HO CHI MINH CITY, Vietnam — After more than a year of watching patients sicken and die of bird flu, Dr. Tran Tinh Hien of the Hospital for Tropical Diseases here thought he understood the illness.
Then last month, he learned of an unsettling study. Japanese researchers retested samples from 30 Vietnamese patients whose lab tests showed no signs of the disease. They discovered that seven had actually been infected.
"We are especially worried, because it may mean we missed some patients," said Tran, the hospital's deputy director.
Tran is part of a growing consensus that the extent of human bird flu infection in Southeast Asia may have been significantly underestimated.
In the last few months, scientists have begun to believe that the inaccuracy of laboratory tests, the wide variation of symptoms and the inability of public health agencies to combat the disease may have created the erroneous perception that bird flu is still rare among humans.
The number of infections is key. The more there are, the greater the chance the virus will mutate into a form that can easily be passed between people, who would have little immunity to the new disease. Scientists believe that nearly all infections so far have been caused by contact with sick or dead poultry.
Officially, the tally doesn't sound alarming. The virus has killed 14 people since December and 46 over the last 15 months. All but one of them were from Vietnam and Thailand. Altogether, there are 69 lab-confirmed cases, according to the World Health Organization.
Yet doctors and public health officials point to a glaring oddity in the statistics that underscores the belief that the case count is too low to be true.
Vietnam and Thailand have reported the overwhelming majority of recent cases. Yet Laos, which is sandwiched between the countries, has reported no cases among people or birds this year. Cambodia, which is also flanked by Vietnam and Thailand, has confirmed only a single human case.
"People are not trying to cover it up, but given how widespread the infection is in poultry in Southeast Asia, it's hard to believe people have gotten ill in only [three] countries," said Jeremy Farrar, a University of Oxford flu researcher at the Hospital for Tropical Diseases.
Philippe Buchy, head of virology at the Pasteur Institute, the only testing center in Cambodia, said, "The best way not to find something is not to look for it."
The current bird flu, a variety formally known as H5N1, was first detected in Hong Kong in 1997. After the slaughter of millions of chickens and ducks throughout Southeast Asia, public health officials thought the disease was under control.
But since December, the virus has seen a resurgence in birds in the region. Some public health officials now believe the disease is ingrained in wild birds and would be virtually impossible to wipe out.
For the most part, doctors searching for infected humans looked for classic pneumonia-like symptoms such as coughing, lung damage and often death.
But the Japanese study and several other recent scientific papers suggest that doctors haven't been looking for all the right clues.
A study published last month in the New England Journal of Medicine described a Vietnamese boy who was diagnosed with encephalitis -- brain swelling, fever and vomiting, followed by coma.
After the boy died in February 2004, H5N1 was found in his spinal fluid, making his the first known case of bird flu that did not resemble pneumonia.
At the other end of the spectrum, tests this month showed that two people who cared for sick relatives -- an 81-year-old man whose grandchildren were ill and a 61-year-old woman whose husband died -- were infected with H5N1 but showed no symptoms.
The recent discovery of mild cases debunks the widely accepted idea that bird flu kills about 70% of its victims, which would make it one of the most lethal infectious diseases.
But that seemingly good news has a disturbing implication. "In global health terms, people who don't die can be more of a problem," Farrar said, because they provide havens for the virus and may help it circulate in their communities and beyond.
The recent rise in human cases in Vietnam points to another potential problem with the official statistics. Seven new cases have been reported since March 7, including a nurse who treated a bird flu patient. Two additional cases are being considered probable infections: a veterinarian from hard-hit Thai Binh province near Hanoi and a man who died Sunday at the provincial hospital in Rach Gia.
Henry L. Niman, an infectious-disease expert whose Pittsburgh-based company, Recombinomics Inc., studies genetic changes in viruses, believes a pattern has emerged.
A week or so after a person in a remote region has died of apparent pneumonia, a family member comes down with similar symptoms and ultimately tests positive for bird flu. That suggests that more people have died from bird flu than the official tallies reflect.