SUNDAY, May 1 (HealthDay News) -- Doctors are erroneously diagnosing autism spectrum disorder in many 18-month-old toddlers who were born extremely premature, a small new study suggests.

Later testing showed that a high proportion of the toddlers diagnosed with autism spectrum disorder at that age were not autistic, the researchers found. Instead, the youngsters simply had a cognitive or language delay, both of which are common in children who were preterm babies, the researchers said.

Part of the problem, the researchers suggested, is that children who were premature may be undergoing testing for autism too early in life.

"We need to continue to watch these kids very closely, keeping in mind that they're at high risk for multiple, different kinds of developmental difficulties and delays, but not necessarily jumping to the conclusion that this is definitely autism at such an early age," said study lead author Dr. Bonnie Stephens.

"We need to monitor them over time, knowing we may not be able to diagnosis them [accurately] until they're a little bit older," added Stephens, a neonatologist and developmental and behavioral pediatrician at Women & Infants Hospital of Rhode Island.

Many neonatal specialists follow the recommendation of the American Academy of Pediatrics that all children aged 18 to 24 months should be screened for autism spectrum disorder. Those who score below-average on the test are then referred for more diagnostic testing.

Stephens said rates of autism are on the rise in the United States and, "there seemed to be higher rates in the extremely preterm population. The first couple of studies showed rates that were higher than seemed possible and these studies were coming out right after the AAP recommendation."

But many things can go wrong in the early life of preterm infants, such as language and cognitive delays, Stephens said.

"We bet that ... there might be a lot of false-positives," she added.

That hypothesis turned out to be correct -- at least in this small study sample of infants born prematurely at Stephens' hospital and monitored for the first 30 months of their lives. Babies in the study had been born before 28 weeks of gestation and were assessed for autism spectrum disorder using three different tests at 18 months and 30 months of age (with the ages corrected for prematurity).

At the first assessment, 18 percent of 152 toddlers tested positive for autism spectrum disorder, but at 30 months, only 10 percent of 116 youngsters did.

"We saw a significant decrease in positive screens," said Stephens, who stressed that the findings were preliminary and that follow-up research is continuing with her team planning a larger trial.

As it turned out, only 3 percent of the youngsters failed all three screening tests at both 18 months and 30 months of age. Later testing showed these children did end up being diagnosed with autism spectrum disorder.

But toddlers who tested positive on some screens for autism spectrum disorder at both 18 and 30 months of age did have cognitive and language delays, Stephens said.

"Early diagnosis is important, but we were becoming concerned that a high rate of false-positives had at least the potential to cause unnecessary stress for parents. In this group of [premature] kids, we followed them so closely they all got early intervention anyway [for other problems]," Stephens said. "We felt that the early identification of a positive screen may not be as critical as making an accurate diagnosis."

Stephens was scheduled to present the findings Sunday at the annual meeting of the Pediatric Academic Societies, in Denver. Research presented at such meetings should be viewed as preliminary because it hasn't undergone scrutiny similar to studies published in peer-reviewed journals.

The study findings run counter to one trend in autism spectrum disorder research these days, which is that being born early may be a risk factor for autism, said Cynthia Johnson, director of the Autism Center at the Children's Hospital of Pittsburgh, University of Pittsburgh Medical Center.

Taken together, she said, "this speaks to the lack of our measures being sensitive enough to differentiate preemies who may end up having some cognitive differences in language but [ones] not specific to autism."

"Deferring a diagnosis makes sense. Sometimes we have to do a little more wait-and-see," Johnson said. "I think if providers are suspicious of an autism diagnosis, it warrants mentioning to families. These families know their children are at risk for a number of things, not just autism, and the science is not where we can be more definitive right now."

More information

The March of Dimes has more on premature babies.