Eventually, a billion people -- half the world's population -- were infected with the flu, and 40–50 million died. People did not go gently. Faces turned blue for lack of oxygen. Victims gasped for air as they hemorrhaged internally, literally drowning in their own blood.

Then it was over. In less than 12 months, the waves subsided. And until recently, so had the world's memory of the deadliest pandemic in history. Today the 1918 flu haunts us as one of public health's greatest unsolved mysteries. And it's possibly a harbinger of the next flu pandemic -- a "when," not "if," proposition, experts warn.

In investigations resembling episodes from the popular "CSI: Crime Scene Investigation" television shows, scientists have searched for the 1918 virus in bodies buried in Alaskan permafrost and autopsied lung tissue stored in wax. So far, the bits and pieces of genetic evidence they've salvaged suggest that its ancestors may have been an avian flu virus, but that it evolved within some other as-yet-unidentified species into a wholly new germ -- one horrifyingly well adapted to humans.

But as Harvard School of Public Health epidemiologists Christina Mills, James Robins, and Marc Lipsitch have shown, you don't have to exhume bodies to unlock important truths about this disease.

By digging through dusty public health records and plugging raw data into a sophisticated mathematical model, the HSPH trio succeeded in estimating how contagious the 1918 bug really was, publishing their findings in the December 16, 2004 issue of Nature. More precisely, they calculated the virus's transmissibility, or reproductive number -- the average number of people infected by a single individual.

HSPH epidemiologists, led by Lipsitch and Assistant Professor Megan Murray, have honed to a fine point estimations of reproductive number for infectious disease. All microorganisms behave like plants or animals when you look at their behavior in populations. They find ecological niches, crowd out competitors, reproduce with assistance from other living things, and mutate in response to evolutionary pressures. Lipsitch is part of a group that is analyzing this Darwinian demimonde and abstracting it into mathematical models, yielding insights that include the reproductive number.

You'd think the "Spanish Lady" -- the pandemic's sobriquet, derived from the mistaken belief that it originated in Spain -- would have been terrifyingly contagious. For decades, many experts thought its reproductive number could be as high as 20 -- meaning that the virus spread from each infected person to 20 others. Depending on the circumstances, a disease like measles has a reproductive number almost that high. But as the HSPH researchers' calculations showed, the reproductive number of the 1918 flu was actually only between two and three.

That's good news, says Lone Simonsen, a senior epidemiologist at the National Institute of Allergy and Infectious Diseases. To stop a pandemic, public health measures need to drive the reproductive number below one, so that each case will spawn less than one new case. If the reproductive number starts out at between two and three, you can do that, notes Simonsen. "If it were 20," she says, "you might as well just sit back and let it hit you."

The HSPH breakthrough couldn't have come at a better time. Because of avian flu in Asia, flu experts say we're closer to a new pandemic now than we've been in a generation. "We know the recipe, and all the ingredients are there," Klaus Stöhr, the World Health Organization's head of flu surveillance, has said. But the HSPH findings could help stop that putative pandemic in its tracks.

Digging for data

There's nothing like a nice, big dataset to get an epidemiologist's juices flowing. But to unravel the 1918 epidemic, Lipsitch, Robins, and Mills were obliged to build theirs from scratch. At first, they hoped to scan graphs showing weekly death tolls for pneumonia and influenza during 1918 into a computer.

But the graphs were drawn by hand, so that didn't work. Mills, who is simultaneously earning an MD from Harvard Medical School and a ScD in epidemiology from HSPH, started researching studies, working her way backward through the citations. "I knew the raw data existed somewhere, because the studies kept referring to it -- and because there were all these graphs."

Soon she found herself in the basement of Harvard's Countway Medical Library, poring over every issue of Public Health Reports from 1918 to 1922, jotting down weekly mortality statistics, and entering a couple of thousand numbers into an Excel spreadsheet. To buttress the analysis, she gathered heaps of other data culled from autopsy reports and other sources.

More slog than "Eureka!" moment, the work was still exciting for Mills: "Through a lot of digging through the dust at Countway, I was able to find the actual numbers used to generate those graphs -- nice, raw data."

Like experts before her, Mills figured she'd find that the 1918 virus was highly contagious. But no matter the city, the reproductive number hovered between two and three, and never rose much above five.

So many deaths in so little time

So how can a not-so-contagious virus cause so many deaths, so quickly? The answer lies in what epidemiologists call "serial interval," the rate of transmission. A virus that infects a lot of people but does so slowly can be less harmful than one that jumps from a single case to a few others at daunting speed. Most flu viruses have serial intervals of just a few days. They spread rapidly, Mills explains, even when their reproductive number is low.

But the 1918 flu virus was exceptionally deadly. Instead of killing one or two out of 1,000 on average, this germ took about 20 lives. Researchers still don't know why it killed such a high percentage of cases. The answer likely lies in the unique way in which the genes of that particular virus were shuffled into a foe never before seen by the human immune system. Stress and overcrowding in a world engulfed by war were probably contributing factors. According to Alfred Crosby's classic book Epidemic and Peace, 1918, roughly 550,000 Americans died from the infection and related complications. That's more than perished in Vietnam, Korea, and the first and second World Wars combined.

What now?

Today, fast deployment of antiviral drugs coupled with a large-scale vaccination campaign could be the only way to rein in a pandemic and prevent a nightmarish replay of 1918.

Fortunately, the flu season of 2004-05 was mild, and the U.S. wound up with a vaccine surplus. But initial shortages caused by contamination at a single factory in Liverpool, England, revealed the precariousness of the American vaccine supply.

Technological advances might help. Normally, flu vaccine is made several months in advance, based on predictions about which strains are most likely to be circulating. Researchers are developing ways to speed up that process so that mass vaccination can begin soon after a pandemic takes off. They're experimenting with "reverse genetic" techniques that back-translate flu virus RNA into DNA, then insert those tiny bits into circles of DNA called plasmids. These plasmids -- artificial versions of a flu virus -- would serve as the ready-to-go stock for a vaccine for the next pandemic. Researchers are also figuring out ways to grow flu virus in cell cultures instead of the traditional medium, chicken eggs.

Antiviral medicines could be especially valuable in pandemic prevention because they're not strain-specific. They cripple flu viruses by blocking enzymes they need to replicate inside a cell. Large reserves could be used to put the brakes on a pandemic, but the location of stockpiles is crucial, notes Lipsitch: "Having them only here in the United States isn't going to do it, because once a pandemic starts, nobody is going to want to ship their limited supply of potentially life-saving drugs overseas. Antivirals need to be in each of the countries where the pandemic is most likely to start -- Asia."

As Christina Mills points out, people unwittingly spread flu virus before they experience symptoms. That means public health measures based on tracking down sick people -- isolating cases, quarantining their contacts, treating everyone with antivirals -- are going to be a step or two behind the virus. And even in the best of circumstances, it will take time to gear up a mass vaccination campaign.

"To buy time, you have to think about ways to limit social contacts," by closing schools, obliging people to work from home, and so on, says Mills. In epidemiologists' vernacular, these are "social distance interventions" akin to steps taken by Philadelphians in 1918.

An urgent proposal

HSPH Dean Barry Bloom sees a larger cloud looming over vaccine production that neither technology nor the best-laid plans can dispel. He says market forces and public expectations -- "We're used to vaccines that cost a dollar apiece" -- tend to work against the reliable production of vaccines by industry. As a result, a surge in infectious disease will be a public health disaster in the making, whether from a "natural" flu pandemic or a bioterrorism attack.

Now, as in 1994, when he first proposed it in Science magazine, Bloom is calling on the federal government to create a National Vaccine Authority (see Dean's Message). Its role would be to build in economic incentives for vaccine suppliers and buffer them from marketplace forces that can render vaccine manufacturing dicey as a business enterprise.

A flu pandemic could kill 89,000 to 207,000 Americans (in an average year, flu is blamed for about 36,000 deaths) and cost from $71.3 to $166.5 billion, according to estimates from the Centers for Disease Control and Prevention (CDC). Last year, Congress earmarked $50 million for pandemic flu planning, and a draft plan was circulated.

Is enough time, money, and attention being spent on this threat? Not by a long shot, say HSPH's epidemiologists and others. In Asia, efforts to control the avian flu are falling woefully short, critics say. If the disease were circulating in the U.S. or Europe, we'd be spending much more to control it, some contend.

Lipsitch and Mills criticized the federal government's draft pandemic plan for not emphasizing surveillance nearly enough. Spotting those first cases of any emerging infection -- the proverbial canaries in a coal mine -- is critical, they say. So too is speedy data collection, which permits epidemiologists to calculate reproductive numbers and other variables useful for allocating limited resources, including vaccine, to halt the spread of disease.

The flip side of the "be prepared" coin is overreaction. In 1976, CDC officials were alarmed by comparisons of swine flu to the 1918 epidemic, and ordered a mass vaccination campaign. That move is now widely seen as a gigantic blunder.

But apathy and jaded outlooks are traps, too. Public health officials rightly see the containment of severe acute respiratory syndrome (SARS) in 2003-04 as a triumph of scientific ingenuity and smart public health policy (see "Taking on the SARS Challenge," online at http://www.hsph.harvard.edu/review/review_fall_03/sars.html, or in the Fall 2003 issue of the Review). After some initial stumbles, including suppression of information by the Chinese government, quick action to quarantine cases averted a more serious epidemic. But too many Americans view SARS as hype -- another threat that never materialized.

"That's the problem with public health," says an exasperated Mills. "If you don't do your job well, it's a disaster. But do your job well, and everybody thinks it was overblown."

"If I had to bet on one strain that would give rise to a pandemic, it would be the avian flu strain now circulating in Asia," says Lipsitch. "It is widespread. It infects humans. And there's evidence that, in limited ways, it can spread from human to human. And it is novel. " But whether it's avian flu or something else, we need to get ready."

Peter Wehrwein writes about science, medicine, and public health.

Photo: Kent Dayton



This page is maintained by Development Communications in the Office of Resource Development.
To contact us with suggestions, comments, and questions, please e-mail: editor@hsph.harvard.edu

Copyright, 2005, President and Fellows of Harvard College