Three new unpublished studies indicate an infrequent association between certain birth defects and a "morning sickness" drug taken by or available to an estimated 30 million women worldwide since 1956.

The drug is trade-named Bendectin and Debendox elsewhere. U.S. physicians in 1978 wrote 3.4 million new and refill Bendectin prescriptions. Debendox is sold over the counter in Canada and other countries.

The Food and Drug Administration has set Sept. 15 and 16 for a meeting of outside expert advisers to evaluate the new studies and other evidence. The sessions will be held at agency headquarters in Rockville.

Dr. J. Richard Court, director of the FDA's Bureau of Drugs, said yesterday, "My guess is that there is going to be a lot of controversy." He said that there are inconsistencies among the studies and questions about the scientific reliability of at least one of them.

At a trial earlier this year in Orlando, Fla., world-renowned experts were split on whether Bendectin is a so-called low-grade teratogen, a weak birth-deforming agent.

Whatever the ultimate decision, Crout said, "Bendectin is very clearly over-prescribed" by physicians who sometimes show "great abandon." For many women afflicted with the nausea of pregnancy, relief is available with unquestioned safety from dry crackers or toast and very cold or very hot liquids, he said.

In Wilton, Conn., Bendectin's manufacturer, Richardson-Merrell Inc., said it has not yet seen the new studies and consequently could not comment on them. "We are firmly convinced that the drug is safe and have seen no scientific evidence, at this point, to cause us to change our position," said communications chief Robert B. Irvine. The company's 1978 Bendectin revenues were about $14.4 million.

For almost 20 years, Bendectin was a combination of an antispasm agent, an antihistamine to control nausea, and a vitamin supplement. In 1976, the company removed the antispasm ingredient, Bentyl. Sales of the three-ingredient version continued abroad.

The three unpublished studes were conducted by the respected Boston Collaborative Drug Surveillance Program in cooperation with the 280,000-member Group Health Cooperative of Puget Sound, Wash.; the Birth Defects Branch of the federal Center for Disease Control in Atlanta, and British epidemiologists Jean Golding and J. A. Baldwin of the University of Oxford.

In the U.S. studies, a key concern was limb deformities, particularly "reductions" in which part of an arm or leg is missing. A limb reduction was at issue in the Orlando trial in which $10 million in damages were sought from the manufacturer by a boy whose mother, Betty Mekdeci, had taken Bendectin in the period of about 24 to 40 days after conception, when embryonic limbs develop.

The trial ended in March with each side claiming victory. The jury found for the Mekdeci family -- but awarded only $20,000. The presiding federal judge, saying the outcome was inconsistent with his instructions to the jury, then decreed a new trial, which is set for January.

The family's attorneys say they have learned of 160 limb-reduction cases allegedly related to Bendectin. An additional 75 cases, according to a summary by the American Trial Lawyers Association, are in or under consideration for litigation.

The Boston study found five limb deformities among the offspring of 2,416 Seattle-area women who had been exposed to Bendectin, as compared with four among 3,836 nonusers. Among users, the overall relative risk as compared with nonusers was 2.1 times greater, which was termed statistically insignificant. But the risk increased by 50 percent to 3.2 when mothers who filled more than one Bendectin prescription were taken into account.

Of the total of nine deformities, four were limb reductions. Only one of the four reductions was among Bendection babies. The other five deformities were of other kinds, such as an extra thumb or toe.

The data base was unique, for these reasons: virtually all of the Puget Sound cooperative's members get all of their medicines from it. Information on their prescriptions is computerized, along with their hospital diagnoses, childbirth and medical follow-up.

In the CDC study, Dr. Jose F. Cordero and three collagues found "no significant association" between limb reductions and use of the original three-ingredient Bendectin. A birth defect rate of less than one per 1,000 births for a particular kind of congenital amputation and another anomaly was found in women exposed to the drug after October 1977.

In the British study, 240 women whose babies had cleft lips or palates were matched with 480 "control women whose babies did not have these defects.