Maternal caffeine intake during pregnancy and the risk of delivering a small for gestational age baby: Kuopio Birth Cohort

In KuBiCo, pregnant women were recruited (majority,  > 90% during the first trimester) by personnel at the prenatal clinics of outpatient healthcare centers in Northern Savo, Finland. The only inclusion criteria in the study were that the participants were expected to give birth after 22nd gestational week in Kuopio University Hospital (KUH) and to be proficient in Finnish.

During the first and third trimesters of pregnancy, three questionnaires were filled out by the study participants (Environmental exposure questionnaires, Food frequency questionnaires (FFQ1 and FFQ2) and Edinburgh Postnatal Depression Scale questionnaires). In addition to these, all pregnant women filled out various questionnaires evaluating the background risk for ongoing pregnancy in Kuopio University Hospital’s Pikku-Haikara system intended for pregnant women. By agreeing to the KuBiCo study, the women give permission for the use of all those and hospital data related to pregnancy and childbirth in the KuBiCo study.

In this study, we included those who had given birth in KUH and whose dietary data were available. In total, 2007 (the first trimester) and 4362 (the third trimester) pregnant women, who had completed the FFQ between March 2013 and December 2022, were included in the study. Of these women, 1262 had completed both the first and third trimester FFQ. All women gave informed consent during recruitment. The study data were collected year-round with standardized, supervised methods according to the study protocol and it represent Finnish pregnant women well [35]. The overall participation in KuBiCo during 2013–2022 (7944 women) has been 35.2% of all parturients giving birth in KUH.

Maternal diet was estimated using an FFQ filled by the participants during the first (up to 13rd week) and the third trimester of gestation (from 28th gestation week to delivery). The FFQ was designed to cover the whole diet during the preceding three months. Average daily nutrient and energy intakes were calculated using the Finnish national food composition (Fineli®) database in February 2023.

The FFQ used was a slightly modified version from that validated in the Kuopio Breast Cancer study [36]. To reflect the increased selection of food items available, for example information on the use of energy drinks and bars, were collected in the modified FFQ. In the KuBiCo, the participants completed the web-based FFQ which also produced automatized personal feedback to all study participants within 24 h after they returned the online FFQ.

The FFQ comprises a food list of 160 food and drink items with 9 frequency response options, ranging from ‘never’ to ‘6 or more times per day’. The portion sizes were fixed and specified with natural units (e.g., mug, glass, slice) or grams. The average daily intake of listed food items and nutrient intakes were calculated by multiplying the frequency of the food intake by the portion size (grams). The output provided more than 60 nutrients and 100 food groups. The caffeine intake was estimated by summarizing the daily intake of coffee (caffeine 70 mg/100 ml), cola drinks (20 mg/100 ml), tea (15 mg/100 ml), energy drinks (32 mg/100 ml), cocoa (3 mg/100 ml), and chocolate (30 mg/100 g).

In this study, the following descriptive characteristics were collected from the KUH birth registry: maternal age, body mass index (BMI; kg/m²) in the first trimester, weight change during pregnancy, smoking during pregnancy, the duration of education, relationship status, parity, the mode of delivery, and maternal diseases and complications. Also, newborn sex, the gestational age at birth, birth weight, head circumference, Apgar Scores (at 5 min), the duration of hospital stay after birth, the need of neonatal intensive care, and congenital anomaly (ICD-10, Q Codes) at birth, were recorded. According to WHO [37], low birth weight was defined as birth weight  < 2500 g, and preterm birth was defined as delivery before 37 weeks of gestation. Macrosomia was classified as birth weight  > 4500 g [38]. Further, we divided the birth weights into two subgroups according to birth weight for gestational age: SGA and large for gestational age (LGA). SGA and LGA were defined as birth weights corrected for gestational age beyond  ± 2 standard deviations (SD) from the mean, according to the sex-specific Finnish fetal growth curves [39]. Maternal hypertension and pre-eclampsia were defined according to ICD-10 codes (O10, O11, O13, O14, O15 and O16).

The statistical analyses were conducted using IBM SPSS Statistics version 25 (SPSS Inc., Chicago, IL, USA). P-values  < 0.05 were considered as significant. Caffeine intake was classified into three categories: low \(\le \)50 mg/d, moderate 51–200 mg/d and high caffeine intake  > 200 mg/d. Caffeine and caffeine containing food intakes were expressed as the mean values with SD and the median values with interquartile range (IQR). Differences between the first and third trimester caffeine intakes were estimated by Wilcoxon test. Maternal and neonatal characteristics were expressed as the mean values with SDs in appropriate units or as percentages (%). Differences between caffeine intake groups were analyzed using ANOVA test for continuous variables and by Chi- square (χ²) test for categorical variables.

Logistic regression models were used to examine associations between maternal caffeine intake in categories and SGA. The results were expressed as odds ratios (ORs) with 95% confidence intervals (CIs) and P-values. Low-caffeine consumers were used as the reference group. Three regression models were used in the analyses. Model 1 was adjusted for maternal age and early pregnancy BMI. Model 2 was adjusted for model 1, smoking and energy intake. Model 3 was adjusted for model 2, the consumption of vegetables, fruits and berries and fish, and hypertension or pre-eclampsia during pregnancy. The association between the first trimester caffeine intake and SGA was also analyzed by dividing caffeine intake into quartiles, quintiles, deciles and into two groups based on the caffeine intake recommendation (\(\le/>\)200 mg). Also, coffee consumption was classified into tertiles and into two groups (\(\le/>\)100 ml), and the associations with SGA were examined by logistic regression. Pearson’s correlation was used to estimate the strength and the direction of the associations.

In our study, the risk of having an SGA newborn was 2% of all participants and 2.5–2.6% among women who consumed caffeine moderate or high amounts. Although the prevalence of SGA children born in this study population was low, a significantly higher prevalence was observed in the group of those who consumed the most caffeine (2.6%, n = 16/627) as compared to those in the lowest intake group (0.9%, n = 6/655) in the first trimester. The prevalence of SGA was at the similar level as in a Norwegian cohort study [21] and in an earlier Finnish birth cohort study [42]. The definition of SGA varies, and in some countries and studies, SGA has typically been defined as a newborn who is below the 10th percentile of birth weight for gestational age [17]. In our study, the definition of SGA was defined as having a birth weight more than two SDs below the mean weight for the gestational age which is stricter and gender and region-based way to select the cases. Compared to birth weight alone, SGA is more versatile concept for describing restricted fetal growth. In addition to SGA as a neonatal outcome, SGA newborns are in the risk for several health problems later in life.

In the analyses of this study, the associations between caffeine intake and SGA were strengthened when adjusted regression models were used. The association were most pronounced in the model with SGA-related covariates such as dietary factors and hypertension or pre-eclampsia. It is well known that caffeine intake is associated with unhealthy lifestyle habits, such as smoking [5, 43] which is also the known risk factor for SGA. Thus, we also adjusted the statistical analyses by smoking.

Exposures in early pregnancy may be more relevant for SGA because early pregnancy is an essential time for the development of the placenta and fetus. The placenta starts to develop in the very early pregnancy and maternal lifestyle can detrimentally influence early placental development and its function later [44] and contribute to restricted fetal growth [45]. Caffeine exposure during a sensitive time frame of pregnancy may also induce epigenetic changes in the developing fetus [46]. It is possible that the consumption of caffeine later in pregnancy does not affect fetal growth if the placenta has developed normally in the early pregnancy.

In Finland, coffee intake is the highest worldwide [47]. The Finnish recommendations for pregnant women state that caffeine should not be consumed more than 200 mg/day and energy drinks are not recommended during pregnancy at all [48]. Based on our recent and previous studies [5], pregnant women consume caffeine a lot. In this study, median-caffeine intake was 121 mg/day in early pregnancy and even higher, 134 mg/day in late pregnancy. In comparison to one recent Nordic study [21], maternal caffeine intake was much lower: median 58 mg/day during mid-pregnancy. The wide distribution of caffeine intake increases the reliability of our study. We also evaluated all dietary sources of caffeine, not only coffee. However, coffee was the primary source of caffeine covering three quarters of intake. Because the consumption of energy drinks was low during pregnancy (2–4% of pregnant women), it was not possible to assess its independent association with SGA.

According to our results, caffeine intake from various sources increases from the first to third trimester of pregnancy. We found that total caffeine intake during the first, but not the third trimester, was significantly associated with the increased risk of SGA. Because our FFQ covered the three preceding months, the first trimester questionnaire extended to the pre-pregnancy period in some participants, and since the first trimester FFQ was closed already at the end of the 13th week of pregnancy, not all study subjects completed the first nutrition questionnaire. Thus, if the results of our study were repeated in other studies, it would be important to extend caffeine recommendations also for women who are planning for pregnancy.

Nutrition studies have some limitations, such as self-reporting and recall bias related to the dietary intake methods. In caffeine-containing drinks, factors like the brand, the type of beans and leaves, preparation, and the amount of caffeine may also cause bias, considering our approximation of maternal caffeine intake [49]. However, Gleason et al. [9] found that the results from self-reported and biomarker-measured caffeine exposure are consistent with each other. As a dietary assessment method, the FFQ is reliable for epidemiological cohort studies like KuBiCo [50]. In this study, nutritional factors were assessed by using the validated and pregnancy-modified FFQ. The presented factors and limitations were addressed by using relevant statistical analyses with adjustments for several potential confounders. The strengths of this study were the population-based prospective study design and large range in caffeine intake due to the high consumption in Finland. Also, careful adjusting with potential cofounding factors brought novelty value to the research.