Review
A review of the epidemiologic evidence concerning the reproductive health effects of caffeine consumption: A 2000–2009 update

https://doi.org/10.1016/j.fct.2010.06.019Get rights and content

Abstract

This review of human studies of caffeine and reproductive health published between January 2000 and December 2009 serves to update the comprehensive review published by Leviton and Cowan (2002). The adverse reproductive outcomes addressed in this review include: (1) measures of subfecundity; (2) spontaneous abortion; (3) fetal death; (4) preterm birth; (5) congenital malformations; and (6) fetal growth restriction. Methodologic challenges and considerations relevant to investigations of each reproductive endpoint are summarized, followed by a brief critical review of each study. The evidence for an effect of caffeine on reproductive health and fetal development is limited by the inability to rule out plausible alternative explanations for the observed associations, namely confounding by pregnancy symptoms and smoking, and by exposure measurement error. Because of these limitations, the weight of evidence does not support a positive relationship between caffeine consumption and adverse reproductive or perinatal outcomes.

Introduction

This review of the literature on consumption of caffeine-containing products and reproductive health is an update of the comprehensive report previously published by Leviton and Cowan (2002). As such, this review is restricted to human studies of caffeine and reproductive health published in English between January 2000 and December 2009. The search strategy consisted of a Pubmed search using the keywords caffeine, coffee or paraxanthine in combination with the following terms: pregnancy, reproduction, fetal development, miscarriage, spontaneous abortion, pregnancy loss, fetal death, stillbirth, congenital malformations, birth defects, fetal growth, growth restriction, growth retardation, small for gestational age, low birthweight, low birth weight, IUGR, preterm, fertility, fecundity, time to pregnancy, sperm, semen, twins, twinning, multiple gestation, or multiple pregnancy. The references cited in all original studies and review papers identified were also examined to ensure completeness.

The reproductive outcomes addressed in this review are organized into six categories: (1) measures of subfecundity; (2) spontaneous abortion; (3) fetal death; (4) preterm birth; (5) congenital malformations; and (6) fetal growth. For each topic, we begin by summarizing study design considerations relevant to investigations of the specific reproductive endpoint. In keeping with the unique format of the previous summary, we then provide a detailed critical review of each study, identifying strengths as well as methodological limitations that may influence results and restrict inferences that can be drawn from individual studies. Individual studies are discussed by topic in chronological order. Table 1 lists the publications reviewed in this report by reproductive outcome evaluated. Two post-1999 publications (Cnattingius et al., 2000, Grosso et al., 2001) were reviewed previously in Leviton and Cowan (2002) and, thus, are not summarized in this review.

We begin with a discussion of general methodological concerns that should be considered when reviewing studies of consumption of caffeine-containing products and reproductive health.

Total caffeine consumption is difficult to measure accurately. Caffeine exposure can occur from various sources including beverages (coffee, tea, soft drinks), chocolate, and some medications. Furthermore, the caffeine content of individual beverage servings varies considerably by method of preparation, product brand, and cup size (Bracken et al., 2002). The most common justification for assessing caffeine exposure from coffee alone or from coffee and tea is that coffee is the predominant source of caffeine exposure and fewer women report high doses from other beverages, foods or pharmaceuticals. Regardless of the population proportions consuming large quantities from other sources, consumption from all sources is important for accurate classification of total caffeine intake for individuals, since low to moderate intake from multiple sources can result in high cumulative caffeine exposure. Furthermore, because coffee also contains many chemicals other than caffeine, it is difficult to disentangle the potential effects of caffeine from those that may be attributed to other compounds.

Relying on coffee intake alone would likely result in underestimations of total caffeine exposure. The influence of this measurement error on observed associations would depend on whether use of caffeine from other sources was more, less or equally common among women experiencing the outcome under investigation. Exposure measurement error is commonly assumed to be similar among those with and without each reproductive disorder, resulting in an underestimation of any coffee/caffeine relationship with the reproduction adversity. This underestimation (bias toward the null) tends to be predictable only when the exposure is dichotomous and the association is independent of other errors (Greenland and Gustafson, 2006). Much of the caffeine literature assesses more than two categories of exposure and thus, misclassification of caffeine intake could produce a bias either toward or away from the null, depending on the nature of the errors.

Other critical aspects of caffeine exposure assessment include the importance of measuring exposures during the relevant exposure time window and the need to capture changing intake patterns throughout pregnancy. Caffeine consumption tends to decrease during the early weeks of pregnancy, coinciding with increasing pregnancy symptoms and aversions (Gadsby et al., 1993, Cnattingius et al., 2000, Lawson et al., 2004). Retrospective reports of caffeine intake collected at a single time point as the average number of daily servings across a large time span such as the first trimester or entire pregnancy (typically converted to caffeine in mg/day) will not accurately characterize true exposure fluctuations. This type of measurement error is especially relevant when the critical window of exposure for selected outcomes occurs during the time interval when consumption patterns are changing in early gestation.

Although a few studies designed for the purpose of investigating caffeine exposure have implemented detailed improvements in exposure assessment, variations in caffeine exposure by source, portion size, brewing method, metabolism and fluctuations over the course of pregnancy continue to result in exposure misclassification. Furthermore, the comparison of findings across studies is difficult due to the use of different categories of caffeine intake and different reference groups.

Self-reported caffeine exposure is not only imprecise, it also fails to account for variability in rates of degradation. The measurement of caffeine metabolites in biologic fluids provides better information about biologic dose in part because it reflects individual differences in caffeine metabolism (Klebanoff et al., 1998c), but such methods are not without limitations. As the major metabolite of caffeine, serum paraxanthine has a short half-life, ranging from 2 to 5 h in early pregnancy to 10 h in late pregnancy (Aldridge et al., 1981). Thus, serum paraxanthine concentrations reflect recent exposures within the day immediately preceding specimen collection. Moreover, studies that incorporate caffeine biomarkers are typically limited to specimens collected at a single time point. Therefore, biomarker concentrations would accurately represent previous caffeine intake patterns only when consumption remains relatively constant over time. The problem with this assumption is that caffeine consumption is known to decrease throughout the first months of pregnancy (Lawson et al., 2004). Thus, biomarker concentrations may also be susceptible to exposure misclassification when a single measurement is intended to represent long-term or usual patterns of exposure during pregnancy.

The rate at which caffeine is cleared from the body may influence biologic dose and exposure interval. Caffeine metabolites might be more important than caffeine in producing a biologic effect. Caffeine clearance rates differ between individuals and are affected by pregnancy and genetics as well as environmental factors such as cigarette smoking, drugs and diet (Aldridge et al., 1981, Kalow and Tange, 1991, Carrillo and Benitez, 2000, Lampe et al., 2000). Variations in caffeine metabolism between individuals are mostly attributed to differences in cytochrome P450 1A2 (CYP1A2) enzyme activity (Arnaud, 1994). Assessments of CYP1A2 phenotypes and genotypes have been used to characterize study populations according to high or low metabolic activity. A comprehensive review of the metabolic considerations for caffeine exposure assessment during pregnancy was published by Grosso and Bracken (2005).

Pregnancy symptoms, including aversions to taste and smells, nausea, and vomiting are more common in healthy pregnancies that result in live births and occur less frequently among women whose pregnancies end in spontaneous abortions (Weigel and Weigel, 1989, Weigel et al., 2006). This relationship is attributed to a stronger pregnancy signal produced by higher concentrations of pregnancy hormones in viable pregnancies (Stein and Susser, 1991, Lawson et al., 2002). Caffeine consumption has been shown to decrease with increasing pregnancy signal symptoms during the early weeks of pregnancy (Lawson et al., 2004, Cnattingius et al., 2000). For example, Lawson et al. (2004) reported that mean onset of nausea, vomiting and appetite loss occurred between 5 and 6 weeks from the last menstrual period, accompanied by a 59% decrease in caffeine intake from coffee between weeks 4 and 6. Thus, women experiencing viable pregnancies are more likely to reduce their caffeine intake in response to the pregnancy signal than women who go on to have a spontaneous abortion. As a result, reduced caffeine consumption may be a consequence of pregnancy viability rather than increased consumption causing any reproductive adversity. “Reverse causation” is the term used to describe such errors in causal inference.

Control for confounding by pregnancy signal symptoms is critical for caffeine studies of spontaneous abortion and fetal death, and may have importance for studies of other adverse pregnancy outcomes. This task, however, is complicated by the difficulty of measuring pregnancy signal symptoms. Studies that include only dichotomous indicators for nausea and vomiting fail to capture the severity, frequency and duration of the symptoms. Furthermore, aversions to specific foods or beverages, which may be equally or more relevant to decreased caffeine consumption, are rarely assessed. Studies have revealed that women who decrease their coffee consumption during early pregnancy commonly attributed these changes to an acknowledged aversion to the taste, smell or thought of coffee (Lawson et al., 2004); whereas, nausea reported as number of hours per week was not statistically associated with reduced coffee consumption (Lawson et al., 2002). Thus, separate assessment of coffee aversion has been recommended for studies of caffeine and pregnancy outcomes (Lawson et al., 2002, Lawson and LeMasters, 2006). Efforts to disentangle this complex relationship will require improved, prospective measurement of all relevant dimensions of the pregnancy signal.

Most investigators recognize the importance of controlling for confounding by smoking when evaluating the reproductive effects of caffeine. Smoking and caffeine use are strongly associated, as heavier smokers tend to consume more caffeine than others (Schreiber et al., 1988, Zavela et al., 1990). Furthermore, smoking is considered a risk factor for many adverse reproductive outcomes such as infertility, spontaneous abortion, fetal growth restriction, stillbirth, and preterm birth (Cnattingius, 2004). Controlling for self-reported smoking status, however, may not provide adequate control for confounding when smoking is measured inaccurately. The stigma of smoking during pregnancy may lead to inaccurate reporting of smoking status or under-reporting of the amount smoked per day (Ford et al., 1997, Klebanoff et al., 1998b, Lindqvist et al., 2002). Morrison (1984) described the potential for residual confounding to occur when the more socially acceptable behavior of caffeine consumption is more accurately reported than the less acceptable behavior of smoking tobacco. Because the two behaviors are highly correlated, the more accurately reported caffeine consumption conveys information about tobacco consumption. Furthermore, many investigators control only for smoking status (yes/no) without consideration of amount smoked. Thus, incomplete control for the effects of smoking may explain observed associations commonly attributed to caffeine use. Some authors have attempted to improve upon these measurements by incorporating cotinine measurements as biomarkers of nicotine exposure. It is important, however, to acknowledge that these biochemical markers reflect recent tobacco exposures and may or may not accurately control for actual smoking patterns during the relevant window of exposure.

Measures of association (e.g., odds ratios and relative risks) and confidence intervals are frequently reported to two decimal places when, in fact, study precision is rarely sufficient for this degree of detail to be meaningful. Thus all results discussed in this review are rounded to one decimal place.

Section snippets

Subfecundity

Studies assessing the impact of caffeine on fertility potential have evaluated a variety of outcomes including time to pregnancy, infertility, semen quality and selected endpoints of assisted reproductive technologies.

Spontaneous abortion

A good review of studies that have evaluated the relationship between caffeine consumption and spontaneous abortion has been published by Signorello and McLaughlin (2004).

Fetal death

Clinical convention distinguishes fetal death, defined as fetal demise after 20 weeks of gestation, from spontaneous abortion defined as pregnancy loss ⩽20 weeks of gestation. The distinction is typically drawn at the mid-point of pregnancy because it approximates the point of fetal viability, which is generally regarded as occurring close to 23–24 weeks of gestation. Because both outcomes address fetal loss, but at different points along the continuum of pregnancy, many of the methodological

Heterogeneity

Preterm birth includes a heterogeneous group of disorders, suggesting heterogeneous etiologies (Savitz et al., 1991, Savitz et al., 2005, Klebanoff and Shiono, 1995, Klebanoff, 1998a, Pennell et al., 2007, Berhman and Butler, 2007, McElrath et al., 2008, Savitz, 2008a). The so-called “spontaneous preterm delivery” group includes preterm labor, pre-labor premature rupture of membranes, placental abruption, and cervical insufficiency, which are associated with intrauterine inflammation. Medically

Heterogeneity

All malformations are not etiologically identical. Even those studies limited to one organ or related structures (e.g., lip and palate) are likely evaluating heterogeneous entities. Some studies have assessed single malformations (Mongraw-Chaffin et al., 2008, Torfs and Christianson, 2000, Browne et al., 2007, Miller et al., 2009, Slickers et al., 2008) or utilized more refined sub-classifications of observed phenotypes (Browne et al., 2007, Bille et al., 2007, Johansen et al., 2009, Schmidt et

Fetal growth restriction

As a perinatal outcome of interest, fetal growth is considered a marker of healthy intrauterine development and a predictor of postnatal morbidity and mortality (Savitz et al., 2002). Most studies of caffeine and fetal growth have assessed intrauterine growth restriction (IUGR) (also referred to as SGA) defined as birth weight <10th centile for gestational age according to a standard growth curve from a selected reference population appropriate for the infant (by sex and race). Other measures

Subfecundity

Of the nine publications since 2002, one evaluated multiple outcomes associated with fertility treatment, one considered self-reported ovulatory infertility, three addressed time to conception, and most (4) assessed the relationship between caffeine and semen parameters.

The only study to assess the effect of caffeine on endpoints of assisted reproductive technology reported no influence of previous or current caffeine intake on oocyte retrieval, fertilization, embryo transfer or the occurrence

Conflict of Interest

The authors declare that there are no conflicts of interest.

Acknowledgements

This work was supported by the Caffeine Working Group of the North American Branch of the International Life Sciences Institute (ILSI). ILSI North America is a public, non-profit foundation that provides a forum to advance understanding of scientific issues related to the nutritional quality and safety of the food supply by sponsoring research programs, educational seminars and workshops, and publications. ILSI North America receives support primarily from its industry membership. Dr. Peck

References (149)

  • A.H. Karypidis et al.

    Association of cytochrome P450 1B1 polymorphism with first-trimester miscarriage

    Fertil. Steril.

    (2006)
  • M.A. Klebanoff et al.

    Response to Letter by Grosso et al

    Ann. Epidemiol.

    (2004)
  • M.A. Klebanoff et al.

    Serum caffeine and paraxanthine as markers for reported caffeine intake in pregnancy

    Ann. Epidemiol.

    (1998)
  • R. Lacroix et al.

    Nausea and vomiting during pregnancy: A prospective study of its frequency, intensity, and patterns of change

    Am. J. Obstet. Gynecol.

    (2000)
  • C.C. Lawson et al.

    Regarding “Caffeine metabolism, genetics and perinatal outcomes: a review of exposure assessment considerations during pregnancy”

    Ann. Epidemiol.

    (2006)
  • C.C. Lawson et al.

    Changes in caffeine consumption as a signal of pregnancy

    Reprod. Toxicol.

    (2004)
  • A. Leviton et al.

    A review of the literature relating caffeine consumption by women to their risk of reproductive hazards

    Food Chem. Toxicol.

    (2002)
  • D.-K. Li et al.

    Reply to caffeine and miscarriage: case closed?

    Am. J. Obstet. Gynecol.

    (2008)
  • R.L. Naeye

    Do placental weights have clinical significance?

    Hum. Pathol.

    (1987)
  • N. Natsume et al.

    Maternal risk factors in cleft lip and palate, case control study

    Br. J. Oral Maxillofac. Surg.

    (2000)
  • M.A. Okon et al.

    Serum androgen levels in women who have recurrent miscarriages and their correlation with markers of endometrial function

    Fertil. Steril.

    (1998)
  • J. Ørskou et al.

    Maternal characteristics and lifestyle factors and the risk of delivering high birth weight infants

    Obstet. Gynecol.

    (2003)
  • K.L. Adeney et al.

    Coffee consumption and risk of gestational diabetes

    Acta Obstet. Gynecol. Scand.

    (2007)
  • A. Aldridge et al.

    The disposition of caffeine during and after pregnancy

    Semin. Perinatol.

    (1981)
  • M.J. Arnaud

    Metabolism of caffeine and other components of coffee

  • D.D. Baird et al.

    Use of time to pregnancy to study environmental exposures

    Am. J. Epidemiol.

    (1986)
  • O. Balat et al.

    The effect of smoking and caffeine on the fetus and placenta in pregnancy

    Clin. Exp. Obstet. Gynecol.

    (2003)
  • B.H. Bech et al.

    Coffee and fetal death, a cohort study with prospective data

    Am. J. Epidemiol.

    (2005)
  • B.H. Bech et al.

    Stillbirth and slow metabolizers of caffeine, comparison by genotypes

    Int. J. Epidemiol.

    (2006)
  • B.H. Bech et al.

    Effect of reducing caffeine intake on birth weight and length of gestation, randomised controlled trial

    BMJ

    (2007)
  • Berhman, R.E., Butler, A.S. (Eds.), 2007. Preterm birth. Causes, Consequences, and Prevention. The National Academies...
  • C. Bille et al.

    Oral clefts and life style factors – a case-cohort study based on prospective Danish data

    Eur. J. Epidemiol.

    (2007)
  • Boylan, S.M., Cade, J.E., Kirk, S.F.L., Greenwood, D.C., White, K.L.M., Shires, S., Simpson, N.A.B., Wild, C.P., Hay,...
  • M.D. Bracken et al.

    Heterogeneity in assessing self-reports of caffeine exposure, implications for studies of health effects

    Epidemiology

    (2002)
  • M.B. Bracken et al.

    Association of maternal caffeine consumption with decrements in fetal growth

    Am. J. Epidemiol.

    (2003)
  • M.L. Browne et al.

    Maternal caffeine consumption and risk of cardiovascular malformations

    Birth Defects Res. A Clin. Mol. Teratol.

    (2007)
  • Bukowski, R., Smith, G.C., Malone, F.D., Ball, R.H., Nyberg, D.A., Comstock, C.H., Hankins, G.D., Berkowitz, R.L.,...
  • A.R. Burdi et al.

    Morphogenesis of the palate in normal human embryos with special emphasis on the mechanisms involved

    Am. J. Anat.

    (1967)
  • M.E. Campbell et al.

    Biotransformation of caffeine, paraxanthine, theophylline, and theobromine by polycyclic aromatic hydrocarbon-inducible cytochrome(s) in human liver microsomes

    Drug Metab. Dispos.

    (1987)
  • M.E. Campbell et al.

    A urinary metabolite ratio that reflects systemic caffeine clearance

    Clin. Pharmacol. Ther.

    (1987)
  • CARE Study Group, 2008. Maternal caffeine intake during pregnancy and risk of fetal growth restriction, a large...
  • J. Carrillo et al.

    Clinically significant pharmacokinetic interactions between dietary caffeine and medications

    Clin. Pharmacokinet.

    (2000)
  • J.E. Chavarro et al.

    Caffeinated and alcoholic beverage intake in relation to ovulatory disorder infertility

    Epidemiology

    (2009)
  • F. Chiaffarino et al.

    Coffee drinking and risk of preterm birth

    Eur. J. Clin. Nutr.

    (2006)
  • B. Clausson et al.

    Effect of caffeine exposure during pregnancy on birth weight and gestational age

    Am. J. Epidemiol.

    (2002)
  • Cnattingius, S., 2004. The epidemiology of smoking during pregnancy, smoking prevalence, maternal characteristics, and...
  • S. Cnattingius et al.

    Caffeine intake and the risk of first-trimester spontaneous abortion

    N. Engl. J. Med.

    (2000)
  • B.A. Cohn et al.

    Epidemiologic studies of human semen quality, considerations for study design

    Am. J. Epidemiol.

    (2002)
  • S.A. Collier et al.

    Maternal caffeine intake during pregnancy and orofacial clefts

    Birth Defects Res. A Clin. Mol. Teratol.

    (2009)
  • L.G. Dales et al.

    An improper use of statistical significance testing in studying covariables

    Int. J. Epidemiol.

    (1978)
  • Cited by (99)

    • Caffeine: An evaluation of the safety database

      2021, Nutraceuticals: Efficacy, Safety and Toxicity
    • GR/HDAC2/TGFβR1 pathway contributes to prenatal caffeine induced-osteoarthritis susceptibility in male adult offspring rats

      2020, Food and Chemical Toxicology
      Citation Excerpt :

      Caffeine is a methylxanthine alkaloid that is widely present in coffee, tea, soft drinks, food and some drugs (Katheria et al., 2015). Epidemiological studies revealed that caffeine ingestion during pregnancy is a definite cause of intrauterine growth retardation (IUGR) (Boylan et al., 2008; Peck et al., 2010). Our previous studies reported that prenatal caffeine exposure (PCE) could induce skeletal growth retardation in fetal offspring rats with IUGR, then leading to poor-quality articular cartilage in adults (Luo et al., 2015; Tan et al., 2012).

    • Serum metabolic profile characteristics of offspring rats before and after birth caused by prenatal caffeine exposure

      2019, Toxicology
      Citation Excerpt :

      In clinical application, caffeine is gradually replacing theophylline as the drug of the first choice for the prevention and treatment of apnea in preterm infants (Aranda et al., 2010; Picone et al., 2012). A large number of population and laboratory studies have confirmed that caffeine is the definite cause of IUGR (Boylan et al., 2008; Peck et al., 2010), and the offspring induced by prenatal caffeine exposure (PCE) at increased risk of cognitive impairment and obesity after birth (Galera et al., 2016; Li et al., 2015). Our previous studies have confirmed that PCE can cause IUGR in rats, accompanied by developmental toxicity organs such as liver, kidney, and adrenal gland, and the studies proposed systematically that “maternal glucocorticoid overexposure” mediated the neuroendocrine metabolic programming mechanism of diseases susceptibility in IUGR offspring (Ao et al., 2015; Chen et al., 2018; Liu et al., 2012; Wang et al., 2014a, b; Xu et al., 2012a, b).

    View all citing articles on Scopus
    View full text