Benzyl alcohol, a preservative commonly used to kill germs in intravenous solutions, may have contributed to the deaths of 16 premature infants in two medical centers, the Food and Drug Administration said yesterday.

The agency, in a letter being mailed today, has urged 50,000 pediatricians, hospital pharmacists and hospital administrators to discontinue its use when treating premature and newborn infants. The FDA said the preservative is considered safe in intravenous solutions administered to children and adults.

The warning letter was prompted by two unpublished studies, one by a research team in New Orleans and another performed independently by a group in Portland, Ore., that were conducted after neonatal specialists in both cities had grown suspicious of a pattern of symptoms in the deaths of infants with a birth weight of four pounds or less.

"These infants were critically ill to begin with, but we noticed that their patterns were different from what you normally see in newborn illness," said Dr. Juan J. Gershanik, director of neonatology at Southern Baptist Hospital in New Orleans.

"The one particular finding that tied them together was a gasping syndrome. With most newborns, the respiratory problems tend to get better over time. With these, they would get worse. And there was also a general unresponsiveness and a tendency to seizures."

Gershanik said that over a period of roughly 18 months in 1980-81, 10 premature babies at the hospital exhibited such symptoms prior to their death. All had been given benzyl alcohol in their intravenous solution (as were all other premature babies who had an intravenous hookup), and all had traces of benzyl alcohol in their blood and urine.

Since June, 1981, the hospital has stopped using the preservative in intravenous solutions administered to premature infants, and, Gershanik said, none of the babies who have died have exhibited those particular "gasping" symptoms.

"The data is certainly suggestive, but there has to be more study," he said.

Benzyl alcohol is most commonly found in intravenous solutions that are drawn more than once from the same bottle or container. Its function is to retard bacterial growth once the seal of the container has been broken.

The familiar sterile saline or water solution that most people associate with intravenous feeding does not contain the preservative. That is because it is in a bottle whose seal is broken only once.

However, there are occasions when doctors draw repeatedly from the same container of saline solution, and these circumstances frequently arise with premature infants.

With these babies, a catheter is commonly inserted into the umbilical artery in order to take blood and to provide food and medication. When blood is taken, it has a tendency to clot in the catheter, so it is necessary to flush it out with a saline solution. The flushing solution is typically drawn repeatedly from the same bottle, and it thus contains the benzyl alcohol to ward off germs.

Dr. Paula Botstein, assistant to the director of the bureaus of drugs and biologics at the FDA, said it is not known how many hospitals use the solution with benzyl alcohol and how many use a straight saline solution and discard the bottle after each use. She added that she assumes the latter practice will become standard operating procedure once the letters are received.

She said the agency is considering whether to push for a labeling requirement on the benzyl alcohol to warn against its use with premature babies, but added she believed the letter was the fastest and most effective way to alert doctors.

The infant mortality rate among premature infants has been falling steadily. Gershanik said that at his clinic in New Orleans, which takes in roughly 500 premature infants a year, the mortality rate is about 15 percent. Roughly one-seventh of those deaths, he said, had evidence linking benzyl alcohol as a contributing cause.