President, Consultants in Epidemiology & Occupational Health, Washington, DC, Associate, Department of Health Policy and Management, School of Hygiene and Public Health, Baltimore, Maryland, Adjunct Professor, Preventive Medicine and Biometrics, Uniformed Services University of the Health Sciences, Bethesda, Maryland, Clinical Assistant Professor, Pediatrics, Georgetown University Medical School, Washington, DC, U.S.A.
Bendectin was widely used for the treatment of nausea and vomiting of pregnancy (NVP) in the United States until 1983 when production was discontinued. At that time it was alleged to have caused birth defects. Numerous studies have been published containing data relevant to that hypothesis. Our epidemiological analysis of available studies in 1984, which was prepared for the court in a case against the manufacturer, found that there was no association between Bendectin use and birth in toto, nor was there any association between Bendectin use and birth defects by organ system.1 Furthermore, analysis of nationwide data showed no change in the birth defect prevalence rates for any birth defect after the sales of Bendectin significantly decreased in 1980 and subsequent years.
A meta-analysis of 16 cohort and 11 case-control studies was performed as an update of the 1984 analysis.2 In the meta-analysis, each person in each study was classified into one of four categories: an exposed case, an exposed non-case, an unexposed case, an unexposed non-case. For all malformations, and for each malformation group, a 2x2 table was developed from the data in the four categories. Analyses were separately done for cohort studies and case-control studies, with odds ratios and confidence limits calculated. Analyses were also performed with data from the cohort and case-control studies pooled. Finally, the meta-analyses were performed chronologically by publication date to determine the increased utility of the additional studies. It was seen that the conclusions developed from the totality of all the studies published through 1991 did not differ from those reached on the basis of all the studies published through 1977. That is, the additional studies in the intervening years had added little new information.
Our meta-analysis showed epidemiologically no association between birth defects and the use of Bendectin. The pooled relative risk was 0.95 (95% CL 0.88-1.04) for congenital malformations with exposure to Bendectin in the first trimester, as summarized from nearly 30 years of published data. None of the individual studies showed a significant association between Bendectin exposure and the risk of all malformations combined. A test for heterogeneity resulted in no heterogeneity among the studies, thus all studies were measuring the same relative risk.
Analysis of data by organ system found no significant increased risk for any or all of the seven congenital malformation categories. There was not enough information on individual malformations to be more specific. The following were the calculated odds ratios:
We then turned to an ecological analysis in order to answer the question of whether there was a change in birth defect rate after Bendectin was withdrawn. In the period of 1970 to 1980, Bendectin was used by about 30% of all pregnant women in the United States (U.S.). Between 1980 and 1984, the product disappeared from the marketplace. After 1984, essentially no patients in the U.S. were treated with Bendectin. Was there a difference in the birth defect rates over the various time periods?
There were three measures of exposure to Bendectin: new therapy starts, the number of tablets of Bendectin produced (Bendectin production and use data for the U.S. were obtained on an annual basis since 1970), and an estimate of the treatment frequency of various conditions (from drug use data from the National Ambulatory Medical Care Survey). We demonstrated the temporal pattern of Bendectin usage estimates was similar for all three measures.
For outcome measures, we obtained the birth defect prevalence rates in the U.S. for individual congenital malformations from the Birth Defect Monitoring Program at the Centers for Disease Control (BDMP, CDC, Atlanta) for the period 1970-1990. Graphic analysis showed that there was no association between the incidence of the following birth defects in the U.S. and the use of Bendectin: facial clefts, cardiac, central nervous system, and limb reduction. Nor was there an association between the use of Bendectin and any of the specific malformations surveyed.
One of the arguments against these findings was that there might have been a replacement treatment prescribed for NVP whose birth defect risks were similar to those of Bendectin. Analysis of drug use data from the National Ambulatory Medical Care Survey (1980-1990) shows no replacement of Bendectin use by use of other antiemetic drugs. Essentially, the use of antiemetic medications during pregnancy decreased similarly to the reduced use of Bendectin during pregnancy.
An analysis of hospitalization rates for NVP was undertaken to assess whether the reduction of antiemetic use during pregnancy affected the rate of hospitalization for NVP. Data from the public use data tapes of the Hospital Discharge Survey (1974-1994, U.S. National Center for Health Statistics) demonstrated a doubling in the hospitalization rate for NVP contemporaneously with the general withdrawal of antiemetic therapy for NVP and the disappearance of Bendectin from the marketplace. At the same time, the rates for birth defects did not change. These data appear to indicate on a population basis the safety and efficacy of Bendectin in the treatment of NVP.
Back to home