Sex captures the attention of the news media so that while medical conferences don’t get much coverage in the lay press, a paper challenging the popular view about the safety of oral sex did get a few headlines. At the recent retroviruses conference in San Francisco, a presentation titled Primary HIV Infections Associated with Oral Transmission from the University of California at San Francisco suggested that oral sex is a significant route of HIV transmission.

The study was a retrospective analysis of 122 people who had recently become HIV+ and it sought to find the probable routes of infection. Once all of the other possible routes of infection had been eliminated (such as unprotected anal sex, anal sex with condom breakage, substance-induced "blackouts" where behaviour could not be determined), eight of the individuals (6.6%) were thought to have got HIV from oral sex. All eight men had considered oral sex to be an acceptable risk.

Such a paper got the inevitable reaction from the news media that oral sex is unsafe, whereas the study authors concluded that "oral sex may be an important mode of transmission due to its frequency" but that "standardized investigation of HIV transmission via oral sex is needed". What these conclusions mean is that the relative risk of oral sex versus unprotected anal sex cannot be determined from these data. Only a prospective study looking at a large group of people over a long period of time and carefully following all of their sexual practices, especially oral sex and protected and unprotected anal and vaginal sex, could determine the comparative risks. The 13-year Vancouver Lymphadenopathy Study (VLAS) of gay men showed an increased risk of transmission for anal sex but did not show a significant increased risk for those who reported only oral sex. That did not mean that oral sex was not a risk but the study could not show the relationship of transmission with oral sex whereas it could for anal sex. This finding suggested that anal sex was more risky than oral sex, but it did not say by how much.

A retrospective study can show how people got HIV, but it does not show the relative risk of a single activity unless it can determine the frequency of the activity. For example, if with gay men, oral sex occurred ten times more often than unprotected anal sex and 7% of all sexually transmitted HIV was from oral sex and 93% from anal sex, that would mean that anal sex was 133 times more likely to transmit HIV. If a population had a very high frequency of a low risk activity and a low frequency of a high risk activity, quite possibly the low risk activity could cause more infections. A similar relationship could be shown between choice of transportation: car accidents cause far more deaths than skydiving each year, yet few people would argue skydiving was safer than being in a car.

The determination of the true risks of oral versus anal sex is problematic due to confounding issues such as the unquantified risks of transmission from protected anal sex, the difficulty in getting accurate reporting from individuals (who would want to admit in 2000 that they had unprotected anal or vaginal sex with someone who could be HIV+?), and by other hidden risks such as injection drug use. Determining the relative risks of different sexual activities is important for developing public health strategies that can reduce the risks to all sexually active people. If there is a large difference (say, 100 times) of the risk between oral and anal sex, a safer sex message that says that both anal and oral sex are risky could cause more infections because people might then take a chance with anal sex because they feel they have already taken a chance with oral sex. A message that promotes abstinence from both oral and anal sex for gay men is not likely to be successful because adherence to such an edict will be short.

The study from San Francisco looked primarily at gay men. It is unlikely that the information can be extrapolated to oral sex between two women or a man and a woman, where the risk could theoretically be higher if menstrual blood were involved. Other factors such as oral health (bleeding, sores, time of brushing or flossing), viral load, ejaculation, ingestion of semen were not considered and they could influence the risk for exposure at an individual level.