Valendar Turner, in Correspondence (“HIV drug remains unproven without placebo trial” Nature 435, 137; 200510.1038/435137a), argues that there is no evidence for antiretrovirals reducing the transmission of HIV from mother to child. He points out that HIV transmission in people taking the antiretroviral drug nevirapine was 13.1% in the HIVNET 012 study in Uganda, whereas only 12% of women in a Rwandan study were found to have transmitted HIV to their babies in the absence of antiretroviral treatment.

Despite a rebuttal in Correspondence by the authors of the Ugandan study, Brooks Jackson and Thomas Fleming (“A drug is effective if better than a harmless control” Nature 434, 1067; 200510.1038/4341067a), Turner's letter continues to be cited by AIDS denialists (for example, C. Farber Harper's Magazine 37–52; March 2006).

The Rwandan study referred to by Turner enrolled 561 pregnant women, of whom 286 were HIV-positive. Of the children born to HIV-positive mothers, 158 were tested for HIV and 19 (12%, as Turner states) were found to be HIV-positive. Why were only 158 children assessed? The answer, conveniently ignored by the denialists, is that follow-up was interrupted by the events of the Rwandan civil war (J. Ladner et al. J. Acquir. Immun. Def. Syndr. Hum. Retrovirol. 18, 293–298; 1998). Given that this interruption was sufficiently lengthy for many HIV-positive children and their mothers to die of AIDS in the interim, the surviving sample of the initial cohort cannot be regarded as representative. The actual figure for HIV transmission was almost certainly much higher. Failing to acknowledge this important caveat to the study appears to us to be inconsistent with accepted academic standards.