Wednesday, May 6, 2009 | Two numbers are crucial in the fight to contain a flu outbreak. The first is the number of confirmed cases. The second is the number of people who have the disease but have not been identified.
Not knowing that second number is what caused the swine flu hysteria of the past few weeks — a reaction that some are now saying became an overreaction.
Some health experts are now wondering whether closing schools for weeks, elevating the World Health Organization’s pandemic threat gauge to its second-highest level and virtually shutting down Mexico City was justified for a disease that has so far killed far less people than the normal seasonal flu.
San Diego State University informatics researcher Faramarz Valafar says it doesn’t have to be this way the next time a virus hits. He claims to have developed a computer system that would give the Centers for Disease Control and Prevention a good estimate of the second number, which in turn will allow officials to mount a more accurate response.
The system, which he calls Real-Time Disease Tracker, would be integrated into the CDC’s website, and track people’s searches about specific diseases like the flu. And then, by overlaying this information with geographic and other data, the system would allow epidemiologists — at a much earlier point — to detect a possible outbreak.
By monitoring the pages being visited and questions being asked by visitors to the website, officials would have far more information about people who are feeling flu symptoms but have not yet have gone to the doctor, Valafar said. And by incorporating the geographic data epidemiologists could, for example, see how a disease might be spreading from San Diego to Anaheim based on searches coming from those cities.
“If we had this information, we would have known what we should close down, what we shouldn’t close down — we would have had much better answers,” said Valafar, an associate professor who runs SDSU’s Bioinformatics and Medical Informatics Research Center.
The system is still in its demo stage, and Valafar has yet to officially present it to the CDC, but the concept is a powerful one, say public health experts. “It would be a great leap forward in terms of tracking infectious diseases and knowing the predicted course of the prevalence of disease. Almost like having a crystal ball,” said Kevin M. Kelly, associate director of emergency medicine at Alvarado Hospital.
Officials at the CDC, however, are not yet sold on the system. They acknowledge its potential, but worry that it could send epidemiologists on goose chases after phantom outbreaks. “I think it is a good experiment to do,” said Leslie Lenert, director of the CDC’s National Center for Public Health Informatics. “But I would like to see the results, and know how many times the information was wrong.”
The debate over Valafar’s system highlights the convergence of the pillars of internet 2.0, such as social networking and community intelligence, with the centuries-old fight to contain infectious diseases. No one is denying that these societal changes will play an increasing role in public health. The argument is over how far things should go.
Valafar said his system would provide the foundation for the CDC’s website to evolve into a Wikipedia-like environment where epidemiologists and lay people can post real-time information on the latest regarding diseases from the flu to cancer. Lenert and his CDC colleagues are not ready to turn over one of the world’s most authoritative sources on critical health information to the masses.
The key concept behind the disease tracker is that people in the wired world increasingly turn to the internet before their doctor when they have questions about their health; and, most importantly, when they are feeling ill.
By tracking that usage, public health officials would know when the number of queries about a specific virus in a specific area spiked above a baseline level. This would give them a heads up on a possible outbreak weeks, even months, before they would through the medical system.
“I would love to be able to get a paper from CDC that says the early warning system is predicting a spike in the swine flu and you must prepare for an influx of patients now,” Kelly said. “That is much more solid that what we have now which is primarily guesswork.”
Valafar and his colleagues did a simulation of the system using data from the 2003 SARS outbreak in Asia. Using partial information on phone and internet inquiries to the WHO about SARS, Valafar said the researchers were able to accurately show how the disease progressed from China to Taiwan.
Valafar acknowledges limitations to the system. The most obvious one being that it only works in the industrialized world where internet connections are prevalent. But, he said, even in the developing world doctors and nurses use the internet for information on the diseases they treat, and spikes in those searches could be captured as well. And, of course, the system will only work if the CDC buys into it.
Lenert has not seen the SARS demo but said the CDC has experimented with Google’s flu tracker, which has some similarities to Valafar’s program. He said his main concern with a system based on internet searches is the “false positives” it would generate.
“It is important that when the system detects an outbreak, that it is really there,” Lenert said. “When there is a new thought or idea going around the internet about health, it doesn’t necessarily mean an outbreak. It could mean that someone was on Oprah.”
The opinions of the CDC official and SDSU researcher further diverge when it comes to taking the Wikipedia concept to the CDC’s website. Valafar contends that by allowing the website to be a repository of the “collective intelligence” of the community, the CDC will be giving the public more timely, and therefore more useful, information on public health issues.
Like it or not, he said, the world is moving away from the Encyclopedia Britannica model. “Whether Wiki should be used or not is obsolete,” Valafar said. “The information age has changed the way people behave and the way information is being communicated, for good or for bad.”
Lenert said the CDC has to remain the authoritative source for health information, and can’t afford to spread misinformation, which the Wiki model is vulnerable to doing, even if it is soon corrected.
“We can’t tolerate inaccuracies during the editorial process,” Lenert said of the Wiki model. “We can’t have, for example, the CDC saying ‘it’s dangerous to take aspirin after a [heart attack]’ even if it would eventually be corrected. As a government agency responsible for public health, we have to be the Encyclopedia Britannica — trusted and authoritative.”