Finding out how prescription drugs affect children isn't easy, even for pediatricians, a new study says.

That's because very little research on children and drugs is published in medical journals that help guide doctors on treatment. The result is that some prescribe the wrong dose or use drugs that could be harmful to kids.

"Ironically, some of the times when drugs do work (in children), they're still not getting published," said Dr. Danny Benjamin, an associate professor at Duke University who led the study and also works for the U.S. Food and Drug Administration.

He said an FDA program meant to encourage drug companies to test how drugs affect children has led to hundreds of studies. The problem is that about half the time, the results don't get published in peer-reviewed medical journals, mainly because researchers and sponsors don't submit them for publication, Benjamin said.

Drug companies that conduct or sponsor pediatric research are motivated mostly to get their products on the market, "not to tend to the public health concerns," Benjamin said.

Also, parents often are reluctant to let their children participate in studies. So the research often involves many institutions with a few children at each location, which complicates compiling data and submitting them for publication, Benjamin said.

Examples the authors cited include unpublished data suggesting that an anesthesia drug might increase children's risk of death when used for sedation. Also, unpublished data has suggested that some steroid creams used for skin rashes in adults could cause a hormone imbalance in children.

"People slather this on children, particularly babies," said study co-author Dr. Dianne Murphy, director of the FDA's office of pediatric therapeutics.

In both cases, precautions are listed on the drug label but not in much detail. They also appear on the FDA's Web site, but that's not where doctors usually look for such information, the researchers said.

"We've just got to get the data out to people who are caring for children," Benjamin said.

His study appears in Wednesday's Journal of the American Medical Association.

JAMA is among the top peer-reviewed journals but editor-in-chief Dr. Catherine DeAngelis said few studies submitted to JAMA involve the effects of medication on children.

The researchers studied the impact of 1997 legislation authorizing a measure that grants drug companies longer patent protection when they agree to study a medication's effects in children.

Between 1998 and 2004, 253 pediatric studies were submitted to the FDA under this program but only 45 percent were published in peer-reviewed journals, the researchers found.

Dr. Peter Lurie of the watchdog organization Public Citizen's Health Research Group said drug companies and academics need to push harder to publish.

"It really is like the tree falling in the woods. The study is of no use whatsoever if it never reaches the practicing physician," Lurie said.

Scott Lassman of Pharmaceutical Research and Manufacturers of America, an industry trade group, said drug companies shouldn't be faulted.

While he agreed publication in a peer-reviewed journal is "the gold standard for getting information out," Lassman said companies often present data at medical conferences and or post them in an online industry database, http://www.clinicalstudyresults.org.