Infectious Disease Surveillance Goes High-Tech
New developments include syndrome-based surveillance and nationwide electronic systems
By Kelly M. Pyrek
The surveillance of emerging infectious diseases has long been the cornerstone of public health preservation, and is one of the most important responsibilities of a hospital's epidemiologist and/or infection control practitioner. Surveillance is defined by the Centers for Disease Control and Prevention (CDC) as "The ongoing, systematic collection, analysis, and interpretation of health data essential to the planning, implementation, and evaluation of public health practice, closely integrated with the timely dissemination of these data to those who need to know. The final link of the surveillance chain is the application of these data to prevention and control practices."1
Despite morbidity and mortality rates associated with the transmission of infectious agents, public-health interventions have been a major impetus behind the 30-year increase in life expectancy since 1900.2 Surveillance provides crucial data to support the CDC's mission of improving public health as well as extending individuals' lifespans. Since the CDC's founding in 1946, public health surveillance has evolved from activities that focused primarily on the prevention and control of acute infectious diseases to include chronic diseases, injuries, risk factors, and health practices.
Disease surveillance systems provide for the ongoing collection, analysis, and dissemination of data to prevent and control disease. Disease surveillance data are used by public health professionals, medical professionals, and members of private industry to identify cases for investigation and follow-up; to estimate the magnitude of a health problem and follow trends in its incidence and distribution; to formulate and evaluate control and prevention measures; to detect outbreaks or epidemics and generate appropriate interventions; to monitor changes in infectious agents (such as antibiotic resistance or emerging infections); to facilitate epidemiologic and laboratory research; and to detect changes in health practice, such as the impact of use of new diagnostic methods on case counts.3
When it comes to surveillance, the challenge for the CDC lies in coordinating the numerous sources of data, varying data collection requirements, and managing the many different partners with whom the agency collaborates. The CDC currently maintains more than 100 surveillance and health information systems, but the majority of them was designed to detect a single organism or condition, and they are largely independent of one another.
Currently there is no single surveillance system in the country, but that may change this year when the CDC launches its National Electronic Disease Surveillance System (NEDSS) in 20 states. The goal is to better manage and enhance the large number of current surveillance systems and allow the public health community to respond more quickly to public health threats such as outbreaks of emerging infectious diseases or threats of bioterrorism. When completed, NEDSS is designed to electronically integrate and link together a wide variety of surveillance activities and will facilitate more accurate and timely reporting of disease information to CDC and state and local health departments. (See sidebar on page 36.) NEDSS will include data standards, an Internet-based communications infrastructure built on industry standards, and policy-level agreements on data access, sharing, burden reduction, and protection of confidentiality.
In November 2001, the House of Representatives heard testimony from a number of scientists and public health officials who charged that an updated infectious disease warning system could prevent needless illnesses and even deaths, especially in light of recent anthrax exposures and the threat of bioterrorism. Alan Zelicoff, a senior physicist at Sandia National Laboratories in Albuquerque, NM, was one of those who testified before the Committee on Energy and Commerce's Subcommittee on Oversight and Investigations. He said the public health system and the traditional medical care delivery system were "minimally prepared" to detect the early manifestations of disease that is intentionally introduced into a community.4
"There are many dirty little secrets in medicine," Zelicoff said in his testimony. "One of them is this: practicing physicians don't report unusual diseases to local public health officials (including signs and symptoms that could be due to bioterrorism), and public health officials don't have the ability to provide timely disease information to physicians working in clinics and hospitals. In my 10 years of medical practice, I never--not once--saw a physician or physician assistant pick up the phone to report a so-called 'reportable' disease. Even in areas of the country where reporting of a small set of key infectious diseases is a legal requirement, physicians rarely comply. Why? The process is burdensome, inefficient, and most importantly, almost never gives anything back to the physician that is of relevance to the patient he or she is caring for."
Zelicoff says that reporting efforts relies on the physician to first identify the fact he or she is dealing with an unusual disease, then knowing who to call, waiting for a public health officer to become available, and then following up a long stream of queries with documentation. Zelicoff emphasized that what is needed is a tool that will establish and maintain communication between overworked clinicians and out-of-reach public health officers, as well as a tool that is easy and intuitive to use, fast, responsive, and cost-effective. In light of recent anthrax exposures, Zelicoff insisted that a surveillance system must show whether or not the exposure to an infectious agent is widespread or localized.
"This is critical information for decision makers," Zelicoff said. "It goes directly to the question of how many people need to be tested, how many people need prophylaxis, and how many people should be followed up. Mark Twain had it right when he said, 'It ain't so much knowing about that what is, but not knowing about what ain't.'" Zelicoff explained that scientists at Sandia were able to determine that if public health officials can be apprised of the earliest cluster of illness that occurs after a large scale exposure or outbreak, the vast majority of individuals exposed-even to anthrax or smallpox-can be saved.
Zelicoff and his colleagues at Sandia, the University of New Mexico, Los Alamos National Laboratories, and the New Mexico Department of Health, developed the Rapid Syndrome Validation Process (RSVP), a syndrome-driven, infectious disease surveillance and reporting system, which was introduced last spring, refined during the summer, and installed at seven hospitals throughout New Mexico. One of the team members who helped create RSVP, Gary Simpson, MD, medical director of infectious diseases for the New Mexico Department of Health, says that RSVP was designed to assist busy physicians.
"Most of the infectious disease surveillance in this country is laboratory based, which is reliable and precise, but it's not always timely," Simpson said. "The lab can take days, weeks, or months to determine the ideology of a cluster of suspicious diseases or clusters of illness that might suggest a threat to public health. Physicians are just too busy and with a few exceptions like TB or the plague, they will rarely report because it means making a lot of phone calls and being placed on terminal hold and it's a pain."
Simpson continues, "The challenge was designing a system that would capture physicians' clinical judgment and experience by reporting syndromes that might represent infectious diseases. This isn't profoundly new, but if we could detect fever and altered mental status in a cluster of 10 cases in a community, that's probably an outbreak of meningococcemia, and yes, it would be timely to be aware of that before the lab got around to isolating it. The problem is, who's going to report the syndromes and how would you define them? Epidemiologic surveillance case definitions are extraordinarily complex. With hepatitis C, for example, the case definition is about a page long. Our approach was to say, when you see a patient with undifferentiated febrile illness, or influenza-like illness, that carries a lot of weight; it might be the flu, it might be an early case of Hantavirus, or it might be a case of early inhalational anthrax."
Simpson explains that a system like RSVP can help clinicians assemble pieces of the puzzle early enough to better determine the presence of a possible outbreak. He refers to the sudden presentation of Hantavirus in New Mexico in 1993.
"We were trying to leverage our experience with the Hantavirus outbreak to imagine a reporting system that would have allowed us to detect that outbreak more quickly," Simpson says. "We became acutely aware after the outbreak that there had been cases of what we retrospectively determined to be Hantavirus occurring weeks or even months prior to the outbreak's detection. We knew we needed a system that physicians would buy into because it would be valuable to the care of their patients, and it would collect epidemiologic data that had sufficient credibility. We also knew we had to make it a system that would work in the busiest emergency room, and we're testing it in high-volume settings."
Simpson says the development team examined syndromes that would manifest themselves through bioterrorism events as well as syndromes that had heralded outbreaks of emerging infectious diseases like Hantavirus, although the latter were the more common occurrences and the most appropriate database for the RSVP system.
"In most communities it will be unequivocably true that emerging infectious diseases would be a much more likely threat than a bioterrorism-linked threat," Simpson observes. "We concentrated on syndromes that characterized the presentation of vaccine-preventable diseases because by far, the most likely threats come from these garden-variety, everyday diseases. For instance, in New Mexico right now, we have two simultaneous pertussis outbreaks as well as a hepatitis A outbreak, and those happen all the time. We realized the core value of RSVP was vaccine-preventable infectious diseases because if we could detect their syndromes and link the detection of clusters of unusual illness in the community, we could link them to a statewide immunization-information system that we're building in New Mexico."
RSVP involves a sophisticated informatics program. According to Simpson, the reportable syndromes are identified and plotted by specific geographic location. The system was designed with a graphic user-interface that would appeal to busy clinicians and could facilitate the entering of epidemiologic information in under 30 seconds via touch-screen technology.
"It was an arduous but worthwhile process to design the touch-screen fields so they have minimal data sets to characterize the syndromes," Simpson explains. "The syndromes are grouped into the categories of influenza-like illness, acute bloody diarrhea, acute hepatitis, acute respiratory distress syndrome, fever and suspected CNS infection, and fever with rash. Inputting is elegant in its simplicity, in that physicians simply locate the screen buttons representing the epidemiologic data categories that best match the information they have to report."
Simpson says RSVP is designed to provide instant reporting of a syndrome that represents a public health threat. "If there are signs and symptoms of fever or rash, I would get paged the instant this case data was inputted so I could respond quickly," Simpson says. "We are working on creating a comprehensive set of signs and syndromes that represent all levels of urgency of reporting. This is important because quick response is key. Most clinicians view reporting as a unidirectional flow-it all goes away and they never hear anything back. So we promised we would provide to the clinician at least 10 times as much information back to them than they were being asked to input. If you input a syndrome into the system, you would get instant feedback, such as a geo-located map of your community, with your data represented by a dot on the map. You would also see the dots of any other syndromes reported in the area. For example, West Nile virus was detected only because the clinician had two elderly patients with encephalitis at the same time and said, 'Gee, that's unusual.' This is a way to see if there are other cases, at the hospital across town or 100 miles away."
Simpson adds that RSVP will undergo further testing throughout the 2001-2002 influenza season, when physicians will undoubtedly have a flurry of cases presenting flu-like signs and symptoms. "We'll be able to see the geographic clustering of these flu-like signs and symptoms, and be able to track them."
Recognizing an Outbreak and Taking Action
One of the most critical steps in investigating a potential outbreak, although elemental, is being able to recognize the presence of an outbreak. To do so, infection control practitioner (ICP) should know the endemic rate of disease at their facility, or the baseline rate.5 An outbreak or epidemic is defined as an increased rate of disease that is significantly higher than the expected baseline rate of disease during a specific time period in a specific location. To determine if a rate of disease occurrence is higher than normal, a reliable surveillance system must be in place before the start of an outbreak.
If an outbreak is suspected, the ICP should immediately visit the affected areas of the hospital, observing and reviewing patient-care practices that might be triggering the transmission of pathogens. It may be possible to interrupt the outbreak by correcting obvious procedural problems, thereby avoiding the need to conduct a full-scale outbreak investigation.
Here are a few tips from public health authorities on what to do during an outbreak investigation:
- Review the medical records of suspected affected patients and create a line-listing to determine which characteristics and potential exposures are most common among them. The line listing should contain the following information: age, gender, race/ethnicity, patient location, underlying diagnosis, medications, device exposures, personnel on shift, and any other symptoms or factors that may relate to the patient's risk of acquiring an infection.
- Develop a preliminary case definition that includes the people affected, the time of disease onset, and the location of the disease acquisition.
- Identify additional cases by searching records from departments including infection control, microbiology, epidemiology, and surgery.
- Construct epidemic curves (the distribution of case-patients by time) once all affected patients have been identified. The shape of the curve can help to hypothesize the mode of transmission and to find the source of the epidemic.
- Save isolates from the affected patients and possible sources of disease.
- Review procedures, observe practices, and interview staff in the affected areas.
- Develop an initial hypothesis to explain how the epidemic arose based upon available data and a review of current literature on similar outbreaks.
If there is a known epidemic, the following action steps should be taken:5
1. Notify local, state, or federal authorities based on reporting requirements. Suspected intrinsic contamination of sterile products, fatal blood transfusion reactions, infections caused by contaminated blood products, and infections associated with defective devices must be reported to the CDC.
2. Institute emergency control measures.
3. Save isolates from affected patients.
4. Quarantine suspected contaminated products/devices.
5. Conduct an initial assessment.
6. Conduct a full epidemiologic study.
7. Obtain samples from suspected equipment.
8. Institute more specific control measures.
9. Assess the efficacy of these control measures.
10. Maintain active surveillance for adverse reactions.