their feedback. He said they would go in alphabetical order.
Anderson would have started, but he was running late. So Cox, the tough-minded virologist from the CDC, was first out of the box. Wasting no time, she went on the attack, calling the tests into question.
“The results I’m most concerned about,” Cox said, “are the antibody test.”
She and her colleagues at the CDC had previously done extensive work developing various tests for antibodies and had discovered that some of these techniques picked up false positives for H5N1. Some even purported to show evidence of the virus in specimens from U.S. blood donors who had never been within thousands of miles of an actual H5N1 outbreak, whether in people or birds. In particular, Cox was skeptical about the reliability of Western blot testing. To avoid misleading findings, the sensitivity and precision of this kind of analysis had to be calibrated using the results of other tests.
On the call, she grilled Li about his techniques in performing Western blot and interrogated him about whether he had used proper scientific controls to gauge the accuracy of his findings.
“Very limited controls,” Li acknowledged. He was already on his heels. “The number of controls is very low.”
Cox pressed on. She said the CDC had tested similar serum samples from Vietnam, and they’d all come back negative for virus antibodies. She called the rate of positives in Li’s test, or assay, “exceedingly high.”
“I really wonder if your assay is picking up false positives,” she continued. She urged that the samples be retested using a more respected technique for detecting antibodies, called microneutralization. “The serology is very much in doubt and must be repeated with another test in another lab,” she insisted.
“I understand the risks of Western blot,” Li responded defensively. He stressed that his findings were only preliminary. He admitted they might have overestimated the extent of infection in Vietnam.
Cox wanted everyone on the call to take a deep breath and think hard before rushing to conclude there was evidence of an emerging pandemic. So she continued to pound away, questioning the overall reliability of Li’s approach.
“The results are in question because they were obtained with an assay that hasn’t been validated,” she argued.
“I agree,” Li conceded, wishing his results had never been passed to the agency. “I didn’t even want that circulated.”
None of the other experts had yet weighed in, and already they could feel Li squirming on the other end of the line.
For a moment, he seemed to win a respite. The call shifted to a discussion of the genetic changes Li had detected and whether these were making the virus less lethal. He said that the mutations had been found in most of the human samples that his team analyzed. He suggested that either the virus was evolving into a less deadly form, or an entirely new flu virus was now circulating in Vietnam.
But he promptly came under fire again, this time from a new quarter. Anderson, the British epidemiologist, had joined the call and pushed Li on whether those mutations alone would determine the lethality of the virus. Anderson personally didn’t think so. He also noted that the analysis had examined only an “exceedingly small” number of samples.
Cox joined back in, suggesting that a whole series of genetic changes would be needed to alter the lethality, not just one. And even if these changes did make the virus less deadly in animals, as some scientists suspected, she was unconvinced these changes would do the same in people. She wasn’t even sure the mutations had occurred at all.
“The results need to be verified,” she said, repeating her refrain. “It’s quite possible they are true, but they need to be verified.”
Horvath, the Australian chief medical officer, now entered the fray. He wondered whether the researchers had tried to make sense of their highly unusual lab findings by comparing them with the actual experience of patients in the hospital.
“It’s difficult to understand, naturally,” Li offered meekly.
Then, abruptly, the tone changed. Having roughed up Li, the expert panel shifted gears and began to ponder what it would mean if he actually proved to be correct.
“The problem is serious, very urgent,” Anderson conceded. “We need access to information. It’s urgent that an independent lab confirms the changes.” But Vietnam might not share that sense of urgency, and Hanoi’s record of cooperation was poor. “It may be necessary to elevate the pandemic level to get the information,” Anderson suggested.
Stohr pressed the issue of whether to in fact raise the level. “How concerned should we be?” he queried.
“On face value, it’s a very serious report,” Horvath answered. “But it’s not sufficient to take action. I’d be concerned with raising the level of pandemic because it might shut Vietnam down.”
“We’d have to weigh that risk,” Anderson said. The move could be highly unpopular in Southeast Asia. He said the agency would have to weigh “the urgency to understand more versus the danger of upsetting the political balance.”
With time on the call running down, the experts turned to the third and final set of findings. These were the results of the PCR tests on the samples from the Thai Binh health-care facilities, which had revealed a 10 percent rate of infection. They were the most troubling findings. PCR was a more reliable test than Western blot.
“Ten percent is extremely high,” Cox noted. During a seasonal flu outbreak, far fewer than 10 percent of people tested under similar circumstances would have been positive. “But if this is a pandemic beginning, it’s not impossible. It’s urgent that the results be verified.”
Stohr concurred. “There’s an urgent need to verify the accuracies of the findings,” he concluded. WHO would rapidly assemble a new team of experienced virologists and epidemiologists, he decided, and dispatch them to Hanoi. WHO’s chief representative to Vietnam would hope to meet early the following week with senior Vietnamese officials and pressure them to let the team in. The Vietnamese government would be urged to turn over serum and other samples and related material for further testing.
A decision on sounding the pandemic alarm, he said, would be temporarily postponed.
At two o’clock Geneva time, Stohr adjourned the call.
Uncertainty is what often separates public health from laboratory science. Lab researchers select questions they believe they can answer and bypass those they can’t. They design experiments to produce definitive proof. Success means generating results that can be reproduced by other scientists in other labs. Public health, by contrast, rarely gets to choose the questions it must answer. People get sick. Doctors try to diagnose. They prescribe treatment and hope it will work. Symptoms, family history, and even test results are suggestive but often not definitive. The whole purpose of getting a second opinion on medical care is not to reproduce the conclusions of the initial physician but to discover if there are different ones. If the affliction is the common cold, uncertainty can be annoying. If it’s cancer, it can be tragic. If it’s flu, it can be catastrophic.
Medical science as a modern discipline is little more than a century old. Molecular biology goes back barely two generations. In that short period, the field has reshaped human knowledge and answered some of the basic questions about our very essence. Yet in confronting pandemic influenza, the greatest human killer in history, science still comes up short.
The difficulties begin with diagnosing the virus. Scientists have had trouble coming up with a fast way of determining whether someone has H5N1, and this lack of an accurate rapid test has repeatedly fostered confusion and even panic, most notably when the disease came knocking on Europe’s door. In the final days of 2005, villagers in the frigid mountains of eastern Turkey began falling sick with a mysterious respiratory ailment. Flu specialists suspected the novel strain. But initial tests by Turkish health officials came up negative. Days later, they reversed themselves, announcing that further tests had confirmed bird flu in two patients. More would follow. Of eight patients ultimately confirmed with bird flu and treated at a hospital in the regional center of Van, rapid flu tests came back negative every time. So did a separate enzyme-linked immunosorbent assay, or ELISA, test. Even the initial PCR test for four of the cases failed to identify the virus, though they were already very sick. Only after additional samples were taken did they test positive.
Then, as fear spread, Turks from around the country flooded hospitals at the first sniffle or cough. Day after day, hundreds of test samples poured in and overwhelmed the country’s national lab, which was equipped to
