GAITHERSBURG, Md (Reuters) - A federal advisory panel, in a close vote, declined Thursday to support easing restrictions on blood donations from gay men, a policy criticised as discriminatory and outdated.

Many members of the US Food and Drug Administration panel said they would like to see a change in the policy, which was adopted to minimise risks of spreading the HIV virus that causes AIDS. But the majority voted that they did not have enough scientific evidence to back the FDA's proposed revisions.

Under FDA rules, men cannot give blood if they have had sex with another man at least once since 1977. The panel voted 7-6 against allowing men to donate if they had not had sex with another man in the past five years.

FDA officials said they were revisiting the policy because new blood tests are more reliable and track HIV infection much quicker than previous ones. Also, gay rights groups and others have complained that the policy differs from the treatment of other high-risk donors. Men who have had sex with a prostitute, for instance, are "deferred" from giving blood for only one year after their last encounter.

Further, critics say asking people whether they have had multiple partners or unprotected sex is a better way to tell who is most at risk for carrying HIV and would not exclude gay men who practice safe sex or abstinence, or others who may have experimented with gay sex years ago.

The restriction for gay men "seems very discriminatory and it seems very arbitrary," said panel member Dr. Mark Mitchell. "I feel very strongly it needs to be changed."

But most members said the FDA did not present enough scientific data to convince them the change would not lead to more HIV-positive units slipping through the extensive blood testing and screening. The FDA usually follows its panels' advice.

"It's all an assumption. There is no evidence," said panel chairman F. Blaine Hollinger of Baylor College of Medicine.

Safety measures have made exposure to HIV from blood donations extremely rare.

Calculating the impact of a five-year deferral for gay men was difficult, said FDA medical officer Andrew Dayton. He estimated that 62,300 men would go to blood centres to give blood if the policy changed. Highly accurate testing likely would catch any HIV-positive, but risks would come from workers who might accidentally release units that were positive, Dayton said.

Still, fewer than two HIV-positive blood units per year would slip out, Dayton estimated, but admitted his numbers "are very iffy, and unfortunately it's all we have to go on."

Panel member Jeanne Linden said she "did not have the evidence" to back changing restrictions on gay donations now, but she "very much encouraged" the FDA to continue reviewing the matter and consider changes.