CDC: Technology May Prevent Pandemic
The Federal Centers for Disease Control and Prevention (CDC) thinks it has found a better way to detect, track and share information about a possible flu pandemic should one break out in the U.S. But first, the CDC will have to convince a skeptical medical community that its technology-based solution, called BioSense, is a prescription worth filling.
Designed as a network that sits atop participating hospitals' existing systems, BioSense gathers and analyzes their data in real time.
The idea is to obtain a constantly refreshed flow of patient information from the field, enabling CDC epidemiologists to immediately detect the early signs of an outbreak of avian flu or other flu-like disease, and help marshal forces to control it.
"Part of a larger bird flu initiative at CDC, BioSense is designed to give us situational awareness of what is happening in a hospital or a city," says Barry Rhodes, associate director for technology and informatics at the CDC and the project's co-leader.
Started in 2003 and made operational in 2004, the BioSense network first gathered medical information on a daily basis from Defense Department and Veterans Affairs hospitals and one large medical laboratory. But late last year, the CDC began an initiative to expand BioSense to, in effect, take the pulse of a much broader spectrum of society, connecting civilian hospitals around the country.
The CDC, which spent $50 million for hospital recruitment and technology last year, will spend another $50 million to expand the network this year to connect more hospitals. The federal agency also expects to start sharing, in turn, its analysis of local and regional influenza-like illness trends with health-care and other public agencies in affected areas.
But the CDC is having a hard time selling BioSense to the civilian medical community for a host of reasons that include security and privacy concerns, the need for hospitals to recode patient and medical records, the feeling of some health-care officials that they are already overburdened with information-technology projects, and the skepticism of hospital administrators who simply don't think they need the system.
Since the CDC began actively recruiting hospitals last fall, only a few dozen non-government hospitals have been connected.
"Like many I.T. projects, BioSense is more difficult to get running than we expected and it is taking longer to get it to work right," says Dr. Jerome Tokars, a medical epidemiologist working on the project.
The people who work at or for CDC see BioSense as a national imperative that will become widely used by hospitals throughout the U.S. sooner or later. "We had people from the medical
community last fall telling us what we're doing is impossible," says David Groves, vice president for public health informatics and CDC project head at SAIC, a BioSense contractor.
"We got the system up and running at 11 health-care
organizations representing 33 hospitals in a few months," Groves says, "and we've got hundreds to go this year. We will never connect all 5,600 hospitals in America, and even connecting the 1,000 hospitals in large cities will still take two to three years or more. But it's not impossible."
Since Jan. 1, those 33 hospitals have been feeding information about patients coming through their doors complaining of flu-like symptoms to the CDC via the network. So far, though, the flow of information has been a one-way street.
As of mid-February, no hospitals in the network were yet receiving reports on influenza trends in their local areas. The CDC expects to start making its local and regional analyses of influenza outbreaks and related-illness trends available to participating hospitals starting this spring.
The CDC now wants to connect at least 300 more hospitals to BioSense by year-end. But, according to Tokars, "I don't know if we will get to our goal." In hindsight, he says, "maybe it would have been good to get it up a few years ago."
CDC: Technology May Prevent Pandemic
The agency has signed on the Constella Group, a health-care information-technology consultant, to try to bring
hospitals along. "We explain to them the benefits of having this data in a standardized form, and that they are consciously participating in a project that is in both their interest and for the public good," says Wayne Myers, director of Constella's Health Sciences division.
But that still might not be enough for the CDC to overcome hospital resistance. "There are significant practicality issues," says Dr. Susan Fernyak, director of the communicable disease control and prevention program at the San Francisco Department of Public Health. "It is no small thing for a hospital to take on another I.T. project these days."
For instance, simply to get a local hospital to kick the habit of faxing data to the city health department in favor of transmitting it electronically is a major challenge, Dr. Fernyak points out. "You'd think that would be a no-brainer, right?" she asks rhetorically. "But a lot of work has to go into it to make it happen. It takes software, hardware and personnel resources to make the transition."
To ease the information-technology burden on participating hospitals, the BioSense network cleverly depends only on medical data that has already being gathered routinely at most large hospitals. This includes information such as emergency-room admissions records, radiology test results, laboratory test findings and prescriptions.
To bypass the need for additional data entry, the CDC installs on each participating hospital's network custom software that the agency developed. This system, in effect, "listens" to the facility's network traffic, filtering out the non-flu-oriented diagnoses and test results and capturing relevant patient records, diagnoses and prescription information.
The software converts this information to the health-care industry's standard HL7 data messaging format, aggregating and transmitting it over the Web to the CDC using public key infrastructure (PKI) encryption.
The system collects and transmits the hospital data every 15 minutes to a central data repository at the CDC. The data to be captured and transmitted includes patient demographics, the chief medical complaint, onset of illness, diagnoses, medical procedures, laboratory results including microbiology information, radiology results and medications prescribed.
But in order for BioSense to capture data for transmission, a hospital must standardize patient and other medical information. Most hospitals use their own data codes.
This means that once a hospital agrees to participate, the CDC and its contractors must work to modify the hospital's data codes as well as map and bundle the data so it can be transmitted and received. "To standardize the data and do all the data validation steps is a huge technological challenge," Rhodes asserts.
CDC technicians first need to learn how a hospital identifies relevant medical conditions and then recode them according to the agency's standards for each symptom, malady or drug. This allows the CDC's software to listen to the hospital's network and identify relevant pieces of information about a patient who came in complaining of cough, fever and headache, as well as the patient's diagnosis and any prescribed medications.
Making this process a little easier is the fact that many large hospitals already use an "interface engine," essentially a hub, to route medical information around the organization's network. "We're using a similar tool that we call the BioSense integrator," Groves of SAIC explains. "It stuffs through the same data flows, filtering the information we want, changing its format, putting it in the right code and batching it every 15 minutes to the CDC."
The federal disease control agency, in effect, taps into the hospital's network, opportunistically taking the data it wants for analysis.
Few would argue with the CDC's ultimate goal: to permit a faster, better informed, more effective response that could save millions of lives. Yet the medical community questions whether the BioSense network is worth it.
"If there is a pandemic flu, we are not going to know about it from a system like this," Dr. Fernyak insists. "In the case of a pandemic flu, we're not going to need an electronic data system to tell us there's a new strain of flu out there."
Adds Dr. John Rosenberg, director of the Infectious Disease Laboratory at the State of California's Department of Health Services in Richmond, Calif., "I think it would be particularly useless for influenza, because for any influenza-like illness, there is no way to say someone has it for sure." Physicians may record six or seven different diagnoses for influenza-like symptoms, which usually include fever, nasal congestion, muscle and bone aches, and include chest X-ray results, he says.
Dr. Rosenberg says that if an epidemic broke out, "You'd know it before the data rolled in. When your emergency rooms fill up, you make a phone call; this is probably a better measure. We're not advocating or supporting the BioSense network, but if a health-care organization wants to participate in it, that's fine."
CDC: Technology May Prevent Pandemic
Big Picture on Contagion">
But a hospital's medical staff might not know they have a serious problem on their hands when patients start coming in with symptoms, Groves says, while CDC scientists and analysts will be able to see a major flu outbreak as it develops over a broader geographic area. Having this bigger picture will help public agenciesnot just medical but police, emergency and National Guard units, if neededto marshal a better-organized response.
Another issue that CDC must deal with to obtain wider acceptance of BioSense is turf poaching. Although participation on the part of hospitals is voluntary, many physicians and health officials see the CDC's impending spread of BioSense as a sign of the federal government encroaching on what has traditionally been the domain of local health-care providers and organizations.
They say it's their responsibilitynot the CDC'sto respond to and manage a local or regional flu outbreak. "The local health departments and hospitals are responsible for dealing with a pandemic," Dr. Fernyak says. "I think the CDC is not taking into consideration the local situation. CDC is not going to be responding to a pandemic on the ground. The local hospitals and physicians are."
Another challenge for the agency is convincing hospitals to allow the federal government to, in effect, listen in on their daily treatment of patients on a real-time basis. The privacy and security of the patient information that hospitals are being asked to surrender is one reason many are reluctant to sign up. The Health Insurance Portability and Accountability Act (HIPAA) restricts their use of patient data and stipulates the need for medical organizations to rigorously safeguard patient information.
HIPAA, though, makes an exception in this case, since the law allows hospitals to share information with public-health agencies such as the CDC, which guarantees that no data is used that would enable identification of individual patients.
Even so, physicians are decidedly chary about surrendering patient medical information to Uncle Sam's databases.
"The CDC has not told us how they plan to deal with the issues of confidentiality and privacy of this information," Dr. Fernyak says. "We have a lot of concerns about the handling of that data and how it will be used by the CDC."
Surprisingly, until the first of the year, the CDC had been receiving regular patient information once a day only from naval and Army hospitals, VA hospitals and a major medical laboratory. This data is still coming in, but the CDC expects this year to gradually wean itself from depending only on this dated information as a larger number of hospitals' patient medical information comes onstream via BioSense.
The agency has also been depending for early detection on a variety of largely ad hoc sources of patient and medical information. For instance, CDC analysts monitor mortality rates in various cities. If a spike in mortality rates occurs in a particular city, agency staff investigate further, checking for other indications such as a corresponding jump in outpatient visits or an increase in reports of viruses.
But gathering this data was slow and labor intensive. "Before now, there never was an attempt to aggregate this much data in near-real time to see what is happening in a hospital or in a city," the CDC's Rhodes explains.
The medical information transmitted daily from government hospitals, for instance, was extremely limited in scope, excluding all civilian medical centers. It was also stale in medical termsthree days old on average. That latency makes the information nearly useless for anyone trying to detect and stem an outbreak of an infectious diseasea situation that experts say can change dramatically each day, if not hourly.
That's why the CDC feels there's no time to waste.
Humans have been stricken with avian influenza in China, Turkey and Nigeria, with more than 90 fatalities since 2003. Most recently, on Feb. 10, Indonesian officials reported that a
23-year-old woman from West Java province died of bird flu.
Experts, including Dr. Fernyak, are concerned that sooner or later, the virus could mutate and spread widely and
rapidly among humans, causing a pandemic. "It could happen in four months or 10 years," she says. "Statistically it's going to happen sometime; we just don't know when."
CDC: Technology May Prevent Pandemic
CDC Base CaseHeadquarters: 1600 Clifton Rd., Atlanta, GA 30333
Phone: (404) 639-3311
Business: Federal agency charged with preventing the spread of diseases.
Associate Director, Technology: Barry Rhodes
Annual Budget for BioSense Project: $50 million
Challenge: Establish a system to detect, track and share information with hospitals if there's a flu pandemic.