transmission of HIV is also suggested by the occasional isolation of HIV from amniotic fluid and cells. Finally, the isolation of HIV from, or the detection of the HIV genome in, blood samples obtained at birth from some HIV-infected infants also suggests intrauterine HIV transmission (Rogers et al., 1989; Ehrnst et al., 1991; Luzuriaga et al., 1993). The proportion of infants infected during each trimester of pregnancy is unknown.

Indirect evidence suggests that 70% to 75% or more of vertical HIV transmission can occur during delivery (Ehrnst et al., 1991; Luzuriaga et al., 1993; Rogers et al., 1994). Negative diagnostic studies in the first two days of life followed by the detection of infection after one week of age are compatible with intrapartum transmission (Luzuriaga et al., 1993). Increased risk of vertical HIV transmission has been correlated with increased duration of rupture of the membranes prior to delivery, particularly in the presence of acute chorioamnionitis (Landesman et al., 1996; Popek et al., 1997). A higher risk of transmission to the firstborn twin, particularly following prolonged labor (Duliege et al., 1995), also supports the concept of intrapartum transmission. The mechanism(s) of intrapartum transmission are unknown, but might include transplacental microtransfusion or infection through mucocutaneous exposure to maternal blood or cervical secretions. The establishment of infection following the inoculation of the simian homolog of HIV (simian immunodeficiency virus) into the conjunctival sac or oropharynx of newborn macaque monkeys also supports the mucocutaneous route of human neonatal infection (Baba et al., 1994).

Vertical HIV transmission can also occur through breast-feeding (Ziegler et al., 1985; Bulterys et al., 1995). HIV RNA and proviral DNA have been detected using the polymerase chain reaction (PCR) in breast milk; viral load appears to be particularly high in colostrum (Ruff et al., 1994). Large, prospective cohort studies suggest an increased risk of transmission associated with breast-feeding. In a meta-analysis, Dunn and colleagues (1992) have estimated that the proportion of transmission attributable to breast-feeding worldwide from an HIV-seropositive woman is 14% (95% confidence interval, 7% to 22%). The risk of breast milk transmission appears to be particularly high when maternal primary infection occurs within the first few months following delivery (Palasanthiran et al., 1993). For these reasons, HIV-seropositive women in industrialized countries are advised not to breast-feed their babies (AAP, 1995a). In July 1998, the World Health Organization recommended that HIV-infected women in developing countries be given information about the benefits and risks of breast-feeding, and an opportunity to make an informed choice about breast-feeding (WHO, 1998).

The observed variability in reported transmission rates probably reflects the multiple factors that influence perinatal HIV transmission. Several studies have