In a surprise twist, authors here retracted findings of a study that found N95 respirators were better than surgical masks at preventing flu.
After a re-analysis prompted by questions from reviewers, the findings were no longer significant, said Holly Seale of the University of New South Wales in Sydney, Australia.
The original study, presented earlier this year, formed the basis of some recommendations on the use of masks in a health care setting.
The retraction -- near the end of a presentation at the annual meeting of the Infectious Diseases Society of America -- prompted a "rush to the microphones" by those involved in flu prevention, one expert said.
The findings appeared to differ -- not only from previous reports, but also from the abstract submitted to this meeting, said Dr. Andrew Pavia of the University of Utah.
Seale acknowledged those differences and agreed that the original results no longer stand. She was not immediately available for additional comment.
The lead author of the study, Raina MacIntyre, also of the University of New South Wales, did not attend the meeting here.
The retraction took experts here by surprise, although many had been critical of some statistical aspects of the study, according to Dr. Neil Fishman of the University of Pennsylvania.
"I think there was little bit of shock that there was such a large change (in the results)," he told MedPage Today after the session.
The study was first presented in San Francisco earlier this year.
The N95 respirators are more costly than simple surgical masks and are in shorter supply. As well, there are costs involved if hospitals try to comply with guidance, he said.
The original Australian finding was a surprise to many experts, Fishman said, because it did not accord with other findings or with clinical experience.
A Canadian study in the Journal of the American Medical Association in October found no difference between fit-tested N95 masks and surgical masks.
The analysis, led by Mark Loeb of McMaster University in Hamilton, Ontario, was a true randomized trial among nurses in eight tertiary care hospitals in Canada.
In contrast, the Australian study, conducted in China, was a cluster randomized trial in which the unit of analysis was the hospital, Fishman said.
But the Australians were criticized because their control group of nine hospitals was not selected randomly, he said.
When the data were re-analyzed excluding those hospitals, the differences between N95 respirators and surgical masks were no longer significant, Seale reported.
"When the P-value was adjusted for clustering and multiple testing," she reported, there was "no significant difference between N95 respirators and medical masks."
Fishman said most clinical experience suggests that flu is spread by large droplets and should be blocked by even a simple barrier such as a surgical mask.