ABSTRACT
Background
Caffeine is the most frequently used psychoactive substance in the United States and >90% of reproductive-age women report some amount of intake daily. Despite biological plausibility, previous studies on caffeine and fecundability report conflicting results. Importantly, prior studies measured caffeine exposure exclusively by self-report, which is subject to measurement error and does not account for factors that influence caffeine metabolism.
Objectives
Our objective was to examine associations between preconception serum caffeine metabolites, caffeinated beverage intake, and fecundability.
Methods
Participants included 1228 women aged 18–40 y with a history of 1–2 pregnancy losses in the EAGeR (Effects of Aspirin in Gestation and Reproduction) trial. We prospectively evaluated associations of preconception caffeine metabolites (i.e., caffeine, paraxanthine, and theobromine) measured from 1191 serum samples untimed to a specific time of day, self-reported usual caffeinated beverage intakes at baseline, and time-varying cycle-average caffeinated beverage intake, with fecundability. Using Cox proportional hazards models, we estimated fecundability odds ratios (FORs) and 95% CIs according to each metabolite. Follow-up was complete for 89% (n = 1088) of participants.
Results
At baseline, 85%, 73%, and 91% of women had detectable serum caffeine, paraxanthine, and theobromine, respectively. A total of 797 women became pregnant during ≤6 cycles of preconception follow-up. After adjusting for potential confounders, neither serum caffeine [tertile (T)3 compared with T1 FOR: 0.87; 95% CI: 0.71, 1.08], paraxanthine (T3 compared with T1 FOR: 0.92; 95% CI: 0.75, 1.14), nor theobromine (T3 compared with T1 FOR: 1.15; 95% CI: 0.95, 1.40) were associated with fecundability. Baseline intake of total caffeinated beverages was not associated with fecundability (>3 compared with 0 servings/d adjusted FOR: 0.99; 95% CI: 0.74, 1.34), nor was caffeinated coffee (>2 compared with 0 servings/d adjusted FOR: 0.93; 95% CI: 0.45, 1.92) or caffeinated soda (>2 servings/d adjusted FOR: 0.92; 95% CI: 0.71, 1.20).
Conclusions
Our findings are reassuring that caffeine exposure from usual low to moderate caffeinated beverage intake likely does not influence fecundability.
This trial was registered at clinicaltrials.gov as NCT00467363.
Keywords: caffeine, paraxanthine, theobromine, pregnancy, fecundability, time to pregnancy
Introduction
Caffeine, 1,3,7-trimethylxanthine, is the most widely used psychoactive substance in the United States and >90% of reproductive-age women report some amount of intake daily (1). The American College of Obstetricians and Gynecologists recommends that women limit caffeine intake to <200 mg/d (approximately two 6-oz cups of coffee per day) during pregnancy, possibly to reduce risk of miscarriage and preterm birth (2, 3). However, consensus on caffeine safety while attempting pregnancy is lacking.
Caffeine is a nonselective adenosine receptor antagonist that may affect fecundity via mechanisms involving ovulation, hormonal milieu, and uterine receptivity (4–10). Caffeine appears to inhibit oocyte maturation and steroid hormone production by inhibiting phosphodiesterase, which increases intracellular cyclic adenosine monophosphate (4, 5). Caffeine may also reduce aromatase activity, impairing the conversion of androgens to estrogens (6). Even moderate levels of caffeine exposure may delay oviductal transport of preimplantation embryos in animal models, resulting in higher rates of implantation failure (7). In humans, caffeine intake is variably associated with concentrations of estradiol and testosterone (4, 6, 10), ovulatory function (8), and menstrual cycle length (9), suggesting effects on processes necessary for successful pregnancy establishment.
Despite biological plausibility, findings of prospective epidemiologic studies examining associations of caffeine intake and fecundability have yielded conflicting results (11–19). Importantly, these prior studies classified caffeine exposure exclusively by self-report, which is potentially prone to substantial misclassification. Estimation of caffeine exposure from self-reported beverage intake is imprecise owing to issues with recall and variability in caffeine content across individual foods and beverages (20). Also not captured by self-report are important genetic and nongenetic factors that influence the rate of caffeine metabolism, such as smoking and polymorphisms affecting expression of cytochrome P450 1A2 (CYP1A2), the hepatic enzyme responsible for caffeine clearance (21, 22).
Serum caffeine metabolites, in contrast, assess the biological dose of caffeine. After ingestion, ∼84% of caffeine is demethylated by CYP1A2 to paraxanthine, the primary metabolite, and to a lesser extent, theobromine (12%) and theophylline (4%) are also produced (23). Importantly, serum caffeine and paraxanthine have been shown to be relatively stable over time and appropriately distinguish levels of caffeine consumption among reproductive-age women (8, 24). To assess whether caffeine exposure is related to fecundability, prospective studies that measure caffeine exposure via serum caffeine metabolites, in conjunction with self-reported intake, are needed.
The primary objective of our study was to examine associations of preconception caffeine metabolites and self-reported intake of caffeinated beverages with fecundability among healthy reproductive-age women attempting to conceive.
Methods
Study population
This was a secondary analysis of the EAGeR (Effects of Aspirin in Gestation and Reproduction) trial (NCT00467363), a multicenter, block-randomized, double-blind, placebo-controlled clinical trial designed to determine the effect of preconception-initiated daily low-dose aspirin (LDA) on reproductive outcomes in women with a history of pregnancy loss (25, 26). The trial design and overall results have been described previously (26).
Study participants were 1228 women beginning or continuing to attempt pregnancy, aged 18–40 y, with 1–2 prior pregnancy losses at any time in the past. Women with a known history of infertility treatment, pelvic inflammatory disease, tubal occlusion, endometriosis, anovulation and polycystic ovary syndrome (PCOS), or uterine abnormality were excluded from participation. Participants were recruited at 4 clinical sites (Salt Lake City, UT; Denver, CO; Buffalo, NY; and Scranton, PA) in the United States from 2007 to 2011 and followed for ≤6 menstrual cycles while attempting pregnancy and then throughout pregnancy for women who conceived. The majority of participants (>82%) were recruited from the Salt Lake City, UT study site.
The institutional review board at each study site (Salt Lake City, UT; Denver, CO; Buffalo, NY; Scranton, PA) and data coordinating center approved the trial protocol and all participants provided written informed consent before enrolling.
Caffeine biomarkers
Upon enrollment and before randomization, women provided blood samples untimed to a specific time of day in the clinic, which were frozen at −80°C until analysis. Caffeine, paraxanthine, and theobromine, the 3 major caffeine metabolites (27), were measured in serum from 1191 available samples. Caffeine metabolites were measured using LC-MS (Agilent Technologies). Of the 1191 samples available for analysis, 184 (15%), 324 (27%), and 109 (9%) were below the limit of detection (LOD) of 0.04 µg/mL for caffeine, paraxanthine, and theobromine, respectively. Values below the LOD were imputed as LOD/√2 (28).
The approximate half-lives of serum caffeine, paraxanthine, and theobromine are 5, 5, and 2–3 h, respectively (23). Paraxanthine is considered a relatively stable caffeine metabolite because its rate of formation is approximately equal to its rate of elimination under usual intake patterns (29). Previous studies have demonstrated that a single sample of serum paraxanthine (timed to a specific time of day and untimed) can appropriately distinguish between levels of caffeine consumption among reproductive-age women and is sensitive enough to detect associations with other reproductive outcomes (8, 30).
Caffeinated beverage intake
On baseline questionnaires, participants self-reported their usual intakes of specific caffeinated beverages (i.e., brewed coffee, instant coffee, Frappuccino, espresso drinks, black tea, green tea, Lipton tea, instant tea, cola, citrus soda, orange soda, root beer, and energy drinks) over the past 12 mo, their usual serving size (oz), and frequency of intake (per month, week, or day). Servings were converted to standard servings per day, which corresponds to a 6-oz (177-mL) cup of caffeinated coffee or tea and a 12-oz (355-mL) can of caffeinated soda. Total preconception caffeinated beverage intake for each woman was estimated by summing the intakes of all caffeinated beverages.
Upon enrollment in the study, participants were provided with preconception diaries and instructions on how to complete diary entries. During active preconception follow-up (cycles 1 and 2), participants recorded the number of caffeinated beverages (open response) they consumed each day in the daily diary. During passive follow-up (cycles 3–6), participants completed questionnaires during end-of-cycle clinic visits that queried the average number of 6-oz cups of coffee (response options: none, 1, 2, 3, 4, 5, or >5 cups/d) and 12-oz cans of soda (response options: none, 1, 2, 3, 4, 5, or >5 cans/d) consumed each day over the previous month. Time-varying cycle-average caffeinated beverage intake was estimated from participants’ responses on diaries and end-of-cycle questionnaires. Among 1228 women, 1132 (92%) recorded total caffeinated beverages during the ≤6 menstrual cycles of follow-up.
Outcome assessment
The outcome of interest for our study was fecundability, defined as the monthly probability of conception. Study-provided fertility monitors (Clearblue Easy Fertility Monitor; Inverness Medical Innovations) were used to time intercourse and estimate date of ovulation. Participants who reported missing menses at the end-of-cycle visit took a urine human chorionic gonadotropin (hCG) test (Quickvue, Quidel Corp), which was sensitive to 25 mIU hCG/mL. Higher-sensitivity urine hCG testing was also performed on stored daily first-morning urine samples from the last 10 d of participants’ first and second menstrual cycles after enrollment. The end of cycle was determined by the onset of menses or a positive hCG pregnancy test.
Covariate assessment
Baseline demographic characteristics (i.e., age, self-identified race/ethnicity, income, marital status, multivitamin use, physical activity, alcohol consumption, and smoking) and reproductive history information (i.e., number of previous losses and number of previous live births) were obtained via questionnaire during a clinic visit on day 2–4 of the menstrual cycle. Height and weight were also recorded during the baseline visit, which were used to calculate BMI (kg/m2). Time-varying covariates including cycle-average self-reported stress (no stress, little stress, moderate stress, and a lot of stress), alcohol intake (continuous; servings/d), smoking (none, actively smoked, or passively smoked), and sexual intercourse frequency (continuous; times/d) were obtained from preconception daily diaries and end-of-cycle questionnaires.
Statistical analysis
We compared demographic and lifestyle characteristics by tertiles of preconception serum caffeine, paraxanthine, and theobromine using generalized linear models to estimate means ± SDs for continuous variables and χ2 tests for categorical variables. We assessed correlations between each preconception caffeine biomarker and baseline caffeinated beverage type using Spearman rank correlation tests.
Caffeine biomarkers were analyzed in tertiles. Baseline and time-varying total and individual caffeinated beverages (i.e., caffeinated coffee, caffeinated soda, and caffeinated tea) were analyzed in categories (0, >0 to 1, >1 to 2, and >2 servings/d). We examined the possibly nonlinear relation between each baseline continuous exposure and fecundability using restricted cubic spline models (with 5 knots specified) and evaluated the individual spline term contributions to the model fit and overall test for nonlinearity.
We used time-fixed (for baseline caffeine biomarkers and caffeinated beverage intakes) and time-varying (for cycle-average total caffeinated beverage intake) discrete Cox proportional hazards regression to estimate unadjusted and adjusted fecundability odds ratios (FORs) and 95% CIs. Women contributed cycles at risk until hCG-detected pregnancy or censoring (loss to follow-up, withdrawal from study, or 6 unsuccessful cycles). Models accounted for left-truncation (cycles trying to become pregnant before study entry) by summing the number of cycles attempting pregnancy before and after study entry, and specifying the entry time as the number of cycles attempting pregnancy before study entry in PROC PHREG in SAS. The FOR represents the per-cycle probability of pregnancy associated with exposure, conditional on not becoming pregnant in the preceding cycle. An FOR < 1 indicates reduced fecundability and longer time to pregnancy, whereas an FOR > 1 indicates improved fecundability and shorter time to pregnancy.
Potential confounders were identified from prior studies on caffeine and fecundability and final covariate selection was informed by use of directed acyclic graphs (31) (Supplemental Figure 1). Models evaluating baseline time-fixed exposures were adjusted for age (continuous), BMI (continuous), smoking status (smoker or nonsmoker), multivitamin use (yes or no), alcohol consumption (drinker or nondrinker), number of previous losses (1 or 2), number of previous live births (0, 1, or ≥2), and exercise (low, moderate, or high). Models evaluating time-varying cycle-average caffeinated beverage intake were adjusted for the aforementioned baseline covariates, plus time-varying cycle-average alcohol intake (continuous; drinks/d), smoking (none, active smoking, or passive smoking), sexual intercourse frequency (continuous; times/d), and stress (no stress, little stress, moderate stress, a lot of stress). Less than 8% of participants were missing data on exposures and covariates. Multiple imputation with the fully conditional specification method was used to address missing data (32). We generated 10 imputed data sets, analyzed each data set individually, and reported combined estimates.
Because smoking is known to influence caffeine metabolism (33), we conducted stratified analyses by baseline smoking status (smoker compared with nonsmoker). We also conducted stratified analyses by time trying to become pregnant at study entry (≤3 compared with >3 cycles) and by treatment arm (LDA compared with placebo). During passive follow-up (i.e., cycles 3–6) time-varying caffeine exposure was assessed by end-of-cycle questionnaires, which could introduce recall bias if women systematically over- or under-reported their caffeinated beverage intake with respect to pregnancy status. To evaluate any impact of recall bias on our estimates for time-varying caffeinated beverage intake in cycles 3–6, we conducted sensitivity analyses restricted to active follow-up (i.e., cycles 1 and 2). Statistical analyses were conducted using SAS version 9.4 (SAS Institute) and DAGitty (version 3.0, Johannes Textor; http://dagitty.net/).
Results
At baseline, 85%, 73%, and 91% of women had detectable concentrations of serum caffeine, paraxanthine, and theobromine, respectively. Median values for caffeine metabolites were 0.27 µg/mL (range: 0.03–17.9 µg/mL; 10th percentile = 0.03, 90th percentile = 3.07 µg/mL) for caffeine, 0.11 µg/mL (range: 0.03–8.96 µg/mL; 10th percentile = 0.03, 90th percentile = 0.72 µg/mL) for paraxanthine, and 0.59 µg/mL (range: 0.03–94.1 µg/mL; 10th percentile = 0.04, 90th percentile = 2.38 µg/mL) for theobromine. Most women (75%) reported consuming caffeinated beverages during the past year, with 28% and 67% reporting any intake of caffeinated coffee and soda, respectively. Median [IQR] time since last loss was 3.7 mo [1.9–9.2 mo]. Serum caffeine and paraxanthine concentrations were positively associated with age, BMI, smoking, employment, and alcohol consumption, and inversely associated with educational attainment and physical activity. Theobromine concentrations were positively associated with employment and physical activity (Table 1).
TABLE 1.
Baseline study population characteristics by tertile of preconception serum caffeine metabolites among 1191 participants in the EAGeR (Effects of Aspirin in Gestation and Reproduction) trial, 2007–20111
| Caffeine | Paraxanthine | Theobromine | |||||||
|---|---|---|---|---|---|---|---|---|---|
| T1 | T2 | T3 | T1 | T2 | T3 | T1 | T2 | T3 | |
| N | 409 | 386 | 396 | 420 | 375 | 396 | 395 | 393 | 403 |
| Median, µg/mL | 0.04 | 0.27 | 2.08 | 0.03 | 0.11 | 0.52 | 0.09 | 0.59 | 1.80 |
| Range, µg/mL | 0.03–0.10 | 0.12–0.78 | 0.80–17.9 | 0.03–0.04 | 0.05–0.23 | 0.25–8.96 | 0.03–0.31 | 0.32–1.03 | 1.05–94.1 |
| Age, y | 27.7 ± 4.3 | 28.4 ± 4.7 | 30.1 ± 5.1 | 27.8 ± 4.2 | 28.3 ± 4.6 | 30.1 ± 5.2 | 28.3 ± 4.7 | 29.1 ± 4.9 | 28.8 ± 4.8 |
| BMI, kg/m2 | 25.4 ± 6.0 | 26.1 ± 6.5 | 27.4 ± 6.7 | 25.2 ± 5.9 | 26.4 ± 6.8 | 27.2 ± 6.6 | 25.7 ± 6.0 | 26.3 ± 6.2 | 26.8 ± 7.1 |
| Race/ethnicity | |||||||||
| White | 385 (94) | 369 (96) | 375 (95) | 399 (95) | 355 (95) | 375 (95) | 365 (92) | 377 (96) | 387 (96) |
| Other | 24 (6) | 17 (4) | 21 (5) | 21 (5) | 20 (5) | 21 (5) | 30 (8) | 16 (4) | 16 (4) |
| Education | |||||||||
| > High school | 371 (91) | 329 (86) | 329 (83) | 383 (91) | 315 (84) | 331 (84) | 332 (84) | 341 (87) | 356 (88) |
| Vitamin use | |||||||||
| No | 30 (7) | 21 (6) | 38 (10) | 30 (7) | 24 (7) | 35 (9) | 33 (8) | 29 (8) | 27 (7) |
| Yes, with folic acid | 332 (82) | 304 (80) | 309 (79) | 348 (83) | 285 (77) | 312 (80) | 308 (79) | 305 (79) | 332 (83) |
| Yes, with no folic acid | 44 (11) | 53 (14) | 44 (11) | 40 (10) | 60 (16) | 41 (11) | 50 (13) | 50 (13) | 41 (10) |
| Smoking in the past year | |||||||||
| Never | 383 (94) | 335 (88) | 317 (81) | 395 (94) | 326 (88) | 314 (80) | 346 (88) | 337 (87) | 352 (88) |
| <6 times/wk | 9 (2) | 29 (7) | 46 (12) | 13 (3) | 25 (7) | 46 (12) | 24 (6) | 28 (7) | 32 (8) |
| Daily | 17 (4) | 18 (5) | 28 (7) | 12 (3) | 18 (5) | 33 (8) | 24 (6) | 24 (6) | 15 (4) |
| Household income, $ | |||||||||
| ≥100,000 | 160 (39) | 162 (42) | 154 (39) | 168 (40) | 154 (41) | 154 (39) | 143 (36) | 171 (44) | 162 (40) |
| 75,000–99,999 | 42 (10) | 41 (11) | 63 (16) | 41 (10) | 46 (12) | 59 (15) | 46 (12) | 45 (11) | 55 (14) |
| 40,000–74,999 | 49 (12) | 53 (14) | 73 (18) | 53 (13) | 48 (13) | 74 (19) | 54 (14) | 64 (16) | 57 (14) |
| 20,000–39,999 | 122 (30) | 100 (26) | 80 (20) | 123 (29) | 98 (26) | 81 (20) | 117 (30) | 91 (23) | 94 (23) |
| ≤19,999 | 36 (9) | 30 (8) | 25 (6) | 35 (8) | 29 (8) | 27 (7) | 35 (9) | 21 (5) | 35 (9) |
| Employed | |||||||||
| Yes | 293 (72) | 269 (70) | 309 (78) | 301 (72) | 264 (70) | 306 (77) | 276 (70) | 279 (71) | 316 (78) |
| Time trying to conceive at study entry, mo | |||||||||
| 0 | 83 (22) | 65 (18) | 51 (14) | 93 (24) | 48 (14) | 58 (16) | 67 (19) | 66 (18) | 66 (17) |
| 1–3 | 190 (50) | 196 (55) | 216 (60) | 199 (51) | 190 (56) | 213 (59) | 196 (55) | 199 (55) | 207 (55) |
| 4–6 | 60 (16) | 54 (15) | 53 (15) | 53 (13) | 65 (19) | 49 (14) | 60 (17) | 43 (12) | 64 (17) |
| >6 | 47 (12) | 42 (12) | 40 (11) | 48 (12) | 39 (11) | 42 (12) | 36 (10) | 51 (14) | 42 (11) |
| Time since last loss, mo | |||||||||
| ≤4 | 214 (53) | 200 (52) | 216 (55) | 217 (52) | 189 (52) | 224 (57) | 207 (53) | 205 (53) | 218 (55) |
| 5–8 | 74 (18) | 77 (20) | 61 (16) | 77 (19) | 72 (20) | 63 (16) | 66 (17) | 61 (16) | 85 (21) |
| 9–12 | 31 (8) | 39 (10) | 28 (7) | 35 (8) | 31 (8) | 32 (8) | 31 (8) | 37 (10) | 30 (8) |
| >12 | 83 (21) | 65 (17) | 84 (22) | 85 (21) | 75 (20) | 72 (18) | 83 (21) | 84 (22) | 65 (16) |
| Previous losses, n | |||||||||
| 1 | 280 (68) | 256 (66) | 262 (66) | 282 (67) | 251 (67) | 265 (67) | 256 (65) | 265 (67) | 277 (69) |
| 2 | 129 (32) | 130 (34) | 134 (34) | 138 (33) | 124 (33) | 131 (33) | 139 (35) | 128 (33) | 126 (31) |
| Previous live births, n | |||||||||
| 0 | 197 (48) | 167 (43) | 186 (47) | 195 (46) | 173 (46) | 182 (46) | 173 (44) | 174 (44) | 203 (50) |
| 1 | 140 (34) | 148 (38) | 145 (37) | 146 (35) | 139 (37) | 148 (37) | 160 (41) | 147 (37) | 126 (31) |
| 2 | 72 (18) | 71 (18) | 65 (16) | 79 (19) | 63 (17) | 66 (17) | 62 (16) | 72 (18) | 74 (18) |
| Alcohol consumption in the past year | |||||||||
| Never | 339 (83) | 272 (71) | 184 (47) | 351 (84) | 253 (68) | 191 (49) | 267 (68) | 252 (65) | 276 (69) |
| Sometimes | 41 (10) | 66 (17) | 102 (26) | 35 (8) | 72 (19) | 102 (26) | 66 (17) | 78 (20) | 65 (16) |
| Often | 29 (7) | 47 (12) | 106 (27) | 34 (8) | 48 (13) | 100 (25) | 62 (16) | 60 (15) | 60 (15) |
| Physical activity | |||||||||
| Low | 97 (24) | 104 (27) | 110 (28) | 99 (24) | 105 (28) | 107 (27) | 118 (30) | 102 (26) | 91 (23) |
| Moderate | 167 (41) | 148 (38) | 172 (43) | 171 (41) | 145 (39) | 171 (43) | 147 (37) | 173 (44) | 167 (41) |
| High | 145 (35) | 134 (35) | 114 (29) | 150 (36) | 125 (33) | 118 (30) | 130 (33) | 118 (30) | 145 (36) |
| Study site | |||||||||
| Salt Lake City, UT | 359 (88) | 318 (82) | 295 (74) | 369 (88) | 303 (81) | 300 (76) | 315 (80) | 321 (82) | 336 (83) |
| Buffalo, NY | 22 (5) | 28 (7) | 26 (7) | 16 (4) | 32 (9) | 28 (7) | 25 (6) | 34 (9) | 17 (4) |
| Denver, CO | 16 (4) | 20 (5) | 36 (9) | 14 (3) | 26 (7) | 32 (8) | 28 (7) | 22 (6) | 22 (5) |
| Scranton, PA | 12 (3) | 20 (5) | 39 (10) | 21 (5) | 14 (4) | 36 (9) | 27 (7) | 16 (4) | 28 (7) |
| Treatment arm | |||||||||
| Placebo | 200 (49) | 196 (51) | 194 (49) | 215 (51) | 181 (48) | 205 (52) | 194 (49) | 206 (52) | 201 (50) |
| Aspirin | 209 (51) | 190 (49) | 202 (51) | 205 (49) | 194 (52) | 191 (48) | 201 (51) | 187 (48) | 202 (50) |
Values are means ± SDs and n (%) estimated from generalized linear models. T, tertile.
Spearman rank correlation coefficients between caffeine biomarkers and self-reported caffeinated beverages ranged from −0.03 to 0.84 (Table 2). Baseline total caffeinated beverage intake was moderately correlated with both serum caffeine (r = 0.52) and serum paraxanthine (r = 0.54) but was not correlated with serum theobromine (r = 0.03).
TABLE 2.
Spearman rank correlation coefficients (r) for preconception caffeine metabolites and self-reported intake of caffeinated beverages, EAGeR (Effects of Aspirin in Gestation and Reproduction) trial
| Caffeine | Paraxanthine | Theobromine | Total caffeinated beverages | Coffee | Soda | Tea | |
|---|---|---|---|---|---|---|---|
| Median (range) | 0.27 (0.03–17.9) µg/mL | 0.11 (0.03–8.96) µg/mL | 0.59 (0.03–94.1) µg/mL | 0.42 (0–24.5) cups/d | 0 (0–20) cups/d | 0.21 (0–18) cups/d | 0 (0–6) cups/d |
| Caffeine | 1.00 | — | — | — | — | — | — |
| Paraxanthine | 0.84 | 1.00 | — | — | — | — | — |
| Theobromine | 0.32 | 0.31 | 1.00 | — | — | — | — |
| Total caffeinated beverages | 0.52 | 0.54 | 0.03 | 1.00 | — | — | — |
| Coffee | 0.44 | 0.43 | −0.01 | 0.52 | 1.00 | — | — |
| Soda | 0.32 | 0.36 | 0.02 | 0.82 | 0.10 | 1.00 | — |
| Tea | 0.21 | 0.19 | −0.01 | 0.25 | 0.44 | −0.03 | 1.00 |
Out of 1228 women who enrolled in the study, 1088 (89%) participants completed follow-up and 797 became pregnant (Supplemental Figure 2). Loss to follow-up was slightly more frequent among those with high serum caffeine [tertile (T)3: 13.4% compared with T1: 11.2%] and paraxanthine (T3: 14.7% compared with T1: 10.5%), and low theobromine (T3: 10.0% compared with T1: 14.9%). Similarly, those with high total caffeinated beverage intake were somewhat more likely to be lost to follow-up (>3 servings/d: 16.0% compared with 0 servings/d: 11.2%).
Overall, we did not observe strong associations of baseline caffeine metabolites with fecundability (Table 3). In unadjusted models, serum caffeine was inversely associated with fecundability (T3 compared with T1 FOR: 0.80; 95% CI: 0.66, 0.97); however, after adjusting for age, BMI, smoking, and other factors, the association was largely attenuated (T3 compared with T1 FOR: 0.87; 95% CI: 0.71, 1.08). BMI accounted for the most attenuation in multivariable models. Serum paraxanthine and theobromine were not associated with fecundability in unadjusted or adjusted models. Results from restricted cubic spline models were consistent with tertile analyses and revealed no associations of caffeine biomarkers with fecundability—linear or nonlinear (Supplemental Figure 3).
TABLE 3.
FORs and 95% CIs for baseline serum caffeine metabolites and fecundability, EAGeR (Effects of Aspirin in Gestation and Reproduction) trial1
| Median (range), µg/mL | Pregnancies, n | Unadjusted2 FOR (95% CI) | Adjusted2,3 FOR (95% CI) | |
|---|---|---|---|---|
| Caffeine | ||||
| T1 | 0.04 (0.03–0.10) | 292 | 1 | 1 |
| T2 | 0.27 (0.12–0.78) | 253 | 0.88 (0.73, 1.07) | 0.90 (0.74, 1.09) |
| T3 | 2.08 (0.80–17.9) | 252 | 0.80 (0.66, 0.97) | 0.87 (0.71, 1.08) |
| Paraxanthine | ||||
| T1 | 0.03 (0.03–0.04) | 293 | 1 | 1 |
| T2 | 0.11 (0.05–0.23) | 255 | 1.03 (0.85, 1.25) | 1.06 (0.87, 1.29) |
| T3 | 0.54 (0.25–8.96) | 250 | 0.85 (0.70, 1.03) | 0.92 (0.75, 1.14) |
| Theobromine | ||||
| T1 | 0.09 (0.03–0.31) | 261 | 1 | 1 |
| T2 | 0.59 (0.31–1.03) | 254 | 1.04 (0.85, 1.27) | 1.08 (0.89, 1.33) |
| T3 | 1.89 (1.05–94.1) | 282 | 1.08 (0.89, 1.31) | 1.15 (0.95, 1.40) |
FOR, fecundability odds ratio; T, tertile.
FOR (95% CI) estimated from Cox proportional hazards models.
Model adjusted for baseline age (continuous), BMI (continuous), smoking status (smoker or nonsmoker), multivitamin use (yes or no), alcohol consumption (drinker or nondrinker), number of previous losses (1 or 2), number of previous live births (0, 1, or ≥2), and exercise (low, moderate, high).
Analyses of self-reported baseline caffeinated beverage intake and fecundability yielded similar findings to those for serum preconception biomarkers (Table 4). For example, in unadjusted analyses, baseline total caffeinated beverage intake >3 compared with 0 servings/d was imprecisely associated with lower fecundability (FOR: 0.84; 95% CI: 0.64, 1.10). After adjusting for potential confounders, the association was attenuated (FOR: 0.99; 95% CI: 0.74, 1.34). Similarly, intakes of baseline caffeinated coffee, soda, and tea were not strongly associated with fecundability in adjusted models, although relatively few women reported consuming >2 servings/d of coffee and tea. Results from spline models were consistent with categorical analyses and did not support any association of baseline caffeinated beverages with fecundability (Supplemental Figure 4).
TABLE 4.
FORs and 95% CIs for associations of preconception caffeinated beverage intake and fecundability, EAGeR (Effects of Aspirin in Gestation and Reproduction) trial1
| Median, servings/d | Pregnancies, n | Unadjusted2 FOR (95% CI) | Adjusted2,3 FOR (95% CI) | |
|---|---|---|---|---|
| Baseline total caffeinated beverages, servings/d | ||||
| 0 | 0 | 209 | 1 | 1 |
| >0 to 1 | 0.2 | 299 | 0.99 (0.81, 1.21) | 1.08 (0.87, 1.32) |
| >1 to 2 | 1.5 | 121 | 0.97 (0.75, 1.25) | 1.00 (0.76, 1.31) |
| >2 to 3 | 2.6 | 72 | 0.80 (0.59, 1.08) | 0.87 (0.64, 1.19) |
| >3 | 5.0 | 96 | 0.84 (0.64, 1.10) | 0.99 (0.74, 1.34) |
| Time-varying total caffeinated beverages,4 servings/d | ||||
| 0 | 0 | 144 | 1 | 1 |
| >0 to 1 | 0.6 | 369 | 0.94 (0.77, 1.15) | 0.99 (0.80, 1.22) |
| >1 to 2 | 1.9 | 178 | 0.96 (0.75, 1.22) | 1.07 (0.82, 1.38) |
| >2 | 3.0 | 65 | 0.64 (0.47, 0.87) | 0.77 (0.55, 1.07) |
| Baseline caffeinated coffee, servings/d | ||||
| 0 | 0 | 577 | 1 | 1 |
| >0 to 1 | 0.4 | 131 | 1.29 (1.04, 1.61) | 1.64 (0.89, 3.01) |
| >1 to 2 | 1.3 | 53 | 0.76 (0.56, 1.04) | 0.60 (0.29, 1.23) |
| >2 | 2.6 | 36 | 0.79 (0.54, 1.14) | 0.93 (0.45, 1.92) |
| Baseline caffeinated soda, servings/d | ||||
| 0 | 0 | 276 | 1 | 1 |
| >0 to 1 | 0.2 | 334 | 0.95 (0.79, 1.14) | 1.02 (0.85, 1.22) |
| >1 to 2 | 1.5 | 82 | 0.87 (0.66, 1.16) | 0.92 (0.69, 1.23) |
| >2 | 3.2 | 105 | 0.81 (0.63, 1.04) | 0.92 (0.71, 1.20) |
| Baseline caffeinated tea, servings/d | ||||
| 0 | 0 | 733 | 1 | 1 |
| >0 to 1 | 0.1 | 55 | 0.85 (0.63, 1.16) | 0.82 (0.59, 1.15) |
| >1 | 2.5 | 9 | 0.82 (0.39, 1.70) | 1.02 (0.48, 2.16) |
FOR, fecundability odds ratio.
FOR (95% CI) estimated from Cox proportional hazards models.
Models for baseline exposures adjusted for baseline age (continuous), BMI (continuous), smoking status (smoker or nonsmoker), multivitamin use (yes or no), alcohol consumption (drinker or nondrinker), number of previous losses (1 or 2), number of previous live births (0, 1, or ≥2), and exercise (low, moderate, high). Model for time-varying caffeinated beverage intake adjusted for baseline age (continuous), BMI (continuous), multivitamin use (yes or no), number of previous losses (1 or 2), number of previous live births (0, 1, or ≥2), and exercise (low, moderate, or high) plus time-varying smoking (none, actively smoked, or passively smoked), alcohol use (continuous; average servings/d), sexual intercourse frequency (continuous; average times/d), and self-reported stress level (no stress, little stress, moderate stress, a lot of stress).
Cycle-specific daily average total caffeinated beverages was estimated from daily diaries during active follow-up (cycles 1 and 2) and end-of-cycle questionnaires during passive follow-up (cycles 3–6). Analyses excluded 36 women who were completely missing daily diaries and end-of-cycle questionnaires.
During active follow-up, median cycle-specific total caffeinated beverage intake was 0.39 and 0.32 servings/d in cycles 1 and 2, respectively. In each of cycles 3–6, median intake was consistently 1 serving/d. Differences between median intake during cycles 1 and 2 and cycles 3–6 likely reflect the change in exposure assessments between active (i.e., daily diaries) and passive (i.e., end-of-cycle questionnaires) follow-up. In unadjusted models accounting for changes in consumption patterns using time-varying cycle-average caffeinated beverage intake, total intake of >2 servings compared with 0 servings/d was associated with reduced fecundability (FOR: 0.64; 95% CI: 0.47, 0.87). However, after adjustment for baseline and time-varying confounders, this association was somewhat attenuated (adjusted FOR: 0.77; 95% CI: 0.55, 1.07). In exploratory time-varying analyses examining extremely high cycle-average intake of total caffeinated beverages, women who consumed >3 compared with 0 servings/d (n = 15 pregnancies) experienced lower fecundability (category median: 4.1 servings/d; adjusted FOR: 0.56; 95% CI: 0.31, 0.99).
In sensitivity analyses in which we adjusted multivariable models for categories of age (<25, 25 to <30, 30 to <35, ≥35 y), BMI (underweight, normal, overweight, obese), and alcohol intake (never, sometimes, often), and further adjusted for employment status (employed, unemployed), hours of sleep per night (continuous), and season (spring, summer, fall, winter), estimates were materially unchanged (serum caffeine T3 compared with T1 FOR: 0.86; 95% CI: 0.69, 1.06; serum paraxanthine T3 compared with T1 FOR: 0.92; 95% CI: 0.75, 1.14; serum theobromine T3 compared with T1 FOR: 1.15; 95% CI: 0.94, 1.39; baseline total caffeinated beverages >3 compared with 0 servings/d FOR: 1.00; 95% CI: 0.74, 1.35). In stratified analyses, we did not observe substantive differences in associations of baseline caffeine biomarkers and fecundability among smokers compared with nonsmokers, although relatively few women (12%) in this study population reported smoking at baseline (Supplemental Table 1). We also did not observe effect modification by time trying to become pregnant at study entry (≤3 compared with >3 cycles) (Supplemental Table 2) or by treatment arm (LDA compared with placebo) (Supplemental Table 3). The sensitivity analysis examining time-varying total caffeinated beverage intake and fecundability limited to active follow-up produced imprecise estimates that were consistent with the overall analysis (adjusted FOR: 0.76; 95% CI: 0.46, 1.26).
Discussion
In this prospective study, we found little evidence that preconception serum caffeine, paraxanthine, or theobromine concentrations are associated with fecundability among healthy reproductive-age women with low to moderate caffeinated beverage consumption. Similarly, we found that baseline and time-varying intakes of total and individual caffeinated beverages, measured by self-report, were not appreciably associated with fecundability. Collectively, our findings suggest that usual caffeine exposure from low to moderate caffeinated beverage consumption likely does not influence fecundability.
To date, studies on caffeine and fecundability have measured caffeine exclusively by self-report; thus, our study extends the existing literature by utilizing preconception caffeine metabolites, in conjunction with self-reported baseline and time-varying intake of caffeinated beverages. Caffeine metabolites capture the biological dose of caffeine, as opposed to self-report, which is potentially subject to substantial misclassification from varying caffeine content in beverages and variability in caffeine metabolism due to genetic polymorphisms and lifestyle factors, such as smoking (20–22). Caffeine metabolites also encompass caffeine exposure from foods and medications, important components of total caffeine intake often not captured by self-report. Serum caffeine metabolites, particularly paraxanthine, have been shown to appropriately rank both pregnant (24) and nonpregnant (8) women according to self-reported caffeine intake and, importantly, improve precision in estimating associations of caffeine exposure with reproductive outcomes, compared with 24-h recall (8, 10). Thus, the lack of associations observed between serum caffeine metabolites and fecundability in the present article provides evidence that low to moderate usual caffeine consumption while attempting to conceive probably does not increase time to pregnancy.
Although our findings for serum caffeine metabolites cannot be directly compared with previous work, the absence of associations observed for serum caffeine metabolites is consistent with several other prospective studies examining total caffeine intake, measured by self-report, and fecundability (11–15). For example, in 2 studies of Danish and North American women attempting to conceive, caffeine was not associated with fecundability (11, 14). Two other smaller studies also reported no associations between total caffeine intake and fecundability (13, 15) or ovulatory disorder infertility (12). In contrast, Wilcox et al. (17) found that baseline caffeine intake >105 mg/mo was strongly associated with 49% lower fecundability among women attempting to conceive for >3 mo; however, this association was attenuated in the overall cohort. Another study reported an imprecise 67% reduction in fecundability among Dutch nonsmoking women who consumed ≥300 compared with <300 mg/d of caffeine from noncoffee sources (16). Differences in findings between studies reporting inverse associations of caffeine and fecundability, and those reporting no association, including ours, could be explained by lack of control for important confounders such as age or alcohol intake (16, 17).
In analyses examining total caffeinated beverages, we found that baseline and time-varying intakes were not appreciably associated with fecundability. However, in time-varying analyses of total caffeinated beverage intake, we observed some suggestion that extremely high intake (>3 compared with 0 servings/d) was associated with lower fecundability. This is somewhat consistent with 2 previous reports, in which stronger inverse associations were observed in models evaluating time-varying intake of caffeinated soda than in models evaluating baseline intake (11, 14). As others have noted, the time-varying exposure data may reduce misclassification arising from changes in consumption across pregnancy attempts, resulting in stronger estimates than from baseline data (14). Considering the lack of a dose–response relation and the small number of pregnancies in cycles where participants’ daily average caffeinated beverage intake exceeded 3 servings/d (n = 15), this may be a chance finding and should be interpreted cautiously.
We observed no associations of caffeinated soda consumption with fecundability, which conflicts with most previous studies. Soda intake of >3 compared with 0 servings/d was strongly, but imprecisely, associated with fecundability among Danish women attempting to conceive (14). Among American and Canadian women attempting to conceive, caffeinated soda intake of 1 compared with 0 servings/d, but not higher levels of intake, was associated with reduced fecundability (11), which is consistent with other studies (12, 15, 17). Reasons for conflicting findings between our study and those reporting inverse associations are unclear. Inverse associations for caffeinated soda and fecundability could be explained by residual confounding arising from unhealthy lifestyle or metabolic factors affecting fecundity, like smoking or PCOS (11), or from other components in caffeinated soda, such as added sugar. Indeed, 1 study reported that intake of sugar-sweetened sodas, but not diet sodas, was associated with reduced fecundability among American and Canadian women, suggesting that added sugars may be more relevant to fecundability than caffeine itself (34). We lacked the power to estimate associations of extremely high intake of caffeinated sodas with fecundability owing to relatively low consumption of caffeinated sodas in EAGeR compared with other preconception cohorts, which may also explain the discordant findings (34).
The lack of associations observed for caffeinated coffee in our study is consistent with most prior prospective studies (11, 12, 14, 15, 19), but conflicts with 2 others that reported associations at extremely high levels of coffee intake or failed to control for important confounders like alcohol (16, 17). Data are more mixed for tea consumption, with 3 studies suggesting improved fecundability with higher tea intake (14, 15, 19), and 2 studies reporting no associations (12, 18) as we did here. It is possible that tea or noncaffeine components in tea may indeed improve fecundability; however, given the extremely low tea consumption in EAGeR, we were unable to examine associations with fecundability at higher levels of intake.
Strengths of our study include a prospective design, minimal loss to follow-up (<11%), and use of caffeine metabolites to measure the biological dose of caffeine, in conjunction with baseline and updated self-reported intake of caffeinated beverages. Furthermore, high-sensitivity hCG testing of daily urine samples allowed us to detect virtually all pregnancies that reached implantation, including ones recognized only by subtle transient increases in hCG, thus reducing potential misclassification of the outcome.
Nevertheless, our study also has limitations. The half-lives of the serum metabolites, particularly caffeine, are relatively short (5–6 h) (35), and single untimed measurements at baseline may not reflect long-term or usual exposure while attempting pregnancy, likely resulting in some exposure misclassification. In light of previous studies showing that serum paraxanthine is sensitive enough to detect associations with other reproductive outcomes (8, 30), combined with stable caffeinated beverage consumption in EAGeR across cycles, it seems unlikely that drastic changes in caffeine exposure over time would explain our null findings. Further, EAGeR participants had lower intake of caffeinated beverages than other preconception cohorts, possibly because of prior pregnancy loss or geographic differences, limiting our ability to estimate associations of extremely high caffeinated beverage intake with fecundability. Residual confounding arising from omitted variables such as sugar-sweetened beverages or glycemic index, or imprecisely measured variables like stress, is also possible, although it seems unlikely that these would explain our null findings. Finally, because EAGeR women had a history of 1–2 prior pregnancy losses, but were otherwise healthy and fecund, our findings may not be widely generalizable.
In conclusion, serum caffeine metabolites were not associated with fecundability among healthy, fecund, reproductive-age women with low to moderate caffeinated beverage consumption. Women and clinicians should be reassured that typical caffeine exposure from low to moderate caffeinated beverage intake probably does not meaningfully affect time to pregnancy.
Supplementary Material
Acknowledgments
The authors’ responsibilities were as follows—SLM, EFS, and RMS: designed and conducted the research; ACP-S, EAD, and AY: analyzed the data or performed statistical analysis; ACP-S, KK, SNH, EAD, LAS, NJP, EFS, KCS, RMS, and SLM: wrote the paper; ACP-S and SLM: had primary responsibility for the final content; and all authors: read and approved the final manuscript. The authors report no conflicts of interest.
Notes
Supported by Eunice Kennedy Shriver National Institute of Child Health and Human Development (NIH, Bethesda, MD, USA) Intramural Research Program contract numbers HHSN267200603423, HHSN267200603424, and HHSN267200603426.
Supplemental Figures 1–4 and Supplemental Tables 1–3 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/ajcn/.
Abbreviations used: CYP1A2, cytochrome P450 1A2; EAGeR, Effects of Aspirin in Gestation and Reproduction; FOR, fecundability odds ratio; hCG, human chorionic gonadotropin; LDA, low-dose aspirin; LOD, limit of detection; PCOS, polycystic ovary syndrome; T, tertile.
Contributor Information
Alexandra C Purdue-Smithe, Division of Women's Health, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA.
Keewan Kim, Epidemiology Branch, Division of Intramural Population Health Research, Eunice Kennedy Shriver National Institute of Child Health and Human Development, NIH, Bethesda, MD, USA.
Karen C Schliep, Department of Obstetrics and Gynecology, University of Utah, Salt Lake City, UT, USA.
Elizabeth A DeVilbiss, Epidemiology Branch, Division of Intramural Population Health Research, Eunice Kennedy Shriver National Institute of Child Health and Human Development, NIH, Bethesda, MD, USA.
Stefanie N Hinkle, Department of Biostatistics, Epidemiology and Informatics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA.
Aijun Ye, Epidemiology Branch, Division of Intramural Population Health Research, Eunice Kennedy Shriver National Institute of Child Health and Human Development, NIH, Bethesda, MD, USA.
Neil J Perkins, Biostatistics and Bioinformatics Branch, Division of Intramural Population Health Research, Eunice Kennedy Shriver National Institute of Child Health and Human Development, NIH, Bethesda, MD, USA.
Lindsey A Sjaarda, Epidemiology Branch, Division of Intramural Population Health Research, Eunice Kennedy Shriver National Institute of Child Health and Human Development, NIH, Bethesda, MD, USA.
Robert M Silver, Department of Obstetrics and Gynecology, University of Utah, Salt Lake City, UT, USA.
Enrique F Schisterman, Department of Biostatistics, Epidemiology and Informatics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA.
Sunni L Mumford, Department of Biostatistics, Epidemiology and Informatics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA.
Data Availability
Data described in the article, code book, and analytic code will be made available upon reasonable request to the corresponding author.
References
- 1. Fulgoni VL III, Keast DR, Lieberman HR. Trends in intake and sources of caffeine in the diets of US adults: 2001–2010. Am J Clin Nutr. 2015;101(5):1081–7. [DOI] [PubMed] [Google Scholar]
- 2. ACOG Committee Opinion no. 462: moderate caffeine consumption during pregnancy. Obstet Gynecol. 2010;116(2 Pt 1):467–8. [DOI] [PubMed] [Google Scholar]
- 3. Carl J, Hill DA. Preconception counseling: make it part of the annual exam: addressing lifestyle issues, managing chronic conditions, updating immunizations, and screening for genetic disorders before pregnancy. pay big dividends in positive outcomes. J Fam Pract. 2009;58(6):307–14. [PubMed] [Google Scholar]
- 4. Lucero J, Harlow BL, Barbieri RL, Sluss P, Cramer DW. Early follicular phase hormone levels in relation to patterns of alcohol, tobacco, and coffee use. Fertil Steril. 2001;76(4):723–9. [DOI] [PubMed] [Google Scholar]
- 5. Miao Y-L, Shi L-H, Lei Z-L, Huang J-C, Yang J-W, Ouyang Y-C, Sun Q-Y, Chen D-Y. Effects of caffeine on in vivo and in vitro oocyte maturation in mice. Theriogenology. 2007;68(4):640–5. [DOI] [PubMed] [Google Scholar]
- 6. Kotsopoulos J, Eliassen AH, Missmer SA, Hankinson SE, Tworoger SS. Relationship between caffeine intake and plasma sex hormone concentrations in premenopausal and postmenopausal women. Cancer. 2009;115(12):2765–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Qian J, Zhang Y, Qu Y, Zhang L, Shi J, Zhang X, Liu S, Kim BH, Hwang SJ, Zhou Tet al. Caffeine consumption during early pregnancy impairs oviductal embryo transport, embryonic development and uterine receptivity in mice. Biol Reprod. 2018;99(6):1266–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Schliep KC, Schisterman EF, Wactawski-Wende J, Perkins NJ, Radin RG, Zarek SM, Mitchell EM, Sjaarda LA, Mumford SL. Serum caffeine and paraxanthine concentrations and menstrual cycle function: correlations with beverage intakes and associations with race, reproductive hormones, and anovulation in the BioCycle Study. Am J Clin Nutr. 2016;104(1):155–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Fenster L, Quale C, Waller K, Windham GC, Elkin EP, Benowitz N, Swan SH. Caffeine consumption and menstrual function. Am J Epidemiol. 1999;149(6):550–7. [DOI] [PubMed] [Google Scholar]
- 10. Schliep KC, Schisterman EF, Mumford SL, Pollack AZ, Zhang C, Ye A, Stanford JB, Hammoud AO, Porucznik CA, Wactawski-Wende J. Caffeinated beverage intake and reproductive hormones among premenopausal women in the BioCycle Study. Am J Clin Nutr. 2012;95(2):488–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Wesselink AK, Wise LA, Rothman KJ, Hahn KA, Mikkelsen EM, Mahalingaiah S, Hatch EE. Caffeine and caffeinated beverage consumption and fecundability in a preconception cohort. Reprod Toxicol. 2016;62:39–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Chavarro JE, Rich-Edwards JW, Rosner BA, Willett WC. Caffeinated and alcoholic beverage intake in relation to ovulatory disorder infertility. Epidemiology. 2009;20(3):374–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Hakim RB, Gray RH, Zacur H. Alcohol and caffeine consumption and decreased fertility. Fertil Steril. 1998;70(4):632–7. [DOI] [PubMed] [Google Scholar]
- 14. Hatch EE, Wise LA, Mikkelsen EM, Christensen T, Riis AH, Sørensen HT, Rothman KJ. Caffeinated beverage and soda consumption and time to pregnancy. Epidemiology. 2012;23(3):393–401. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Caan B, Quesenberry CP Jr, Coates AO. Differences in fertility associated with caffeinated beverage consumption. Am J Public Health. 1998;88(2):270–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Jensen TK, Henriksen TB, Hjollund NH, Scheike T, Kolstad H, Giwercman A, Ernst E, Bonde JP, Skakkebaek NE, Olsen J. Caffeine intake and fecundability: a follow-up study among 430 Danish couples planning their first pregnancy. Reprod Toxicol. 1998;12(3):289–95. [DOI] [PubMed] [Google Scholar]
- 17. Wilcox A, Weinberg C, Baird D. Caffeinated beverages and decreased fertility. Lancet. 1988;332(8626–8627):1453–6. [DOI] [PubMed] [Google Scholar]
- 18. Wilcox A, Weinberg C. Tea and fertility. Lancet. 1991;337(8750):1159–60. [DOI] [PubMed] [Google Scholar]
- 19. Florack EI, Zielhuis GA, Rolland R. Cigarette smoking, alcohol consumption, and caffeine intake and fecundability. Prev Med. 1994;23(2):175–80. [DOI] [PubMed] [Google Scholar]
- 20. McCusker RR, Goldberger BA, Cone EJ. Caffeine content of specialty coffees. J Anal Toxicol. 2003;27(7):520–2. [DOI] [PubMed] [Google Scholar]
- 21. Peck JD, Leviton A, Cowan LD. A review of the epidemiologic evidence concerning the reproductive health effects of caffeine consumption: a 2000–2009 update. Food Chem Toxicol. 2010;48(10):2549–76. [DOI] [PubMed] [Google Scholar]
- 22. Zaigler M, Rietbrock S, Szymanski J, Dericks-Tan JS, Staib AH, Fuhr U. Variation of CYP1A2-dependent caffeine metabolism during menstrual cycle in healthy women. Int J Clin Pharmacol Ther. 2000;38(5):235–44. [DOI] [PubMed] [Google Scholar]
- 23. Lelo A, Miners J, Robson R, Birkett D. Quantitative assessment of caffeine partial clearances in man. Br J Clin Pharmacol. 1986;22(2):183–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Klebanoff MA, Levine RJ, Dersimonian R, Clemens JD, Wilkins DG. Serum caffeine and paraxanthine as markers for reported caffeine intake in pregnancy. Ann Epidemiol. 1998;8(2):107–11. [DOI] [PubMed] [Google Scholar]
- 25. Schisterman EF, Silver RM, Lesher LL, Faraggi D, Wactawski-Wende J, Townsend JM, Lynch AM, Perkins NJ, Mumford SL, Galai N. Preconception low-dose aspirin and pregnancy outcomes: results from the EAGeR randomised trial. Lancet. 2014;384(9937):29–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Schisterman EF, Silver RM, Perkins NJ, Mumford SL, Whitcomb BW, Stanford JB, Lesher LL, Faraggi D, Wactawski-Wende J, Browne RW. A randomised trial to evaluate the effects of low-dose aspirin in gestation and reproduction: design and baseline characteristics. Paediatr Perinat Epidemiol. 2013;27(6):598–609. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Kalow W, Tang B-K. Use of caffeine metabolite ratios to explore CYP1A2 and xanthine oxidase activities. Clin Pharmacol Ther. 1991;50(5/1):508–19. [DOI] [PubMed] [Google Scholar]
- 28. Hornung RW, Reed LD. Estimation of average concentration in the presence of nondetectable values. Appl Occup Environ Hyg. 1990;5(1):46–51. [Google Scholar]
- 29. Lelo A, Birkett D, Robson R, Miners J. Comparative pharmacokinetics of caffeine and its primary demethylated metabolites paraxanthine, theobromine and theophylline in man. Br J Clin Pharmacol. 1986;22(2):177–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Klebanoff MA, Levine RJ, DerSimonian R, Clemens JD, Wilkins DG. Maternal serum paraxanthine, a caffeine metabolite, and the risk of spontaneous abortion. N Engl J Med. 1999;341(22):1639–44. [DOI] [PubMed] [Google Scholar]
- 31. Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Med Res Method. 2008;8(1):70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. van Buuren S. Multiple imputation of discrete and continuous data by fully conditional specification. Stat Methods Med Res. 2007;16(3):219–42. [DOI] [PubMed] [Google Scholar]
- 33. Brown CR, Jacob P 3rd, Wilson M, Benowitz NL. Changes in rate and pattern of caffeine metabolism after cigarette abstinence. Clin Pharmacol Ther. 1988;43(5):488–91. [DOI] [PubMed] [Google Scholar]
- 34. Hatch EE, Wesselink AK, Hahn KA, Michiel JJ, Mikkelsen EM, Sorensen HT, Rothman KJ, Wise LA. Intake of sugar-sweetened beverages and fecundability in a North American preconception cohort. Epidemiology. 2018;29(3):369–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Kohlmeier M. Xenobiotics. In: Kohlmeier Meditor. Nutrient metabolism. 2nd ed. San Diego, CA: Academic Press; 2015. p. 95–110. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data described in the article, code book, and analytic code will be made available upon reasonable request to the corresponding author.