Article InfoMaterials and methods
Study Design
From
January 2014 through August 2016, 359 women undergoing a fresh IVF
cycle at a tertiary university-affiliated hospital were recruited into a
study on environmental exposures and fertility. Cryopreserved cycles
were excluded from our analysis. The study was approved by our local
Institutional Review Board, and all patients signed informed consents.
Participants were enrolled during ovarian stimulation and followed
through one fresh IVF cycle. For the analysis of intermediate IVF
outcomes, exclusion criteria included women with missing embryology (n =
3) or exposure (n = 2) information, women who froze their oocytes (n =
11), women using egg donors (n = 1), and women missing information on
oocyte retrieval (n = 2). Thus, the final data set consisted of 340
women. For the analysis of clinical IVF outcomes, we further excluded
one woman who was lost to follow-up after identification of a clinical
pregnancy, bringing the final analytic sample to 339 women.
Exposure Assessment
Women
reported their usual intake of caffeinated and noncaffeinated beverages
on the first day of stimulation and/or on the day of oocyte retrieval.
The questionnaire specifically asked women, “Do you drink any of the
following 14 beverages: filtered coffee, instant coffee, boiled black
coffee, mud coffee, decaffeinated coffee, cappuccino, espresso,
caffeinated tea, herbal tea, chocolate drinks, caffeinated soda,
caffeinated diet sodas, noncaffeinated diet sodas, and energy drinks
and, if so, in what quantity (in cups).” Women were also provided with
information on converting common serving sizes to cups (e.g., 1 mug = 2
cups). Total caffeine intake was estimated by summing the caffeine
content for each specific beverage multiplied by their frequency of
intake. We assumed the following caffeine concentrations for each
caffeinated beverage: filtered coffee, 95 mg/cup; instant coffee,
63 mg/cup; boiled black and mud coffee, 115 mg/cup; decaffeinated
coffee, 2 mg/cup; cappuccino, 64 mg/cup; espresso, 64 mg/shot;
caffeinated tea, 26 mg/cup; chocolate drinks, 5 mg/cup; caffeinated
sodas, 16 mg/cup; and energy drinks, 111 mg/cup.
Covariate Assessment
Height and weight, measured at the start of the IVF cycle by a trained nurse, were used to calculate body mass index (BMI; kg/m2).
A woman's age, smoking status, number of previous pregnancies and
deliveries, duration of infertility, and IVF attempt number were
abstracted from patients' medical records. On the same questionnaire as
the one about the beverages, women also provided information on their
country/region of birth, years of education, smoking history, and field
of employment.
Outcome Assessment
Patients
were treated with controlled ovarian stimulation using one of three
protocols (GnRH antagonist, GnRH agonist suppressive protocol, or GnRH
agonist flare-up protocol) as clinically indicated. Patients were
monitored during gonadotropin stimulation for serum E2,
follicle size measurements and counts, and endometrial thickness through
2 days before oocyte retrieval. HCG was administered approximately
36 hours before the scheduled oocyte retrieval procedure to induce
oocyte maturation. Women received conventional insemination or
intracytoplasmic sperm injection (ICSI) as clinically indicated.
Embryologists classified oocytes as germinal vesicle, metaphase I,
metaphase II (MII), or degenerated. Embryologists determined
fertilization 16–18 hours after insemination as the number of oocytes
with two pronuclei. The resulting embryos were assessed for cell number,
symmetry, and fragmentation (25).
Top-quality embryos were considered to be embryos with 7–8 cells on day
3 (or in cases of day 2 transfer, 4 cells) and <10% fragmentation.
Positive β-hCG (i.e., successful implantation) was defined as a serum
β-hCG level >25 mIU/mL typically measured 14 days after oocyte
retrieval. Embryos were scheduled for transfer on day 3 in
non–preimplantation genetic diagnosis (PGD) patients. In cases for which
day 3 was a holiday, transfers were performed on day 2 (n = 12 cycles).
For PGD patients, embryos were biopsied on day 3 and transferred on day
4. Clinical pregnancy was defined as the presence of an intrauterine
gestational sac and fetal heartbeat confirmed by ultrasound by 7 weeks
of gestation, and live birth as the delivery of a live neonate on or
after 24 weeks of gestation. All clinical information was abstracted
from medical records.
Statistical Analysis
Women
were stratified into quartiles of total caffeine intake and categories
of beverage consumption based on the distribution of consumption in the
population. Descriptive statistics, calculated for demographic and
reproductive characteristics in the entire cohort and by quartile of
total caffeine intake, were presented as mean (SD) or number of women
(%). For continuous and categorical variables, analysis of variance
(ANOVA) and χ2 tests were used, respectively, to test for associations across categories of total caffeine intake.
To
evaluate the association between caffeine and beverage intake and
number of total oocytes, mature oocytes, fertilized oocytes, and
top-quality embryos (all count data), we used a multivariable Poisson
regression with log link. Adjusted marginal mean counts for each
quartile or category were obtained. For the clinical outcomes, we used a
multivariable logistic regression model to derive the adjusted
proportion of initiated cycles resulting in implantation, clinical
pregnancy, and live birth for each quartile or category. Risk of
pregnancy loss was evaluated among women with an implantation and was
defined as any loss of pregnancy before live birth. To test associations
between caffeine and beverage intake and pregnancy loss, we used a
Poisson regression with log link and present results as risk ratios (95%
confidence intervals [CIs]). Tests for trend were conducted across
quartiles or categories using the median level of intake in each
category as a continuous variable in the regression models.
Confounding
was evaluated using prior knowledge and descriptive statistics from our
cohort. Variables retained in the final multivariable models were age,
BMI, smoking status, and country of origin. Specific beverages also were
further adjusted for coffee, caffeinated tea, herbal tea, sugared soda,
and energy drink intakes. To test for potential effect modification by
PGD, we included a cross-product term in the final multivariable model.
All analyses were conducted using SAS Software package 9.4.
Results
Between
January 2014 and August 2016, 340 women who underwent ovarian
stimulation for a fresh IVF cycle completed questionnaires on their
recent consumption of coffee, tea, hot chocolate, soda, and energy
drinks. The women were 31.5 ± 4.0 years old and had a BMI of 23.4 ±
4.5 kg/m2 (Table 1).
Their primary reason for undergoing IVF was PGD for autosomal recessive
disorders (n = 129 women, 37.9%) followed by male factor infertility
(33.8%), unexplained infertility (16.8%), and known female factors
(i.e., anovulation or endometriosis, 11.5%). The majority (62.4%) of
women were undergoing their first IVF cycle, 20.9% were undergoing their
second IVF cycle, and 16.8% were undergoing their third IVF cycle.
There were no differences regarding the number of embryos transferred,
the number of top-quality embryos transferred, and the day of transfer
across the quartiles of caffeine intake. The mean ± SD caffeine intake
of women in our cohort was 163.5 ± 125.3 mg/day, which corresponds to
about 2.5 cups of instant coffee per day. Women with higher caffeine
intake were, on average, slightly older and more likely to be current
smokers (Table 1); all other demographic and reproductive characteristics were similar across quartiles of caffeine intake.
| Characteristic | Total cohort | Caffeine intake (range, mg/d) | P valuea | |||
|---|---|---|---|---|---|---|
| Q1 (0–70) | Q2 (71–141) | Q3 (142–230) | Q4 (231–816) | |||
| Women, n | 340 | 84 | 86 | 87 | 84 | |
| Age, y | 31.5 (4.0) | 31.2 (4.2) | 30.7 (4.0) | 31.9 (4.3) | 32.2 (3.4) | .06 |
| BMI, kg/m2 | 23.4 (4.5) | 23.2 (4.9) | 23.1 (3.7) | 23.3 (4.9) | 24.0 (4.5) | .56 |
| Current smoker, n (%) | 61 (17.9) | 5 (6.0) | 19 (22.1) | 14 (16.1) | 23 (27.7) | .002 |
| Native born, n (%) | 256 (75.3) | 68 (81.0) | 70 (81.4) | 60 (69.0) | 58 (69.9) | .10 |
| Years of education | 15.4 (2.7) | 15.1 (2.8) | 15.1 (2.8) | 15.8 (2.6) | 15.5 (2.9) | .33 |
| Type of work, n (%) | .72 | |||||
| Education | 59 (17.4) | 16 (19.1) | 11 (12.8) | 20 (23.0) | 12 (14.5) | |
| Healthcare | 37 (10.9) | 8 (9.5) | 11 (12.8) | 11 (12.6) | 7 (8.4) | |
| Office work | 161 (47.4) | 38 (45.2) | 45 (52.3) | 37 (42.5) | 41 (49.4) | |
| Other | 83 (24.4) | 22 (26.2) | 19 (22.1) | 19 (21.8) | 23 (27.7) | |
| Gravidity, n (%) | .25 | |||||
| 0 | 145 (42.7) | 35 (41.7) | 35 (40.7) | 44 (50.6) | 31 (37.4) | |
| 1 | 102 (30.0) | 24 (28.6) | 31 (36.1) | 25 (28.7) | 22 (26.5) | |
| ≥2 | 93 (27.4) | 25 (29.8) | 20 (23.3) | 18 (20.7) | 30 (36.1) | |
| Parity, n (%) | .61 | |||||
| 0 | 208 (61.2) | 50 (59.6) | 54 (62.8) | 58 (66.7) | 46 (55.4) | |
| 1 | 96 (28.2) | 23 (27.4) | 25 (29.1) | 23 (26.4) | 25 (30.1) | |
| ≥2 | 36 (10.6) | 11 (13.1) | 7 (8.1) | 6 (6.9) | 12 (14.5) | |
| Years of infertility | 1.4 (1.8) | 1.3 (1.7) | 1.3 (1.5) | 1.3 (1.8) | 1.7 (2.2) | .45 |
| Previous IVF attempts, n (%) | .08 | |||||
| 0 | 212 (62.4) | 48 (57.1) | 56 (65.1) | 62 (71.3) | 46 (55.4) | |
| 1 | 71 (20.9) | 25 (29.8) | 17 (19.8) | 12 (13.8) | 17 (20.5) | |
| ≥2 | 57 (16.8) | 11 (13.1) | 13 (15.1) | 13 (14.9) | 20 (24.1) | |
| Infertility diagnosis, n (%) | .38 | |||||
| Male | 115 (33.8) | 30 (35.7) | 32 (37.2) | 29 (33.3) | 24 (28.9) | |
| Female | 39 (11.5) | 10 (11.9) | 6 (7.0) | 10 (11.5) | 13 (15.7) | |
| Unexplained | 57 (16.8) | 9 (10.7) | 12 (14.0) | 17 (19.5) | 19 (22.9) | |
| PGD | 129 (37.9) | 35 (41.7) | 36 (41.9) | 31 (35.6) | 27 (32.5) | |
| Antagonist protocol, n (%) | 330 (97.1) | 80 (95.2) | 85 (98.8) | 84 (96.6) | 81 (97.6) | .55 |
| FSH ampules | 25.8 (17.8) | 25.6 (20.8) | 23.7 (11.1) | 26.5 (16.1) | 27.4 (21.7) | .57 |
| ET day, n (%) | .16 | |||||
| 2 | 12 (3.5) | 3 (3.6) | 6 (7.0) | 0 (0.0) | 3 (3.6) | |
| 3 | 199 (58.5) | 46 (54.8) | 44 (51.2) | 56 (64.4) | 53 (63.9) | |
| 4b | 129 (37.9) | 35 (41.7) | 36 (41.9) | 31 (35.6) | 27 (32.5) | |
| Embryos transferred, n (%) | .63 | |||||
| 0 | 53 (15.6) | 11 (13.1) | 16 (18.6) | 18 (20.7) | 8 (9.6) | |
| 1 | 114 (33.5) | 29 (34.5) | 30 (34.8) | 29 (33.3) | 26 (31.3) | |
| 2 | 159 (46.8) | 40 (47.6) | 38 (44.2) | 37 (42.5) | 44 (53.0) | |
| 3–4 | 14 (4.2) | 4 (4.8) | 2 (2.3) | 3 (3.5) | 5 (6.0) | |
| Top-quality embryos transferred (≥1), n (%) | 236 (69.4) | 64 (76.2) | 57 (66.3) | 56 (64.4) | 59 (71.1) | .34 |
Note: Data
are presented as mean (SD) or number of women (%) unless otherwise
specified. BMI = body mass index; ET = embryo transfer; FSH =
follicle-stimulating hormone; IVF = in vitro fertilization; PGD =
preimplantation genetic diagnosis; Q = quartile.
aDifferences across categories were tested using an ANOVA test for continuous variables and a χ2 test for categorical variables.
bAll day 4 ETs were PGD cycles.
There
were no associations between total caffeine intake and number of total
or mature oocytes, fertilized oocytes, or top-quality embryos (Table 2).
Similarly, total coffee intake was not associated with these outcomes.
Higher consumption of instant coffee was associated with a slightly
lower number of fertilized oocytes. However, none of the other coffee
beverages were related to this or other oocyte or embryo outcomes. On
the other hand, women with higher caffeinated tea intake had a lower
number of total (P trend = .001) and mature oocytes (P trend = .003) and a lower number of fertilized oocytes (P trend = .05). In contrast, herbal tea intake, was associated with higher total and mature oocyte yield (P trend = .001 and .02, respectively).
| Beverage intake | n | Adjusted mean (95% CI)a | |||
|---|---|---|---|---|---|
| Total oocytes | Mature oocytesb | Fertilized oocytesb | Top-quality embryos | ||
| Total caffeine intake (mg/d) | |||||
| Q1 (0–70) | 84 | 10.7 (10.0, 11.4) | 8.4 (7.8, 9.1) | 6.0 (5.5, 6.6) | 2.7 (2.4, 3.1) |
| Q2 (71–141) | 86 | 10.2 (9.6, 11.0) | 7.8 (7.2, 8.4) | 5.2 (4.8, 5.8)c | 2.4 (2.1, 2.8) |
| Q3 (142–230) | 87 | 9.8 (9.1, 10.4) | 7.5 (6.9, 8.1)c | 5.5 (5.1, 6.0) | 2.4 (2.1, 2.7) |
| Q4 (231–816) | 83 | 9.9 (9.3, 10.7) | 7.7 (7.1, 8.3) | 5.4 (5.0, 6.0) | 2.6 (2.2, 2.9) |
| P trendd | .13 | .11 | .21 | .69 | |
| Total coffee, cups/d | |||||
| 0 | 70 | 11.1 (10.2, 11.8) | 8.8 (8.1, 9.5) | 6.4 (5.8, 7.0) | 2.6 (2.3, 3.1) |
| 0.1–1 | 70 | 8.8 (8.1, 9.5)c | 6.8 (6.3, 7.5)c | 4.8 (4.4, 5.4)c | 2.4 (2.0, 2.8) |
| 1.1–2 | 92 | 10.2 (9.6, 10.9) | 8.1 (7.5, 8.7) | 5.4 (4.9, 5.9)c | 2.4 (2.1, 2.8) |
| 2.1–3 | 52 | 10.2 (9.3, 11.1) | 7.1 (6.4, 7.8)c | 5.5 (4.9, 6.2) | 2.5 (2.1, 3.0) |
| 3.1–10 | 56 | 10.0 (9.2, 10.9) | 7.8 (7.1, 8.6) | 5.5 (4.9, 6.2)c | 2.5 (2.1, 3.0) |
| P trend | .64 | .14 | .24 | .79 | |
| Instant coffee, cups/d | |||||
| 0 | 115 | 10.8 (10.2, 11.5) | 8.4 (7.8, 8.9) | 6.1 (5.7, 6.6) | 2.8 (2.5, 3.1) |
| 0.1–1 | 73 | 9.6 (8.9, 10.3)c | 7.5 (6.9, 8.1)c | 5.1 (4.6, 5.7)c | 2.5 (2.1, 2.9) |
| 1.1–2 | 85 | 9.3 (8.7, 10.0)c | 7.2 (6.7, 7.8)c | 5.0 (4.6, 5.5)c | 2.1 (1.8, 2.4)c |
| 2.1–10 | 67 | 10.1 (9.4, 10.9) | 7.8 (7.1, 8.5) | 5.5 (5.0, 6.1) | 2.6 (2.2, 3.0) |
| P trend | .06 | .10 | .04 | .18 | |
| Filtered coffee, cups/d | |||||
| 0 | 325 | 10.0 (9.7, 10.4) | 7.7 (7.4, 8.1) | 5.5 (5.3, 5.8) | 2.5 (2.3, 2.7) |
| 0.1–2 | 15 | 10.2 (8.6, 12.0) | 8.3 (6.9, 10.0) | 5.6 (4.4, 7.0) | 2.0 (1.4, 2.9) |
| Boiled or mud coffee, cups/d | |||||
| 0 | 317 | 9.9 (9.6, 10.3) | 7.7 (7.4, 8.0) | 5.5 (5.3, 5.8) | 2.5 (2.3, 2.6) |
| 0.1–5 | 23 | 11.3 (10.0, 12.8)c | 8.4 (7.3, 9.7) | 5.6 (4.7, 6.6) | 2.8 (2.2, 3.6) |
| Cappuccino, cups/d | |||||
| 0 | 281 | 9.9 (9.6, 10.3) | 7.8 (7.5, 8.1) | 5.5 (5.3, 5.8) | 2.5 (2.3, 2.7) |
| 0.1–4 | 59 | 10.4 (9.6, 11.3) | 7.6 (6.9, 8.3) | 5.4 (4.8, 6.1) | 2.5 (2.1, 3.0) |
| Espresso, cups/d | |||||
| 0 | 315 | 10.1 (9.7, 10.4) | 7.8 (7.5, 8.1) | 5.5 (5.2, 5.8) | 2.5 (2.3, 2.6) |
| 0.1–4 | 25 | 9.4 (8.3, 10.7) | 7.1 (6.1, 8.2) | 5.9 (5.0, 6.9) | 2.7 (2.1, 3.5) |
| Decaffeinated coffee, cups/d | |||||
| 0 | 324 | 10.0 (9.7, 10.4) | 7.8 (7.5, 8.1) | 5.5 (5.3, 5.8) | 2.5 (2.3, 2.7) |
| 0.1–3 | 16 | 9.9 (8.5, 11.5) | 6.8 (5.7, 8.2) | 4.9 (3.9, 6.1) | 2.5 (1.8, 3.4) |
| Caffeinated tea, cups/d | |||||
| 0 | 168 | 10.3 (9.8, 10.8) | 8.0 (7.5, 8.4) | 5.8 (5.4, 6.2) | 2.6 (2.3, 2.8) |
| 0.1–1 | 111 | 10.1 (9.5, 10.7) | 7.9 (7.4, 8.5) | 5.7 (5.3, 6.2) | 2.5 (2.2, 2.8) |
| 1.1–8 | 61 | 9.2 (8.5, 10.0)c | 6.9 (6.3, 7.6)c | 4.5 (4.0, 5.0)c | 2.2 (1.9, 2.6)c |
| P trend | .03 | .03 | .002 | .15 | |
| Herbal tea, cups/d | |||||
| 0 | 242 | 9.7 (9.3, 10.1) | 7.6 (7.2, 7.9) | 5.4 (5.1, 5.7) | 2.5 (2.3, 2.7) |
| 0.1–1 | 66 | 10.5 (9.7, 11.3) | 8.0 (7.3, 8.7) | 5.6 (5.1, 6.3) | 2.5 (2.1, 2.9) |
| 1.1–6 | 32 | 11.6 (10.4, 12.8)c | 8.8 (7.9, 10.0)c | 6.1 (5.3, 7.0) | 2.1 (1.7, 2.7) |
| P trend | .001 | .02 | .11 | .23 | |
| Hot chocolate, cups/d | |||||
| 0 | 295 | 9.9 (9.6, 10.3) | 7.7 (7.4, 8.0) | 5.4 (5.2, 5.7) | 2.4 (2.2, 2.6) |
| 0.1–5 | 45 | 10.7 (9.8, 11.7) | 8.1 (7.3, 9.0) | 6.0 (5.3, 6.8) | 3.0 (2.6, 3.6)c |
| Sugared soda, cups/d | |||||
| 0 | 243 | 10.4 (10.0, 10.9) | 8.1 (7.8, 8.5) | 5.8 (5.5, 6.1) | 2.6 (2.4, 2.8) |
| 0.1–1 | 60 | 8.9 (8.2, 9.7)c | 6.9 (6.3, 7.6)c | 4.7 (4.2, 5.3)c | 2.4 (2.0, 2.8) |
| 1.1–10 | 37 | 9.3 (8.3, 10.3)c | 6.9 (6.1, 7.8)c | 5.2 (4.5, 6.0) | 2.0 (1.6, 2.5)c |
| P trend | .002 | <.001 | .01 | .03 | |
| Diet soda, cups/d | |||||
| 0 | 249 | 10.1 (9.7, 10.5) | 7.7 (7.4, 8.1) | 5.5 (5.2, 5.8) | 2.4 (2.2, 2.6) |
| 0.1–1 | 40 | 10.0 (9.1, 11.1) | 8.2 (7.3, 9.2) | 5.7 (5.0, 6.5) | 3.0 (2.5, 3.6)c |
| 1.1–15 | 51 | 9.7 (8.8, 10.6) | 7.7 (6.9, 8.5) | 5.6 (4.9, 6.3) | 2.3 (1.9, 2.8) |
| P trend | .42 | .78 | .66 | .82 | |
| Energy drink, cups/d | |||||
| 0 | 320 | 10.4 (10.0, 10.7) | 8.0 (7.7, 8.3) | 5.6 (5.4, 5.9) | 2.5 (2.4, 2.7) |
| 0.1–2 | 20 | 6.0 (5.0, 7.2)c | 5.1 (4.2, 6.3)c | 4.0 (3.2, 5.0)c | 1.9 (1.4, 2.7) |
Note: Models were run using Poisson regression with log link. All data are presented as adjusted mean counts. Q = quartile.
aTotal
caffeine and caffeinated beverages were adjusted for intake age, BMI,
smoking status, and country of origin. Specific beverages were further
adjusted for coffee, caffeinated tea, herbal tea, sugared soda, and
energy drink intake.
bFor
one patient, the embryologist dropped a dish with oocytes, and as a
result the exact number of mature/fertilized oocytes is missing.
cP <.05 for that specific category compared with lowest category or nondrinkers.
dP trend was calculated by using the median value in each category as a continuous variable in the multivariable model.
Intake
of sugared soda, but not of diet soda, was associated with lower total
and mature oocytes, fertilized oocytes, and top-quality embryos. Women
who consumed sugared soda had, on average, 1.1 fewer oocytes retrieved,
1.2 fewer mature oocytes retrieved, and 0.6 fewer fertilized oocytes
compared with women who did not consume sugared soda (P for
trend = .002, <.001, and .01, respectively). Similarly, inverse
associations were seen for these same outcomes comparing energy drink
consumers with nonconsumers (P value for comparisons ≤.001, <.001, and .005, respectively).
Among
the 339 women who were included in analysis of clinical outcomes, 283
(83.5%) had an ET, 116 (34.2%) had positive β-hCG 14 days after ET, 102
(30.1%) had a clinical pregnancy, and 83 (24.5%) had a live birth. Total
caffeine intake was not associated with probability of positive β-hCG,
clinical pregnancy, or live birth after IVF (Table 3).
Of all the beverages examined, only sugared soda intake was related to
clinical outcomes. Higher intake of sugared sodas was inversely
associated with clinical pregnancy (P-trend = .01) and live birth (P-trend
= .01). Specifically, compared with women who reported no sugared soda
intake (0 cups/day), the adjusted difference in percent of cycles
resulting in live birth for women consuming 0.1–1 cups/day and >1
cup/day were −12% and −16%, respectively (P trend = .01). When the analysis was restricted to only women who underwent ET, this association strengthened (Supplemental Table 1).
The adjusted differences in percent of transfers resulting in live
birth for women consuming 0.1–1 cups/day and >1 cup/day compared with
nonconsumers were −15% and −19%, respectively (P trend = .01).
Among women with implantation, the risk of pregnancy loss before
delivery was 3.51 (95% CI, 1.46, 8.45) times higher among women
consuming >1 cup/day of sugared soda compared with women consuming no
sugared soda (P trend = .02).
| Beverage intake (mg/d) | n | Adjusted proportions (95% CI) | ||
|---|---|---|---|---|
| Positive β-hCG | Clinical pregnancy | Live birth | ||
| Total caffeine intakea | ||||
| Q1 (0–70) | 84 | 0.36 (0.26, 0.47) | 0.33 (0.24, 0.44) | 0.33 (0.24, 0.44) |
| Q2 (71–141) | 86 | 0.30 (0.22, 0.41) | 0.25 (0.17, 0.35) | 0.20 (0.13, 0.30) |
| Q3 (142–230) | 86 | 0.35 (0.26, 0.46) | 0.31 (0.22, 0.41) | 0.23 (0.15, 0.33) |
| Q4 (231–816) | 83 | 0.35 (0.25, 0.46) | 0.31 (0.22, 0.42) | 0.22 (0.14, 0.32) |
| P trendb | .47 | .94 | .18 | |
| Total coffee, cups/d | ||||
| 0 | 70 | 0.35 (0.24, 0.47) | 0.33 (0.23, 0.45) | 0.32 (0.22, 0.44) |
| 0.1–1 | 70 | 0.33 (0.23, 0.45) | 0.24 (0.15, 0.36) | 0.19 (0.12, 0.31) |
| 1.1–2 | 92 | 0.33 (0.24, 0.43) | 0.28 (0.20, 0.38) | 0.23 (0.15, 0.33) |
| 2.1–3 | 51 | 0.38 (0.25, 0.52) | 0.31 (0.20, 0.46) | 0.22 (0.12, 0.35) |
| 3.1–10 | 56 | 0.32 (0.21, 0.45) | 0.31 (0.20, 0.45) | 0.22 (0.13, 0.35) |
| P trend | .89 | .89 | .32 | |
| Instant coffee, cups/d | ||||
| 0 | 115 | 0.32 (0.24, 0.42) | 0.28 (0.21, 0.37) | 0.24 (0.17, 0.33) |
| 0.1–1 | 73 | 0.37 (0.27, 0.49) | 0.32 (0.22, 0.44) | 0.25 (0.16, 0.36) |
| 1.1–2 | 84 | 0.36 (0.26, 0.47) | 0.29 (0.21, 0.40) | 0.23 (0.15, 0.34) |
| 2.1–10 | 67 | 0.30 (0.20, 0.42) | 0.28 (0.19, 0.40) | 0.23 (0.14, 0.34) |
| P trend | .78 | .96 | .82 | |
| Filtered coffee, cups/d | ||||
| 0 | 324 | 0.33 (0.28, 0.38) | 0.29 (0.24, 0.34) | 0.23 (0.19, 0.28) |
| 0.1–2 | 15 | 0.54 (0.30, 0.77) | 0.46 (0.23, 0.71) | 0.32 (0.13, 0.59) |
| Boiled or mud coffee, cups/d | ||||
| 0 | 316 | 0.34 (0.29, 0.39) | 0.29 (0.24, 0.35) | 0.24 (0.19, 0.29) |
| 0.1–5 | 23 | 0.34 (0.18, 0.56) | 0.29 (0.14, 0.51) | 0.21 (0.09, 0.42) |
| Cappuccino, cups/d | ||||
| 0 | 281 | 0.34 (0.28, 0.40) | 0.30 (0.24, 0.35) | 0.24 (0.19, 0.29) |
| 0.1–4 | 58 | 0.34 (0.22, 0.47) | 0.28 (0.18, 0.41) | 0.22 (0.13, 0.35) |
| Espresso, cups/d | ||||
| 0 | 314 | 0.33 (0.28, 0.39) | 0.29 (0.24, 0.34) | 0.24 (0.19, 0.29) |
| 0.1–4 | 25 | 0.42 (0.24, 0.62) | 0.36 (0.20, 0.57) | 0.24 (0.12, 0.45) |
| Decaffeinated coffee, cups/d | ||||
| 0 | 323 | 0.34 (0.29, 0.39) | 0.29 (0.25, 0.35) | 0.24 (0.20, 0.29) |
| 0.1–3 | 16 | 0.33 (0.15, 0.59) | 0.26 (0.10, 0.53) | 0.12 (0.03, 0.39) |
| Caffeinated tea, cups/d | ||||
| 0 | 168 | 0.36 (0.28, 0.43) | 0.30 (0.23, 0.37) | 0.22 (0.17, 0.30) |
| 0.1–1 | 111 | 0.29 (0.21, 0.38) | 0.26 (0.18, 0.35) | 0.22 (0.15, 0.31) |
| 1.1–8 | 60 | 0.38 (0.26, 0.51) | 0.35 (0.23, 0.48) | 0.30 (0.19, 0.43) |
| P trend | .98 | .70 | .35 | |
| Herbal tea, cups/d | ||||
| 0 | 242 | 0.37 (0.31, 0.43) | 0.32 (0.26, 0.38) | 0.26 (0.21, 0.32) |
| 0.1–1 | 65 | 0.25 (0.16, 0.37) | 0.21 (0.13, 0.33) | 0.15 (0.08, 0.26) |
| 1.1–6 | 32 | 0.31 (0.17, 0.48) | 0.27 (0.14, 0.44) | 0.23 (0.12, 0.40) |
| P trend | .18 | .20 | .24 | |
| Hot chocolate, cups/d | ||||
| 0 | 294 | 0.33 (0.28, 0.39) | 0.28 (0.23, 0.33) | 0.22 (0.17, 0.27) |
| 0.1–5 | 45 | 0.39 (0.25, 0.54) | 0.40 (0.26, 0.55) | 0.36 (0.23, 0.52) |
| Sugared soda, cups/d | ||||
| 0 | 242 | 0.38 (0.32, 0.44) | 0.35 (0.29, 0.41) | 0.28 (0.23 0.34) |
| 0.1–1 | 60 | 0.21 (0.13, 0.33)c | 0.16 (0.09,0 0.28)c | 0.16 (0.09 0.27) |
| 1.1–10 | 37 | 0.28 (0.16, 0.45) | 0.20 (0.10, 0.37) | 0.12 (0.05 0.27)c |
| P trend | .06 | .01 | .01 | |
| Diet soda, cups/d | ||||
| 0 | 248 | 0.33 (0.28, 0.40) | 0.29 (0.24, 0.35) | 0.22 (0.17, 0.28) |
| 0.1–1 | 40 | 0.41 (0.26, 0.57) | 0.38 (0.24, 0.55) | 0.32 (0.19, 0.48) |
| 1.1–15 | 51 | 0.31 (0.19, 0.46) | 0.23 (0.14, 0.37) | 0.23 (0.13, 0.38) |
| P trend | .97 | .68 | .63 | |
| Energy drink, cups/d | ||||
| 0 | 319 | 0.34 (0.29, 0.39) | 0.29 (0.24, 0.35) | 0.24 (0.19, 0.29) |
| 0.1–2 | 20 | 0.34 (0.16, 0.59) | 0.31 (0.13, 0.56) | 0.23 (0.09, 0.49) |
Note: Models were run using logistic regression. All data are presented as adjusted mean proportions. Q = quartile.
aTotal
caffeine and caffeinated beverages were adjusted for intake age, BMI,
smoking status, and country of origin. Specific beverages were further
adjusted for coffee, caffeinated tea, herbal tea, sugared soda, and
energy drink intake.
bP trend was calculated by using the median value in each category as a continuous variable in the multivariable model.
cP value is <.05 for that specific category compared with lowest category or nondrinkers.
There was no evidence of effect modification by PGD for the association between caffeine and sugared soda and live birth (P value for interaction = .43 and .85, respectively).
Discussion
In
our prospective cohort of women undergoing IVF we found that higher
preconception intake of sugared sodas was associated with a lower number
of total and mature oocytes retrieved, a lower number of fertilized
oocytes, and a lower proportion of cycles resulting in clinical
pregnancy and live birth. Intake of caffeinated tea and energy drinks
was also associated with poorer oocyte and embryo outcomes, but these
associations did not translate into poorer clinical outcomes. Contrary
to our hypothesis, however, intake of caffeine, coffee, specific
caffeinated beverages, or diet sodas failed to show consistent
associations with any of the outcomes examined.
Regarding
our initial hypotheses that caffeine and/or the increased glycemic
impact of sugar in beverages might alter IVF outcomes, the one most
supported by our data is the second one pertaining to sugar. This
proposed mechanism not only fits with the sugar soda findings, but also
explains the results regarding energy drinks, which often contain high
levels of either sucrose or high-fructose corn syrup. Moreover, it is
possible that a high proportion of women add sugar to their caffeinated
tea, which might explain why we also found a similar inverse association
with this beverage and intermediate IVF outcomes.
Consumption
of sugared soda has been linked to abnormal markers of glycemic control
such as insulin resistance, metabolic syndrome, and type 2 diabetes (26, 27, 28, 29).
Insulin resistance is a situation in which the responsiveness of the
body to the hormone insulin is diminished, resulting in metabolic
dysregulation (30, 31).
Insulin resistance can alter the maternal metabolic environment and
follicular fluid microenvironment, leading to lower quality oocytes and
embryos (9, 11, 32). In mice, this condition leads to altered mitochondrial function and abnormal spindle formation (33).
Interestingly, another marker of abnormal glycemic control,
glycosylated hemoglobin (HbA1C), which increases in diabetes, was
previously reported to be associated with reduced fertility (34).
We
found that the proportion of cycles resulting in clinical pregnancy and
live birth was lower among women who consumed higher levels of sugared
sodas compared with those who drank lower levels of these beverages.
However, to the best of our knowledge, the only other study that has
investigated the association between female intake of sugared sodas and
IVF outcomes found no association between usual intake of soda in the
year before infertility treatment and probability of live birth (24).
However, this study did not distinguish between diet and sugared soda
and was limited by a low number of high soda consumers (only 8.3%
reported consuming one or more serving of any type of soda, while in our
study 27.9% of women reported consuming ≥1 cup of any type of soda).
This previous study also only focused on subfertile women undergoing ART
for infertility treatment as compared with our cohort, which also
included women undergoing ART for PGD. Taken together, these differences
in study population could help explain some of our seemingly discrepant
findings. In contrast, and more in line with our current findings, a
previous study among women conceiving spontaneously showed that women
who consumed high levels of sugar sweetened beverages had decreased
fecundity. Specifically, among 3,628 women planning a pregnancy in
Denmark, the adjusted fecundability ratios were 0.89 (0.80–0.98), 0.85
(0.71–1.02), 0.84 (0.57–1.25), and 0.48 (0.21–1.13) for <1, 1, 2, and
3+ servings per day, respectively, compared with none (16).
This inverse association with fecundity, however, was not confirmed in a
more recent prospective study among 2,135 North American pregnancy
planners (11). Moreover, in this same cohort of women, there was no link between prepregnancy soda consumption and pregnancy loss (35).
In
contrast with our hypothesis, we failed to find an association between
caffeine consumption and number of total, mature, or fertilized oocytes
or embryo quality. In addition, preconception caffeine intake was not
associated with implantation, clinical pregnancy, or live birth. Our
results are in line with a previous study of 221 women undergoing IVF,
in which no associations were observed between recent caffeine intake
and IVF outcomes (23).
A lack of association between current consumption of caffeine and
number of oocytes retrieved, fertilization rate, implantation rate, or
live-birth rate was also reported by Al-Saleh et al. and Choi et al.,
who evaluated 2,474 women undergoing 4,716 IVF treatment cycles from
1994 to 2003 (21, 22).
A third study among 619 women undergoing IVF also found no significant
associations between current caffeine intake and pregnancy rate, despite
having a median caffeine intake of 455.8 mg/day (21).
Finally, the most recent study on caffeine intake and IVF outcomes (n =
300 women, 493 IVF cycles) found no association between usual caffeine
intake over the previous year and intermediate or clinical outcomes of
IVF (24).
Our
study was subject to some limitations. Although women reported their
intake of specific beverages (in cups), there is still likely error in
their self-report as information on serving size and amount consumed was
limited. Furthermore, the brewing method for the specific coffee types
was not collected, which could impact the quantity of caffeine in a
given serving size. However, misclassification of beverage intake is
unlikely to be linked with IVF outcome given the prospective nature of
our study and would therefore tend to attenuate our associations toward
the null. We also lacked information regarding beverage habits during
the pregnancy, and it is possible that patients change their drinking
beverage habits throughout pregnancy, especially by decreasing their
caffeine and diet soda consumption. Beverage consumption could also be
correlated with other diet and lifestyle factors as well as dietary
consumption that were not assessed and that may confound the relation
with fertility. Thus, we cannot exclude the possibility that the lack of
association seen in our data is an artifact produced by unmeasured
factors.
Finally, given the sample size of our study, our
results cannot rule out modest effect sizes, which we were underpowered
to detect.
Despite the above limitations, our analysis
had several important strengths. First, collection of prospective data
from a relatively large number of women undergoing IVF enables us to get
accurate data regarding their usual drinking habits before oocyte
retrieval. Second, to generalize our results, we included also fertile
women undergoing IVF for PGD. Moreover, our cohort included a wide range
of caffeine intake, which allowed us to examine more relevant levels of
exposure given the high intake of caffeine by reproductive-age women in
many countries (36, 37).
Last, all patients underwent their IVF treatment in the same program,
enabling standardization of the laboratory conditions and limiting
possible interobserver variations.
In conclusion,
prepregnancy consumption of sugared sodas seems to have the most
detrimental impact on IVF outcomes compared with other commonly consumed
beverages. Given our findings, it is possible that sugar, rather than
the caffeine, is a stronger reproductive toxicant. Although we failed to
find an association between caffeine consumption and IVF outcomes,
these results should to be interpreted with caution and deserve further
evaluation, including measurement of urinary caffeine metabolites to
assess the accurate level of exposure.
