Introduction

More than one-half of women with uncomplicated pregnancies suffer from nausea and vomiting during pregnancy (NVP). The condition usually occurs during weeks 7–12 of pregnancy and rarely needs intervention. NVP occurs at an increased rate in young women, in multiparous women, twin pregnancies, and if the fetus is a girl [1]. The cause of the condition is not clear, but proposed etiologic mechanisms are increased levels of gonadotropic and thyroid hormones, dysregulation of the gastric rhythm and autonomous nervous system as well as psychological factors [2]. In addition, it has recently been proposed that placental development regulates NVP [3].

NVP which substantially reduces the pregnant woman’s life quality may require intervention with antiemetics until it subsides. The possible teratogenic potential of the chosen drug must then be considered since NVP may occur during the period of organogenesis.

It has been proposed that antihistamine drugs as a group lack teratogenic effects [4–6]. However, there is very little clinical information on the antiemetic dopamine blocking agents with respect to teratogenicity. Neither metoclopramide [7] nor ondansetron [8] has been found to have a teratogenic effect; however, the latter study had very low power to detect any effect.

The present study was undertaken in order to describe antiemetic use during pregnancy in Sweden and to study delivery outcome after such therapy.

Materials and methods

The Swedish Medical Birth Registry

Data were obtained from the Swedish Medical Birth Registry, which collects data on antenatal care, delivery, and the neonatal outcome of nearly all deliveries in Sweden (a few percentages are missing [9]). Liveborn infants and stillbirths (according to the Swedish definition, 28 completed gestational weeks) are included. This registry is based on copies of the original medical records, which are identical all over Sweden. On July 1, 1994, a computerized search of the antenatal care records was initiated for information on drug use. The standard practice in Sweden is that during the first antenatal care visit (usually between weeks 10 and 12), the pregnant woman is interviewed by a midwife. Among the many questions asked is one which refers to drug use since the woman has become pregnant. This information thus basically refers to first trimester drug use. Drugs prescribed during the antenatal care visit are recorded separately from those which refer to first trimester use. Drug names (including over-the-counter drugs and herbal drugs) are recorded in clear text that is later translated to ATC codes which are stored in the register. Drugs without an ATC code are stored in the register with the names in clear text. Possible information (often incomplete) on dosage and the timing of drug use is also kept in clear text. In the present study, all births registered in the Medical Birth Registry from July 1, 1995 through to December 31 2002 were studied. During this time, 29,804 pregnant women with 31,130 infants reported the use of antiemetic drugs from a total of 665,572 pregnant women with 676,198 infants that were registered.

Selection of drugs with antiemetic properties

Antiemetic drug use was studied in the pregnant women cohort derived from the Swedish Medical Birth Registry. These antiemetics included ones used in clinical practice with different pharmacological activities, such as antihistamines (meclozine, cyclizine, promethazine, thiethylperazine, dimenhydrinate), dopamine modulators (metoclopramide, prochlorperazine, dixyrazine) and 5-HT3 receptor antagonist (ondansetron).

Among the drugs studied, two are no longer available on the Swedish market (cyclizine and dixyrazine). Corticosteroids were not included because they have a broad range of clinical application and are rarely used in Sweden as an antiemetic during pregnancy.

Variables studied

The following descriptive variables were studied:

  • Maternal age at delivery

  • Maternal parity (sum of previously born children +1; thus, a woman having her first infant is said to be parity 1).

  • Maternal smoking, as recorded at the interview during the first antenatal visit: unknown, none, <10 cigarettes per day, 10 or more cigarettes per day.

  • Unwanted childlessness, as a measure of subfertility: number of years the couple had tried to have a baby, as reported during the first antenatal visit.

  • Concomitant drug use, as reported at the first antenatal care visit.

  • Country of birth of mother: obtained by linkage with population register (Statistics Sweden).

  • County of residence of the mother at the time of delivery.

  • Educational level of mother as a proxy for social level, available only up to and including births in 2001: education in 2002, obtained by linkage with central register of education (Statistics Sweden).

The following outcome variables were studied:

  • Number of infants in birth

  • Infant sex.

  • Preterm birth (<37 completed weeks) among singleton infants.

  • Low birth weight (<2500g) among singleton infants.

  • Small-for-gestational age (<–2 standard deviations from expected weight [9]).

  • Congenital malformations identified from the Medical Birth Registry, the Registry of Congenital Malformations, and the Hospital Discharge Register [10]

Statistical methods

Women who had used antiemetics were compared with all women who had given birth during the study period, and the infants of mothers of the first group were compared with all infants born. Comparisons were made using the Mantel-Haenszel technique following adjustments, as will be described, and risks were expressed as odds ratios (OR) with 95% confidence intervals (95% CI), estimated using Miettinen’s method. When the expected numbers were low, risk ratios as observed over expected numbers (after making the adjustments as specified) were determined instead, and the 95% CI was determined from exact Poisson distributions. Trends in ORs were studied using weighted linear regression analysis of the log (ORs). Frequencies were compared with chi-square statistics. The 95% CI of sex ratios was based on binomial distributions.

Results

Exposure to antiemetic drugs registered in Sweden

Among all women registered in the Medical Birth Register, 29,807 (4.5%) reported the use of antiemetics. Of these, 23,396 (78%) reported antiemetic use at the first antenatal care visit only, 4,017 (13%) reported no antiemetic use at the first antenatal care visit but had an antiemetic prescribed during later pregnancy, and 2,394 (8%) reported both early use of an antiemetic and a later prescription. Table 1 shows the individual drugs used, sometimes a combination of drugs. Two antiemetic drugs were used by 3,207 women (11% of drug users), three antiemetic drugs by 454 (1.5%) women, and four or more such drugs by 72 (0.2%) women. As seen in Table 1, the most frequently reported antiemetic drug was meclozine (68%). Promethazine alone and in combination with caffeine was the second most frequently used drug, followed by cyclizine, dixyrazine, and metoclopramide in decreasing frequency. Only a few women reported the use of thiethylperazine, dimenhydrinate, prochlorperazine, or ondansetron.

Table 1 Reported use of specific antiemetic drugs and the period of pregnancy during which the drug was taken
Full size table

Maternal characteristics

Table 2 shows maternal age, parity, smoking in early pregnancy, and years of unwanted childlessness among women using antiemetics compared with all women giving birth. For each variable, adjustments were made for year of delivery and for the other three variables. There is a clear-cut trend with maternal age, with the highest drug use among young women. Primiparous women show a lower use than multiparous women, and women who smoked use these drugs less than women who did not. Women who reported unwanted childlessness had a slightly higher use of antiemetics than women who did not, and this was most evident at 1–2 years of involuntary childlessness.

Table 2 Characteristics of women reporting use of antiemetics compared with population
Full size table

We compared the characteristics of the women according to which drug they had used. The only statistically significant heterogeneity between drugs, although small, was for parity 1 and meclozine, where the OR was 0.85 (95% CI: 0.83–0.88), while for all drugs it was 0.79 (95% CI: 0.77–0.81).

Other drug use reported by women using antiemetic drugs

Women using antiemetics were compared with women not using antiemetics after adjustment of year of birth, maternal age, parity, smoking, and involuntary childlessness. The OR for using any other drug was 2.56 (95% CI: 2.50–2.62). Table 3 shows some selected groups of concomitantly used drugs. Increased ORs are seen for drugs used for stomach ulcer and gastric reflux, for multivitamins and folic acid, analgesics, neuroleptics, antidepressants, and cough medicines. Significantly decreased ORs are seen for various other groups of drugs: insulin, oral contraceptives, progesterone and ovarian stimulations, thyroxine, anticonvulsants, and anti-asthmatics.

Table 3 Comparison of women using drugs other than antiemetics concomitantly with antiemetics with use of these other drugs among women
Full size table

Maternal country of birth and educational level

The country of birth of the woman can affect the use of antiemetic drugs. When all women born outside Sweden are compared with Swedish-born women, the former have a higher use of antiemetics. After adjustment for year of birth, maternal age, parity, smoking, and years of involuntary childlessness, the OR is 2.01 (95% CI: 1.95–2.08). When different sub-groups of non-Swedish women were studied, it appears that women born in the other Nordic countries closely resemble the women born in Sweden (OR=0.96, 95% CI: 0.87–1.05).

Table 4 presents the association between maternal educational level as a proxy for social level and the use of antiemetics. There is a clear-cut trend, in which a decreasing use of antiemetics correlates positively with increasing educational level. The group “unknown” is mainly made up of immigrants.

Table 4 Antiemetic use and a woman´s educational level. Information is only available up to and including 2001
Full size table

Characteristics of offspring of women exposed to antiemetic drugs

Sex ratio

The normal sex ratio is 1.06 and little influenced by year of birth, maternal age, parity, smoking, or involuntary childlessness, which is the reason why no adjustment for these variables were made. The sex ratio among singleton infants born of women using antiemetics is 0.92 (95% CI 0.90–0.94): 14,880 boys and 16,186 girls. As the number of antiemetic drugs used increases, the sex ratio decreases: from 0.93 with one drug, 0.89 with two drugs, and 0.83 with three or more drugs. This trend is of marginal statistical significance (p=0.049).

The singleton infant sex ratio after maternal use of various antiemetics varies between 0.60 (diphenhydramine) and 1.17 (ondanseron). This variation is explained by low numbers for some drugs and random variations. A chi-square analysis of heterogeneity between the proportions of males and females following exposure to various drugs gives chi-square=7.8 at 9 df (p=0.55).

Multiple births

The women exposed to antiemetics had 681 twin deliveries and seven deliveries of triplets or more. In the population studied, the odds ratio for twinning when exposed to one or more antiemetics is 1.51 (95% CI: 1.40–1.64), adjusted for year of birth, the age of the mother, and years of involuntary childlessness. The odds ratio for twinning is 1.42 when only one antiemetic was used, 1.79 when two had been used, 2.67 when three had been used, and 2.55 when four or more antiemetics had been used. A test for a linear trend gives z=3.9, p<0.001. An increased risk of twin births was observed for all antiemetic drugs.

Preterm birth, birth weight, and small-for-gestational age in singleton births

Table 5 summarizes the results for preterm birth, low birth weight, and small-for-gestational age according to antiemetic drug use. Moderate but statistically significant decreases in risks were seen for all three variables when all antiemetic drugs were studied together. The exclusion from the analysis of infants of mothers born outside of the Nordic countries barely affected the estimates.

Table 5 Preterm birth, low birth weight, and small-for-gestational age according to antiemetic drug use
Full size table

The rate of preterm births varies significantly according to drug used: from 3.9 to 6.2%. This heterogenity is statistically significant (chi-square=26.5 at 7 df, p<0.001), and some drugs (diphenhydramine, promethazine, and promethazine combinations) show ORs significantly over 1.0.

Similarly, the rate of low-birth-weight infants varies from 0 to 5%. This heterogenity is also statistically significant (chi-square=22.9 at 7 df, p=0.002), with statistically high ORs for promethazine and thiethylperazine.

The rate of small-for-gestational age infants also varies: from 0 to 2.7%, but this variation is not statistically significant (chi-square=10.6 at 7 df, p=0.16).

Congenital malformations

Congenital malformations were studied in all infants, including babies of multiple births. Table 6 summarizes the results of an analysis of all congenital malformations registered in the Medical Birth Registry. The registered malformation rate varies for the different drugs, with 3.6% for cyclizine being the highest. However, this variability may be random (chi-square=4.9 at 6 df, p=0.56).

Table 6 Congenital malformations registered in the Medical Birth Registry among all infants born
Full size table

Many of the registered malformations are mild conditions and have not been registered uniformly; for example, coloboma of the eye, preauricular tags, branchial cysts or fistules, patent ductus arteriosus at preterm birth, single umbilical artery, undescended testicle, unstable hip, and skin malformations (mainly naevus). After these mild conditions are removed from the analysis, 684 infants with malformations remain in the antiemetic-exposed group (2.2%) and 16,994 in the population (2.5%). The adjusted OR remains at 0.90 (95% CI 0.83–0.97).

The use of all sources for congenital malformations (see Materials and methods) gave a better ascertainment and also identified malformations observed after the neonatal period – 4.7% of all infants born have a recorded congenital malformation on this basis. Among infants exposed to antiemetics, the rate is 4.3% and the crude OR is 0.93 (95% CI 0.88–0.98). This material was used in order to identify specific groups of congenital malformations (Table 7). The ORs vary between the different groups, but a chi-square analysis indicates that the heterogeneity has only a marginal statistical significance (chi-square=14.1 at 7 df, p=0.049). Hypospadias, however, shows intrinsically a significantly low OR.

Table 7 Number of infants with specific groups of congenital malformations, identified from all available sources and odds ratios (OR) with 95% confidence intervals (95% CI)
Full size table

Discussion

The data presented here originates from information on antiemetic drug use during pregnancy assembled in the Swedish Medical Birth Register and was recorded before the outcome of the pregnancy was known. The information on drug use, smoking, and years of unwanted childlessness was obtained from interviews performed in early pregnancy and is therefore prospective with respect to pregnancy outcome; consequently, the problem with recall and interviewer bias was eliminated. Additional information, such as maternal education, country of birth, and county of residency, was obtained by record linkage and is therefore also unbiased. This information and other data (year of birth, maternal age, and parity) made it possible for us to adjust for these variables in the analysis. The register is population-based with a participation rate of more than 97%.

The register also has weaknesses. It is unlikely that all drug use is reported or recorded. In this study, about 5% of all women reported the use of antiemetics during pregnancy. In a questionnaire study [11] from a region of Sweden, 15% of the women reported the use of drug treatment of NVP and about 13% reported the use of antihistamines. The lower rate in the present study may not only be due to a lack of reporting/recording but also to the fact that women after the first antenatal care visit may use over-the-counter drugs which will not be registered. However, the incomplete identification of women who used antiemetics will only slightly influence the risk estimates because the relatively few unidentified cases will only marginally dilute statistics of the large control population [12].

Another draw-back of the register is that little information is available on the timing or duration of drug use or the amount taken. Data for women who only used a single tablet is confounded with data for women who used drugs for an extended period. This tends to bias the estimated ORs towards unity.

No information is obtainable for spontaneous or induced abortions. According to present Swedish law, induced abortions cannot be registered with personal identification data, which makes it impossible to study such cases for, among other things, drug use. A number of abortions will have been induced because of congenital malformations detected during prenatal diagnosis and will not be included in the analysis. While this will scarcely affect the risk estimates, it will reduce the power of the study. If the malformation in question is always or nearly always aborted (like anencephaly or bilateral kidney agenesis), obviously a possible teratogenic effect of the studied drugs cannot be identified. If the use of a drug (e.g., anticonvulsant) can affect the probability or degree of a detailed prenatal diagnosis, biased results can be obtained. It is very unlikely that the use of antiemetics will have such an effect.

The antiemetic drugs included in this study were among those documented in the literature for use in treating NVP and they were registered in Sweden between 1995 up to and including 2002: antihistamines (meclozine, cyclizine, promethazine, thiethylperazine, dimenhydrinate), dopamine antagonists (metoclopramide, prochlorperazine and dixyrazine) or 5-HT3 receptor antagonist (ondansetron). Antihistamine use dominated during the first trimester, when NVP is most common, primarily in the form of meclozine, followed by promethazine both singly and in combination with caffeine/ephedrine. Women born outside the Nordic countries used antiemetic drugs more often than Swedish-born women. There was no major difference between women of Swedish origin and women born in other Nordic countries.

Women who reported antiemetic drug use were more likely to be younger, multiparous, and non-smokers. They were also more prone to give birth to a girl and the rate of twin births was higher. These features are in agreement with previous reports on NVP and exposure to antiemetic drugs during pregnancy [1, 2]. It should be stressed that the analyses in the present paper compare women using antiemetics with women who did not use antiemetics – even though the majority of the latter will also have had NVP. Despite this, the differences that were observed indicate that women using antiemetics represent a selected group, most probably because they have more severe NVP than the majority of pregnant women.

Another variable which indicates an increased usage of antiemetics is the presence of unwanted childlessness as an indicator of couple subfertility. In our data, women with a low education used antiemetics more often than women with a higher education. We can only speculate why women with a higher education use antiemetics less frequently. A possible explanation is that women with a relatively higher education are more aware of the risks with drug use during early pregnancy. They may also be more informed about the good prognosis of NVP and therefore avoid pharmacological treatment.

Infants born by women who used antiemetics during pregnancy were characterized by having a better outcome than expected, even when the above-mentioned confounders were taken into consideration. The odds ratios for preterm birth, low birth, weight, small-for-dateness and congenital malformations were significantly low. Similar findings were made in a previous study that was restricted to the use of meclozine [7]. We found no signs that the drugs had a teratogenic effect, but for some of the drugs studied, the number of exposures was low. Furthermore, anti-emetic therapy does not usually begin during the most sensitive part of organ formation, and we were unable to identify women who had used antiemetic drugs very early in their pregnancy (for example, due to motion sickness).

The better-than-expected infant outcome after maternal use of antiemetics is hardly a direct drug effect. More likely, the presence of NVP indicates good placental function. Previous studies have verified this for different pregnancy outcomes, such as for miscarriages and preterm births [11, 13, 14]. Placental hormones are thought to play a role in the etiology of NVP [2, 3], and a well-functioning placenta may both increase the probability of NVP and of a good pregnancy outcome. Smoking in early pregnancy was associated with a lower usage of antiemetics. A simple explanation would be that women who experience NVP reduce their smoking, but one study [11] showed that the percentage of women who stopped smoking or reduced smoking during pregnancy was about the same, irrespective of the presence or absence of NVP, and that smoking before pregnancy had a similar “protective” effect on NVP. A possible explanation is that maternal smoking negatively influences placental development and actually reduces NVP rate or severeness. For instance, endothelin has been proposed to influence placental blood flow and nicotine plays a role in modulating endothelin and its receptors, lending some support to a pharmacological effect of smoking on placental development [15–17].

A similar explanation could exist for the negative association we found between the use of antiemetics and certain concomitant drug use: insulin, progesterone and ovarian stimulation, thyroxine, anticonvulsants, and anti-asthmatics. This negative association indicates a reduced NVP when the pregnant woman has a chronic disease. In such cases, the majority of the pregnant women had a typical sub-optimum pregnancy outcome.

Certain drugs were used more often by women using antiemetics than they were by other women. One complicating factor is that an under-reporting of drug usage occurs, and if some women are more prone than others to report drug use, a false association – specifically between commonly used drugs – will be obtained. This may explain the finding for multivitamins and folic acid, analgesics, and cough medicines. Other associations may be due to the usage of the concomitant drug in NVP, notably drugs used for stomach ulcers and gastric reflux . The increased use of neuroleptic and antidepressant drugs may indicate a high rate of NVP or a more severe NVP, or a reduced tolerance to NVP, among women with psychiatric problems. The importance of psychological factors for NVP has been widely discussed [2].

In summary, pregnant women exposed to the antiemetics selected for this study, in particular to antihistamines, showed an overall better neonatal outcome with respect to several variables, including prevalence at birth of malformations.