Catherine Strommen, an elementary school teacher in Fremont, California, padded to the kitchen computer and took one last look at her favorite chat room, Teachers.Net. It was midnight, February 9, 2003. At that late hour, two or three stray chatters were scrolling down the screen—likely Asians or Australians, who usually logged on when Americans were turning in.
One of the posts came from China—someone who called himself “Ben” (most adopted Anglo names). But his tone was distinctly different from the usual exchanges about pedagogy, food, and literature. In tenuous English, he described an illness that started like a cold, but killed its victims in days. A friend who worked in a hospital had died, as had the friend’s mother. In Guangdong Province, where Ben resided, hospital doors were locked.
Alarmed, Strommen emailed an old neighbor and friend, Stephen Cunnion, M.D., a retired Navy physician and epidemiologist who now lived in Maryland. A practical, no-nonsense man, Cunnion started searching the web. With no success, he tried a new tack—sending an email to ProMED-mail, a global electronic reporting system for outbreaks of emerging infections and toxins. After quoting Strommen’s missive, he asked: “Does anyone know anything about this problem?”
The tiny ProMED staff conducted its own web search. It, too, came up empty-handed. On February 10, it sent out to tens of thousands of subscribers a posting headed: “PNEUMONIA – CHINA (GUANGDONG): RFI,” or Request for Information.
Thus did the world first learn of SARS, the new and deadly infection that would kill 774 people and infect 8,000 in 27 countries. The next day, the World Health Organization issued a belated bulletin on the raging epidemic.
Welcome to modern disease surveillance. When a serendipitous exchange on a teacher’s chat room morphs into a global outbreak alert, public health practice will never be the same. Until a few years ago, we had to wait for bureaucrats to announce disease emergencies within their borders. Today, abetted by the Internet, outbreak news is highly contagious. Web crawlers, automated translation, GIS mapmaking, ISP tracking, and cell phone imagery have created a world where, for good or ill, rumors of outbreaks outpace confirmation.
And it’s not just fast-moving infections that are flushed out by high technology. Social epidemics, such as the next new fad in drug abuse, are also being monitored through canny—and perfectly legal—eavesdropping on public electronic forums. The Internet has even been used to intervene in public health emergencies—by emailing, for example, online daters exposed to sexually transmitted diseases.
If Hippocrates were alive today, he’d probably write a treatise on the Internet. But even in his time, he had a plugged-in perspective. In the classic On Airs, Waters, and Places, he described in vivid detail how disease was influenced by the seasons, winds, water, geography, and what today we would call “lifestyle.” Armed with those facts, aspiring healers, he wrote, “must proceed to investigate everything else.” In 2008, that’s just what’s happening online. The father of medicine would probably need three laptops to keep up with it all.
ProMED-mail, a program with the International Society for Infectious Diseases, is the iconic example of how the Internet is advancing public health. Launched in 1994, it arrived at a moment when CompuServe and AOL were fueling personal email, and when scientists were becoming aware of the intensifying emergence of new infectious diseases, such as Ebola hemorrhagic fever. It was also a time when government health officials still tightly guarded information on disease outbreaks. According to Larry Madoff, M.D., ProMED’s current editor, the thinking among its founders was: “Wouldn’t it be great if people who had access to the Internet in far-flung places, and were seeing something unusual, could send an email to this list? We wouldn’t have to wait for a public health laboratory to notify the Under Minister of Public Health, who would notify the Minister, who would notify the World Health Organization. We would all know at once.”
ProMED’s virtues are those of the Internet itself: It is fast, nimble, egalitarian, bottom-up, open, and transparent. These are also the hallmarks of an effective public health campaign: delivering the facts swiftly to those who need to know. China’s SARS epidemic actually surfaced in November 2002—three months before the fateful ProMED posting. If a ProMED subscriber in Guangdong had written directly to the listserv in, say, mid-December—by which time the epidemic had spread to three cities—who knows? Perhaps the scourge would never have vaulted beyond the country’s borders.
Despite the explosion of web innovation, ProMED remains conspicuously low-tech and hands-on. Individuals feed outbreak information into the system through email, while ProMED’s small staff gleans material from its own searches of media and official reports. The seven or eight daily postings, available through email or on the website, are thoughtfully moderated and commented upon by a panel of experts—a unique step that adds meaning and context to otherwise unmediated facts or assertions. Each posting is limited to 25 KB bandwidth—to ensure that it slips through an old-fashioned dial-up modem in the most remote areas of the world (where new infectious threats tend to smolder). “We use technology that was state-of-the-art in 1994. We use email—plain-text email at that. We don’t use fancy fonts,” Madoff says. “The power of the Internet is its ubiquity and speed; it’s not necessarily in all the neat things you can do.”
More technologically awesome is GPHIN—the Global Public Health Intelligence Network, part of the Public Health Agency of Canada. GPHIN is a secure Internet-based early warning system that gathers preliminary reports of public health significance in nine languages, round the clock. Launched as a prototype in 1998, it relies on news aggregators to scour 20,000 media sources—from wires to websites to blogs. Every 15 minutes, its software pulls articles and assigns them a relevancy score based on keywords and syntax. Non-English articles are machine translated. Expert analysts further parse the material, which ranges from infectious diseases to natural disasters to product safety.
Each week, GPHIN sends hundreds of reports to its main client: the World Health Organization, as well as to government officials, NGOs, and other subscribers, who pay tens or hundreds of thousands of dollars annually. Unlike ProMED—which operates on a shoestring, is free of charge, and is open to the public—GPHIN is government-affiliated, pricey, and restricted to organizations with an established public health mandate: the loaded Lexus compared to the stripped-down Corolla.
Both services face the challenge of “noise.” Flooded by disease news—some vitally important, most not—what to publish? GPHIN actually snagged and sent out early Chinese-language news items about what later became SARS. In November 2002, it posted a report that an unusual number of people were showing up in an emergency room in Guangdong, appearing to suffer from atypical pneumonia. In January 2003, GPHIN’s computers found an English-language financial report from a pharmaceutical company, describing a spike in sales for its antiviral drugs, which the firm conjectured was due to an unusual outbreak in the Chinese province.
At the time, these reports didn’t stand out; only after SARS was front-page news did investigators comb the record and discover these unheeded warnings. In the massive daily GPHIN trawl, stories about respiratory symptoms during flu season in a country of 1 billion people were understandably overlooked.
Which is to say that the power of the Internet to amass ever-larger haystacks of information makes it more and more difficult to find the needles. According to Associate Editor Marjorie Pollack, M.D., ProMED’s lead moderator during the SARS outbreak, “Do you post every individual unknown case of everything? What’s your threshold? What will your system tolerate? Is a single undiagnosed case of something unusual of global importance?”
Sometimes, it is. A strongly suspected case of smallpox would demand instant attention—because it would have to have been terrorist-sown. So would pneumonic plague, which is spread person-to-person through the air, outside areas where it is naturally endemic in wild rodents.
Yet a mass of data culled from news aggregators is riddled with false positives and false negatives. “We deal with a spectrum of electronic information. It’s a constant tradeoff between sensitivity and specificity,” says John Brownstein, Ph.D., an assistant professor of pediatrics at Harvard University Medical School.
Brownstein has fashioned a different form of electronic surveillance: real-time disease mapping. With computer scientist Clark Freifeld, he is scraping news aggregators such as Google News, as well as ProMED, WHO reports, and other sources, using software that matches a glossary of infectious disease terms with words related to geography. He then superimposes his results against a background of gorgeous Google satellite maps.
Brownstein’s HealthMap is the first real-time disease alert that conveys facts graphically rather than verbally—a comprehensive and dynamic portrait of the world’s well-being. Viewers can scan cities, countries, or continents for hourly updated reports. Teardrop-shaped “heat” markers indicate the level of urgency: red for, say, ongoing outbreaks of Ebola in Uganda or human “bird flu” cases in Egypt, yellow for older reports of food recalls, and the like. Viewers can click on news updates, and cut the data according to news feed, disease, country, or timeliness.
Brownstein’s quest to deep-mine the Internet sprang from the realization that just about everything a disease cartographer needs is already on the web. “We use tools that are freely available and data sets that are open to the public,” he says.
Who clicks on HealthMap? Not only web surfers, but also governments and public health officials. In a world where pathogens need no passports, bureaucrats often don’t know what’s fulminating just across the border—and could soon jump into their own backyard. Brownstein’s next ambition is to track increases in keyword searches by geographic region. With algorithms linking Google Trends—which displays the top cities, regions, and languages in which users type in a given search term—to Google Maps, he hopes to lexically locate nascent emergencies. Such an approach may have uncovered, for instance, a surge in searches for “pneumonia” and “flu” in Guangdong in late 2002.
The Internet’s most salutary effect on public health is that governments can no longer hide outbreaks—though, as SARS proved, they can try. Ever since the WHO passed its revised International Health Regulations in 2005—an agreement to contain emergencies at the source, not only at national borders—the agency has relied more and more on “informal sector reports” from media and websites, and less on official candor. Before these new regulations, the WHO itself used to publish a weekly Outbreak Verification List, known in the trade as the Wednesday “rumor list,” a tally of unconfirmed disease reports from all over the world that only a select group was permitted to read. Now the Internet generates its own rumor lists—24/7, for anyone to peruse.
A GPHIN or ProMED posting is a wedge for the WHO to convince reticent nations to come clean with the truth—or for outside authorities to investigate. As GPHIN chief Abla Mawudeku puts it, “WHO can knock on a country’s door and say, ‘We have some information about an event occurring in your backyard. It’s from the media. Could you help verify this?’”
As communications technology gains definition, so will disease surveillance. “We can locate outbreaks within countries and provinces, but we can’t locate them within a farm or village,” says Madoff. “In a blue sky future, I can envision being able to know where individual cases are.”
Which is to say that the power of the Internet to amass ever-larger haystacks of information makes it more and more difficult to find the needles. According to Associate Editor Marjorie Pollack, M.D., ProMED’s lead moderator during the SARS outbreak, “Do you post every individual unknown case of everything? What’s your threshold? What will your system tolerate? Is a single undiagnosed case of something unusual of global importance?”
Sometimes, it is. A strongly suspected case of smallpox would demand instant attention—because it would have to have been terrorist-sown. So would pneumonic plague, which is spread person-to-person through the air, outside areas where it is naturally endemic in wild rodents.
Yet a mass of data culled from news aggregators is riddled with false positives and false negatives. “We deal with a spectrum of electronic information. It’s a constant tradeoff between sensitivity and specificity,” says John Brownstein, Ph.D., an assistant professor of pediatrics at Harvard University Medical School.
Brownstein has fashioned a different form of electronic surveillance: real-time disease mapping. With computer scientist Clark Freifeld, he is scraping news aggregators such as Google News, as well as ProMED, WHO reports, and other sources, using software that matches a glossary of infectious disease terms with words related to geography. He then superimposes his results against a background of gorgeous Google satellite maps.
Brownstein’s HealthMap is the first real-time disease alert that conveys facts graphically rather than verbally—a comprehensive and dynamic portrait of the world’s well-being. Viewers can scan cities, countries, or continents for hourly updated reports. Teardrop-shaped “heat” markers indicate the level of urgency: red for, say, ongoing outbreaks of Ebola in Uganda or human “bird flu” cases in Egypt, yellow for older reports of food recalls, and the like. Viewers can click on news updates, and cut the data according to news feed, disease, country, or timeliness.
Brownstein’s quest to deep-mine the Internet sprang from the realization that just about everything a disease cartographer needs is already on the web. “We use tools that are freely available and data sets that are open to the public,” he says.
Who clicks on HealthMap? Not only web surfers, but also governments and public health officials. In a world where pathogens need no passports, bureaucrats often don’t know what’s fulminating just across the border—and could soon jump into their own backyard. Brownstein’s next ambition is to track increases in keyword searches by geographic region. With algorithms linking Google Trends—which displays the top cities, regions, and languages in which users type in a given search term—to Google Maps, he hopes to lexically locate nascent emergencies. Such an approach may have uncovered, for instance, a surge in searches for “pneumonia” and “flu” in Guangdong in late 2002.
The Internet’s most salutary effect on public health is that governments can no longer hide outbreaks—though, as SARS proved, they can try. Ever since the WHO passed its revised International Health Regulations in 2005—an agreement to contain emergencies at the source, not only at national borders—the agency has relied more and more on “informal sector reports” from media and websites, and less on official candor. Before these new regulations, the WHO itself used to publish a weekly Outbreak Verification List, known in the trade as the Wednesday “rumor list,” a tally of unconfirmed disease reports from all over the world that only a select group was permitted to read. Now the Internet generates its own rumor lists—24/7, for anyone to peruse.
A GPHIN or ProMED posting is a wedge for the WHO to convince reticent nations to come clean with the truth—or for outside authorities to investigate. As GPHIN chief Abla Mawudeku puts it, “WHO can knock on a country’s door and say, ‘We have some information about an event occurring in your backyard. It’s from the media. Could you help verify this?’”
As communications technology gains definition, so will disease surveillance. “We can locate outbreaks within countries and provinces, but we can’t locate them within a farm or village,” says Madoff. “In a blue sky future, I can envision being able to know where individual cases are.”
If AIDS emerged in today’s communications-rich era, would we realize it sooner? Not necessarily. The digital divide still leaves Africa comparatively uncovered. (ProMED’s subscription list, for example, has no Nigerian ISPs.) “We would hear news reports if there was some disease eating away at Rwandans and western Ugandans, which caused weight loss and gradual deterioration. But where do you put that down with the burden of malaria and everything else?” asks Jack Woodall, Ph.D., one of ProMED’s founders. “I’m not sure what kinds of bells would ring.”
In other words, extending the reach of the web to every square meter of the earth’s surface means nothing unless the right person in the right place—someone who is curious, who senses that things are awry, who is connected both socially and virtually (a doctor or nurse, perhaps, but then again, maybe a schoolteacher)—cares enough to dash off a few words and press “Send.”
When Catherine Strommen looks back on her unwitting role in alerting the world to SARS, she marvels at the sheer improbability of it all. “It’s very chaos theory or butterfly effect to me,” she says.
And she regrets that she couldn’t tangibly help the stranger named Ben, who never returned to the Teachers.Net site. Strommen is a people person, but the disease-tracking power of the Internet comes from thickly impersonal networks. “I was never able to tell him anything,” she still laments, five years later. “I didn’t get to complete the circuit.”
Madeline Drexler is a Boston-based science journalist, and author of Secret Agents: The Menace of Emerging Infections.