computerwoche.de

Archiv

IT not doing its part in the fight against avian flu

30.01.2006
IT is in a deadly race with the avian flu, which first appeared in China and Southeast Asia and has now reached as far west as Turkey. The use of IT for tracking and potentially defeating a pandemic is being hampered by a lack of best practices, a need for more comprehensive state-of-the-art systems and the absence of a system to share data consistently from the actual sites of outbreaks, according to experts at the Centers for Disease Control and elsewhere.

In North America, agencies at the state, local and federal levels use a wide variety of systems of differing degrees of sophistication to track influenza and share basic data among public health officials. Some of these are homegrown; others are "syndromic surveillance" applications -- dedicated systems created by third parties that track epidemics and help trigger a response from relevant agencies. Even with the information that's collected currently, the systems are unable to automatically do the requisite analysis and make the necessary recommendations to mount the most effective and aggressive response to stop a pandemic, say health officials.

There are virtually endless ways to improve the existing surveillance systems, according to Lynette Brammer, epidemiologist for influenza at the Centers for Disease Control in Atlanta, which uses seven different applications to track flu outbreaks. "The majority of our systems have been running for years and years, and some of the technology is definitely not state of the art," she says. The applications primarily feature Microsoft SQL Server database technology on the back end and an Access database management system on the front. The CDC also employs a SAS Institute Inc. database with custom-written scripts to track flu appearances across 122 cities in the U.S.

The databases handle various activities, says Brammer. For instance, one checks to see if a given city is experiencing a spike in mortality rates. If so, staffers can determine whether there are corresponding jumps in outpatient visits or reports of virus outbreaks. The data could also be checked against another system, which would analyze it to determine whether a given vaccine was effective and why. The agency would be able to see where the outbreak occurred and send alerts to doctors so they could recognize patients who are at risk.

The process of gathering data, which is received by fax or via the Web, tends to be slow and labor-intensive, says Brammer. For instance, one of the ways the CDC gleans its information is from the U.S. Influenza Sentinel Provider Surveillance System. This Web-accessible tool takes feeds from 2,200 volunteer doctors who have collected information from patients exhibiting flulike symptoms. The CDC would like to streamline the collection process, Brammer says. Efforts to establish a way to access hospital data as well are ongoing. The agency also taps into the World Health Organization's FluNet system, a Web-accessible database that lets users query on influenza-related activity by country or region.

Tying together the various systems for rapid analysis could be difficult, because every flu outbreak has a different pattern, Brammer says. Symptoms may be consistent between outbreaks, but geographic patterns aren't, and the peak season is long and variable, running from December through March, she notes. Interpreting the data probably will require some sort of manual intervention.