WASHINGTON (AP) -- Infections may play a bigger role in premature birth than doctors have thought, says a new study that found almost one in seven women in preterm labor harbored bacteria or fungi in their amniotic fluid.

It's a small study, and it doesn't prove that the germs triggered the early labor.

But Monday's research used specialized molecular testing to uncover microbes that ordinary methods miss, and thus uncovered more women with simmering infections than previously estimated.

The more heavily infected the amniotic fluid, the more likely the woman was to deliver a younger, sicker baby, researchers reported in PLoS One, the online journal of the Public Library of Science.

"We don't think any organisms belong in the amniotic sac," said Stanford University microbiologist Dr. David Relman, the study's senior author. "You'd have to presume there's something wrong."

More than half a million babies a year are born premature, before completion of 37 weeks of pregnancy. It's a toll that has steadily risen for two decades, yet doctors don't know the cause of most preterm births or how to prevent them. Every extra week in the womb helps. Those born before 32 weeks face the greatest risk of death or devastating disabilities, but even babies born a few weeks early can face serious problems.

Certain infections, such as vaginal or urinary tract infections, are known to raise the risk of premature birth, presumably by causing inflammation that in turn triggers labor.

But how much of a role infections play, especially those that don't cause outward symptoms, has been a key question, said Dr. Michael Katz of the March of Dimes, who wasn't part of the new study.

So the Stanford team tried a new approach.

They turned to samples of amniotic fluid saved from women who had gone into preterm labor at a Detroit hospital between 1998 and 2002. Doctors at the time tried standard tests to detect infection, and saved the leftover fluid for research.

This time around, Relman and research fellow Dr. Dan DiGiulio used more sophisticated testing known as PCR to find and reproduce bits of genetic material from germs. They used only fluid that had been collected through a syringe in the abdomen -- like routine amniocentesis tests are done -- before the women's water broke, to ensure the tests didn't detect post-labor germs. Ultimately included were 166 women, 113 of whom delivered prematurely.

Some 15 percent of the women harbored bacteria or fungi, and those who did all delivered prematurely. Adding the PCR tests found 56 percent more infected women than standard testing alone detected.

Even that is likely an underestimate, the researchers concluded, because they were using samples so old that the DNA in them had begun degrading.

The standard tests were especially likely to miss infections in women whose babies were born extremely premature, before 25 weeks.

And there was a surprising variety of germs: 17 bacterial species -- including one never-before-seen type -- and one fungus.

"It's a very, very important first step," the March of Dimes' Katz said of the research. But, "there are still many hurdles."

Next researchers will have to prove if harboring these germs really predicts who will go into preterm labor. Relman's team, with funding from the National Institutes of Health, now is studying 2,000 women who get routine amniocentesis in their second trimester, to try to answer that.

If so, then the questions become where those germs originate, whether there's a less invasive way to find who's at risk, and if there's any treatment that might help.

Copyright 2008 The Associated Press. All rights reserved. This material may not be published, broadcast, rewritten or redistributed.