To the Editor.
—In their study of female-to-male transmission of HIV, Padian et al1 draw attention to the considerable methodological difficulties in the design of partner studies, in particular the biases that may lead to overestimation of the risk of transmission and their careful attempts to overcome them. As they discuss, their estimate is lower than many other studies.1,2 In addition to the issues raised by Padian et al, we suggest that another explanation for the differences could be a methodological problem in their own study that produces a bias toward zero in the estimated transmission rate.The study recruited both men and women infected with HIV and tested the serostatus of their heterosexual partners. For concordantly infected couples, the direction of transmission was ascertained by identifying a "well-established" source of risk for one partner. Overall, 379 couples were reported, 62 of whom were concordantly infected—an "either direction" heterosexual
Phillips A, Johnson A. Female-to-Male Transmission of HIV. JAMA. 1992;268(14):1855-1856. doi:10.1001/jama.1992.03490140063024