ATLANTA—Zika infection during pregnancy can lead to birth defects except, of course, when it does not. Now scientists are wondering why the virus catastrophically affects some fetuses but not others.

In Colombia, where the number of known Zika infections is second only to Brazil, there have been relatively few cases of related birth defects: 57 compared with more than 2,000 in Brazil, according to the World Health Organization. The U.S. has the third-highest number of Zika-related birth defects, with 31 combined cases and lost pregnancies due to miscarriage. The exact ratio of Zika infections to birth defects in each country remains difficult to determine, partly because so many Zika patients do not appear symptomatic, but what is clear is that birth defect rates are uneven.

That inexplicable geographic variability fueled speculation among scientists attending the American Society of Tropical Medicine and Hygiene Annual Meeting in Atlanta this week about what is causing the disparate patterns of Zika-related birth defects. Some answers may come from comparing the northeastern part of Brazil, where birth defects are the most common, to the rest of Brazil and Colombia, says Pedro Fernando da Costa Vasconcelos, director of Brazil’s National Reference Laboratory for Arboviruses.

One notable variance is that in northeastern Brazil few people receive vaccinations against the yellow fever virus whereas that is the norm in the other two locations, da Costa Vasconcelos says. Researchers need to study that disparity, he adds, because the yellow fever and Zika viruses are closely related, so yellow fever vaccination might provide some cross-reactive protection. Such an analysis should be starting soon, he says.

Other theories abound, however. There is no reliable tracking of how many women with Zika in Brazil and Colombia have chosen to terminate their pregnancies out of concern about birth defects. “What we aren’t getting a good handle on are abortions,” says Albert Ko, a professor of epidemiology and medicine at Yale University who conducts research on Zika in Brazil, so experts do not know if rates of termination could explain some of the geographical discrepancy in birth defects.

In the U.S. the future may bring better statistics about Zika-related abortions, says Anne Schuchat, principal deputy director for the U.S. Centers for Disease Control and Protection. The CDC is tracking women infected with Zika throughout their pregnancy and “characterizing the spectrum of what occurs whether there is delivery or a termination,” she said at the conference. Ivan Gonzalez, co-director of the Zika Response Team at the University of Miami, says that he has already heard about several women in his area who have had abortions due to their Zika status. Outside of his health system, he says, “It is very difficult to keep track of how many women are having abortions [in the Miami area].”

Still other factors, or a combination of them, could be at play. For example, it remains unclear if Zika patients’ genetics, prior infections, or even something about the mosquitoes themselves—such as the viral and bacterial load they carry—may influence the dynamics of the virus in pregnant women. To date, there have not been any studies on these factors, so the public is left with theory and speculation.

Greater understanding may come from a large-scale study following thousands of Zika-affected pregnant women and their children. In June it started recruiting 10,000 pregnant women across nine locations in the Americas, Ko says. The research will follow the women from their first trimester onward for at least one year after the baby’s birth, comparing birth outcomes between mothers infected with Zika and those who were not. With this data researchers hope to get a clearer picture in the years to come.