A new warning about possible risks of Bendectin, the only government-approved medicine for the "morning sickness" of early pregnancy, is expected to be added to the prescribing instructions for physicians because of new studies that raise questions about possible birth defects.

Concerned about the results of the studies, the Food and Drug Administration called the manufacturer, Merrell Dow Pharmaceuticals, Inc., to a closed 2 1/2-hour meeting last Tuesday at which the FDA discussed the possibility of stricter labeling. A spokesman for Merrell said the company proposed including "additional information on animal studies" in the labeling "for the sake of completeness, even though the studies are only preliminary." No deadline was set, but sources said the agency is pressing for early July.

Results of a recent lab study indicated a possible link between Bendectin and a life-threatening hole in the diaphragm called diaphragmatic hernia. FDA statistics indicate the defect occurs in about 40 of each 10,000 babies exposed to the drug in the first three months of pregnancy. The defect lets the intestines enter the chest and squeeze the lungs, and it is often fatal even in newborns who undergo successful surgery.

The lab study, done in 1981 for the West German health ministry, found diaphragm hernias in 2 to 5.9 percent of the fetuses of Wistar rats treated with various large doses of the antihistamine ingredient of Bendectin, but no hernias in the fetuses of untreated control rats.

In addition, new reviews of statistical surveys have pinpointed an occurrence of the defect among the infants of American women who took either Bendectin or anti-nausea antihistamines.

These results clash with the current officially approved labeling, which gives physicians this assurance: "Studies in rats and rabbits have revealed no suggestion of drug-induced fetal abnormalities at doses of Bendectin up to 90 times the maximum human dose." Since 1980, the labeling has advised a woman to use Bendectin only if she gets no relief from eating soda crackers or dry toast or drinking hot or cold liquids.

Surgeon General C. Everett Koop first heard about the studies from Susan McFalls, a consultant to Rep. Doug Walgren (D-Pa.), who has researched Bendectin since February, 1980.

"The 36 years I spent in clinical medicine before coming to the government had to do largely with the treatment of congenital anomalies," Koop told FDA Commissioner Arthur Hull Hayes Jr. and seven other agency officials at a tape-recorded April 8 meeting arranged at Walgren's request. "I am not a great believer in the old adage that where there is smoke there is fire, but it seemed to me that on this particular occasion there were some burning embers and a few other things that really made it worthwhile to look at this."

Hayes agreed to order a complete review of Bendectin, which is intended to relieve the nausea and vomiting of the first trimester, when the risk of congenital defects is greatest. During peak sales in the late 1970s 400,000 women annually are estimated to have taken it. Pharmacists in 1981 dispensed 1.9 million new and refill prescriptions, down 25 percent from 1980. The drug is also sold abroad -- sometimes over-the-counter -- as Debendox, Lenotan and Meribental.

Reimar Roll of Berlin, who did the diaphragm study, concluded in his report: "The occurrence of diaphragmatic hernias was striking, because this malformation has never been seen in all our control untreated groups checked so far." At daily doses of at least 100 milligrams per kilogram of body weight, he wrote, the antihistamine ingredient, doxylamine succinate, "looks like a clearcut teratogen birth-deforming agent in Wistar rats."

The Merrell spokesman said that the doses were up to 375 times the human equivalent and that mice exposed to Bendectin did not produce deformed fetuses. Rats metabolize drugs much faster than do humans, and, scientists point out, the absence of an adverse drug reaction in one or another species does not detract from the potential importance of its occurrence in another species.

A second preliminary study, done last year on monkeys in California, disclosed a possible link between Bendectin and a hole in the wall of the heart called ventricular septal defect. Although relatively common, it often closes itself by the end of an infant's first year of life.

The two animal studies are the first done independently of Merrell. They supplement earlier Bendectin data on limb and other birth defects. An agency panel of outside advisers concluded in September, 1980, that the data did not establish cause-effect relationships but cautioned that a "residue of uncertainty" exists as to the possibility of a link between the drug and congenital heart anomalies and cleft lips or palates.

FDA files reveal a quarter-century of apparently lax handling of Bendectin data by the agency and by the company, which Dow Chemical Co. acquired in 1981 from Richardson-Merrell, Inc.

For example, the documents show that Merrell knew of Roll's study from the start and knew by mid-September, 1981, of the "general outcome and conclusion." The company didn't submit it to the FDA for 4 1/2 months -- until two weeks after Walgren had put a recently translated copy into the hands of Commissioner Hayes.

The papers also show that for 15 years, staffers repeatedly warned high FDA officials that the principal human safety study -- done by Merrell and used to promote the drug to physicians -- could be invalid and misleading.

One critical episode involves Rolf Bass, a Berlin physician who did a precision analysis of a 1958-1965 U.S. survey of 50,282 pregnancies. He found that doxylamine may induce an 11-fold increase in the normal rate of diaphragm hernia and a 30-fold increase in a range of diaphragm defects that don't threaten life. He warned against use of pregnancy anti-vomiting drugs, including Bendectin.

Bass published his article -- next to one by a German Merrell executive -- in February, 1979, in a German medical journal. The company cited the Bass article in a bibliography sent to the FDA in June, 1980, but forwarded it too late for distribution to members of the Bendectin advisory panel. FDA scientists apparently didn't notice it for nearly two years -- until the April 8 meeting prompted a search of agency files.

For the FDA's Dr. Franz W. Rosa, Roll's work was a provocative clue that the U.S. survey put human birth defects in broad categories that could be obscuring diaphragm hernias, which in 1980 were reported in about 1,300 infants, one-third of them stillborn. By March he had made two striking but inconclusive findings:

Diaphragm defects, some of them not life-threatening, seemed to be related solely to anti-nauseant antihistamines -- at a rate 6.2 times normal -- and not to the chemically different antistamines used in cold tablets.

In the infants of Bendectin users, the apparent incidence of the diaphragm hernia was 7.8 times higher than the rate of the range of diaphragm defects in the babies of non-users.

At the same time, Barbara B. Manard, a Washington-based health consultant, re-analyzed surveys and sifted federal data on up to 1 million hospital birth records a year. Manard found a strong indication that the hernia incidence increased 64 percent between 1970 and 1980, roughly keeping pace with increases in Bendectin sales.

For 20 years, a key element of the story has been the Merrell study of "more than 4,000 pregnancies (2,218 matched pairs)." The study reported reassuring results: the 11 malformations found in the infants of Bendectin users were actually 10 fewer than in the babies of the non-user controls -- 0.72 percent vs. 0.95 percent. For years afterward the company told physicians the study showed "no correlation between Bendectin and fetal abnormalties."

In 1967, an FDA medical officer found gaps in the back-up data. "I feel that this study is of practically no value and actually could be misleading," he said in an internal memo. "I do not believe the results vindicate Bendectin . . . " In 1979, an FDA expert on medical data said the study "appears to be of poor quality." He added, "A conclusion of no association between the overall malformation rate and Bendectin does not appear well-founded."

Ten years ago, the files show, another FDA medical officer told superiors that Merrell's drug experience reports on deformed limbs and other defects in the babies of Bendectin users were "incomplete" and sometimes illegible. Last October, the FDA's Rosa said that 219 cases of defects appeared to be missing from six volumes of Merrell adverse drug reaction reports. He said that 74 inquiries "are clearly adverse drug reaction reports. More serious is a list of 47 claims and 113 lawsuits, 145 of which have specified birth defects."

In 1956, the FDA papers show, Merrell applied for permission to sell the original three-ingredient Bendectin without submitting animal tests on possible toxicity or peril to the fetus. Human tests had been limited, focusing mainly on efficacy and on short-term safety, not on the safety of the fetus. The FDA approved the application within four weeks.

In addition to doxylamine, which is in several over-the-counter sedatives and cold preparations and is chemically related to other antihistamines for nausea and motion sickness, the mixture contained a vitamin. An anti-spasm compound was dropped in 1976 to overcome the "possibly" effective rating that the National Academy of Sciences and the FDA gave the three-part mixture.