We prospectively examined the relation between the intake of alcoholic and caffeinated beverages and risk of infertility due to ovulation disorders. Only the consumption of 2 or more soft drinks per day was associated with this infertility after accounting for multiple potential confounders. The association between high consumption of soft drinks and ovulatory disorder infertility did not appear to depend on the caffeine content of these beverages. Further, there was no evidence that an association between caffeine, alcohol, or sodas and ovulatory infertility was restricted to subgroups of women defined by age, parity, or smoking status, as has been the case in some previous studies.6,12,18,26
We found a positive association between alcohol intake of approximately 1 drink or more per day and ovulatory disorder infertility in age-adjusted analyses. This association disappeared after accounting for parity and other factors. Alcohol drinking at lower levels was unrelated to ovulatory infertility in all analyses. To date, 16 studies have examined the association between alcohol intake and fertility in women,1,9,10,14,15,17,19–28
many of them reporting deleterious effects of alcohol. However, only 5 of these studies have been prospective.9,17,22,26,27
Among the prospective studies, 2 reported decreased fecundability,17,22
1 reported a higher risk of infertility,27
1 found a positive relation between alcohol and infertility among women older than 30 years of age but also a similarly strong inverse association among younger women,26
and 1 found no association between alcohol intake and fecundability.9
Only 1 previous study has examined alcohol intake in relation to specific infertility diagnoses and found alcohol to be associated with a higher risk of infertility due to ovulatory factor or endometriosis.10
Our results are not in agreement with this report. When our results are considered with those of all other prospective studies, there is an evenly divided field between null and positive studies. Clearly, more large prospective studies are necessary to clarify the role of moderate alcohol consumption on human fertility.
We did not observe a positive association between intake of caffeine, coffee, tea, or decaffeinated coffee and risk of ovulatory infertility. Indeed, supplementary analyses
suggested a lower, rather than increased, risk of ovulation-related infertility with very high levels of coffee consumption, although these findings were based on only 7 cases and could be the result of chance. Ten of the 17 studies conducted to date have found a relation between coffee or caffeine intake and decreased fertility.2,3,5–8,12,13,18,20
Most of these studies are retrospective.3,5–8,12,13,20
When only the 5 prospective studies are considered, 2 reported decreased fecundability2,22
and 3 showed either no association16,17
or slightly improved fertility.9
Leviton and Cowan50
have described how retrospective studies and studies of lower methodologic quality are more likely to report a relation between caffeine intake and impaired reproduction (including decreased fertility) than prospective studies. Furthermore, a positive association between prepregnancy coffee drinking and a higher rate of dizygotic twin pregnancies has been described,51
suggesting that caffeine may, if anything, stimulate ovulation rather than suppress it. Our results are in agreement with the majority of prospective epidemiologic studies conducted to date that suggest no association between caffeine intake and reduced fertility.
Soft drinks were the only beverages positively related to ovulatory infertility. Our analyses suggested that neither caffeine nor fructose was responsible for this association. Extreme comparisons of caffeine and coffee intake (coffee being the most important source of caffeine in this population) suggested no association or an inverse association with ovulatory disorder infertility. Also, including caffeine intake in the same multivariate model with caffeinated sodas made the association between caffeinated sodas and ovulatory disorder infertility stronger, rather than weaker as would be expected if caffeine explained part of the association. In addition, noncaffeinated soft drinks showed a similar relation to this condition, albeit slightly weaker. Similarly, intakes of both sugared and diet soft drinks were positively related to ovulatory infertility and fructose intake was unrelated to this condition. The lack of specificity for a particular type of beverage suggests that this could be a chance finding. Alternatively, this association could be due to soft drink components common to all types of soft drinks or a dietary pattern in which soft drinks (regardless of type) are preferably consumed, which could not be accounted for in this analysis. In support of the later hypotheses, others have previously reported that intakes of caffeinated, decaffeinated, sugared, and diet soft drinks have similar relations to the incidence of impaired fasting glucose and metabolic syndrome;52
a very relevant finding given the role of insulin sensitivity on ovulation and the pathogenesis of polycystic ovary syndrome. In addition, the notion that the relation between soft drinks and infertility is independent of caffeine is further supported by previous studies. For example, Wilcox and colleagues53
reported that an increase of 1 caffeinated soft drink per day was associated with a 50% lower chance of conception each month after accounting for coffee consumption and other relevant variables, and concluded that this association exceeded what would be expected, based on data on coffee and fecundability from the same study, if caffeine were responsible for the association. Also, Hatch and Bracken8
reported a stronger association between soft drinks and delayed conception than between coffee and this outcome at lower levels of caffeine intake from sodas. Both studies were conducted in the United States, where coffee is the main contributor to intake of caffeine. Further research is needed to confirm these findings and to identify what components of soft drinks explain their association with decreased fertility.
The limitations of our study need to be considered. First, in assembling the cohort we assumed that the pregnancies included in the analysis were planned. Although cases and women reporting infertility due to other causes were clearly attempting to conceive, some pregnancy noncases may have conceived accidentally, potentially distorting the results. However, we studied a group who, based on their educational and demographic characteristics, was likely to plan a pregnancy54
and further restricted the study to married women, whose pregnancies are also more likely to be intentional than those of unmarried women.54
In addition, we included in the noncase group women diagnosed with infertility from other causes, making it less likely that pregnancy intention affected our findings. As a related issue, we did not collect data on male partner exposures. However, male exposures are unlikely to have a large effect on their partner’s fertility25,55
and thus their influence on our results is probably minimal. Moreover, any effect of male exposures on female fertility can be accounted for in part by adjusting for the same factor in women, as we did for smoking and other lifestyle characteristics because spouses often share environmental exposures and lifestyle practices56,57
and are likely to change a health-related behavior when their spouse changes it.58,59
Second, the possibility that our findings may be due in part to unmeasured confounding cannot be completely ruled out. Nevertheless, we considered the potential confounding effects of many variables associated with caffeine or alcohol intake and ovulatory infertility, as well as of recognized risk factors for infertility, and found that only statistical adjustment for parity had important impact on the results. Third, our dietary data referred to a period up to 4 years before the reported pregnancy or failed attempt, potentially misclassifying the true exposure to these beverages around the periconceptional period and attenuating the results toward the null. Last, because we examined only the association of caffeinated and alcoholic beverage intake with ovulation-related infertility, we cannot rule out the possibility that these drinks impair fertility by affecting other aspects of reproductive function.
In summary, our results do not support the notion that consuming alcohol or caffeine in moderation affects ovulatory function to the point of increasing the frequency of infertility due to ovulation disorders. Consistent with previous reports,8,53
these data also suggest that soft drinks may be a risk factor for infertility and that this relation is independent of their caffeine content. Because a randomized trial of moderate caffeine or alcohol consumption in relation to fertility may be judged as unethical, further large prospective observational studies, preferably in populations with different patterns of alcohol and caffeine use, are necessary to determine whether moderate consumption of these substances affects fertility in humans.