Literature Review - Systematic Research 4 Pages

1 3

Eur J Nutr (2018) 57:107–117 https://doi.org/10.1007/s00394-016-1301-2

ORIGINAL CONTRIBUTION

Pre‑pregnancy caffeine and caffeinated beverage intake and risk of spontaneous abortion

Audrey J. Gaskins1,3 · Janet W. Rich‑Edwards2,3,4 · Paige L. Williams2,5 · Thomas L. Toth6 · Stacey A. Missmer2,3,7,8 · Jorge E. Chavarro1,2,3

Received: 28 January 2016 / Accepted: 16 August 2016 / Published online: 29 August 2016 © Springer-Verlag Berlin Heidelberg 2016

consuming <50 mg/day (p trend = 0.05). Total coffee intake had a positive, linear association with SAB. Com- pared to women with no pre-pregnancy coffee intake, women consuming ≥4 servings/day had a 20 % (6, 36 %) increased risk of SAB (p trend = 0.01). There was no dif- ference in the association between caffeinated and decaf- feinated coffee and risk of SAB. Pre-pregnancy intake of caffeinated tea, caffeinated soda, and decaffeinated soda had no association with SAB. Conclusions Pre-pregnancy coffee consumption at levels ≥4 servings/day is associated with increased risk of SAB, particularly at weeks 8–19.

Keywords Caffeine · Coffee · Miscarriage · Pregnancy · Spontaneous abortion

Introduction

Studies on maternal caffeine consumption and risk of spontaneous abortion (SAB) have been conducted since

Abstract Purpose To investigate the relation between pre-pregnancy caffeine and caffeinated beverage intake and risk of sponta- neous abortion (SAB). Methods Our prospective cohort study included 15,590 pregnancies from 11,072 women with no history of SAB in the Nurses’ Health Study II (1991–2009). Beverage intake was assessed every 4 years using a validated questionnaire. Pregnancies were self-reported with case pregnancies lost spontaneously at <20 weeks gestation. Multivariable log- binomial regression models with generalized estimating equations were used to estimate the relative risks (RRs) and 95 % confidence intervals (CIs). Results There was a positive linear trend across catego- ries of pre-pregnancy caffeine intake and risk of SAB such that women consuming >400 mg/day had 1.11 (95 % CI 0.98, 1.25) times the risk of SAB compared to women

Electronic supplementary material The online version of this article (doi:10.1007/s00394-016-1301-2) contains supplementary material, which is available to authorized users.

* Audrey J. Gaskins [email protected]

1 Department of Nutrition, Harvard T.H. Chan School of Public Health, Building II 3rd Floor, 655 Huntington Ave, Boston, MA 02115, USA

2 Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA

3 Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA

4 Department of Medicine, Connors Center for Women’s Health and Gender Biology, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA

5 Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA, USA

6 Vincent Obstetrics and Gynecology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA

7 Department of Obstetrics, Gynecology and Reproductive Biology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA

8 Department of Obstetrics, Gynecology and Reproductive Biology, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA

108 Eur J Nutr (2018) 57:107–117

1 3

the 1980s and stands as the most heavily researched dietary factor in regard to pregnancy outcomes. While a recent meta-analysis concluded that for every 100 mg/ day increment in maternal consumption of caffeine during early pregnancy, the risk of SAB increased by 14 % [1], the evidence on pre-pregnancy consumption of caffeine and risk of pregnancy loss is less clear. As the majority of pregnancy losses occur very early in gestation, pre- pregnancy caffeine consumption is of research interest as this may be a relevant time window of exposure for these early losses.

Many previous studies on pre-pregnancy caffeine intake and miscarriage have enrolled women during early preg- nancy and inquired about caffeine intake retrospectively prior to the interview [2–5], raising concerns about differ- ential left truncation and recall biases. Out of the 5 stud- ies that have enrolled women prior to pregnancy, three have been limited by small sample sizes (n = 171, 575, and 113) [6–8]. While the two other studies were much larger, the generalizability of their results to the US population is questionable [9, 10]. Specifically, the Tolstrup et al. study from Denmark included 1,381 pregnancies but recorded a mean daily caffeine intake of almost 500 mg (more than twice the average intake of American woman [11]) and <1 % of their women reported no coffee consumption [9]. The Hahn et al. study, also from Denmark, involved 5,132 women [10], but only included pregnancy planners, rais- ing questions about generalizability given that than half of pregnancies in the US are unplanned [12] and the known links between pre-pregnancy health behaviors, pregnancy intention, and pregnancy outcomes [13, 14].

Therefore, we sought to expand on the existing research by evaluating the association of pre-pregnancy caffeine and caffeinated and decaffeinated beverage intake with sponta- neous abortion in a large, prospective cohort of women in the USA.

Methods

Study population

We used data from the Nurses’ Health Study II (NHS2), an ongoing prospective cohort of 116,480 female nurses, ages 24–44 years at the study’s inception in 1989. Ques- tionnaires are distributed every 2 years to update life- style and medical characteristics and to capture incident health outcomes. Diet was first assessed in 1991 and has been updated every 4 years thereafter. Response rates for each questionnaire cycle have exceeded 90 %. Women were eligible for this analysis if they had no history of pregnancy loss in 1991 and reported at least one preg- nancy during 1992–2009. Participants contributed eligible

pregnancies until their first pregnancy loss or the end of follow-up. Women were censored after their first preg- nancy loss to avoid reverse causation bias resulting from behavioral changes which may occur after experiencing an adverse outcome. Of the 19,451 eligible pregnancies, we excluded from the analysis those with missing data on diet (n = 2,475), implausible or missing gestational age (n = 111), missing year of pregnancy (n = 619), as well as those among women with a diagnosis of type II diabe- tes (n = 69), cardiovascular disease (n = 86), or cancer (n = 141) prior to the pregnancy (Supplemental Figure 1). The excluded women did not differ substantially in age, BMI, parity, smoking status, race, history of infertility, and cumulative incidence of spontaneous abortion compared to included women. The final sample consisted of 15,950 pregnancies from 11,072 women. This study was approved by the institutional review board of the Partners Health Care System, Boston, Massachusetts, with the participants’ consent implied by the return of the questionnaires.

Caffeine assessment

Diet was evaluated using a validated 131-item food fre- quency questionnaire (FFQ) [15, 16]. Women were asked to report how often, on average, they consumed speci- fied amounts of each food and beverage included in the questionnaire during the previous year. There were nine response options ranging from “none or less than 1/month” to “6+/day.” The nutrient content and portion size of each item is obtained from a nutrient database derived from the US Department of Agriculture and additional informa- tion from manufacturers [17]. Specific caffeine-containing items were caffeinated coffee (137 mg caffeine/cup) and tea (47 mg caffeine/cup), caffeinated sodas (46 mg caf- feine/bottle or can), and chocolate (7 mg caffeine/serving). We calculated the total intake of caffeine by summing the caffeine content for specific items multiplied by weights proportional to the frequency of use. In a similar popula- tion of nurses, high correlations were found between caf- feinated beverage intakes assessed with the FFQ and four 1-week diet records collected over 1 year (coffee: 0.78, tea: 0.93, and soda: 0.84) [16, 18]. To maintain a prospec- tive analysis, diet information from 1991 was related to pregnancies in 1992 to 1995; the 1995 diet information was used for pregnancies in 1996 to 1999; and so forth. If a woman was missing the most recent diet questionnaire prior to her pregnancy (<5 % of women), the most recent previous dietary data were carried forward.

Outcome assessment

Women were asked to report their pregnancies at study enrollment and in each biennial follow-up questionnaire.

109Eur J Nutr (2018) 57:107–117

1 3

On the 2009 questionnaire, women also reported informa- tion on the year, length, complications, and outcomes of all previous pregnancies. Options for pregnancy outcomes were a singleton live birth, multiple birth, miscarriage or stillbirth, tubal or ectopic pregnancy, or induced abor- tion. Gestational lengths were reported in categories: <8, 8–11, 12–19, 20–27, 28–31, 32–36, 37–39, 40–42, and ≥43 weeks gestation. Self-reported pregnancy outcome and gestation length have been previously found to be val- idly reported [19]. Spontaneous abortion was defined as a spontaneous loss occurring before 20 completed weeks of gestation. We also considered early- to mid-first trimester losses (<8 weeks), late-first trimester losses (8–11 weeks), early second trimester losses (12–19 weeks), and still- births (≥20 weeks) as separate outcomes. The validity of maternal recall of pregnancy loss has not been assessed in this population; however, the sensitivity of reporting a loss is estimated to be around 75 % [20, 21]. In the vali- dation study by Wilcox et al., it was shown that the accu- racy of report depends greatly on the gestational age at the time of SAB (accuracy was 54 % at <7 weeks vs. 93 % at>13 weeks) and on the length of time since SAB (accu- racy was 82, 79, and 73 % for SABs occurring <10, 10–19, and ≥20 years ago) [20]. The comparison group for our analyses was all pregnancies that did not end in SAB (live births [n = 12,298], induced abortions [n = 634], or tubal/ ectopic pregnancies [n = 142]).

Covariate assessment

Information on potential confounding variables was assessed at baseline and during follow-up. For variables that were updated over follow-up, the most recent value prior to pregnancy was used. Maternal age was computed as the difference between year of birth and the year of preg- nancy. Weight, smoking status, multivitamin use, hormonal contraceptive use, duration of rotating night shift work, and history of infertility were self-reported at baseline and updated every two years thereafter. Race and height were reported in 1989. Pre-pregnancy body mass index (BMI) was calculated as self-reported weight in kilograms divided by self-reported height in meters squared. In a previous validation study, self-reported weight was highly corre- lated with weight measured by a technician among a simi- lar group of nurses (r = 0.97) [22]. Physical activity was ascertained in 1991, 1997, 2001, and 2005 using a vali- dated questionnaire [23], from which metabolic equivalent task–hours per week were derived. History of ovulation- inducing medication use was self-reported starting in 1993 and updated every 2 years. Marital status was reported in 1989, 1993, and 1997. Employment outside the home was assessed in 1989, 1993, and 2001.

Statistical analysis

Baseline characteristics were derived from the 1991 ques- tionnaire for all women contributing eligible pregnancies, and differences in baseline characteristics by pre-pregnancy caffeine intake were compared. Overall caffeine intake was analyzed as a continuous exposure and as categories based on predetermined cutoffs. Pre-pregnancy intake of caffein- ated and decaffeinated beverages was categorized accord- ing to the distribution in the cohort.

The relative risk (RR) of SAB in relation to pre-preg- nancy caffeine and beverage intake was estimated using log-binomial regression. Generalized estimating equations with an exchangeable working correlation structure were used to account for the within-person correlation between pregnancies. For the primary analysis, we examined the relation between the continuous measure of caffeine intake and risk of SAB nonparametrically with restricted cubic splines [24]. Tests for nonlinearity used the likelihood ratio test, comparing the model with only the linear term to the model with the linear and the cubic spline terms. For all other analyses, the RR was computed as the risk of SAB in a specific category compared with the risk in the lowest caffeine or caffeinated beverage intake category. Tests for linear trend across categories were conducted by using the median values in each category as a continuous variable. In addition to age-, calorie-, and year-adjusted models, multi- variable models were further adjusted for a priori selected pre-pregnancy covariables. These included BMI, smoking status, physical activity, history of infertility, marital status, employment outside the home, duration of rotating shift work, race, alcohol intake, and supplemental folate intake. All caffeinated beverages were also further adjusted for one another. Decaffeinated beverages were further adjusted for caffeinated coffee intake. Fully adjusted models were run both with and without adjusting for nulliparity since adjust- ing for reproductive history might lead to overadjustment [25, 26]. Categorical covariables included an indicator for missing data, if necessary.

To assess the robustness of our findings, we investigated whether the relation of pre-pregnancy caffeine and bever- age intake with SAB differed by gestational age at loss. We also performed an analysis restricted to pregnancies from women who never smoked to address the potential of resid- ual confounding by smoking. To minimize uncontrolled confounding by behaviors related to pregnancy planning and recognition, we restricted analyses to married women not using hormonal contraception [27, 28]. Last, to address the potential of misclassification of exposure due to the interval between diet assessments, we restricted analyses to pregnancies in 1992, 1996, 2000, and 2004, the years clos- est to diet assessment, and we performed an analysis where

110 Eur J Nutr (2018) 57:107–117

1 3

we did not carry forward previous diet data if the most recent diet data were missing.

Results

Overall, 11,072 women met our inclusion criteria, contrib- uting 15,950 pregnancies to this analysis during 18 years of follow-up. The cumulative risk of pregnancy loss in our cohort was 6.3 % by week 7, 13.6 % by week 11, 17.3 % by week 20, and 18.0 % by week 43+. On average, women in our cohort consumed 184 mg/day of caffeine. The big- gest contributor to pre-pregnancy caffeine intake was caf- feinated coffee (71 %), followed by caffeinated tea (14 %), caffeinated soda (12 %), and chocolate (3 %). In 1991, the most common method of preparing caffeinated and decaffeinated coffee was filtered (81 and 66 %) followed by instant (8 and 16 %) and espresso or percolator (5 and 3 %).

Women consuming the highest amounts of caffeine were, on average, slightly older and of higher BMI, more likely white, nulliparous, employed outside the home, and to have worked rotating night shifts in the past 2 years (Table 1). Women with higher caffeine intake were also less likely to be never smokers and never oral contraceptive users. Women consuming the highest amounts of caffeine were less likely to consume a multivitamin, had higher intake of calories, fat, and alcohol, and had lower intake of carbohydrates, protein, and folate.

The association between pre-pregnancy caffeine intake and risk of spontaneous abortion showed sug- gestive evidence of a nonlinear association (P value for nonlinearity = 0.06), but this was primarily driven by associations with losses at <8 weeks (P value for nonline- arity = 0.004) as the association with losses at 8–11 weeks and 12–19 weeks appeared linear (P value for nonlinear- ity = 0.50 and 0.53, respectively). When caffeine intake was categorized, there was a positive linear trend across categories of pre-pregnancy caffeine intake and risk of SAB such that women consuming >400 mg/day had 1.11 (95 % CI 0.98, 1.25) times the risk of SAB compared to women consuming <50 mg/day (p trend = 0.05) (Table 2). When specific beverages were examined, there was a positive, linear association between pre-pregnancy cof- fee intake and risk of SAB. The adjusted RR (95 % CI) for <1, 1, 2–3, and ≥4 servings/day was 1.02 (0.92, 1.14), 1.11 (0.99, 1.24), 1.09 (1.00, 1.19), and 1.20 (1.06, 1.36), respectively, compared to women consuming none (P trend = 0.004). This translated into an adjusted absolute difference in SAB risk of 3.6 % (95 % CI 1.0, 6.2 %) com- paring women consuming ≥4 servings/day to women con- suming none. This association became slightly attenuated after adjustment for pre-pregnancy caffeine intake (P trend

for coffee = 0.06); however, caffeine intake was not asso- ciated with SAB in this model (P = 0.77). The adjusted RR (95 % CI) with further adjustment for caffeine was 1.02 (0.91, 1.14), 1.11 (0.99, 1.24), 1.09 (0.98, 1.20), and 1.18 (1.00, 1.40) for women consuming <1, 1, 2–3, and ≥4 servings/day of coffee, respectively, compared to women consuming none.

Both caffeinated and decaffeinated coffee intakes had a positive association with SAB (P trend = 0.01 and 0.04, respectively) (Table 2). When these exposures were included in the same multivariate model, there was no significant difference in the effect of a one serving/day increase on risk of SAB as the relative risks of caffein- ated and decaffeinated coffee were almost identical (RR for caffeinated coffee: 1.04 (1.01, 1.06); RR for decaffein- ated coffee: 1.04 (0.99, 1.09); P for comparison = 0.97). Pre-pregnancy intake of other caffeinated beverages such as caffeinated tea and soda and decaffeinated soda had no association with risk of SAB.

Overall, while there was no clear linear association between pre-pregnancy caffeine intake and SAB at <8 or 8–11 weeks gestation, there was a positive linear asso- ciation between pre-pregnancy caffeine intake and SAB at 12–19 weeks gestation (Table 3). This appeared to be driven by caffeinated coffee intake which had a positive association with SAB at 8–11 and 12–19 weeks gestation but not at <8 weeks gestation. The associations between caffeinated tea, caffeinated soda, decaffeinated coffee, and decaffeinated soda and pregnancy loss were nonsignificant across all time points (Table 3 and Supplemental Table 1). There were no associations between pre-pregnancy caffeine or beverage intake and risk of stillbirth.

In our sensitivity analyses, the magnitude of association between total and decaffeinated coffee remained relatively consistent (Supplemental Table 2); however, there was little relation between pre-pregnancy caffeine intake and SAB. The positive association between pre-pregnancy caffein- ated coffee intake and SAB persisted when analyses were restricted to never smokers, the most likely pregnancy plan- ners, and with no diet update but became attenuated when analyses were restricted to pregnancies in the years follow- ing dietary assessment.

Discussion

In our study, women who consumed ≥4 servings/day of coffee prior to pregnancy had a 3.6 % higher absolute risk of SAB compared to woman consuming none. This asso- ciation was driven by losses occurring at 8–19 weeks of gestation. Both pre-pregnancy caffeinated and decaffein- ated coffee intakes were associated with risk of SAB, sug- gesting that some component in coffee other than caffeine

111Eur J Nutr (2018) 57:107–117

1 3

was driving this relationship. Moreover, none of the other caffeinated beverages, including tea and soda, were associ- ated with risk of SAB. The increased risk of SAB at high levels of caffeine intake (e.g., >400 mg/day) appeared to

be driven by pre-pregnancy caffeinated coffee intake; how- ever, this association was hard to disentangle given the high correlation between caffeine and coffee intake in this population.

Table 1 Baseline demographic characteristics by category of caffeine intake in 1991 among 11,072 Women in the Nurses’ Health Study II

BMI body mass index, MET-h metabolic equivalent task hours, OC oral contraceptives

Pre-pregnancy caffeine intake in 1991

Range of intake <50 mg/day 50–99 mg/day 100–199 mg/day 200–299 mg/day 300–399 mg/day ≥400 mg/day

Number of women 2909 1584 2867 845 1564 1282

Maternal age, years 31.0 [29.0, 34.0] 31.0 [28.0, 33.0] 31.0 [29.0, 34.0] 31.0 [29.0, 33.0] 32.0 [30.0, 35.0] 32.0 [29.0, 35.0]

Pre-pregnancy BMI, kg/m2

22.2 [20.3, 25.0] 22.2 [20.5, 25.0] 22.3 [20.5, 25.0] 22.7 [20.8, 25.6] 22.1 [20.4, 24.3] 22.5 [20.8, 25.2]

Physical activity, MET-h/week

15.3 [6.4, 30.6] 14.6 [6.1, 30.4] 16.1 [6.9, 32.2] 14.7 [5.3, 32.1] 16.1 [6.8, 33.4] 15.7 [6.3, 33.6]

Smoking status, n (%)

Never smoker 2433 (83.6) 1292 (81.6) 2061 (71.9) 573 (67.8) 877 (56.1) 637 (49.7)

Former smoker 385 (13.2) 224 (14.1) 572 (20.0) 170 (20.1) 485 (31.0) 333 (26.0)

Current smoker 85 (2.9) 65 (4.1) 229 (8.0) 101 (12.0) 199 (12.7) 311 (24.3)

White, n (%) 2673 (91.9) 1476 (93.2) 2678 (93.4) 794 (94.0) 1477 (94.4) 1214 (94.7)

Married, n (%) 2217 (76.2) 1181 (74.6) 2015 (70.3) 554 (65.6) 1049 (67.1) 777 (60.6)

Ever OC use, n (%)

Never 602 (20.7) 230 (14.5) 439 (15.3) 109 (12.9) 234 (15.0) 178 (13.9)

Past 1761 (60.5) 944 (59.6) 1698 (58.9) 497 (58.8) 1005 (64.3) 787 (61.4)

Current 546 (18.8) 410 (25.9) 739 (25.8) 239 (28.3) 325 (20.8) 316 (24.7)

Employed, n (%) 2667 (91.7) 1490 (94.1) 2712 (94.6) 810 (95.9) 1478 (94.5) 1238 (96.6)

Rotating night shift work in past 2 years, n (%)

None 2240 (77.0) 1174 (74.1) 2074 (72.3) 555 (65.7) 1165 (74.5) 878 (68.5)

1–4 months 240 (8.3) 138 (8.7) 280 (9.8) 91 (10.8) 119 (7.6) 132 (10.3)

5–14 months 206 (7.1) 125 (7.9) 221 (7.7) 81 (9.6) 132 (8.4) 99 (7.7)

≥15 months 221 (7.6) 140 (8.8) 285 (9.9) 114(13.5) 135 (8.6) 169 (13.2) History of infertility,

n (%) 471 (16.2) 216 (13.6) 404 (14.1) 129 (15.3) 227 (14.5) 210 (16.4)

Parity, n %

Nulliparous 1199 (41.2) 654 (41.3) 1294 (45.1) 417 (49.4) 709 (45.3) 634 (49.5)

1 887 (30.5) 490 (30.9) 756 (26.4) 216 (25.6) 402 (25.7) 270 (21.1)

2 508 (17.5) 288 (18.2) 567 (19.8) 150 (17.8) 318 (20.3) 237 (18.5)

3+ 227 (7.8) 108 (6.8) 176 (6.1) 46 (5.5) 101 (6.5) 109 (8.5) Multivitamin use,

n (%) 1681 (57.8) 870 (54.9) 1493 (52.1) 402 (47.6) 745 (47.6) 556 (43.4)

Calories, kcal/day 1721 [1387, 2081] 1816 [1433, 2203] 1755 [1412, 2132] 1868 [1471, 2337] 1705 [1386, 2077] 1867 [1473, 2265]

% Calories from fat 30.0 [26.4, 33.8] 30.5 [27.1, 33.8] 30.6 [27.1, 34.3] 31.4 [28.1, 35.7] 31.0 [27.4, 34.5] 31.9 [28.0, 35.7]

% Calories from carbohydrates

51.3 [46.9, 56.2] 51.3 [47.3, 56.0] 50.8 [46.1, 55.4] 49.4 [44.6, 54.4] 49.9 [45.1, 54.0] 48.6 [44.3, 53.7]

% Calories from protein

19.5 [17.5, 21.7] 18.8 [16.7, 21.0] 19.1 [16.8, 21.2] 18.7 [16.9, 20.8] 19.1 [17.0, 21.3] 18.7 [16.7, 21.0]

Alcohol, g/day 0.0 [0.0, 1.9] 0.9 [0.0, 2.8] 1.8 [0.0, 4.4] 1.8 [0.0, 4.8] 2.8 [0.9, 6.9] 2.8 [0.9, 6.9]

Folate, μg/day 511 [318, 826] 455 [295, 746] 417 [294, 720] 381 [277, 620] 385 [290, 643] 361 [273, 564]

Supplemental folate, μg/day

134 [0, 400] 56 [0, 400] 56 [0, 400] 0 [0, 228] 0 [0, 267] 0 [0, 228]

112 Eur J Nutr (2018) 57:107–117

1 3

Table 2 Pre-pregnancy Caffeine and Caffeinated and Decaffeinated Beverage Intake and Relative Risks of Spontaneous Abortion among 11,072 Women (15,950 pregnancies) in the Nurses’ Health Study II

Categories of intake (range) Cases/total % Age-, year-, and energy-adjusted RR (95 % CI)a

Multivariate-adjusted RR (95 % CI)b

Caffeine

<50 mg/day 759/4533 16.7 1.00 (Ref) 1.00 (Ref)

50–99 mg/day 333/2192 15.2 0.93 (0.83, 1.05) 0.92 (0.82, 1.03)

100–199 mg/day 705/4115 17.1 1.01 (0.92, 1.11) 0.99 (0.91, 1.09)

200–299 mg/day 185/1091 17.0 0.99 (0.86, 1.14) 0.96 (0.83, 1.11)

300–399 mg/day 428/2302 18.6 1.06 (0.95, 1.17) 1.04 (0.93, 1.16)

≥400 mg/day 346/1717 20.2 1.14 (1.02, 1.28) 1.11 (0.98, 1.25) P for trend 0.008 0.05

Total coffee

Per 1 serving/day 1.04 (1.02, 1.06) 1.04 (1.01, 1.06)

Never 1065/6633 16.1 1.00 (Ref) 1.00 (Ref)

<1 serving per day 347/2145 16.2 1.01 (0.91, 1.13) 1.02 (0.92, 1.14)

1 serving per day 317/1758 18.0 1.09 (0.98, 1.22) 1.11 (0.99, 1.24)

2–3 servings per day 782/4281 18.3 1.08 (1.00, 1.18) 1.09 (1.00, 1.19)

≥4 servings per day 245/1133 21.6 1.21 (1.07, 1.37) 1.20 (1.06, 1.36) P for trend 0.002 0.004

Caffeinated coffee

Per 1 serving/day 1.04 (1.01, 1.07) 1.04 (1.01, 1.06)

Never 1151/7117 16.2 1.00 (Ref) 1.00 (Ref)

<1 serving per day 490/2883 17.0 1.05 (0.95, 1.15) 1.06 (0.96, 1.16)

1 serving per day 404/2254 17.9 1.08 (0.98, 1.20) 1.09 (0.98, 1.20)

2–3 servings per day 568/3035 18.7 1.10 (1.00, 1.20) 1.09 (1.00, 1.20)

≥4 servings per day 143/661 21.6 1.23 (1.06, 1.42) 1.20 (1.03, 1.41) P for trend 0.003 0.01

Decaffeinated coffee

Per 1 serving/day 1.04 (0.99, 1.10) 1.04 (0.98, 1.09)

Never 1797/10690 16.8 1.00 (Ref) 1.00 (Ref)

1–4 servings per month 431/2573 16.8 0.98 (0.89, 1.08) 0.97 (0.88, 1.07)

2–6 servings per week 237/1175 20.2 1.19 (1.05, 1.34) 1.21 (1.07, 1.36)

≥1 serving per day 291/1512 19.3 1.06 (0.95, 1.18) 1.06 (0.95, 1.18) P for trend 0.04 0.04

Caffeinated tea

Per 1 serving/day 0.99 (0.95, 1.03) 0.98 (0.95, 1.02)

Never 1106/6241 17.7 1.00 (Ref) 1.00 (Ref)

<1 serving per day 1085/6311 17.2 1.01 (0.94, 1.09) 0.99 (0.92, 1.06)

1 serving per day 305/1817 16.8 1.02 (0.91, 1.14) 1.00 (0.89, 1.12)

≥2 servings per day 260/1581 16.5 0.98 (0.87, 1.11) 0.96 (0.85, 1.09) P for trend 0.84 0.60

Caffeinated soda

Per 1 serving/day 1.04 (1.00, 1.07) 1.02 (0.99, 1.06)

Never 748/4230 17.7 1.00 (Ref) 1.00 (Ref)

<1 serving per day 1256/7437 16.9 1.02 (0.94, 1.11) 1.00 (0.92, 1.08)

1 serving per day 403/2341 17.2 1.06 (0.95, 1.18) 1.02 (0.91, 1.14)

≥2 servings per day 349/1942 18.0 1.12 (1.00, 1.26) 1.06 (0.94, 1.19) P for trend 0.04 0.26

Decaffeinated soda

Per 1 serving/day 0.97 (0.93, 1.02) 0.97 (0.93, 1.01)

113Eur J Nutr (2018) 57:107–117

1 3

Previous studies on pre-pregnancy caffeine or coffee intake and risk of SAB are summarized in Supplemental Table 3. Overall, the majority of previous studies found an increased risk of SAB comparing women in the high- est versus lowest category of pre-pregnancy caffeine intake, although the magnitude of these effects varied and many were not statistically significant. Moreover, in a sub-anal- ysis of a recent meta-analysis, the association between pre- pregnancy caffeine intake and pregnancy loss was not sta- tistically significant (RR = 1.02, 95 % CI 0.97, 1.07) [29]. Our results suggesting an increased risk of SAB at high levels of caffeine intake (e.g., >400 mg/day) are consistent with the most recent study which found a relative risk of 1.10 (95 % CI 0.91, 1.32) comparing women consuming >300 mg/day to <100 mg/day. Moreover, women in their study with a preconception coffee intake ≥3 servings/day had 1.23 (95 % CI 0.91, 1.65) times the risk of SAB com- pared to women consuming none which was similar to our estimate of 1.20 (95 % CI 1.03, 1.41) for women consum- ing ≥4 servings of coffee per day compared to none. The associations between preconception coffee intake and SAB in the Danish study were also stronger for pregnancy losses at ≥8 weeks of gestation, parallel to our results. Finally, these authors also found no association between other caf- feinated beverages and SAB. Taken together, these results provide consistent evidence that pre-pregnancy coffee intake at levels ≥3–4 servings per day is associated with slightly increased risk of SAB between gestation weeks 8–19.

Our results which showed a possible association between pre-pregnancy decaffeinated coffee intake and risk of SAB and no association between caffeinated tea and

soda intake and SAB suggest that a component other than caffeine could be driving the association between coffee and SAB. This theory is supported by the one other study that assessed pre-pregnancy decaffeinated coffee intake and risk of SAB [2]. In this study, the adjusted odds ratio (95 % CI) of SAB for pre-pregnancy decaffeinated coffee intake of 0.5–1, 2, and ≥3 servings per day was 1.17 (0.83, 1.66), 0.77 (0.41, 1.45), and 1.52 (0.85, 2.72) compared to no intake [2]. Alternatively, the association we found between pre-pregnancy decaffeinated coffee and SAB could be driven by women at higher risk of SAB switching from caffeinated to decaffeinated coffee due to concerns over caffeine’s potential reproductive health effects. While we tried to reduce this type of bias by excluding women with a history of SAB, adjusting for many well-known risk fac- tors for SAB (e.g., age and history of infertility), and per- forming various sensitivity analyses, it is possible this bias could still be explaining all or part of this association.

One of the biggest limitations of our current analysis is that we lacked information on caffeine and beverage intake during early pregnancy. While it is well documented that the majority of women decrease or quit drinking coffee during the first trimester of pregnancy, women with higher pre-pregnancy coffee consumption tend to decrease their coffee consumption to a lesser extent [30–32]. Thus, it is likely that the women in our highest category of pre-preg- nancy coffee intake were also the highest coffee consum- ers during early pregnancy. If early pregnancy coffee intake (e.g., between weeks 1 and 6) is causally related to SAB at weeks 8–19, then this could be driving the association we observed between pre-pregnancy intake and pregnancy loss due to the high correlation between pre-pregnancy

Analyses run using a log-binomial generalized linear model with an exchangeable working correlation structure to compute relative risk esti- mates. Tests for linear trend were conducted by using the median values in each category as a continuous variable

RR relative risk, CI confidence interval a Adjusted for age (continuous), total energy intake (continuous), and year (continuous) b Age-, year-, and energy-adjusted model further adjusted for body mass index (<18.5, 18.5–24.9, 25–29.9,≥30, and missing), smoking sta- tus (never, former, current, and missing), physical activity (<3 MET-h/week, 3–8.9 MET-h/week, 9–17.9 MET-h/week, 18–26.9 MET-h/week, 27–41.9 MET-h/week, >42 MET-h/week, and missing), history of infertility (no, yes, and missing), marital status (married, not married), employment status (yes, no), duration of rotating shift work in past 2 years (none, 1–4 months, 5–14 months, ≥15 months), race (white, other), alcohol intake (quintiles), supplemental folate intake (quartiles). All caffeinated beverages were adjusted for one another. Decaf coffee and soda intake were further adjusted for caffeinated coffee intake

Table 2 continued

Categories of intake (range) Cases/total % Age-, year-, and energy-adjusted RR (95 % CI)a

Multivariate-adjusted RR (95 % CI)b

Never 895/4670 19.2 1.00 (Ref) 1.00 (Ref)

1–4 servings per month 727/4309 16.9 0.95 (0.87, 1.04) 0.94 (0.86, 1.03)

2–6 servings per week 716/4345 16.5 0.95 (0.87, 1.04) 0.95 (0.87, 1.04)

1 serving per day 242/1533 15.8 0.92 (0.81, 1.05) 0.92 (0.81, 1.04)

≥2 servings per day 176/1093 16.1 0.93 (0.80, 1.07) 0.90 (0.78, 1.04) P for trend 0.26 0.16

114 Eur J Nutr (2018) 57:107–117

1 3

T ab

le 3

P

re -p

re gn

an cy

C af

fe in

e an

d C

af fe

in at

ed a

nd D

ec af

fe in

at ed

C of

fe e

In ta

ke a

nd R

el at

iv e

R is

ks o

f S

po nt

an eo

us A

bo rt

io n

an d

S ti

ll bi

rt h

S tr

at ifi

ed b

y G

es ta

ti on

al L

en gt

h am

on g

11 ,0

72

W om

en (

15 ,9

50 P

re gn

an ci

es )

in t

he N

ur se

s’ H

ea lt

h S

tu dy

I I

C at

eg or

ie s

of i

nt ak

e P

re gn

an cy

l os

s <

8 w

ee ks

P re

gn an

cy l

os s

8– 11

w ee

ks P

re gn

an cy

l os

s 12

–1 9

w ee

ks P

re gn

an cy

l os

s 20

+ w

ee ks

C as

es /t

ot al

A dj

us te

d R

R (9

5 %

C I)

a C

as es

/t ot

al A

dj us

te d

R R

(9 5

% C

I) a

C as

es /t

ot al

A dj

us te

d R

R (

95 %

C I)

C as

es /t

ot al

A dj

us te

d R

R

(9 5

% C

I) a

C af

fe in

e, m

g/ da

y

< 50

30 3/

45 33

1. 00

( R

ef )

31 0/

41 47

1. 00

( R

ef )

14 6/

37 98

1. 00

( R

ef )

34 /3

63 2

1. 00

( R

ef )

5 0–

99 11

2/ 21

91 0.

76 (

0. 61

, 0 .9

4) 15

1/ 20

20 1.

01 (

0. 84

, 1 .2

2) 70

/1 84

5 1.

02 (

0. 77

, 1 .3

5) 16

/1 76

5 0.

93 (

0. 52

, 1 .6

9)

1 00

–1 99

25 0/

41 15

0. 84

( 0.

71 , 0

.9 9)

29 4/

37 52

1. 02

( 0.

87 , 1

.1 9)

16 1/

33 98

1. 25

( 1.

00 , 1

.5 7)

30 /3

21 5

0. 83

( 0.

50 , 1

.3 6)

2 00

–2 99

72 /1

09 1

0. 84

( 0.

65 , 1

.0 8)

69 /9

91 0.

90 (

0. 70

, 1 .1

6) 44

/9 01

1. 29

( 0.

92 , 1

.8 0)

6/ 85

3 0.

53 (

0. 22

, 1 .2

7)

3 00

–3 99

13 2/

23 02

0. 74

( 0.

60 , 0

.9 0)

20 2/

20 93

1. 20

( 1.

01 , 1

.4 4)

94 /1

84 4

1. 32

( 1.

01 , 1

.7 1)

14 /1

74 3

0. 59

( 0.

31 , 1

.1 3)

≥ 40

0 13

2/ 17

17 0.

93 (

0. 75

, 1 .1

5) 13

5/ 15

06 1.

11 (

0. 91

, 1 .3

7) 79

/1 32

5 1.

54 (

1. 16

, 2 .0

5) 20

/1 23

5 1.

07 (

0. 59

, 1 .9

3)

P f

or t

re nd

0. 23

0. 07

0. 00

1 0.

54

T ot

al c

of fe

e, s

er vi

ng s

N ev

er 40

7/ 66

33 1.

00 (

R ef

) 43

9/ 60

80 1.

00 (

R ef

) 21

9/ 55

66 1.

00 (

R ef

) 46

/5 31

7 1.

00 (

R ef

)

< 1

pe r

da y

12 1/

21 45

0. 94

( 0.

77 , 1

.1 4)

16 5/

19 78

1. 17

( 0.

99 , 1

.3 9)

61 /1

77 9

0. 89

( 0.

67 , 1

.1 8)

13 /1

70 5

0. 91

( 0.

49 , 1

.6 8)

1 p

er d

ay 10

9/ 17

58 0.

97 (

0. 79

, 1 .1

9) 12

7/ 15

90 1.

08 (

0. 89

, 1 .3

1) 81

/1 44

3 1.

46 (

1. 13

, 1 .8

7) 20

/1 35

8 1.

51 (

0. 89

, 2 .5

7)

2 –3

p er

d ay

27 0/

42 81

0. 94

( 0.

81 , 1

.1 0)

33 0/

38 69

1. 13

( 0.

98 , 1

.3 0)

18 2/

34 53

1. 33

( 1.

08 , 1

.6 2)

30 /3

24 9

0. 82

( 0.

50 , 1

.3 3)

≥ 4

pe r

da y

94 /1

13 3

1. 11

( 0.

88 , 1

.3 8)

10 0/

99 2

1. 26

( 1.

02 , 1

.5 7)

51 /8

70 1.

39 (

1. 03

, 1 .8

9) 11

/8 14

1. 02

( 0.

52 , 2

.0 1)

P f

or t

re nd

0. 81

0. 03

0. 00

3 0.

54

C af

fe in

at ed

c of

fe e,

s er

vi ng

s

N ev

er 45

1/ 71

17 1.

00 (

R ef

) 47

4/ 65

11 1.

00 (

R ef

) 22

6/ 59

56 1.

00 (

R ef

) 51

/5 69

9 1.

00 (

R ef

)

< 1

pe r

da y

15 9/

28 83

0. 87

( 0.

73 , 1

.0 3)

22 6/

26 59

1. 17

( 1.

00 , 1

.3 6)

10 5/

23 91

1. 18

( 0.

94 , 1

.4 8)

21 /2

27 0

1. 04

( 0.

62 , 1

.7 3)

1 p

er d

ay 14

6/ 22

54 0.

96 (

0. 80

, 1 .1

5) 15

1/ 20

32 0.

99 (

0. 83

, 1 .1

9) 10

7/ 18

48 1.

56 (

1. 24

, 1 .9

6) 15

/1 73

0 0.

87 (

0. 48

, 1 .5

6)

2 –3

p er

d ay

18 9/

30 35

0. 87

( 0.

73 , 1

.0 3)

24 8/

27 39

1. 17

( 1.

01 , 1

.3 7)

13 1/

24 22

1. 40

( 1.

12 , 1

.7 5)

25 /2

27 8

0. 91

( 0.

55 , 1

.5 0)

≥ 4

pe r

da y

56 /6

61 1.

09 (

0. 82

, 1 .4

3) 62

/5 68

1. 38

( 1.

06 , 1

.7 9)

25 /4

94 1.

26 (

0. 84

, 1 .9

1) 8/

46 6

1. 24

( 0.

57 , 2

.6 9)

P f

or t

re nd

0. 62

0. 01

0. 00

9 0.

95

D ec

af fe

in at

ed c

of fe

e

se rv

in gs

N ev

er 66

0/ 10

69 0

1. 00

( R

ef )

74 8/

97 32

1. 00

( R

ef )

38 9/

88 32

1. 00

( R

ef )

80 /8

39 3

1. 00

( R

ef )

1 –4

p er

m on

th 14

6/ 25

73 0.

94 (

0. 79

, 1 .1

3) 20

1/ 23

54 1.

08 (

0. 93

, 1 .2

6) 84

/2 11

0 0.

83 (

0. 66

, 1 .0

5) 23

/2 01

4 1.

25 (

0. 78

, 2 .0

2)

115Eur J Nutr (2018) 57:107–117

1 3

and early first trimester pregnancy coffee consumption. However, considering many pregnancies in the USA are unplanned and the most pregnancies are not recognized until 4–6 weeks of gestation, the distinction between pre- pregnancy and early pregnancy consumption, while inter- esting from a causal standpoint, is minor in terms of clini- cal and public health recommendations.

Filtered coffee consists of many biologically active compounds including caffeine, polyphenols (such as chlo- rogenic acids), melanoidins, and trigonelline [33]. Plau- sible biological explanations for an association between pre-pregnancy coffee intake and risk of SAB do exist and primarily relate to coffee’s effects on endogenous sex hor- mone metabolism. In a separate study, women with the highest coffee (but not total caffeine) intake at their last menstrual period had 32 % lower early pregnancy preg- nanediol-3-glucuronide levels, a hormone critical to preg- nancy maintenance [34]. In two separate studies from the NHS2 cohort, higher coffee intake was associated with lower total and free luteal plasma estradiol levels [35] and higher 2-catechol estrogen metabolites, 2-hydroxyestrone and 2-hydroxyestradiol [36]. Coffee has many constitu- ents that may influence sex hormone metabolism, includ- ing caffeine. Metabolism of caffeine is catalyzed primarily by CYP1A2 enzymes, and laboratory and human studies suggest that caffeine is as an inducer of CYP1A2 activity [37]. This hepatic enzyme also plays a key role in 2-, 4-, and 16-hydroxylation of estrogen [38]. Independent of caf- feine, coffee may also act as an inducer of CYP1A2 [39], potentially by polycyclic aromatic hydrocarbons produced by high brewing temperatures [40]. Laboratory evidence also suggests that caffeic acid and chlorogenic acid, two polyphenols found in coffee, can inhibit 2-catechol and 4-catechol methylation [41].

The limitations of our study are worth considering in lieu of our significant findings. Misclassification of caf- feine and beverage intake is likely because diet informa- tion was only updated every four years. It is also likely that there was misclassification of total caffeine consumption as caffeine content can differ widely based on the brew- ing method, serving size, and portion of serving consumed [42]. However, misclassification of caffeine and beverage intake is unlikely to depend on SAB status given the pro- spective nature of our study. This type of non-differential misclassification would therefore tend to attenuate our associations toward the null.

There is also concern that many early losses were unrec- ognized and that this could be differential with respect to pregnancy intention or pre-pregnancy caffeine or beverage intake. If women attempting to conceive limited their cof- fee intake and were more likely to recognize early miscar- riages, the early miscarriage rate would be higher in the lowest coffee intake group and would bias the association A

na ly

se s

ru n

us in

g a

lo g-

bi no

m ia

l ge

ne ra

li ze

d li

ne ar

m od

el w

it h

an e

xc ha

ng ea

bl e

w or

ki ng

c or

re la

ti on

s tr

uc tu

re t

o co

m pu

te r

el at

iv e

ri sk

e st

im at

es . T

es ts

f or

l in

ea r

tr en

d w

er e

co nd

uc te

d by

u si

ng

th e

m ed

ia n

va lu

es i

n ea

ch c

at eg

or y

as a

c on

ti nu

ou s

va ri

ab le

. T he

r ef

er en

ce g

ro up

f or

p re

gn an

cy l

os se

s <

8 w

ee ks

w er

e al

l pr

eg na

nc ie

s, t

he r

ef er

en ce

g ro

up f

or p

re gn

an cy

l os

se s

8– 11

w ee

ks w

er e

al l

pr eg

na nc

ie s

la st

in g

be yo

nd 8

w ee

ks ,

th e

re fe

re nc

e gr

ou p

fo r

pr eg

na nc

y lo

ss es

1 2–

19 w

ee ks

w er

e al

l pr

eg na

nc ie

s la

st in

g be

yo nd

1 2

w ee

ks ,

an d

th e

re fe

re nc

e gr

ou p

fo r

pr eg

na nc

y lo

ss es

2 0+

w

ee ks

w er

e al

l pr

eg na

nc ie

s la

st in

g be

yo nd

2 0

w ee

ks

R R

r el

at iv

e ri

sk , C

I co

nfi de

nc e

in te

rv al

a A

dj us

te d

fo r

ag e

(c on

ti nu

ou s)

, en

er gy

i nt

ak e

(c on

ti nu

ou s)

, bo

dy m

as s

in de

x (<

18 .5

, 18

.5 –2

4. 9,

2 5–

29 .9

, ≥

30 ,

an d

m is

si ng

), s

m ok

in g

st at

us (

ne ve

r, f

or m

er ,

cu rr

en t,

a nd

m is

si ng

), p

hy si

ca l

ac ti

vi ty

( <

3 M

E T

-h /w

ee k,

3 –8

.9 M

E T

-h /w

ee k,

9 –1

7. 9

M E

T -h

/w ee

k, 1

8– 26

.9 M

E T

-h /w

ee k,

2 7–

41 .9

M E

T -h

/w ee

k, >

42 M

E T

-h /w

ee k,

a nd

m is

si ng

), y

ea r

of p

re gn

an cy

( co

nt in

uo us

), h

is to

ry

of i

nf er

ti li

ty (

no ,

ye s,

a nd

m is

si ng

), a

nd m

ar it

al s

ta tu

s (m

ar ri

ed ,

no t

m ar

ri ed

), e

m pl

oy m

en t

st at

us (

ye s,

n o)

, du

ra ti

on o

f ro

ta ti

ng s

hi ft

w or

k in

p as

t 2

ye ar

s (n

on e,

1 –4

m on

th s,

5 –1

4 m

on th

s,

≥ 15

m on

th s)

, ra

ce (

w hi

te ,

ot he

r) ,

su pp

le m

en ta

l fo

la te

i nt

ak e

(q ua

rt il

es ),

a nd

a lc

oh ol

i nt

ak e

(q ui

nt il

es ).

C af

fe in

at ed

c of

fe e

w as

a dj

us te

d fo

r ca

ff ei

na te

d te

a an

d so

da i

nt ak

e. D

ec af

c of

fe e

w as

fu

rt he

r ad

ju st

ed f

or c

af fe

in at

ed c

of fe

e in

ta ke

T ab

le 3

co

nt in

ue d

C at

eg or

ie s

of i

nt ak

e P

re gn

an cy

l os

s <

8 w

ee ks

P re

gn an

cy l

os s

8– 11

w ee

ks P

re gn

an cy

l os

s 12

–1 9

w ee

ks P

re gn

an cy

l os

s 20

+ w

ee ks

C as

es /t

ot al

A dj

us te

d R

R (9

5 %

C I)

a C

as es

/t ot

al A

dj us

te d

R R

(9 5

% C

I) a

C as

es /t

ot al

A dj

us te

d R

R (

95 %

C I)

C as

es /t

ot al

A dj

us te

d R

R

(9 5

% C

I) a

2 –6

p er

w ee

k 82

/1 17

5 1.

18 (

0. 94

, 1 .4

7) 10

0/ 10

61 1.

24 (

1. 02

, 1 .5

2) 55

/9 43

1. 27

( 0.

97 , 1

.6 8)

4/ 88

4 0.

47 (

0. 17

, 1 .2

7)

≥ p

er d

ay 11

3/ 15

12 1.

12 (

0. 92

, 1 .3

6) 11

2/ 13

62 1.

01 (

0. 83

, 1 .2

2) 66

/1 22

6 1.

08 (

0. 84

, 1 .4

0) 13

/1 15

2 0.

99 (

0. 55

, 1 .7

8)

P f

or t

re nd

0. 10

0. 45

0. 16

0. 50

116 Eur J Nutr (2018) 57:107–117

1 3

between pre-pregnancy coffee intake and SAB downwards, particularly for early losses. With this in mind, it is hard to determine whether the null association we observed between coffee intake and losses at <8 weeks is due to this phenomenon or due to a lack of true biological effect. Recall errors in the self-report of spontaneous abortion are also likely as validation studies conducted in other cohorts suggest that the accuracy of recalling an SAB is 75 % [20]. Fortunately, since information on caffeine intake was assessed prior to outcomes assessment, we would not expect these recall errors to result in recall bias.

As in all observational studies, despite our adjustment and stratification for a variety of potential confounders, we cannot rule out the possibility of residual confounding including confounding by exposures of the participant’s partners which were not assessed in this study. In addition, while we controlled for pre-pregnancy smoking status and conducted a sensitivity analysis in never smokers, there may be residual confounding by misreporting of smoking status and passive smoking. Our study also does not dis- tinguish chromosomally normal from abnormal miscar- riages. If pre-pregnancy caffeine or beverage intake affects chromosomally abnormal versus normal miscarriages dif- ferently, then this heterogeneity in outcome would tend to mask associations that would otherwise be evident in one of these two subgroups. Finally, while the Nurses’ Health Study II cohort does not represent random samples of US women, the distribution of pre-pregnancy caffeine intake and the sources of caffeine were similar to reproductive women in the USA during this same time period [43].

Despite these limitations, our study had many strengths including a large number of pregnancies, prospective design, nearly complete follow-up over the 18 years, abil- ity to assess specific caffeinated and decaffeinated bever- ages, inclusion of both planned and unplanned pregnancies, use of a validated diet assessment, and inclusion of early pregnancy losses. By studying pre-pregnancy caffeine and beverage consumption, we were also able to study an etio- logically relevant exposure time window in regard to risk of SAB without concern of confounding due to nausea and food aversions.

In conclusion, we found that pre-pregnancy coffee con- sumption at levels ≥4 servings/day was associated with increased risk of SAB, particularly at gestation weeks 8–19. The magnitude of coffee’s effect on risk of SAB (3.6 % absolute difference comparing ≥4 servings of coffee per day vs. none) was comparable to the detrimental effects of class 1 and 2 obesity on SAB (4.2 % absolute difference comparing women ≥35 kg/m2 vs. 18.5–24.9 kg/m2) [44] and the beneficial effect of consuming high amounts of supplemental folate (3.1 % absolute difference comparing women consuming no supplemental folate vs. >730 mcg/ day) in this cohort [45]. Since coffee represents the largest

source of caffeine in the US diet, our results are consist- ent with current guidelines from the American Congress of Obstetricians and Gynecologists which suggest women who are pregnant and capable of pregnancy limit their caf- feine intake to <200 mg/day [46]. Our data, however, also suggest that consumption of decaffeinated coffee could be a risk factor for spontaneous abortion. Future research is needed to disentangle the effects of caffeine versus other components of coffee that could be driving this association with pregnancy loss.

Compliance with ethical standards

Conflict of interest On behalf of all authors, the corresponding author states that there is no conflict of interest.

References

1. Greenwood DC, Thatcher NJ, Ye J, Garrard L, Keogh G, King LG, Cade JE (2014) Caffeine intake during pregnancy and adverse birth outcomes: a systematic review and dose-response meta-analysis. Eur J Epidemiol 29(10):725–734

2. Fenster L, Hubbard AE, Swan SH, Windham GC, Waller K, Hiatt RA, Benowitz N (1997) Caffeinated beverages, decaffeinated coffee, and spontaneous abortion. Epidemiology 8(5):515–523

3. Infante-Rivard C, Fernandez A, Gauthier R, David M, Rivard GE (1993) Fetal loss associated with caffeine intake before and dur- ing pregnancy. JAMA 270(24):2940–2943

4. Axelsson G, Rylander R, Molin I (1989) Outcome of pregnancy in relation to irregular and inconvenient work schedules. Br J Ind Med 46(6):393–398

5. Savitz DA, Chan RL, Herring AH, Howards PP, Hartmann KE (2008) Caffeine and miscarriage risk. Epidemiology 19(1):55–62 b9

6. Wen W, Shu XO, Jacobs DR Jr, Brown JE (2001) The associa- tions of maternal caffeine consumption and nausea with sponta- neous abortion. Epidemiology 12(1):38–42

7. Pollack AZ, Buck Louis GM, Sundaram R, Lum KJ (2010) Caf- feine consumption and miscarriage: a prospective cohort study. Fertil Steril 93(1):304–306

8. Wilcox AJ, Weinberg CR, Baird DD (1990) Risk factors for early pregnancy loss. Epidemiology 1(5):382–385

9. Tolstrup JS, Kjaer SK, Munk C, Madsen LB, Ottesen B, Berg- holt T, Gronbaek M (2003) Does caffeine and alcohol intake before pregnancy predict the occurrence of spontaneous abor- tion? Hum Reprod 18(12):2704–2710

10. Hahn KA, Wise LA, Rothman KJ, Mikkelsen EM, Brogly SB, Sorensen HT, Riis AH, Hatch EE (2015) Caffeine and caffein- ated beverage consumption and risk of spontaneous abortion. Hum Reprod 30(5):1246–1255

11. Fulgoni VL 3rd, Keast DR, Lieberman HR (2015) Trends in intake and sources of caffeine in the diets of US adults: 2001– 2010. Am J Clin Nutr 101(5):1081–1087

12. Finer LB, Zolna MR (2014) Shifts in intended and unintended pregnancies in the United States, 2001–2008. Am J Public Health 104(Suppl 1):S43–S48

13. Dott M, Rasmussen SA, Hogue CJ, Reefhuis J (2010) Associa- tion between pregnancy intention and reproductive-health related behaviors before and after pregnancy recognition, National Birth

117Eur J Nutr (2018) 57:107–117

1 3

Defects Prevention Study, 1997–2002. Matern Child Health J 14(3):373–381

14. Mohllajee AP, Curtis KM, Morrow B, Marchbanks PA (2007) Pregnancy intention and its relationship to birth and maternal outcomes. Obstet Gynecol 109(3):678–686

15. Willett WC, Lenart E (1998) Chapter 6: Reproducibility and validity of Food Frequency Questionnaires. In: Willett WC (ed) Nutritional Epidemiology, 2nd edn. Oxford University Press, New York, New York

16. Rimm EB, Giovannucci EL, Stampfer MJ, Colditz GA, Litin LB, Willett WC (1992) Reproducibility and validity of an expanded self-administered semiquantitative food frequency questionnaire among male health professionals. Am J Epidemiol 135(10):1114–1126 (discussion 1127‑1136)

17. U.S. department of agriculture agricultural research service (2012) USDA National nutrient database for standard reference, Release 25. http://www.ars.usda.gov/ba/bhnrc/ndl

18. Salvini S, Hunter DJ, Sampson L, Stampfer MJ, Colditz GA, Rosner B, Willett WC (1989) Food-based validation of a dietary questionnaire: the effects of week-to-week variation in food con- sumption. Int J Epidemiol 18(4):858–867

19. Olson JE, Shu XO, Ross JA, Pendergrass T, Robison LL (1997) Medical record validation of maternally reported birth character- istics and pregnancy-related events: a report from the children’s cancer Group. Am J Epidemiol 145(1):58–67

20. Wilcox AJ, Horney LF (1984) Accuracy of spontaneous abortion recall. Am J Epidemiol 120(5):727–733

21. Kristensen P, Irgens LM (2000) Maternal reproductive history: a registry based comparison of previous pregnancy data derived from maternal recall and data obtained during the actual preg- nancy. Acta Obstet Gynecol Scand 79(6):471–477

22. Rimm EB, Stampfer MJ, Colditz GA, Chute CG, Litin LB, Wil- lett WC (1990) Validity of self-reported waist and hip circumfer- ences in men and women. Epidemiology 1(6):466–473

23. Wolf AM, Hunter DJ, Colditz GA, Manson JE, Stampfer MJ, Corsano KA, Rosner B, Kriska A, Willett WC (1994) Reproduc- ibility and validity of a self-administered physical activity ques- tionnaire. Int J Epidemiol 23(5):991–999

24. Durrleman S, Simon R (1989) Flexible regression models with cubic splines. Stat Med 8(5):551–561

25. Weinberg CR (1993) Toward a clearer definition of confounding. Am J Epidemiol 137(1):1–8

26. Howards PP, Schisterman EF, Heagerty PJ (2007) Potential con- founding by exposure history and prior outcomes: an example from perinatal epidemiology. Epidemiology 18(5):544–551

27. Kost K, Forrest JD (1995) Intention status of U.S. births in, dif- ferences by mothers’ socioeconomic and demographic character- istics. Fam Plann Perspect 27(1):11–17

28. Glei DA (1999) Measuring contraceptive use patterns among teenage and adult women. Fam Plann Perspect 31(2):73–80

29. Chen LW, Wu Y, Neelakantan N, Chong MF, Pan A, van Dam RM (2016) Maternal caffeine intake during pregnancy and risk of pregnancy loss: a categorical and dose-response meta-analysis of prospective studies. Public Health Nutr 19(7):1233–1244

30. Lawson CC, LeMasters GK, Wilson KA (2004) Changes in caffeine consumption as a signal of pregnancy. Reprod Toxicol 18(5):625–633

31. Cnattingius S, Signorello LB, Anneren G, Clausson B, Ekbom A, Ljunger E, Blot WJ, McLaughlin JK, Petersson G, Rane A, Granath F (2000) Caffeine intake and the risk of first-trimester spontaneous abortion. N Engl J Med 343(25):1839–1845

32. Hook EB (1978) Dietary cravings and aversions during preg- nancy. Am J Clin Nutr 31(8):1355–1362

33. Ludwig IA, Clifford MN, Lean ME, Ashihara H, Crozier A (2014) Coffee: biochemistry and potential impact on health. Food Funct 5(8):1695–1717

34. Lawson CC, LeMasters GK, Levin LS, Liu JH (2002) Pregnancy hormone metabolite patterns, pregnancy symptoms, and coffee consumption. Am J Epidemiol 156(5):428–437

35. Kotsopoulos J, Eliassen AH, Missmer SA, Hankinson SE, Twor- oger SS (2009) Relationship between caffeine intake and plasma sex hormone concentrations in premenopausal and postmeno- pausal women. Cancer 115(12):2765–2774

36. Sisti JS, Hankinson SE, Caporaso NE, Gu F, Tamimi RM, Ros- ner B, Xu X, Ziegler R, Eliassen AH (2015) Caffeine, coffee, and tea intake and urinary estrogens and estrogen metabolites in premenopausal women. Cancer Epidemiol Biomarkers Prev 24(8):1174–1183

37. Butler MA, Iwasaki M, Guengerich FP, Kadlubar FF (1989) Human cytochrome P-450PA (P-450IA2), the phenacetin O-deethylase, is primarily responsible for the hepatic 3-demeth- ylation of caffeine and N-oxidation of carcinogenic arylamines. Proc Natl Acad Sci USA 86(20):7696–7700

38. Lee AJ, Cai MX, Thomas PE, Conney AH, Zhu BT (2003) Char- acterization of the oxidative metabolites of 17beta-estradiol and estrone formed by 15 selectively expressed human cytochrome p450 isoforms. Endocrinology 144(8):3382–3398

39. Djordjevic N, Ghotbi R, Bertilsson L, Jankovic S, Aklillu E (2008) Induction of CYP1A2 by heavy coffee consumption in Serbs and Swedes. Eur J Clin Pharmacol 64(4):381–385

40. Vakharia DD, Liu N, Pause R, Fasco M, Bessette E, Zhang QY, Kaminsky LS (2001) Effect of metals on polycyclic aromatic hydrocarbon induction of CYP1A1 and CYP1A2 in human hepatocyte cultures. Toxicol Appl Pharmacol 170(2):93–103

41. Zhu BT, Wang P, Nagai M, Wen Y, Bai HW (2009) Inhibition of human catechol-O-methyltransferase (COMT)-mediated O-methylation of catechol estrogens by major polyphenolic components present in coffee. J Steroid Biochem Mol Biol 113(1–2):65–74

42. Bracken MB, Triche E, Grosso L, Hellenbrand K, Belanger K, Leaderer BP (2002) Heterogeneity in assessing self-reports of caffeine exposure: implications for studies of health effects. Epi- demiology 13(2):165–171

43. Frary CD, Johnson RK, Wang MQ (2005) Food sources and intakes of caffeine in the diets of persons in the United States. J Am Diet Assoc 105(1):110–113

44. Gaskins AJ, Rich-Edwards JW, Colaci DS, Afeiche MC, Toth TL, Gillman MW, Missmer SA, Chavarro JE (2014) Prepreg- nancy and early adulthood body mass index and adult weight change in relation to fetal loss. Obstet Gynecol 124(4):662–669

45. Gaskins AJ, Rich-Edwards JW, Hauser R, Williams PL, Gillman MW, Ginsburg ES, Missmer SA, Chavarro JE (2014) Maternal prepregnancy folate intake and risk of spontaneous abortion and stillbirth. Obstet Gynecol 124(1):23–31

46. The American Congress of Obstetricia ns and Gynecologists (ACOG) Committee Opinion No. 462 (2010) Moderate caf- feine consumption during pregnancy. Obstet Gynecol 116(2 Pt 1):467–468

Reproduced with permission of copyright owner. Further reproduction prohibited without permission.