To the Editor.
—I read with interest the article by Dr Boyer and colleagues.1I have a few comments regarding the analysis in this otherwise excellent study. The authors considered 68 live-born infants, including two sets of twins, from 66 pregnancies in 63 women as 68 independent mother-infant pairs in the analysis. As multiple pregnancies in one woman as well as a set of twins are not independent, this assumption is erroneous.The data presented in Table 2 of their article should have been stratified by zidovudine treatment. For determining the effect of zidovudine in preventing transmission, multiple regression analysis controlling for confounders (maternal CD4 cell count, viral titer, duration of human immunodeficiency virus [HIV] infection, acquired immunodeficiency syndrome— defining illnesses for in utero transmission and maternal blood exposure, and genital lesions for intrapartum transmission) should have been used.