The number of babies born hooked on their mothers' opioids has been soaring for almost two decades, yet experts still are not sure whether morphine or methadone is best to gradually break the infants' drug dependence.

A rigorous new government-funded study has found that methadone has a slight advantage over morphine, modestly reducing babies' length of treatment and hospitalization.

But the study, published Monday in JAMA Pediatrics, further complicates the question of how to safely and effectively treat babies going through withdrawal, called neonatal abstinence syndrome (NAS). At the behest of the U.S. Food and Drug Administration, the researchers spent a year developing a methadone formulation that could be custom-prepared at each of the eight participating hospitals. This was done to avoid using commercial methadone solutions, which are not approved for pediatric use and contain alcohol as a preservative.

No one can say whether the alcohol is harmful for newborns, but the researchers and the FDA wanted to eliminate the theoretical possibility.

"In commercial methadone solutions, 15 percent is alcohol," said lead researcher Jonathan M. Davis, a Tufts Medical Center pediatrician who chairs the FDA's neonatal advisory committee. "That's double the amount of alcohol in a glass of wine. Most people I tell about this, they're dumbfounded, partly because no one looks at the label."

Davis said he hoped the results prompt a pharmaceutical company to seek FDA approval of an alcohol-free pediatric methadone product.

About 80 percent of hospitals treat newborn withdrawal with morphine, the only opioid approved for pediatric use. Like adults in withdrawal, infants feel the effects in every organ system; their symptoms include tremors, muscle spasms, shrill crying, irritability, diarrhea, vomiting, poor sleeping, and seizures.

Methadone and another medication, buprenorphine, are milder opioids used to help people stop using painkillers or illicit opiates such as heroin. Even pregnant women are put on methadone or buprenorphine to cut cravings without delivering euphoric effects because abruptly quitting opioids increases the risk of miscarriage. However, their newborns may suffer withdrawal.

Buprenorphine, which is not approved for babies, has not caught on for NAS treatment, even though Thomas Jefferson University researchers published a study a year ago that found it reduced hospital stays compared to morphine. Jefferson now relies on buprenorphine to treat NAS.

Walter Kraft, the physician and clinical pharmacologist who led the Jefferson study, said the buprenorphine formulation contains 30 percent alcohol.

"We did measure the levels [of blood alcohol] in the babies," Kraft said Friday. "It turns out infants clear it quickly, quicker than adults. But in terms of the safety of alcohol in neonates, it's not well-established."

Kraft, who was part of a group that monitored the safety of the new study, called the results "exciting."

"I believe, although it hasn't been demonstrated, that the results would be comparable with commercial methadone" solutions, he said.

The new study — the largest of its kind and funded by the National Institute on Drug Abuse — involved 116 infants who were randomly assigned to methadone or morphine. Their doses were based on their weight and the severity of their withdrawal symptoms. Medical staff did not know which drug the babies were receiving.

The babies on methadone needed treatment for an average of about 14 days during an average hospital stay of 22 days. On morphine, babies averaged about 16 days of treatment and 23 days in the hospital.

Another advantage of methadone: It's longer-acting, so the babies needed only three doses a day, compared to six doses of morphine. (The methadone regimen included three placebo doses so it was indistinguishable from morphine.)

The new study, like the Jefferson study, did not consider costs. But reducing hospitalization by even a day or two could translate to big health-care savings. A recent study calculated hospital care costs $1.5 billion for the estimated 23,000 infants a year who go through withdrawal.

And that count is rising. In Pennsylvania, for example, the number of drug-exposed newborns has skyrocketed from 788 in 2000 to 3,289 last year; about 60 percent suffered from NAS, according to an analysis by the Pennsylvania Health Care Cost Containment Council.

"Our study suggests methadone is a more effective agent," Davis said. "And three times a day is a better approach than six times a day. There's less chance for error. But I'd love to see a safer formulation."