.
The state also has a system that can look into the electronic medical records of patients in real time. This system scans every day not just for bioterrorism but also for everyday public health issues. Epidemiologists at the state level then can deploy if onsite assistance is needed.
The state has a number of task forces and task-oriented workgroups that allow potential collaborators to meet each other and work together. For example, the Food-Borne Disease Task Force, created by the governor in 2003, combines staff from 20 different organizations who meet six times a year. These groups make it possible to convert data into actionable information, said Maillard. These task forces, which originally started as a means to convene relevant stakeholders around specific issues, also serve the purpose of maintaining connections and relationships.
Finally, North Carolina uses checklists, situation reports, and an incident command system for larger events of public health significance. It has established memoranda of understanding among the departments of health, agriculture, and environmental and natural resources (the latter of which has since been combined with public health). It consults and works with CDC, the Food and Drug Administration, and the Department of Agriculture. It even has a team of student volunteers at the University of North Carolina who are available to help with call lines, interviews, and case control studies.
The System in Action
An example that illustrates the operations of the system was the recall of a brand of chili associated with botulism. Even though a holiday limited the number of staff who were available, Maillard was able to
immediately find 12 patients who met the botulism syndrome case definition. He then further examined these cases to see if they were associated with the food product. The information was shared with the department in charge of the recall, which found that many stores were still displaying the product. Personnel “visited 16,000 stores and removed 35,000 cans of product, which was more than the rest of the country combined.”
Another example involves vectorborne diseases such as Rocky Mountain spotted fever, which has a high incidence in North Carolina. In 2003 the West Nile Virus Task Force was created to deal with the spread of the West Nile virus, and the task force was later expanded to be the Vector-Borne Disease Task Force. The task force meets regularly with not only public health staff but also entomologists, local health departments, academicians, and members of the public to share surveillance data, talk about what is known and not known, and share study findings. Similarly, the Disaster Epidemiology Group has worked on hurricanes, floods, wildfires, heat waves, and other events to enhance the epidemiology capacity for disaster response.
The state has learned through successive deployment how to conduct surveillance projects, community assessment projects, and environmental studies quickly, said Maillard. An infection control program in the state has detected disease outbreaks associated with reuse of single-use products that should not be shared among patients. “It works both ways—we tell them about the cluster of outbreaks we are informed about and investigate, and health service regulation tells us about the findings they noticed during inspections.”
The North Carolina Intelligence Sharing and Analysis Center provides two-way information sharing with law enforcement through public health staff with security clearances. The state also has signed a memorandum of understanding with the local health departments of the Eastern Band of Cherokee Indians that describes who will do what with regard to information sharing.
In response to a question about the usefulness of memoranda of understanding, Maillard responded that they have the benefit of laying out who could get the information and how information will be shared. “We all work from the same large base as quickly as possible,” he said.
Regarding a question about collecting information from veterinary
clinics as well as emergency departments, Maillard pointed to several collaborations with the agricultural sector, though funds for such activities have been diminishing. Animal surveillance generally has been at a more basic level than human surveillance, he said, but animal surveillance in North Carolina has had extensive experience with mapping and has been willing to share its mapping structure with public health agencies.