The study involved measuring -- in umbilical cord blood samples -- a protein known as placental alkaline phosphatase as an indicator of maternal-fetal microtransfusion.
"This enzyme, made in the placenta, is very big -- usually too big to pass through the natural barrier that protects babies from disease-causing organisms mothers sometimes carry before birth," Meshnick said. "We figured that if we found it in blood from the umbilical cord, which links the placenta to the baby, that would be an indication of mixing, that something tore in the placenta and leaked the mothers' virus-contaminated blood to the infant."
Researchers then compared what they found in cord blood to whether or not infants were infected with HIV, he said.
"We looked for a statistical relationship between the presence of the enzyme and the risk of infection in the babies," Meshnick said. "We found a very strong correlation between the enzyme and the risk of infection."
It appears that transmission of the virus occurs during labor -- when contractions occur -- rather than during passage of the baby through the birth canal, he said.
"It has been known for a long time that HIV-infected women who undergo caesarean section before they go into labor do not transmit the virus, whereas those who undergo emergency C-sections after they go into labor do transmit it," Meshnick said. "What we saw was consistent with this. It looks like direct mother-to-child blood mixing occurring during labor leads to infection."
The new findings indicate that intervening just before birth by such suggested methods as sanitizing the birth canal may not work, he said. A better alternative might be to have pregnant women start taking anti-retrovirals before going into labor.