In this investigation, we have examined the relationship between lifestyle factors/occupational exposures and pregnancy outcomes resulting from treatments for unexplained infertility in a large, prospective multicenter trial. Supplemental Table 5
provides a summary of our findings. Given the high prevalence of exposure to these factors in modern society, it is imperative to have a better understanding of the relationship between these factors and outcomes in order to better counsel women regarding lifestyle modifications that may improve the chances of conception while undergoing treatment. Of the lifestyle factors and exposures evaluated in this investigation, only coffee, tea, or alcohol use was significantly associated with pregnancy and live-birth outcomes. Specifically, past users of coffee, tea, or alcohol had significantly higher chances of conception and live-birth compared to never and current users. Other factors that have been related to impaired natural fertility in previous investigations such as smoking, high BMI, illicit drug use, and exposure to environmental toxins (25
) were not significantly associated with the outcomes of fertility treatments. These findings were consistent in both the bivariate and the logistic regression analyses. Any relationship between illicit drug use and pregnancy outcomes would have been difficult to ascertain in this investigation, as the variables related to illicit drug use (“women marijuana trying” and “women cocaine trying”) only captured any use of marijuana and cocaine rather than specified current or past use, or the use of any other substance. Moreover, because both of these drugs are illegal, actual use may be underreported. With regard to age, we found that the live birth rates were lower in the 30–40 year old women as compared to women in the 20–29 year-old age group (13.9% vs. 18.3%, p=0.167, ). One explanation for the lack of significant difference in pregnancy or live birth rates among different age groups is that this is a preselected group of women with ‘unexplained’ infertility. It is possible that the younger women have subclinical reduced ovarian reserve or some other unmeasured variable that makes them similar to the older women resulting in infertility; thus, the younger women behave similarly to the older women with respect to pregnancy and delivery (29
The effect of alcohol use on natural fertility in women has not been clearly established. In a prospective study of 7,393 women, Eggert and colleagues identified an increased risk of infertility (relative risk = 1.6; 95% CI: 1.1–2.3) in high consumers of alcohol (≥ 2 drinks/day) relative to moderate consumers (30
). Conversely, other investigations have not identified a significant relationship between alcohol use in women and fecundability (3
), but have shown an increase in first trimester pregnancy loss (8
). Within the context of infertility treatments such as COS-IUI, we are unaware of prior studies investigating the relationship between alcohol use and outcomes; however, consumption of at least four drinks per week was associated with a decrease in the IVF live-birth rate in one investigation (19
As with alcohol use, we are unaware of previous investigations evaluating the impact of coffee or tea drinking on outcomes following infertility treatments such as COS-IUI. Given that both coffee and tea contain significant amounts of caffeine, it seems likely that the relevant exposure is caffeine. We identified no significant relationship between soda drinking and either pregnancy or live-birth rates; however, soda contains significantly less caffeine than either coffee or tea. High caffeine use (> 5–7 cups/day) has been associated with decreased natural fertility in some investigations (3
), an effect which may be dose-related (32
). However, others have failed to identify a significant relationship (4
). It has been shown in some studies that moderate to heavy caffeine use increased the rate of pregnancy loss (33
). One may hypothesize that the higher pregnancy and live birth rates observed in the “past” users of coffee or tea may be due to higher pregnancy loss rates in the “current” users. However, this is not supported by our data. In fact, the pregnancy loss rate in the “past” users was the highest among the three groups (“past” at 6.7%, “current” at 5.4%, and “never” at 2.7%).
Given that previous investigations have generally shown a negative impact of female smoking and obesity on the time to spontaneous conception (25
) and outcomes following IVF treatment (17
), we were surprised that no significant relationship was identified between these variables and either pregnancy or live-birth rates. Consistent with our findings, Farhi and colleagues did not identify significant differences in pregnancy rate between smokers and nonsmokers (16.3% and 15.8%, respectively) in a retrospective review of 885 couples undergoing COS-IUI, although a higher dose of gonadotropins was required in smokers (35
). Similarly, a retrospective review of the outcomes of 333 ovulatory women undergoing COS-IUI identified no significant difference in cycle fecundity among different BMI groups ranging from underweight to obese (36
). It is possible that the observation of impaired natural fertility in obese women is partially related to ovulatory dysfunction.
Our observation of increased pregnancy and live-birth rates in past users of coffee, tea or alcohol relative to current and never users requires further evaluation and validation. Although we did not have a prior knowledge for this finding nor did we have an external dataset to validate it, there are reasons to believe its validity. If these exposures had long-lasting negative effects on conception, one would expect to observe a similar negative impact on outcomes in both current and past users compared to never users. Alternatively, if exposure to these factors resulted in only short-term effects, then one would expect past and never users to have similar pregnancy rates, both of which would be superior to current users. However, neither of these outcomes was observed. It is possible that women who discontinue drinking coffee, tea or alcohol in anticipation of attempting conception possess characteristics that are associated with positive health outcomes, such as an internal locus of control (i.e. a belief that their ability to conceive can be self-managed and controlled), as it is generally considered that consumption of caffeine containing beverages and alcohol are not healthy habits prior to conception. Perhaps women who have recently discontinued the use of coffee, tea, or alcohol in an attempt to improve their chances of achieving a pregnancy are also making other lifestyle changes that were not measured or not fully adjusted for in this investigation. Since the discontinuation of coffee or tea or alcohol increase both the pregnancy and live birth rate, the possible undetected positive lifestyle changes along with the discontinuation of these habits may have beneficial effects on both pregnancy and live birth (37
). One of the factors is smoking status. Smoking has been shown to increase or decrease the effect of coffee or tea drinking on pregnancy outcome (23
). The lack of effect of coffee or tea drinking on pregnancy outcome among patients who never smoked in this study suggests that smoking and coffee or tea drinking have an interacting relationship with conception and live birth rates. Another possibility is that never users of coffee, tea, or alcohol are simply different in their ability to conceive at baseline than are current and past users. In other words, if exposure to these factors causes a temporary and reversible negative impact on fecundability, then one would expect past users to experience higher pregnancy rates than current users. Never users that would have been susceptible to the negative effects of coffee, tea, or alcohol could have already achieved a pregnancy prior to enrollment. Thus, the remaining “never” users have different underlying etiologies for their infertility. Previous studies have investigated the relationship between social class status and pregnancy outcome, and lower level of social class may have a lower pregnancy rate and higher rate of adverse birth outcome (38
). The lack of significant association between coffee, tea or alcoholic beverage drinking and male or female education level (data not shown), one of the main social class factors, suggests that baseline social class status may not be a potential explanation for the difference in pregnancy and live birth rate observed in this study. Regardless of the mechanism, the magnitude of the effects observed in this investigation (adjusted odds ratio 4.0 for past users of coffee or tea; 1.9 for past users of alcohol) is considerable. Therefore, further prospective investigations are needed to confirm and extend the finding of improved pregnancy and live-birth rates following the recent discontinuation of alcohol, coffee and tea.
Limitations of the current investigation should be noted. First, all data regarding lifestyle factors were self-reported, and it is possible that subjects may have underreported exposures. Particularly this may be true with regards to smoking and alcohol use behaviors. Second, the association between greater pregnancy and live-birth rates noted in past users of coffee, tea, and alcohol compared to current and never users does not necessarily imply a causal relationship between these factors and outcomes. The data do not contain information to infer this causal relationship.
In summary, in a large, prospective multicenter trial investigating the effectiveness of treatments for unexplained infertility, we identified past use of alcohol, coffee and tea as being significantly associated with increased odds of conception and live-birth. Other lifestyle factors and exposures, including smoking, BMI, ever use of illicit drug, and exposure to environmental toxins were not significantly related to outcomes. Additional prospective investigations are necessary to confirm the finding of improved fecundity following the recent discontinuation of alcohol, coffee and tea.