[ Table of Contents ]

Surveillance Systems Give First Alert

By Susan Sample

The first sign of a bioterrorist attack may be as inconspicuous as a flag on a computer screen in Salt Lake City. Yet it could unleash a response as massive as a network of early warning satellite dishes spanning the globe—and help save just as many lives.

Two electronic disease surveillance systems tested during the 2002 Winter Games in Salt Lake City continue to operate at or in conjunction with the University health sciences center. Each has a unique focus: ALERT covers University Hospital and clinics, while RODS monitors University Hospital’s Emergency Department and nine emergency rooms and 19 acute-care clinics owned by Intermountain Health Care (IHC) in seven Utah counties. Both systems share the same goal: to detect in real time patterns of symptoms that would indicate a bioterrorist attack or an outbreak of disease.

“Surveillance systems are a basic public health tool,” said Kurt T. Hegmann, M.D., M.P.H., research associate professor of family and preventive medicine at the U School of Medicine. He is one of the developers of the U of U Community Clinics Bioterrorism Surveillance System, which ended with the Paralympics last March. “What is new is the use of large computer databases for disease surveillance.


“The prior systems were much inferior,” explained Hegmann. “People would go to the ER with a computer, or pad of paper, and enter data from paper logs. You were counting on no errors and no lost patients in the logs. There were problems not only with manpower, but timeliness. You’re talking typically days—not hours—to analyze data.”

When bioterrorism is suspected, time is critical. “An early surveillance system all across America will be an unbelievable benefit for us in combatting natural diseases—and a sheer necessity to respond rapidly and appropriately to biological agents,” said George L. White Jr., Ph.D., M.S.P.H., professor and director of public health programs in the Department of Family and Preventive Medicine.

Hegmann collaborated with statistician Stephen Alder, Ph.D., research associate in the U public health program, to create the surveillance system for Þve of the University’s community clinics: Greenwood, Redwood, Stansbury, Summit, and Westridge. Supported by the Thrasher Fund, the system analyzed numbers of diagnoses, trends over time, and constellations of disorders grouped to detect any patterns. Data were interpreted with 100 different graphs per day for the previous 24-hour period.

“What we showed was a substantial outbreak of inþuenza in the community. It was worse than other years, and it coincided with the Olympics,” said Hegmann. “The benefits were that we communicated this information to the Salt Lake valley and state health departments. We also gave the information back to providers to let them know, so they could improve their patient care.

“The strength of our system is that we knew who saw a doctor, and why, in that time period. The weakness of our system is that we had only 100,000 patients covered in the clinics.”

The ALERT system—Advanced Logic for Event Detection in Real Time—monitors not only patients seen in the University Hospital’s emergency department and outpatient clinics, but also those admitted to the hospital. During the Olympics and Paralympics, the hospital’s Polyclinic also was included in the surveillance system. It was developed by Adi V. Gundlapalli, M.D., Ph.D., a former fellow in the medical school’s Division of Infectious Diseases and associate hospital epidemiologist, in collaboration with Louise J. Eutropius, R.N., B.S.N., C.I.C., hospital infection control practitioner, and Matthew H. Samore, M.D., hospital epidemiologist and associate professor of internal medicine. Technological help was provided by TheraDoc, Inc., of Salt Lake City.

To streamline the hospital epidemiologists’ responsibility to report cases of “notifiable diseases” (a list of 60-70, including diarrheal diseases and inþuenza, compiled by the Council of State and Territorial Epidemiologists), ALERT flags on the computer screen positive tests, in addition to listing other laboratory or radiology tests ordered, results, and basic demographics on each patient: name, age, sex, address, and the ordering physician.
Developed at the University of Utah, the ALERT system (Advanced Logic for Event Detection in Real Time) tracks how many patients are seen daily in the emergency department and outpatient clinics. Graphs, such as teh one above, show not only how these totals match up to expected levels, but also when the rates hwave reached warning limits.

“This is where we’re novel,” noted Gundlapalli. “We’re trying to pick up a number of different signals to alert us to natural disease outbreaks or agents of bioterrorism. We came up with lab-based signals—how many patients had orders for spinal fluid cultures, blood cultures, stool cultures—since ordering patterns can give you an idea of what might be happening.

“We look at radiology reports generated. The computer will read radiologists’ dictated reports and flag X-rays that are positive for pneumonia. This is tremendously important,” he noted, “because respiratory problems can be symptoms of weapons of mass destruction, or an outbreak.”

ALERT also tallies how many patients are seen in the emergency department, what their complaints are, and what their actual diagnoses are. Patients are categorized according to the severity of their condition, ranging from Level 1 to 5 on a scale Gundlapalli devised. For instance, a physician may suspect that a patient in the emergency room complaining of a cough and fever has pneumonia. If results from a chest X-ray are positive, the patient is Level 1. If the physician also orders a blood test and the results show an elevated white count, indicating an infection, the patient is Level 2. If the patient is admitted to the hospital and has a blood culture, “then you know this is a pretty sick patient: Level 4,” said the epidemiologist.

The computer also looks for “nested” signals by combining data. If a stool culture was ordered; the patient had a high white blood count or low white blood count; and the patient had been seen in the emergency department or clinic within 24 hours and a blood culture was ordered, then the “nested” signals would indicate infectious diarrhea.

Gundlapalli monitored ALERT three to four times a day during the Olympics, but, as he noted, “The beauty was that it only took 30 minutes with this system.” After reviewing the signals, he prepared a daily report, which was sent to the Utah Department of Health and the Salt Lake Valley Health Department. No patient names were included. Patient confidentiality is a major concern in all surveillance systems operating at the U. Security safeguards have been incorporated to allow only certain individuals access to data.

Should additional funding become available, Gundlapalli would like to extend ALERT to other hospitals in Utah. Another possibility would be to incorporate some of the system’s unique characteristics into RODS, Real-Time Outbreak and Disease Surveillance. Developed at the University of Pittsburgh Medical Center in Pennsylvania, RODS was activated in Utah during the Olympics and Paralymics, and monitored by Per H. Gesteland, M.D., senior fellow in the medical school’s Department of Medical Informatics and a fellow with the National Library of Medicine.

“ALERT is very sophisticated. It looks at lots of signals in case-level detail,” said Gesteland. “RODS captures large cohorts and looks at one type of information: a patient’s chief self-reported complaint. It’s coarse data, but it’s very informative and very powerful.”

During the Olympics and Paralympics, RODS analyzed in real time the chief complaint—such as a sore throat, fever or rash—of all 2,500-3,000 patients reporting each day at a variety of sites: University Hospital’s emergency department and Polyclinic, and IHC’s nine emergency rooms and 19 acute-care clinics in Salt Lake, Summit, Utah, Davis, Weber, Morgan, and Wasatch counties. RODS also maps the complaint to one of seven syndromes relating to bioterrorism. For example, a respiratory complaint is linked to anthrax, rash to smallpox, hemorrhagic fever to ebola.

Data were processed in Pennsylvania; when the number of cases reported exceeded expected levels, Gesteland and several other public health officials received an alarm, an automated text message on a personal pager. They examined details on their computers and convened a RODS Technical Advisory Group, which included a Utah Department of Health ofÞcial, a local health department official from the county where the alarm was triggered, and a Pittsburgh researcher.

Twice, Gesteland was paged. Fortunately, both were false alarms. If there had been an outbreak, the advisory group would have reported within two hours to another team of officials who would have decided how to respond.

This summer, the U of U, IHC, the Utah Department of Health, and University of Pittsburgh were joined by the Indiana University School of Medicine in submitting a proposal in response to a solicitation from the National Library of Medicine to use RODS in a three-year contract to determine if the early warning surveillance system is “scaleable” and “sharable.”

“By scaleable, we mean: is what we did with RODS for the Olympics something we can we do for the state and the region?” explained Reed M. Gardner, Ph.D., professor and chair of the U Department of Medical Informatics. “If we have a set of rules that work here, will they work in Indiana or another state?”

Bacterial sensitivity and distribution, as well as antibiotic resistance, can differ with geography. “The way bacteria spreads and is treated in a valley like ours is different than in an urban setting,” said Gesteland, who noted there are “other issues” to consider as well, such as different methods health systems use to enter patient data.

The multi-center study would enable RODS to expand “broader and deeper,” said Gardner. “We would expand our coverage by including additional health-care facilities in Utah and by incorporating additional data like laboratory and X-ray reports.”

Not only are the surveillance systems critical to detecting bioterrorism outbreaks, but they also are very important in handling community-wide epidemics. “I’m very interested in systems that will help me as a physician,” noted Gesteland, who is board-certified in both internal medicine and pediatrics. “Knowing which infectious pathogens are circulating in the community at any given time will definitely affect how I treat a child I see in the clinic.

“When you can take the data and return it to the point of care, and affect the delivery of patient care, it’s exciting.”

The ALERT surveillance system is unique in how it enables University Hospital epidemiologists to track numerous signals in case-level detail, including patients' compliants and diagnoses, as well as laboratory and radiology tests, which could alert physicians to a natural disease outbreak or agents of bioterrorism.

We always welcome your comments about the magazine. Address letters to: Editor, Health Sciences Report, Office of Public Affairs, University of Utah Health Sciences Center, 50 North Medical Drive, Salt Lake City, UT 84132. FAX: (801) 585-5188. E-mail: Susan.Sample@hsc.utah.edu.

[ Table of Contents ]