Toddlers of moms who took the epilepsy drug valproate during pregnancy had lower IQs than the children of women who used other anti-seizure medicines, according to a new study.

The valproate children had IQ scores six to nine points lower by age 3, said the study's lead author, Dr. Kimford Meador of Emory University. The drug, also sold in the U.S. under the brand name Depakote, had previously been linked to birth defects, particularly spina bifida. Women of childbearing age have long been advised to avoid it.

"We've known this drug is a bad actor for a long time," said Dr. Lewis Holmes, director of the North American Antiepileptic Disease Pregnancy Registry, based at Massachusetts General Hospital in Boston.

The new study is important because it's the largest to show a connection between valproate and diminished IQ. Its publication in the prestigious New England Journal of Medicine should alert physicians who until now have ignored the drug's potential dangers to fetuses, added Holmes, who was not involved with the study.

In the United States, about 25,000 children are born each year to women who have epilepsy, a brain disorder that causes people to have recurring seizures. In the study, researchers followed pregnant women in the United States and United Kingdom between 1999 and 2004. The results are based on about 260 of their children.

Toddlers whose mothers had taken valproate had IQs of 92, on average. In contrast, IQ scores were in the range of 98 to 101 for children of women who had taken lamotrigine, phenytoin, and carbamazepine. IQ tests are designed so a child of average intelligence scores 100.

The higher the dosage of valproate a woman had taken, the lower the IQ of the child, the researchers found. For the other drugs, dosage levels made no significant difference.

The number of children in the study is small, and it's possible that other factors influenced the results. However, the researchers accounted for differences in a child's birth weight, the age and IQs of their mothers, the type of epilepsy the mothers had, and other factors that could have influenced the results.

A major drawback is that the study did not include children whose epileptic mothers took no medication during pregnancy, Holmes said.

It's possible that all four epilepsy medications had some effect on mental development, he said. Without such a comparison group, it's hard to know, said Holmes, who is also a professor of pediatrics at Harvard Medical School.

Valproate, also given for migraine headaches and mood disorders, continues to be used by some epileptics because it's the only thing that works, said Meador, the study author.

Doctors say it's often important for epileptic women to keep taking their medications during pregnancy because seizures can lead to injuries.

Women on the drug who want to get pregnant should plan their pregnancies carefully and consult with a doctor, wrote Swedish researcher Dr. Torbjorn Tomson, in an editorial that accompanied the new study.

Switching drugs after a woman realizes she is pregnant is unlikely to reduce the risk of birth defects. And abruptly stopping he medication may endanger the mother and the fetus, he wrote.

"That could be catastrophic," Meador agreed.