Children who are born only a year or so after an older sibling appear to be more likely to be diagnosed with autism than siblings with a bigger age gap, surprising new research suggests.

The study, which appears in the journal Pediatrics, is preliminary and the authors say more research is needed to confirm the finding. But the study is a large one, based on more than a half-million children, which reduces the likelihood that the findings were based on chance.

The study looked at births from 1992 through 2002 in California. The researchers looked at second-born children born to the same parents whose older siblings didn't have autism.

They found the overall prevalence of autism was less than one per cent in the study. But they found that the sooner the second child was conceived, the greater the likelihood that child would later be diagnosed with autism.

When mothers became pregnant with their second child three years or more after the first, about 2.5 out of every 1,000 second siblings were diagnosed with autism. But when the second pregnancy occurred less than a year after the birth of the older sibling, about 7.5 in every 1,000 of the children were diagnosed with autism.

Rates were somewhere in the middle for mothers who became pregnant between one and three years after giving birth to their first kid.

The effect was found for parents of all ages, decreasing the chance that it was the older age of the parents that increased the risk of autism and not the birth spacing.

The results seem to have surprised even the researchers who crunched the data.

"That was pretty shocking to us, to be honest," senior author Peter Bearman of Columbia University in New York told the Associated Press.

The researchers said they took into account other risk factors for autism, such as the mother's age and race, and the child's gender, but the increased risk of close birth spacing remained.

"No matter what we did, whether we were looking at autism severity, looking at age, or looking at all the various dimensions we could think of, we couldn't get rid of this finding," Bearman said.

Still, Bearman's team says more studies are needed to confirm the birth spacing link.

Some data suggests the rates of closely spaced births are increasing, perhaps because women are delaying childbirth and are eager to have children before their fertility declines. U.S. government data show the number of closely spaced births – babies born less than two years apart -- is rising, from 11 per cent of all births in 1995, to 18 per cent in 2002.

Why having children in close succession would raise the risk of autism isn't clear. It could be a simple matter of better diagnosis, the authors say, since parents are more likely to notice developmental differences in their children if they are close in age.

But biological factors could also play a role. It's known that pregnancy depletes a mother's store of key nutrients, such as iron and folate. Prior research has tied close birth spacing to low birth weights and prematurity, possibly because of lack of folate.

"And it could be a combination of effects, not a single explanation but a combination of dynamics," Bearman said.

The researchers looked at births from 1992 through 2002 in California. They analyzed data on second-born children born to the same parents whose older siblings didn't have autism. The information on autism diagnoses came from the state's Department of Developmental Services.

The overall prevalence of autism was less than one per cent in the study. Of all the 662,730 second-born children in the analysis, 3,137 had an autism diagnosis. Of the 156,034 children conceived less than a year after the birth of their older siblings, 1,188 had an autism diagnosis -- a higher rate, but still less than one per cent.

Children with Asperger's syndrome and pervasive developmental disorders, milder forms of autism, weren't included. Government studies indicate about one in 100 children have autism disorders, including the milder forms.

Dr. Diane Ashton, March of Dimes' deputy medical director, called the study results an interesting finding that she hasn't seen in prior research. The results will have to be replicated, she said, but her organization already suggests at least a year between pregnancies.

"That is to allow the mother to rebuild depleted nutritional stores and decrease the risk for low birth weight and prematurity. Surely this evidence would provide additional reasons for those recommendations to be made," she said.

The March of Dimes also recommends that all women of childbearing age take a daily multivitamin containing folic acid, an artificial version of folate. Since half of pregnancies aren't planned, the recommendation includes women who aren't trying to get pregnant.

The new study was funded by the Robert Wood Johnson Foundation and the National Institutes of Health.

Bearman hesitated to give advice to parents planning families because the results are so new and unconfirmed. Older parents may not want to wait two or three years for a second child because of other health concerns, he said.

"The advice for parents is to pay attention to the science," Bearman said.