Data for the present study were derived from the Korean Pregnancy Outcome Study (KPOS), a prospective cohort study. Between March 2013 and January 2017, all pregnant women who visited Cheil General Hospital and CHA Hospital for antenatal care during the first trimester were asked to participate in the KPOS. Women were excluded from enrolment if they were not Korean or were pregnant with triplets or higher-order multiple gestations. Gestational age was determined based on the date of the last menstrual period in women who had conceived naturally, and was confirmed by the first trimester ultrasound. After the first antenatal visit, eligible participants were requested to complete several sets of questionnaires or examinations at each of the following visits: visit 1 in the first trimester (around 12 weeks of gestation); visit 2 in the second trimester (around 24 weeks of gestation); visit 3 in the third trimester (around 36 weeks of gestation); visit 4 at birth; and visit 5 at 4–6 weeks after birth.
As shown in Supplementary Figure 1 [see Additional file 1], after excluding 55 individuals with missing dietary data, we performed a cross-sectional analysis of 3,510 women who had positive pregnancy results. Trained research nurses explained the study in detail, obtained written informed consents, and completed questionnaires. All participants provided written informed consent, and the study protocol was approved by the Institutional Review Board (IRB) of Cheil General Hospital (IRB number: CGH-IRB-2013-10) and CHA Hospital (IRB number: 2013-14-KNC13-018), separately. It was clearly explained to all participants that they were free to withdraw from any part of the study at any point in time.
A face-to-face interview was conducted to evaluate participants’ socio-demographic profiles, medical and family history, reproductive information, health-related behaviors, and psychological health.
Data on socio-demographic status included age, educational level, household income, employment status, marital status, cohabiting family composition, and information on spouses. Family history of hypertension, diabetes, gestational diabetes mellitus, preeclampsia, depression, and other mental illness was also taken.
Participants underwent clinical and laboratory examinations, including anthropometric measurements, blood pressure measurements, and blood and urine laboratory tests, during pregnancy. Asian classification of obesity used in this study using the body mass index (BMI) . Symptoms of depression were assessed using the Korean version of the Edinburgh Postnatal Depression Scale (K-EPDS), which is a reliable measurement for peripartum depression and validated questionnaire with 10 items; those with K-EPDS scores ≥10 were considered to have symptoms of antenatal depression [14, 15]. Those taking anti-depressant drugs and those with a self-reported physician’s diagnosis of depression were considered to have a history of depression. Cigarette smoking, alcohol intake, and supplement intake were evaluated during each visit. Physical activity was assessed during each visit with a self-reported questionnaire.
Dietary intake patterns were evaluated using a questionnaire during the first visit. The coffee consumption pattern before conception was determined through the question, “How often did you drink coffee before the pregnancy?” on the questionnaire. Coffee consumption was categorized into five groups (seldom, 2–3 cups/week, 4–6 cups/week, 1 cup/day, and 2 or more cups/day). In the analysis, participants were divided into four groups based on their reported amount of coffee consumption: ≥2 cups/day, “heavy coffee drinkers”; 1 cup/day, “moderate coffee drinkers”; <1 cup/day, “light coffee drinkers”; and <1 cup/week, “seldom coffee drinkers” (reference group). Preferences for the following types of coffee were noted: black coffee, black coffee with sugar, black coffee with creamer, and instant coffee mix (instant coffee with creamer and sugar).
We obtained information on antenatal pregnancy complications and birth details. First trimester complications, including emesis and bleeding in early pregnancy, were assessed during the first visit. In this study, bleeding in early pregnancy was defined as the occurrence of vaginal bleeding in the first 20 weeks of pregnancy, confirmed using ultrasonographic examinations by a physician [16, 17]. The birth outcomes included gestational age at birth, type of labor (induced or spontaneous), type of birth, indication for Caesarean birth, and birth complications.
Blood pressure was measured during every visit using the automatic oscillometric technique, but a diagnosis of hypertensive disorders of pregnancy was confirmed by manual measurements using blood pressure cuffs and auscultation. Blood samples and placenta were stored in -70 ºC freezers at a controlled temperature and humidity. All biological samples were marked with barcodes and stored in the National Biobank of Korea. We uploaded the data from all questionnaires and examinations to a web-based clinical data management system (iCReaT) managed by the Korea National Institute of Health.
We summarized the general characteristics of study participants using means and standard deviations for continuous variables and observed numbers and percentages for categorical variables. To statistically analyze differences among groups, a general linear model and the chi-square test were used for continuous and categorical variables, respectively. The Bonferroni post-hoc test was used to identify groups showing significant differences. For some analyses, the lower categories of exposure variables were combined into a single stratum because of the small number of subjects in these categories. Multivariate logistic regression analysis was used to estimate odds ratios (ORs) with 95% confidence intervals (CIs) for the association between coffee consumption and bleeding in early pregnancy. Age, BMI, systolic blood pressure, cigarette smoking and alcohol consumption behavior, previous and current physical activity levels, stress levels, history of depression, presence of antenatal depressive symptom during the first trimester, type of emesis, parity, and the number of livebirths, stillbirths, miscarriages, and abortions were considered as covariates in the adjusted model.
An additional sensitivity analysis was performed with stratification according to age and BMI. We also performed an additional multiple logistic regression analysis to estimate the association between the type of coffee preferred and the risk of bleeding in early pregnancy, as well as the association between coffee consumption and the risk of miscarriage and stillbirth. All statistical analyses were performed using the SAS software (version 9.4, SAS; NC, USA), and two-sided p-values less than 0.05 were considered indicators of statistical significance.