Human Reproduction Vol.16, No.12 pp. 2705–2709, 2002
Moderate alcohol consumption and waiting time to pregnancy Mette Juhl1,5, Anne-Marie Nyboe Andersen2, Morten Grønbæk3 and Jørn Olsen4 1Danish
Epidemiology Science Centre at the Department of Epidemiology Research, Statens Serum Institut, 5, Artillerivej, DK-2300 Copenhagen S, 2Department of Social Medicine, Department of Public Health, University of Copenhagen, 3, Blegdamsvej, DK-2200 Copenhagen N, 3Danish Epidemiology Science Centre at the Institute of Preventive Medicine, Copenhagen Hospital Corporation, Kommunehospitalet, DK-1399 Copenhagen K and 4Danish Epidemiology Science Centre at the Department of Epidemiology and Social Medicine, 6, Vennelyst Boulevard, DK-8000 Aarhus C, Denmark 5To
whom correspondence should be addressed. E-mail:
[email protected]
BACKGROUND: Recent research indicates that even a moderate consumption of alcohol in women trying to become pregnant is associated with longer waiting time to pregnancy. The findings, though, are based upon few observations. METHODS: Self-reported data on alcohol intake and waiting time to pregnancy (0–2, 3–5, 6–12 and >12 months) was used for 39 612 pregnant women, recruited to the Danish National Birth Cohort within the first 24 weeks of pregnancy from 1997 to 2000. Main outcome measures were odds ratios (OR) for a prolonged waiting time to pregnancy according to alcohol intake. RESULTS: In nulliparous women neither moderate nor high alcohol intake was related with longer waiting time to pregnancy compared with a low intake. In parous women, a modest association was seen only among those with an intake of >14 drinks per week (subfecundity OR 1.3; 95% confidence interval 1.0–1.7). Women who reported no alcohol intake had a slightly longer waiting time (subfecundity OR 1.2; 95% confidence interval 1.1–1.3) than women with a moderate intake of alcohol. CONCLUSIONS: Our findings do not corroborate recent results suggesting a marked reduction in fecundity associated with a moderate intake of alcohol. Key words: alcohol/birth cohort/fecundity/time to pregnancy
Introduction Various studies have addressed the possible association between alcohol intake and fecundity, defined as the ability for a sexually active couple not using contraception to become pregnant within a given time period. The study by Olsen et al. observing women treated for infertility did not find any negative impact on fecundity with regard to moderate consumption of alcohol (Olsen et al., 1983) in accordance with other studies utilizing different designs (Joesoef, 1993; Florack et al., 1994; Zaadstra et al., 1994; Curtis, 1997). A large European multicentre study found that women with a preconceptional alcohol intake ⬎14 drinks per week waited longer for a pregnancy than women with no alcohol intake (Olsen et al., 1997). Associations between a moderate alcohol consumption and infertility have also been reported by others (Hakim et al., 1998). In one study an effect was seen only in women who had infertility related to ovulatory factors or endometriosis (Grodstein et al., 1994). In a recent follow-up study a ⬎50% reduction in fecundity was found in women with any alcohol intake compared with non-drinkers (Jensen et al., 1998). These results, if true, call for substantial modifications of our alcohol policy for women trying to become pregnant. The results, though, were based upon a small study with highly selected participants. © European Society of Human Reproduction and Embryology
This study aims to examine the association between female consumption of alcohol and waiting time to pregnancy, with particular focus on moderate levels of alcohol intake. The size of the study enables us to examine whether a potential association is modified by body mass index (BMI), smoking and parity.
Materials and methods Study population The study was carried out within the Danish National Birth Cohort, a nationwide study of pregnant women and their offspring. The pregnant women received written information about the cohort study at the first antenatal visit to the general practitioner in gestational week 6–10 and were included when an informed consent form was registered. Inclusion criteria to the cohort, besides being pregnant were: (i) that they lived in Denmark; (ii) that they intended to carry their pregnancy to term and (iii) that they spoke Danish well enough to participate in four telephone interviews during pregnancy and early motherhood. The first interview was scheduled to take place between weeks 12–16 of gestation. About half of all general practitioners in Denmark participated in the study and about 60% of all pregnant women invited chose to participate. We estimate that about 35% of all pregnant women in Denmark participated in the study.
2705
M.Juhl et al.
0.5–2 drinks per week, 2.5–14 drinks per week and ⬎14 drinks per week respectively. Most women reported drinking 0.5–2 drinks per week, and this group was therefore used as the reference exposure. Since no information about smoking habits before pregnancy was available, we used the question whether they had smoked at any time in the first trimester to categorize participants as either smokers or non-smokers. BMI was calculated as weight in kg/height in m2, on the basis of the women’s report on height and weight before pregnancy.
The study included 39 612 women, recruited from October 1, 1997, to March 15, 2000, who had given their first pregnancy telephone interview and who were still pregnant at that time. Measurement of exposure and potential confounders The participants were asked about their drinking habits before pregnancy, specified for each type of beverage; beer, wine and spirits. The question was: ‘How many ordinary beers did you drink per week before you became pregnant?’ and the same question was asked for ‘glasses of wine’ and ‘glasses of spirits’. One bottle of beer contains 11.6 g of alcohol, and 12 g of alcohol is an approximate average for one serving of wine or spirits in Denmark. In the analysis we added each type of beverage to one variable of total alcohol consumption per week. If some alcohol intake was reported, but less than one unit per week, this was coded as half a unit per week. Low, moderate and high alcohol intake was defined according to the following levels;
Measurement of outcome The outcome measure was waiting time to pregnancy based upon the following question: ‘How long a time did you try to become pregnant, before you succeeded?’ followed by fixed answering categories: 0–2 months, 3–5 months, 6–12 months and ⬎12 months. The women were also asked whether the pregnancy was planned, partly planned or not planned. If the pregnancy was unplanned they were not asked about waiting time to pregnancy. Subfecundity is, in this study, defined at two different levels; namely having a waiting to pregnancy of ⬎5 and ⬎12 months.
Table I. Characteristics of the participants. Pregnancy planners only. n ⫽ 29 844 Variables
Value
n
%
Waiting time to pregnancy, months
0–2 3–5 6–12 ⬎12 missing 0 0.5–2 2.5–7 7.5–14 ⬎14 missing ⬍25 25–29 30–34 35⫹ 0 1 2⫹ missing Yes No missing ⬍18.5 18.5–30 ⬎30 missing
14271 6235 4651 4607 80 3679 12429 10910 2368 316 142 2290 11568 11809 4177 13113 12383 4330 18 7239 22598 7 1292 25783 2286 483
48 21 16 15 0 12 42 37 8 1 0 8 39 40 14 44 41 15 0 24 76 0 4 86 8 2
Alcohol, units per week
Age, years
Parity
Smoking (ever in this pregnancy) BMI (kg/m2)
Statistical analysis The association between alcohol intake and waiting time to pregnancy was estimated at two levels for waiting time to pregnancy (⬎5 and ⬎12 months) by means of standard logistic regression (Olsen et al., 1998). Since results were similar at the two cut-off levels, we combined the results by using the method of Kalbfleisch and Prentice (Kalbfleisch and Prentice, 1980). We used a logistic model to estimate the conditional probability of not being pregnant within each waiting time interval given that the woman did not become pregnant in the previous intervals. Alcohol intake, potential confounders such as age, parity, smoking, thinness (BMI ⬍18 kg/m2) and obesity (BMI ⬎30 kg/m2), and a categorical variable to indicate the waiting time interval were included in the model as explanatory variables. Analyses were done in SPSS 9.0. All effect measures are given in subfecundity odds ratios (OR) where a high OR indicates a longer waiting time.
Results Out of the 39 612 recruited pregnancies, 29 933 pregnancies were planned (76%). We excluded 89 women because of endometriosis, ovarian cancer or cervical cancer, which left us with 29 844 cases. Almost half of the women had obtained their pregnancy
BMI ⫽ body mass index.
Table II. The distribution of pregnancy planners according to waiting time to pregnancy and alcohol consumption Average number of drinks per week
Waiting time to pregnancy in months 0–2
0 0.5–2 2.5–7 7.5–14 ⬎14 Total
3–5
Total
n
%
n
%
n
%
n
%
n
%
1750 5963 5264 1106 117 14200
(48) (48) (48) (47) (37)
684 2677 2292 480 71 6204
(19) (22) (21) (20) (22)
573 2007 1640 357 57 4634
(16) (16) (15) (15) (18)
659 1755 1687 416 71 4588
(18) (14) (16) (18) (22)
3666 12402 10883 2359 316 29626
100 100 100 100 100 100
n ⫽ 29 844 and not 29 626 due to missing values in the time-to-pregnancy or alcohol variables.
2706
⬎12
6–12
Alcohol and pregnancy waiting time
within the first 2 months, but 15% waited more than a year to become pregnant. Most women (79%) reported an alcohol intake of 0.5–7 drinks per week (Table I). Twelve percent reported no intake at all and 1% reported a high intake. Women in the highest alcohol group reported on average 18 drinks per week, and 54 women reported an intake of more than 21 drinks per week. Twenty-two percent of women with a high alcohol intake waited ⬎12 months to become pregnant and 14% in the low drinking group (Table II). Almost half of the women became pregnant within the first 3 months of trying, except those with an alcohol intake of ⬎14 drinks per week (37%). Women who had a high alcohol intake were older, more often nulliparous and more often smokers (Table III). The lowest proportion of smokers was found among women with an alcohol intake of 0.5–7 drinks per week. BMI was not associated with alcohol intake. Tables IV and V show the association between alcohol intake and waiting time to pregnancy, firstly by using two often used cut-off points (Table IV), and secondly by using the entire distribution of waiting time in the analysis (Table V). After adjustment for potential confounders the longer waiting time to pregnancy seen in the highest alcohol group disappeared. Age and parity were the adjustment variables that most influenced the crude OR. Women who reported no alcohol intake had longer waiting times than women who reported some alcohol intake. Including women with partly planned pregnancies in the analysis did not alter these estimates. Removing women from the analyses who ever had any kind of pelvic disease did not change the estimates. Neither did
Table III. Distribution of potential confounders according to alcohol consumption Variables
Age, means % nullipara % smokers BMI (kg/m2), means
Alcohol consumption, units per week 0
0.5–2
2.5–7
7.5–14
⬎14
29 36 31 24
30 40 23 24
31 48 21 23
31 57 33 23
32 68 42 23
BMI ⫽ body mass index.
removal of 2338 women with a history of fertility treatment. When data in Table V were analysed for nulliparous and parous women separately none of the OR between alcohol consumption and subfecundity were statistically significant. The highest association was found in parous women with an alcohol intake ⬎14 drinks per week (subfecundity OR 1.3; 95% confidence interval 1.0–1.7) (data not shown). In nulliparous women no association was seen between high alcohol intake and waiting time to pregnancy (Table VI), but smoking, age and obesity were associated with longer waiting times. Non-drinking obese women had the longest waiting time. No significant interaction between alcohol intake and the stratification variables was found.
Discussion Moderate consumption of alcohol was not associated with longer waiting times to pregnancy. In women who reported an intake of ⬎14 drinks per week the elevated risk of a prolonged waiting time disappeared after control for confounders in the nulliparous group. The National Board of Health in Denmark recommends an intake of ⬍14 drinks per week for women. Our results do not provide arguments against this recommendation. The most consistent finding was a longer waiting time to pregnancy in women who reported no alcohol intake. We found no support for the findings of two recent studies suggesting that low levels of alcohol were associated with reduced fecundity (Hakim et al., 1998; Jensen et al., 1998). All epidemiological studies on alcohol intake, including our study, rely on self-reported data, which are expected to be under-reported. If this under-reporting is of similar magnitude at all levels of exposure, our results would indicate no effect at even higher levels of intake. If, on the other hand, heavy drinkers were more likely to deny their intake, waiting times could be longer for the reported non-drinkers, as we found. We expect this problem to be small since very few, if any, alcohol abusers were recruited to the study. We assume to have recruited about 60% of those invited. This response rate could cause selection bias if the decision to participate is associated with both alcohol intake and waiting time to pregnancy. We believe this is unlikely, since studying determinants of subfecundity was not specified as one of the
Table IV. Odds ratios for a waiting time to pregnancy ⬎5 and ⬎12 months according to weekly alcohol intake Drinks per week
Waiting time to pregnancy ⬎5 months
0 0.5–2 2.5–7 7.5–14 ⬎14
⬎12 months
Crude OR
95% CI
Adjusted OR
95% CI
Crude OR
95% CI
Adjusted OR
95% CI
1.16 1 1.01 1.12 1.56
(1.07–1.26) – (0.96–1.07) (1.02–1.23) (1.25–1.96)
1.24 1 0.90 0.84 0.99
(1.14–1.34) – (0.85–0.95) (0.76–0.93) (0.78–1.26)
1.33 1 1.11 1.30 1.76
(1.21–1.47) – (1.04–1.20) (1.16–1.46) (1.34–2.30)
1.48 1 0.94 0.86 0.92
(1.33–1.64) – (0.87–1.01) (0.76–0.98) (0.70–1.23)
Logistic regression. Regression analysis included age (4), parity (2), smoking (2), thinness [body mass index (BMI) ⬍18.5] and obesity (BMI ⬎30). OR ⫽ odds ratio; CI ⫽ confidence interval.
2707
M.Juhl et al.
aims of the cohort. Furthermore, we studied an alcohol effect at intake, which is considered acceptable in Denmark. This study includes only women who actually became pregnant and all effect measures are conditional upon having obtained a pregnancy that survived at least the first 12 weeks of gestation. Only data on women who planned their pregnancies are presented in this paper. If women with a high alcohol intake practice pregnancy planning less often we would expect risk estimates to be biased towards high values, which was not seen. Furthermore, including women with partly planned pregnancies did not have much influence on any of the effect measures. Bias caused by past reproductive experience is an often neglected problem (Olsen, 1999). If women with previous long waiting times unrelated to alcohol intake reduce their drinking habits when they try to become pregnant again, women with a low alcohol intake would have long waiting times to pregnancy. We found the opposite. Even when we restricted the analyses to nulliparous women, who have less reproductive experience to modify their behaviour, no association was found between alcohol intake and waiting time to pregnancy. Studying determinants of subfecundity requires comparable
Table V. Odds ratios (OR) for an increasing waiting time to pregnancy (per increment of the categories stated in Table I) according to weekly alcohol intake Drinks per week
Crude OR
95% CI
Adjusted OR
95% CI
0 0.5–2 2.5–7 7.5–14 ⬎14
1.13 1 1.03 1.12 1.44
(1.07–1.19) – (0.99–1.07) (1.05–1.20) (1.23–1.70)
1.18 1 0.93 0.88 1.00
(1.12–1.25) – (0.90–0.97) (0.82–0.94) (0.85–1.18)
Discrete time survival analysis. Regression analysis included age (4), parity (2), smoking (2), thinness [body mass index (BMI) ⬍18.5] and obesity (BMI ⬎30) and a time-variable to indicate the time interval.
persistence in pursuing a pregnancy attempt among the compared groups (Basso et al., 2000). We expect this problem to be of minor importance after correction for age and parity. We found smoking and BMI to be associated with subfecundity, which is consistent with previous findings (Olsen et al., 1983; Joffe and Li, 1985; Joesoef, 1993; Zaadstra et al., 1994; Curtis, 1997; Bolumar et al., 2000). Throughout our analysis non-drinkers had a consistently higher risk of subfecundity than women with some alcohol consumption. We do not expect alcohol to improve fecundity but a moderate intake may correlate with a higher frequency of intercourse, which may explain the longer waiting times in women who reported no intake. Data on frequency of intercourse were not available. Misclassification on the proper time of exposure may partly explain our results (Weinberg et al., 1994). We asked for average alcohol intake before pregnancy without specifying the time period, but we assume that women with shorter waiting times more often report alcohol use closer to starting time of pregnancy planning than women who waited longer to become pregnant. If women with longer waiting times reduced their alcohol intake to no intake during the planning period, believing that this would increase their chances of becoming pregnant, we would expect the high risks we see for non-drinkers. If women with past fecundity problems totally abstain from drinking alcohol when again trying to become pregnant, this could also explain our findings of longer waiting times among non-drinkers. We do not believe, however, that these methodological problems are important at present, since alcohol is not known to depress female fecundity in general in Denmark. The study showed shorter waiting times for those with a low intake of alcohol compared with non-drinkers. Smaller amounts of alcohol may have a positive impact on the female reproductive system, perhaps by providing some stress control (Negro-Vilar, 1993; Wasser et al., 1993; Sanders and Bruce, 1997; Hjollund et al., 1999).
Table VI. Odds ratios (OR) for an increasing waiting time to pregnancy (per increment of the categories stated in Table I) according to alcohol intake, stratified by age, parity, smoking, body mass index (BMI) and alcohol Values
Drinks per week 0 OR
Age, years:
Parity: Smoking: BMI:
⬍25 25–29 30–34 ⬎34 Nulliparous women Parous women Non-smokers Smokers ⬍18.5 18.5–30 ⬎30
0.85 1.28 1.93 2.83 1.30 0.58 1.16 1.65 1.34 1.21 2.10
⬎14
0.5–14 95% CI (0.74–0.98) (1.18–1.39) (1.76–2.11) (2.41–3.31) (1.20–1.42) (0.54–0.62) (1.09–1.24) (1.51–1.81) (1.10–1.64) (1.14–1.28) (1.83–2.41)
OR 0.81 1 1.55 2.42 1 0.49 1 1.19 1.07 1 1.54
95% CI (0.74–0.87) (1.48–1.62) (2.28–2.57) – (0.47–0.51) (1.14–1.24) (0.98–1.18) – (1.43–1.65)
OR 0.66 0.79 1.63 3.39 0.95 0.62 1.12 1.15 0.82 1.08 1.41
All adjusted for alcohol* 95% CI (0.29–1.51) (0.58–1.08) (1.25–2.12) (2.48–4.62) (0.78–1.17) (0.47–0.83) (0.90–1.39) (0.89–1.48) (0.41–1.66) (0.90–1.28) (0.74–2.70)
OR
95% CI
0.77 1 1.56 2.46 1 0.48 1 1.22 0.93 1 1.46
(0.72–0.83) – (1.50–1.63) (2.33–2.60) – (0.46–0.50) – (1.17–1.27) (0.86–1.02) – (1.31–1.61)
Four separate discrete time survival analyses. All four models included all other variables in the analysis besides the one being examined; parity (2), age (4), smoking (2), thinness [body mass index (BMI) ⬍18.5], obesity (BMI ⬎30) and a time-variable to indicate the time interval. *These values indicate the estimated risk for each age- parity- smoking- and BMI-group. CI ⫽ confidence interval.
2708
Alcohol and pregnancy waiting time
In conclusion our findings suggest that moderate alcohol intake is not strongly associated with subfecundity. Acknowledgements We thank Per Kragh Andersen at the Danish Epidemiology Science Centre for valuable comments on methodology in this study. This work was supported by grants from the Danish National Research Foundation and from the Danish National Board of Health.
References Basso, O., Juul, S. and Olsen, J. (2000) Time to pregnancy as a correlate of fecundity: differential persistence in trying to become pregnant as a source of bias. Int. J. Epidemiol., 29, 856–861. Bolumar, F., Olsen, J., Rebagliato, M. et al. (2000) Body Mass Index and Delayed Conception: A European Multicenter Study on Infertility and Subfecundity. Am. J. Epidemiol., 151, 1072–1079. Curtis, K.M. (1997) Effects of cigarette smoking, caffeine consumption, and alcohol intake on fecundability. Am. J. Epidemiol., 146, 32–41. Florack, E.I., Zielhuis, G.A. and Rolland, R. (1994) Cigarette smoking, alcohol consumption, and caffeine intake and fecundability. Prev. Med., 23, 175–180. Grodstein, F., Goldman, M.B. and Cramer, D.W. (1994) Infertility in women and moderate alcohol use [see comments]. Am. J.Public Health, 84, 1429–1432. Hakim, R.B., Gray, R.H. and Zacur, H. (1998) Alcohol and caffeine consumption and decreased fertility [published erratum appears in Fertil. Steril., 1999, 71, 974]. Fertil. Steril., 70, 632–637.
Hjollund, N.H., Jensen, T.K., Bonde, J.P. et al. (1999) Distress and reduced fertility: a follow-up study of first-pregnancy planners. Fertil. Steril., 72, 47–53. Jensen, T.K., Hjollund, N.H., Henriksen, T.B. et al. (1998) Does moderate alcohol consumption affect fertility? Follow up study among couples planning first pregnancy [see comments]. Br. Med. J., 317, 505–510. Joesoef, M.R. (1993) Fertility and use of cigarette, alcohol, marijuanen, and cocaine. Ann. Epidemiol, 3, 592–594. Joffe, M. and Li, Z. (1985) Male and Female Factors in Fertility. Am. J. Epidemiol., 140, 921–929. Kalbfleisch, J.D. and Prentice, R.L. (1980) Discrete Failure Time Models, an Alternative Discrete Model. In The Statistical Analysis of Failure Time Data. John Wiley and Sons, pp. 37–38. Negro-Vilar, A. (1993) Stress and other environmental factors affecting fertility in men and women: overview. Environ. Health Perspect, 101, 59–64. Olsen, J. (1999) Design options and sources of bias in time-to-pregnancy studies. Scand J. Work Environ Health, 25, 5–7. Olsen, J., Rachootin, P., Schiøtt, A.V. et al. (1983) Tobacco Use, Alcohol Consumption and Infertility. Int. J. Epidemiol., 12, 179–184. Olsen, J., Bolumar, F., Bisanti, L. et al. (1997) Does Moderate Alcohol Intake Reduce Fecundability? Alcohol Clin. Exp. Res, 21, 206–212. Olsen, J., Juul, S., and Basso, O. (1998) Measuring time to pregnancy. Hum. Reprod., 13, 1751–1756. Sanders, K.A. and Bruce, N.W. (1997) A prospective study of psychosocial stress and fertility in women. Hum. Reprod., 12, 2324–2329. Wasser, S.K., Sewall, G. and Soules, M.R. (1993) Psychosocial stress as a cause of infertility. Fertil. Steril., 59, 685–689. Weinberg, C.R., Baird, D.D. and Wilcox, A.J. (1994) Sources of bias in studies of time to pregnancy. Statistics in Medicine, 13, 671–681. Zaadstra, B.M., Looman, C.W., te Velde, E.R. et al. (1994) Moderate drinking: no impact on female fecundity. Fertil. Steril., 62, 948–954. Received on May 4, 2001; accepted on September 6, 2001
2709
