DALLAS – The risk of autism is significantly increased in both very small and very large for gestational age infants, compared with average for gestational age infants, according to findings from a large retrospective population-based cohort study.

The findings, which indicate that the risk for autism is increased by 10% for very small for gestational age (SGA) and 12% for very large for gestational age (LGA) infants confirm similar findings from prior smaller studies, and also provide new information about the influence of birth weight on the risk, Dr. Gaea Moore reported at the annual meeting of the Society for Maternal-Fetal Medicine.

Of the nearly 6 million infants from the 1991-2001 California birth cohort, 21,717 who were eventually diagnosed with autism were included in the study. When stratified based on birth weight, the likelihood of autism in the SGA group was significant only in the preterm very SGA infants (those above the 95th percentile), said Dr. Moore of the University of Colorado at Denver, Aurora.

In the LGA group, the pattern was more complex. A significantly increased risk for autism was found only in very LGA infants (above the 95th percentile) born at term (greater than 39 weeks’ gestation).

"Interestingly, we found a protective effect in the LGA birth weight group, which was significant in the preterm period at less than 32 weeks," Dr. Moore said, noting that the finding was significant for both LGA and very LGA infants.

The findings confirm the results of multiple smaller analyses, but also provide a "new piece of the puzzle," she said.

"That is the finding that very SGA children born in the preterm setting have an increased risk of autism," she said, adding, "We suspect that the association might be the result of a combination of neurologic insults at a key window of development – likely, whatever contributed to the restricted growth in the intrauterine environment, combined with the typical neurological insults of prematurity."

As for the dichotomous pattern of risk in the LGA infants, the finding may explain why LGA birth weight has not been associated with autism in the past, she said, explaining that the protection afforded in the preterm period likely cancelled out the increased risk at term.

The increased risk in the term setting may reflect maternal abnormalities such as obesity and diabetes, which are risk factors both for delivery of an LGA child, and for delivery of a child with autism, she noted.

"The decreased risk in the preterm period (for LGA infants), is interesting. We know that LGA preterm infants have a higher rate of survival and a higher rate of survival without major morbidity. Thus, these larger preterm infants may simply survive prematurity with a lesser degree of neurologic insult," she suggested.

Alternately, the finding may reflect incorrect dating, which would explain both larger infants and infants less likely to sustain perinatal or neurological injury, she said.

The findings have implications for heightened screening for SGA and LGA during pregnancy, as well as for encouraging appropriate weight gain during pregnancy, she concluded.

Dr. Moore said she had no relevant financial disclosures.