Hyperemesis gravidarum (HG) is a serious complication of pregnancy. Simple morning sickness is more common and less serious. When I started out in medicine, we routinely treated morning sickness with Bendectin. It was a safe and effective remedy, a combination of the antihistamine doxylamine and a B vitamin, pyridoxine. Unfortunately the manufacturer, Merrill Dow Pharmaceuticals, was bombarded with numerous lawsuits claiming that Bendectin caused birth defects. There was a clear scientific consensus that the evidence did notshow that Bendectin caused birth defects and there was plentiful evidence of its safety. The lawyers prevailed over the science, and in 1983, Dow voluntarily took Bendectin off the market to avoid further litigation expenses. After the drug’s withdrawal, the rate of birth defects did not decrease, but the rate of hospitalization for hyperemesis gravidarum doubled.
This was frustrating to those of us treating morning sickness, because it denied us a remedy that we knew was safe and left us with only less safe options for a potentially life-threatening condition. Some of us found a way to do an end run. We advised patients to buy doxlyamine (sold over-the-counter as the sleep aid Unisom), and to take it with pyridoxine, effectively reconstituting what we could no longer get as a single pill.
The Daubert Standard
One good thing came of the litigation against Dow. The controversy led to a landmark U.S. Supreme Court decision, Daubert v. Merrill Dow Pharmaceuticals in 1993, which set a new standard for rules for admission of expert testimony, including scientific evidence. Under Daubert, the trial judge assumes a gatekeeper responsibility. The judge decides whether an expert’s testimony is based on a reliable foundation of scientific knowledge and can exclude evidence not based on good science. Daubert is now embodied in Rule 702 of the Federal Rules of Evidence:
A witness who is qualified as an expert by knowledge, skill, experience, training, or education may testify in the form of an opinion or otherwise if:
(a) The expert’s scientific, technical, or other specialized knowledge will help the trier of fact to understand the evidence or to determine a fact in issue;
(b) The testimony is based on sufficient facts or data;
(c) The testimony is the product of reliable principles and methods; and
(d) The expert has reliably applied the principles and methods to the facts of the case
The Daubert standard is now the law in federal court and over half the states. While the original intent of Daubert and the rules of evidence the Supreme Court interpreted in that case was to liberalize the introduction of expert testimony, the general consensus is that Rule 702 has kept a good deal of junk science out of the courtroom. This is demonstrated by the history of the Daubert case itself.
The Dauberts had two children with serious congenital defects that they blamed on Bendectin. In the trial court, an expert for the defense testified that extensive published scientific studies had found no evidence that Bendectin was a risk factor for congenital anomalies, and there was widespread acceptance that it was not. The plaintiffs responded with the testimony of eight experts who disagreed, based on animal studies, chemical structure analyses, and the unpublished “reanalysis” of data from human statistical studies. Their star witness, Dr. William McBride, was the lead author of a study allegedly showing that Bendectin caused birth defects. When the paper was published, his co-authors protested about discrepancies, and he was eventually found guilty of falsifying data and of scientific fraud, and was struck off the register (no longer allowed to practice medicine).
The appellate court ruled (as summarized here) that:
The Plaintiffs’ experts did not conduct their research independent of the litigation and the theories have not been published in scientific journals or reviewed by peers, though there has been ample time to do so because the theories and litigation has been around for a decade…The Plaintiffs cannot show causation directly and attempt to show it through circumstantial evidence provided by their experts. The testimony cannot establish that the defects were not caused by an independent cause, since limb reduction defects occur in the babies of mothers who did not take the drug. The Plaintiffs’ experts cannot say that the drug more than doubled their risk of the defect, only that there was a statistical relationship between the drug and the birth defect.
New evidence against antihistamines?
A new study published in the European Journal of Obstetrics and Gynecology and Reproductive Biology, is titled “Antihistamines and other prognostic factors for adverse outcome in hyperemesis gravidarum.” The lead author is Marlena Fejzo, a respected researcher in the field. It is a case control study with 254 patients with hyperemesis serious enough to require intravenous fluids and/or total parenteral nutrition/nasogastric feeding tube, and 308 control patients. Subjects submitted their medical records and completed an online survey. A large number of demographic characteristics, pre-existing conditions, pregnancy symptoms and treatments, and maternal and fetal outcomes were included.
Objective: The stated purpose of the study was to determine the frequency of adverse perinatal outcomes in women with hyperemesis gravidarum and to identify prognostic factors.
Results: Several demographic characteristics were found to be associated with HG: gestational hypertension, immune problems, anxiety, bipolar disorder, and depression. Carrying a female fetus increased the risk of HG. None of the factors that were significantly different in women with HG were associated with adverse outcomes except for gestational hypertension. Onset of nausea and vomiting at 3-4 weeks gestation was associated with poorer outcome; onset of symptoms at 1-2 weeks and >4 weeks was not.
Adverse fetal outcomes (AFOs): Women with HG were more likely to report lower birth weight and prematurity. The overall rate of adverse fetal outcomes was 16.93% compared to 4.55% in controls. There was no significant difference in the rate of birth defects, perinatal mortality, or weight below the 10th percentile.
Medications/treatments: They explored the effect of treatments by comparing 43 HG subjects with an adverse fetal outcome (AFO) to 211 subjects with no AFO. They looked at 36 different medications and treatments. Acupuncture was associated with significantly fewer AFOs (but it was also reported to be ineffective in relieving symptoms). A statistically significant correlation was found between AFOs and Solumedrol, Phenergan, and some (but not all) antihistamines. For most of the 36 treatments there was no statistically significant correlation with AFO. I was puzzled to see dimenhydinate listed as one of the antihistamines associated with AFO but also listed among the anti-motion sickness medications that were not associated with AFO. Bendectin was in the list of antihistamines correlated with AFO, and they noted a “trend towards adverse outcome” with Bendectin itself, but noted that there were not enough subjects to reach statistical significance. In other words, this study did not show that Bendectin caused adverse fetal outcomes.
What about effectiveness? The only treatments self-reported as effective by more than 50% of patients were cannabis, intravenous fluids, methylprednisolone, and ondansetron (Zofran), and there was no significant association between self-reported medication effectiveness and outcome for any medication including the medications associated with poor outcome.
Problems with the study
Here are a few of the things that bothered me about this study:
- The title may indicate bias. They could have called it “Frequency of adverse perinatal outcomes and prognostic factors in women with hyperemesis gravidarum” which would have been more in line with the stated objectives of the study, but they chose to emphasize antihistamines in the title. Maybe I’m just quibbling, and I admit this is a hot-button issue for me because of what happened with Bendectin.
- The discussion at the end expressed more concern about the use of antihistamines than I felt was warranted. Again, that could be bias on my part.
- They examined 36 different treatments. When you look at that many factors, you are almost guaranteed to find some spurious correlations. They did not do any statistical corrections for multiple endpoints.
- The numbers were far too small to reach reliable conclusions for most treatments. They only looked at 43 subjects with HG and AFOs, and the numbers of subjects using each treatment were small.
- Some of the numbers in the table struck me as odd. For example, for Motilium, there was a total of only 2 patients who got the treatment, presumably one in the AFO group and one in the no-AFO group. They nevertheless calculated a p value of 0.3105%. They calculated p values for several other treatments that were only used by 4-6 subjects total. Is it legitimate to calculate p values based on such small numbers? I don’t know, but it certainly isn’t very meaningful. And calculating out to 4 decimal places for a sample of 2 is beyond silly.
- There are other studies that support the safety of antihistamines. One study even found that they protected against fetal malformations.
- Data were based on patient recall, which is potentially unreliable.
- Early symptoms were associated with more AFOs. I wondered whether patients with early symptoms were more likely to try antihistamines.
What Should We Make of this Study?
The mom/baby websites are already warning against antihistamines for morning sickness. A typical one says:
Poor outcomes were associated with early start of symptoms and treatment with methylprednisolone, promethazine and other antihistamines; Benadryl, Gravol, Unisom, Vistaril/Atarax and Diclectin/Bendectin, independent of effectiveness.
That’s inaccurate. The association they found was for a group of many antihistamines lumped together, and when they examined Bendectin separately, they did not find a significant association.
Should we change our practices based on this study? I think not. Better studies will be required before we can conclude that antihistamines are harmful.
In June, 2013, Bendectin (doxylamine/pyridoxine) returned to the US market under the new brand name Diclegis. We already had good evidence that Bendectin was safe, and this new study failed to find a significant association between Bendectin and adverse fetal outcomes. I see no reason to avoid the new formulation.
Note: Thanks to Jann Bellamy, JD for her assistance with the Daubert section.
Corrections: The Daubert case involved one Daubert child and a child from another family. McBride was not a witness in the original case, but his now-discredited study was relied on by other witnesses.
This article was originally published in the Science-Based Medicine blog.