If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
To determine whether multivitamin supplements modify the relationship between alcohol consumption during pregnancy and the risk of miscarriage.
We used data from a population-based cohort study of pregnant women (n = 1061; response rate = 39%). Participants were asked about their alcohol consumption and vitamin intake during pregnancy.
Among multivitamin nonusers, women who drank alcohol during their pregnancy were more likely to have a miscarriage compared with women who abstained (adjusted hazard ratio, 1.67; 95% confidence interval, 1.04–2.69). However, among multivitamin users, there was no difference in the risk of miscarriage between alcohol consumers and abstainers. Results suggest the volume of alcohol as well as the timing of multivitamin supplementation may also be important.
Our findings suggest that a woman of childbearing years might decrease her risk of miscarriage associated with alcohol intake by taking multivitamin supplements. However, our findings should be interpreted with caution and future research replicating these findings is necessary.
; however, the mechanism through which alcohol causes damage to the fetus resulting in miscarriage remains unknown. It is possible that the toxicity of alcohol occurs partly through its disruption of maternal and/or fetal nutrition. Research on the deleterious effects of alcohol on nutrition has shown that alcohol consumption can lead to various micronutrient deficiencies (such as vitamins C and E, zinc, iron, and potassium).
On the basis of this evidence, multivitamin supplements may potentially provide protection.
We are not aware of any studies that have evaluated whether multivitamin supplements modify the relationship between alcohol consumption during pregnancy and miscarriage. To examine this potential relationship, we used data from a prospective cohort study to evaluate whether: (1) taking multivitamin supplements modifies the effect of any alcohol consumption during pregnancy and miscarriage, (2) taking multivitamin supplements modifies the relationship between the average number of drinks a week during pregnancy (4+ drinks per week, <4 drinks per week, and no alcohol intake) and miscarriage, and (3) the timing of multivitamin supplement exposure modifies the relationship between any alcohol consumption during pregnancy and miscarriage.
Materials and Methods
The current secondary data analysis was conducted using data from a population-based prospective cohort study conducted in the Kaiser Permanente Medical Care Program (KPMCP) in Northern California. All KPMCP women members who lived in San Francisco County and parts of San Mateo County who had a positive pregnancy test at 1 of 2 San Francisco KPMCP facilities from October 1996 to October 1998 were identified through the computerized laboratory database as potential eligible subjects. A woman's second pregnancy, if any, during the study period was not eligible for the study. All women submitting a urine sample for a pregnancy test were given a flyer describing the purpose and procedures of the study and a postage-paid and self-addressed return refusal card. Women with positive pregnancy tests who did not return a refusal postcard were contacted by a trained female interviewer to determine their eligibility for the study. English-speaking women who intended to carry their pregnancy to term and whose gestational age at the pregnancy test was less than or equal to 10 complete weeks were eligible for the study. The median gestational age at study entry was 40 days. The original study recruited women to evaluate the relationship between electromagnetic field exposure during pregnancy and miscarriage.
A total of 2729 pregnant women were identified as eligible for the original study (Figure). Of the eligible women, 1380 (50.6%) initially agreed to participate in the study, of whom 1061 (39%) completed an in-person interview. In-person interviews were conducted by a trained interviewer to obtain detailed information about alcohol consumption, previous pregnancy history, demographic characteristics, and other possible confounders. The current analysis included 1061 women who completed an in-person interview.
Institutional review board approval for this study was obtained from Kaiser Permanente's Human Subjects Committee.
Miscarriage was defined as a fetal loss occurring before 20 complete weeks of gestation and was based on the date of the woman's last menstrual period (LMP). It was ascertained for all participants through 1 of the following methods: electronic KPMCP databases, reviewing medical charts, and telephoning women whose outcomes could not be identified through either of the previous methods.
Participants were asked if they drank any alcoholic beverages “since becoming pregnant or since LMP.” Women who drank any alcohol were then asked the number of beers (1 beer was equal to 12 ounces), the number of glasses of wine or champagne (1 glass was equal to 4 ounces), and the number of mixed drinks (1 drink was equivalent to 1 ounce of hard liquor) consumed since becoming pregnant. A variable approximating the average number of alcoholic drinks consumed per week was calculated by adding the total number of alcoholic beverages consumed since the beginning of pregnancy, and dividing by the gestational age in weeks at the time of the interview. Based on the distribution of the average number of drinks per week, our sample size and previous research, alcohol consumption was further categorized into 3 mutually exclusive categories: (1) no alcohol intake (n = 626), (2) drank <4 drinks per week (n = 403), and (3) drank 4+ drinks per week (n = 32).
Participants were asked if they had taken any vitamins, including multiple, prenatal, and single vitamins or any other type of supplements “since becoming pregnant or their LMP.” Women were asked about the type and brand of supplement for each of the supplements they reported taking. In addition, they were asked when they started taking the vitamin supplement.
All women who reported taking either a multivitamin or prenatal vitamin during their pregnancy were considered multivitamin users (n = 730). Multivitamins and prenatal vitamins are similar in the content of vitamins and other micronutrients. We also considered the timing of the multivitamin exposure window further categorizing women into 3 categories: (1) periconceptional user (began taking multivitamins before pregnancy and continued in pregnancy, n = 475), (2) prenatal user (began taking multivitamins during pregnancy, n = 252), and (3) nonuser (n = 331). Three women who reported taking multivitamins did not report when they began taking them.
Participants were asked about various behaviors during their pregnancy. These included whether they used any illicit drugs, engaged in regular exercise (physical activity for ≥30 minutes at least 3 times a week), smoked at all, or drank any caffeine during pregnancy. Prepregnancy body mass index was calculated (kg/m2) and categorized into 2 categories: (1) underweight/normal ≤24.9, and (2) overweight/obese 25.0+.
Other demographic characteristics considered were race (white, black, Hispanic, Asian, and Native American), education (≤high school, some college/technical training, graduated from college, and attended graduate school), and marital status (married, have a partner but not married, and single). Women were also asked about their income, but 59 (6%) did not respond and therefore were categorized separately (<$35K, $35–$59K, $60K+, and nonresponders). Age was dichotomized as less than or equal to 35 and 36 or older, because pregnancies occurring among women aged 36 or older are considered high risk. Finally, we considered whether this pregnancy was intended and previous miscarriage history.
STATA version 9 (Stata Corp, College Station, TX) was used for all analyses. Pearson χ2 tests were conducted to test differences between categorical variables.
Cox proportional hazards analysis was used to examine whether multivitamin supplements modify the relationship between alcohol exposure and miscarriage at any given gestational age while controlling for other possible confounders. This analysis considered all covariates significantly associated with both alcohol and miscarriage. The period for which a woman was considered at risk began at the gestational age when she had a positive pregnancy test (study entry) and continued until she had a miscarriage, ectopic pregnancy, or induced abortion (3.6%), or was censored at 20 weeks' gestation (80%) because, by definition, a miscarriage occurs through 20 weeks of gestation. The time variable used in the proportional hazards model (gestational age in days) was left-truncated at their positive pregnancy test to reflect participants' actual contribution of their person-time.
Three separate sets of Cox proportional hazards models were conducted. The first set of analyses evaluated any multivitamin use (any use during pregnancy vs no use), any alcohol consumption (any alcohol intake vs no intake), and miscarriage. The second set of analyses evaluated any multivitamin use, the average number of drinks a week (drank 4+ drinks per week, drank <4 drinks per week, and no intake), and miscarriage. The final set of analyses evaluated the timing of multivitamin exposure (nonuse, prenatal, and periconceptional), any alcohol consumption during pregnancy and miscarriage.
Each set of analyses began with a likelihood ratio test comparing the Cox proportional hazards model that included alcohol use, multivitamin use, all covariates, and an interaction term for alcohol use and multivitamin use, to the same model with the exclusion of the interaction term. The interaction term differed for each set of analyses as described previously. On a P value of less than .10 for the likelihood ratio test, additional Cox proportional hazards models were conducted stratified by (1) any multivitamin use (first and second sets of analyses) or (2) timing of multivitamin exposure (third set of analyses) to assess the relationship between alcohol consumption during pregnancy and miscarriage. As tests for interaction generally have less power to test for statistical significance,
before any data analysis, it was determined that a corresponding P value of less than .10 for the likelihood ratio tests would be the cutoff for conducting the stratified analyses.
Sixteen percent (n = 172) of the women in our study had a miscarriage. Forty-one percent of the women reported drinking any alcohol, with 3% drinking 4 or more drinks a week and 38% drinking less than 4 drinks a week (Table 1). The mean number of drinks per week among women who drank alcohol was 1.24 (standard deviation [SD], 2.5). Most women (69%; n = 730) took multivitamin supplements; 475 (65%) began taking them during pregnancy and 252 (35%) began taking them before pregnancy.
TABLE 1Demographic characteristics, pregnancy behaviors, previous miscarriage history, and alcohol consumption by miscarriage and alcohol intake
We found few possible covariates significantly associated with miscarriage in the bivariate analyses (Table 1). Although not statistically significant at P < .05, the data suggest that women who were older than 35 years of age and women who drank any caffeinated beverage during pregnancy were more likely to have a miscarriage. These (maternal age and caffeine intake) and 2 other significant covariates, marital status and unintended pregnancy, were all also significantly associated with alcohol consumption and were included as possible confounders in all multivariate analyses.
In the first set of multivariable models, the results for the likelihood ratio test (P = .07) conducted for the analyses evaluating any multivitamin use and any alcohol consumption indicate multivitamin use modified the relationship between alcohol consumption and miscarriage. Therefore, we conducted additional analyses stratified by multivitamin use. Among women who did not take multivitamins, women who drank alcohol were 1.67 times more likely to have a miscarriage compared with women who abstained, after adjusting for possible confounders (adjusted hazard ratio [aHR], 1.67; 95% confidence interval [CI], 1.04–2.69) (Table 2). However, among multivitamin users, there was no difference in the risk of miscarriage between women who drank alcohol and women who abstained (aHR, 0.98; 95% CI, 0.65–1.48).
TABLE 2Any alcohol consumption during pregnancy and the hazard ratio for miscarriage by multivitamin status
Multivitamin use status Multivitamin nonuser (n = 331)
The P value for the likelihood ratio test assessing the interaction between any multivitamin use and the average number of drinks per week was 0.098. Table 3 shows the HRs for the relationship among the 3 categories of the average number of drinks per week during pregnancy and miscarriage stratified by multivitamin use. Among multivitamin nonusers, women who drank at least 4 drinks a week were more than 6 times as likely to have a miscarriage compared with women who abstained (aHR, 6.30; 95% CI, 2.32–17.05) and those who drank fewer than 4 drinks a week had a nonsignificant increased risk of miscarriage, compared with women who abstained (aHR, 1.51; 95% CI, 0.92–2.47) (Table 3). However, among multivitamin users, the risk of miscarriage associated with alcohol intake during pregnancy significantly diminished for women who drank 4 or more drinks a week (aHR, 1.87; 95% CI, 0.79–4.45) and there was no increased risk of miscarriage for women who drank fewer than 4 drinks a week (aHR, 0.92; 95% CI, 0.60–1.41).
TABLE 3Frequency of alcohol consumption during pregnancy and the hazard ratio for miscarriage stratified by multivitamin exposure
Multivitamin use status Multivitamin nonuser (n = 331)
The P value for the likelihood ratio test assessing the interaction between the timing (none, prenatal, and periconceptional) of multivitamin exposure and any alcohol consumption was .008; therefore we stratified by the timing of multivitamin use. Table 4 displays the HRs for the relationship between alcohol consumption and miscarriage disaggregated by the timing of multivitamin exposure. A trend emerged with the timing of multivitamin use in which the timing of multivitamin use appeared to have a differential but diminished effect on alcohol consumption and the risk of miscarriage, compared with multivitamin nonusers.
TABLE 4Any alcohol consumption during pregnancy and hazard ratio for miscarriage stratified by timing of multivitamin exposure
Multivitamin use status Multivitamin nonuser (n = 331)
Additional analyses were conducted to evaluate the relationship between the timing of multivitamin exposure and the average number of drinks per week and miscarriage. Results followed a similar pattern, regardless of the amount of alcohol consumed (results not shown).
To ensure the robustness of our results, additional analyses were conducted that included other demographic characteristic and pregnancy behavior variables (regular exercise, white vs other races, illicit drug use, and education) significantly associated with alcohol intake and multivitamin use, but not considered traditional confounders. Similar trends and results emerged for all analyses.
Our results indicate multivitamin status is an important modifier in the relationship between pregnancy drinking and miscarriage that should be considered when evaluating the relationship between alcohol consumption and miscarriage. We found the risk of miscarriage was greatest for women who drank alcohol and reported no multivitamin supplementation. Commencement of multivitamin supplementation before pregnancy appeared to have the greatest impact on the relationship between alcohol consumption during pregnancy and miscarriage. Yet, our findings also suggest that multivitamin supplementation started during pregnancy may provide benefits as well.
These findings should be interpreted in light of certain limitations. First, we note that miscarriage was based on a clinical diagnosis and therefore miscarriages that occurred before pregnancy awareness were not included. Second, 105 (61%) of the women were interviewed after they had a miscarriage. However, the proportion of women who reported alcohol use during pregnancy was the same for women interviewed either before or after miscarriage (44% for both), with no indication of differential reporting because of the timing of interview in relation to miscarriage.
Third, the generalizeability of the findings may be limited because of the study's low response rate. We cannot rule out with certainty that participation was not associated with factors related to both alcohol consumption and multivitamin use; however, the requirements of participation in the original study resulted in many refusals. We were able to obtain the percentage of miscarriage among nonparticipants which was similar to that of participants (17.2% vs 16.4%, respectively) somewhat reducing this concern. In addition, other articles published from these data have reported findings consistent with previous research.
Nevertheless, low participation could potentially impact the interpretation of the findings.
The accuracy of self-reported alcohol consumption during pregnancy is a concern in all studies evaluating alcohol-related reproductive and birth outcomes. Comparisons of self-reported pregnancy drinking with the use of vessels (varying sizes of beer, wine, and other glasses) to visualize the amount of alcohol consumed have found an underestimation of alcohol consumption resulting from self-report.
Further, studies comparing pregnancy drinking as measured by antenatal self-report and retrospective self-report have documented an underreporting for antenatal assessment, especially among heavy drinkers.
Finally, we were unable to establish which micronutrients are most important in modifying the relationship between prenatal alcohol exposure and miscarriage. Although information was collected on the brand and type of vitamin supplements consumed, the small number of women taking any particular individual vitamin or micronutrient supplement made analyses of these individual micronutrients unfeasible. Of the 776 women who reported taking vitamins, 78% reported taking only multivitamins, and only 5% reported not taking multivitamins.
Previous research has found an increased risk of miscarriage when alcohol consumption is measured in discrete levels.
To an extent, our findings support previous research regarding this relationship. For example, our findings suggest an increased risk of miscarriage for women who drank 4 or more drinks a week compared with abstainers, among both multivitamin users and nonusers. However, the magnitude of the risk of miscarriage associated with alcohol consumption was smaller for multivitamin users. We note that our results are based on a small sample that may affect the precision of our results. Contrary to the literature that has not found an increased risk of miscarriage at lower levels of alcohol intake, our findings suggest an increased risk of miscarriage for women who drank less than 4 drinks a week, but only among multivitamin nonusers.
Research has shown that optimal nutrition at the start of pregnancy is important for healthy pregnancy outcomes. For example, studies evaluating multivitamin supplementation suggest the timing of supplementation may be important in affecting preterm birth and small-for-gestational-age births.
The results from our study appear to support previous research suggesting the timing of optimal nutrition during pregnancy is important. Our findings suggest that starting multivitamin supplementation during pregnancy may reduce the risk of miscarriage for women who drank alcohol during pregnancy compared with abstainers. Further, although based on a small sample, a nonsignificant, but protective trend was found for periconceptional users. Although further research is needed to replicate this finding, this association is consistent with previous reports that perinatal vitamin use is beneficial to a healthy pregnancy.
The epidemiologic literature lacks research assessing the potential role of nutrition as a modifier in the relationship between alcohol consumption and adverse pregnancy outcomes. However, nutrition has been found to be an important modifier between alcohol consumption and several chronic diseases such as cancer.
Thus, it is plausible that multivitamins may mitigate the risk of adverse pregnancy outcomes, including miscarriage, associated with alcohol use during pregnancy.
Our study has important implications for women of childbearing age in the United States. Nearly half of the pregnancies in the United States are unintended and surveys have shown that binge drinking is prevalent among women of childbearing age.
Our findings suggest that a woman of childbearing years might decrease her risk of miscarriage by taking multivitamin supplements as part of her daily nutrition. However, we acknowledge that our findings should be interpreted with caution and future research replicating these findings is necessary.
We thank Drs Barbara Abrams and Sylvia Guendelman, for their thoughtful comments and suggestions on earlier versions of this article. In addition, we thank Dr Mark Hudes for his statistical consult.
Pregnancy-related substance use in the United States during 1996-1998.
This article was a chapter of a dissertation (L.A.A.) supported by NIAAA training Grant T32-AA007240-29 .
Cite this article as: Ammon Avalos L, Kaskutas LA, Block G, et al. Do multivitamin supplements modify the relationship between prenatal alcohol intake and miscarriage? Am J Obstet Gynecol 2009;201:563.e1-9.