UK Cochrane Centre, London, England, U.K.
This survey describes in brief what clinical trials have been published on the effectiveness of treatments in preventing or controlling nausea and vomiting in pregnancy (NVP). The purpose is to follow the sequence in which the various drugs and other treatments were tried, and identify important gaps in knowledge that remain. The survey was done as part of a larger survey of trials covering nausea and vomiting from all causes (Herxheimer, submitted for publication).
Clinical trials are necessary to determine how effective a treatment is, and to do that one must compare the treatment with what would have happened if it had not been used. We thus need a comparison group, called the control group. To make sure that any differences in outcome between the test group and the control group can be attributed to the test treatment, everything else about the groups should as nearly as possible be the same - the severity and duration of the illness or symptoms, the staff and lay care givers looking after the patients, food, and so on. The best way of making sure that the participants in the treatment and control groups are as similar as possible is to randomise them - that is, to determine by chance which group each trial participant will be in. Then, the larger the groups, the more alike they are likely to be in all their characteristics, though we can never be sure that we have accounted for all the variables which could influence the outcome. Randomisation is the most effective way of minimising bias, that is, of allowing an objective assessment of the effect of a treatment; it is the single most important feature of a trial. Evidence from randomised trials carries far more weight than information from studies in which participants were not randomly allocated to either a treatment or a control group.
An important point in relation to trials is that the likelihood of NVP is influenced by many factors, notably the woman's previous experiences, feelings and expectations. So the use of control groups is also necessary to assess how frequently NVP occurs without the intervention that is being tested. Prevention of NVP thus encompasses much more than the administration of an antiemetic treatment: it includes informing the woman about what to expect, allaying anxiety, and explaining what she can do to minimise the occurrence or intensity of NVP. That suggestion she can control as well as induce NVP is illustrated by the case of a woman with severe NVP who obtained relief by taking ipecacuanha, a notorious emetic.1 Placebo effects on NVP can thus be powerful, and the design and evaluation of trials must allow for them.
Since data from uncontrolled studies are unreliable and often misleading, the survey considers only evidence from randomised clinical trials, i.e., trials in which the participants were (or may have been) randomly assigned to the treatments that were being compared. The trials were those identified by Jewell and Young in their systematic review.2 The abstracts or summaries of the trials were examined to ascertain what questions each had addressed.
Only 25 published reports concerned trials of treatments of nausea and vomiting (NV) in early pregnancy, very few compared with 425 trials on post-operative NV, and 500 on NV caused by cancer chemotherapy published in the same period (Herxheimer, submitted for publication). Table 1 groups these 25 trials by the treatments that they tested.
Antihistamines were the first drugs tested, beginning with dimenhydrinate in 1951,3 mepyramine and promethazine in 1953,4 and meclizine in 1955.5 Subsequently, five other antihistamines were tested - buclizine (1958),6 doxylamine in Debendox/Bendectin (1959, 1971, 1977),7-9 trimethobenzamide (1961),10 thiethylperazine (1964),11 and hydroxyzine (1971).12 A meta-analysis of these trials concluded that antihistamines cause a beneficial reduction in nausea, but that they also cause some sleepiness.2 No trials using an antihistamine have been published since 1977, apart from the N-of-1 trial of promethazine in patients with hyperemesis gravidarum.13
Pyridoxine was used in combination with an antihistamine in seven trials between 1955 and 1977, and these included three trials of Debendox/Bendectin. The two trials in which pyridoxine was used alone appeared only in 1991 and 1995. In both trials nausea was improved, but the trial using 25 mg three times a day significantly reduced vomiting, whereas the one using 10 mg three times a day did not.14,15
The first trial of acupressure for NVP was published in 1988. By 1996, five further trials had appeared, as well as one testing another method of afferent sensory stimulation. All but one of the RCTs showed clear evidence of benefit comparable to that achieved by the antihistamines.2, 23
The early trials of antihistamines, the only ones we have, were simple and direct. They do however leave us with many unanswered questions. Only one drug, meclizine, was tested at more than one dose level. No trial used a variable dose regimen in which the dose could be adjusted according to the response. As a result we do not know whether any of the doses are optimal, except that if they caused troublesome sleepiness they were too high. It therefore seems desirable to do trials with the drugs that have survived, primarily doxylamine and meclizine, to find the lowest effective doses and the highest useful doses. Another question that remains unanswered is whether it is necessary to give the drugs twice or even three times a day. It would be desirable to have information on the pharmacokinetics, especially during the first trimester of pregnancy. If sleepiness is a problem, then it may make sense to give a dose at bedtime, but nausea is rarely troublesome at night. A third unexamined question concerns the duration of treatment. Most of the trials were very short - the longest lasted 28 days. If a drug is effective initially, does it remain effective for the two or three months when it is needed.
We also do not know what is the best way to use pyridoxine, in terms of either dose, frequency or duration of treatment. The evidence that its antinauseant effect is additive or supra-additive with that of doxylamine or meclizine is not clear, and a three-arm trial comparing doxylamine plus pyridoxine with pyridoxine alone and with doxylamine alone is being planned. It would also help if we knew how pyridoxine works.
The attraction of acupressure is that it is non-invasive and cheap, and it is very unlikely to have any adverse effects. It may therefore be a good first choice, even if it is only moderately effective. Only women for whom it does not help adequately need be offered medication. Perhaps trials of antihistamines and pyridoxine should be confined to such women.