As the international epidemic of Zika virus disease has unfolded and led to devastating birth defects for at least 1,300 children in eight countries, an agonizing question has persisted: What is the chance that an infected pregnant woman will have a baby with these defects?

Researchers don't yet have a complete answer, but they are slowly homing in on one.

The largest study to ever look at the question says the risk of one especially severe type of birth defect is "substantial" — in the range of 1 percent to 14 percent. It also reinforces the understanding that women infected in the early stages of pregnancy face the greatest risk.

The range is so unusually wide because researchers are relying on imprecise and incomplete information as they to try to quickly estimate the level of risk in advance of what they say is likely transmission of Zika by mosquitoes in the U.S. later this year.

The study focused on what was seen in just one place, a state in northeast Brazil. And it looks only at microcephaly, a condition in which a baby's skull is much smaller than expected because the brain hasn't developed properly. But health officials say Zika can cause other birth defects, too.

"These numbers are probably only the tip of the iceberg," said Dr. Neil Silverman, a UCLA professor of obstetrics who has been advising the California Department of Public Health on Zika issues.

A study done in Rio de Janiero published earlier this year suggested there is a nearly 30 percent risk for all kinds of Zika-related birth defects and fetal death. The new research doesn't necessarily contradict that, said Silverman, who was not involved in the latest research.

The new study was done by government scientists at the Centers for Disease Control and Prevention. It was published online Wednesday by the New England Journal of Medicine.

The Zika virus causes only a mild and brief illness, at worst, in most people. But in the last year, infections in pregnant women have been strongly linked to fetal deaths and to potentially severe birth defects, mostly in Brazil. Last month, the CDC said there was enough evidence to declare that Zika causes microcephaly and other brain defects.

The virus is spread mainly through the bite of a tropical mosquito called Aedes aegypti. The bug can be found in the southern United States, but there's no evidence that they've been spreading the virus in the U.S. yet.

The new study is based on about 400 babies with microcephaly in the Brazilian state of Bahia, who were diagnosed between July and February.

The researchers made estimates based on what data they could get, but faced challenges. For example, they weren't certain all 400 microcephaly cases were correctly diagnosed. They didn't know for sure how many of the mothers of the affected children had a Zika infection, nor how many pregnant women in Bahia in total were infected with Zika virus during that time period.

But based on other data from Bahia and from two earlier Zika outbreaks, they could make calculations that accounted for different scenarios — including one which assumed most people were infected and another in which Zika infections were less common.

More on this...

A large Zika epidemic would lead to more babies born with birth defects, but the chance that an individual baby would be affected shrinks as an outbreak progresses. That's because a large outbreak would expose more women to the virus and allow them to develop immunity by the time they got pregnant, Silverman said.

The study found the risk of a fetus developing microcephaly ranged from about 1 percent, when most people were infected, to nearly 14 percent, when only a small proportion were.

"This gave us an appreciation that the risk is substantial," said Michael Johansson, a CDC biologist who was the study's lead author.

Previously, the best estimate of the Zika-related risk of microcephaly came from a very small study based on eight cases that occurred during a Zika outbreak in French Polynesia in 2013 and 2014. That study estimated risk of microcephaly at 1 percent for pregnancies in which the mother was infected in the first trimester.

The range reported in the new study was also for women infected in the first trimester. For women infected later in pregnancy, the risks were well below 1 percent no matter how much Zika was spreading in the community, the authors found. The Rio study, though, found other harms in babies born to women infected during the second or third trimesters, experts noted.