Computer Models Calm Fears of Bird Flu Pandemic, Urge Vigilance

By M.L. Baker  |  Posted 08-03-2005
An outbreak of bird flu in humans need not become a pandemic, conclude two papers published this week in prestigious science journals. Scientists used computer models to predict how surveillance and containment efforts could prevent the worldwide spread of disease caused by the H5N1 avian influenza virus.

Though the disease has led to the deaths of tens of millions of chicken and ducks, the virus rarely infects humans; fewer than 150 cases have been reported since 2003. When humans do become infected, however, the disease is highly lethal.

If the virus mutates to a form that is more easily passed from person to person, public officials fear the toll could be much, much higher. The last worldwide flu pandemic began in Hong Kong in 1968. It sickened an estimated 675,000 Americans and killed 34,000. Estimates of deaths from the Spanish flu pandemic of 1918 range well over 20 million.

The scientists' models examined how quarantining and providing antiviral medicines in areas reporting an outbreak could prevent its spread. In some cases they considered the impact of prevaccinating at-risk populations, even if the vaccine was only marginally effective. Both groups of researchers are part of a network called Models for Infectious Disease Agents Study.

"For the first time in human history, we have a chance of stopping the spread of a new influenza strain at the source through good surveillance and aggressive use of public health measures," said Katrina Kelner, deputy editor, Life Sciences, at Science, which published one of the studies.

Patients are willing to pay for electronic medical records. Click here to read more.

The models can be used to plan prevention strategies but can also be applied during an outbreak to help public health officials predict trends and see if strategies are working, said Neil Ferguson, a professor of mathematical biology at Imperial College London, whose study is published in Nature.

Ferguson recommends that infected patients be isolated and government officials notified when the virus is recognized as unusual. Schools and workplaces should be closed to reduce the opportunity for people to transmit the virus. And about 20,000 healthy people closest to the outbreak should receive antivirals to keep them from getting sick and spreading the disease further. He estimated that up to 3 million courses of antiviral treatments would need to be stockpiled to quell the outbreak.

Researchers spent a year gathering the data for their model. "You need to collect as much data as possible ahead of time for this sort of modeling," said Ferguson. Few countries in the region publish information about locations of schools and major companies, how far people travel to get to school or work, and use of public transportation, and even where they do, the data needs to be collated and analyzed before scientists can use it, he explained.

The other study was led by Ira Longini of Emory University. His group simulated a population of half a million people and the contacts between them. The model showed how a single case of human-transmittable virus could spread through social networks.

For a moderately infectious virus, the model found the disease could be contained if intervention began within 21 days of the strain's emergence. If 80 percent of people in the same "mixing groups" received antiviral drugs, the model predicted that the virus could be contained.

If public health officials instead targeted individuals within a certain geographic range to receive antivirals, they would need to reach 90 percent of that population. Longini's group also found that the efficacy of antivirals to contain an outbreak is much reduced 28 days after the first case; the efficacy of quarantine is much reduced 42 days after the first case is reported.

This assumes that the virus is considerably less infectious than the viruses that caused earlier worldwide flu pandemics. However, a newly emerged strain should also be less infectious because it is not adapted for human-to-human contact.

Although most cases of bird flu have occurred in Vietnam, the researchers fed their model with data from rural Thailand because that country had much more complete information, including a 2000 census and a study of social networks.

The researchers plan to test different scenarios as new data becomes available. They will also develop a collection of models so that they can compare outcomes from different interventions.

Public health officials also urge that a vaccine be developed to stop H5N1 in both humans and poultry.