The first 1000 days of life are known as a critical window for the effect of environmental influences (including maternal dietary intake) [1
] on early growth and development and has become a popular domain for researchers and health care professionals. Within this window, much of the focus has been on the infant before and after birth and to a much lesser degree on women during their preconception period, despite the acknowledgment of the importance of the preconception phase on pregnancy and later health outcomes for both mother and child. When women are in good health and nutritional status before pregnancy, this beneficially influences the health of the child in two ways. First, nutrition may already influence the quality of the oocyte and its environment before conception [2
]. Second, after conception, the first trimester of pregnancy is the period when most fetal organs and the placenta are formed, and when many women are not yet aware of being pregnant. If women are in an adequate nutritional status before pregnancy, this may have beneficial effects mainly during this first, important, part of pregnancy.
Previous studies have investigated the associations between macronutrient intake and birth weight [3
]. Although results from these analyses may provide useful insights to generate hypotheses regarding nutrient intake and outcomes, in practice we do not consume isolated macronutrients. Consequently, multiple food groups may contribute to the observed associations and it has indeed been found that food-based scores may have stronger associations with chronic diseases [5
]. The effects of foods likely reflect complex synergistic contributions from and interactions among food structure, preparation methods, macronutrient quality (e.g., glycemic index and fiber content in case of carbohydrates), and micronutrients content [5
]. In addition, as people modify their nutrient intake primarily by their choice of foods, dietary recommendations are generally based on epidemiologic analysis with foods (e.g., food items or groups), as opposed to analysis based on nutrient intake [6
]. Therefore, our objective is to further explore the association between intake of food groups and pregnancy outcome, more specifically the birth weight of the offspring.
We investigate the association of 22 food groups, classified based on the food-based 2015 Dutch Dietary Guidelines [7
], and birth weight in the Perined-Lifelines linked birth cohort [9
]. Subsequently, we examined which macronutrients may contribute to the association found between food groups and birth weight.
A total of 2368 women from The Lifelines Cohort Study were linked with their data in Perined. After excluding women who did not have reliable or missing dietary intake reported (resp. n
= 427 and n
= 168), pre-term births (gestational age <37 weeks; n
= 110) and unknown sex of the child (n
= 1), 1698 women remained available in the analysis. The characteristics of this group of women have been described in detail previously [4
]. In short, the average maternal age was 29 years (25th–75th percentile: 27–32 years). Children were on average born at 39.4 weeks of gestational age with an had a mean birth weight of 3578 g (SD 472 g). Almost all women were of white (east/west European) ethnicity (97.8%), and the majority completed higher education (55.8%). The average maternal BMI was 23.8 kg/m2
(25th–75th percentile: 21.7–26.6), and average energy intake amounted to 1813 kcal/day (25th–75th percentile: 1545–2141). The average time between the FFQ (baseline) and the delivery of the child was 13 months (25th–75th percentile: 11–16 months). As described previously, there were no differences between the group of women with reliable vs. unreliable dietary intake that we consider to may influence our results in terms of selection bias.
3.1. Food Groups Intake in BMI Quintiles
The median consumption per food group in grams per 1000 kcal over groups of low (quintile 1), normal (quintile 2–4), and high (quintile 5) BMI is shown in Table 1
. The intake of food groups legumes and nuts, tea, refined grain products, and sugary products decreased over the groups of BMI (resp. p
= 0.003, p
= 0.008, p
= 0.03, p
< 0.001) (Table 1
). Intake of food groups eggs, red and processed meat, and artificially sweetened products increased over the BMI groups (Table 1
) (resp. p
= 0.05, p
< 0.001 p
< 0.001). The median LLDS in the complete cohort was 24 (IQR: 20–28) and decreased significantly over increasing BMI quintiles (p
3.2. Association of Intake of Food Groups and Birth Weight
First, unadjusted linear regression analysis was performed (data not shown); however, the R-squared of the models was very low and therefore not reliable (R-squared = 0.03). Adjusted linear regression analysis with adjustment for intake of energy (in kcal), maternal BMI, maternal age, smoking, alcohol, education level, urbanization level, parity, sex of the newborn, and maternal ethnicity showed an increase in explained variability of the model (R-squared = 0.15) and was therefore used for further interpretation of the analysis.
Adjusted linear regression analysis in the complete cohort showed that increased intake of food group artificially sweetened products was associated with increased birth weight (β
= 0.001 [95% CI 0.000 to 0.001, p
= 0.002]) (Table 2
). To illustrate, per 10 g/1000 kcal intake of artificially sweetened products a day, birth weight is 0.001 SD higher (1SD = 472 g). As portion sizes of specific food groups are generally bigger than 10 g, this translates into a relevant further actual increased birth weight. To illustrate, a normal glass contains 200 g of, for example artificially sweetened beverages. So, with an intake of 2000 kcal per day, this makes 100 g/1000 kcal, resulting in an effect size of 10 × 0.001 = 0.01 SED = 4.7 g higher birth weight per portion artificially sweetened beverages.
In addition, increased intake of vegetables was associated with increased birth weight (β
= 0.002 [95% CI 0.000 to 0.003, p
= 0.03]) (Table 2
). Increased intake of food group eggs was associated with decreased birth weight (β
=−0.093 [95% CI −0.174 to −0.013, p
= 0.02]) (Table 2
). When linear regression analysis was performed in stratified groups of BMI quintiles, no association between food group intake and birth weight was shown.
3.3. Macronutrient Contribution to Associations between Food Groups and Birth Weight
To investigate which macronutrients from food group artificially sweetened products might contribute to the association found in linear regression analyses, we performed linear regression analysis with the macronutrients from this food group with birth weight. Analysis was performed twice; first only main macronutrients (total protein, fat, total carbohydrates), and second with the sub-macronutrients (nutrient quality) from animal and plant protein (instead of total protein) and mono- and disaccharides, and polysaccharides (instead of total carbohydrates) were included.
Total and animal protein from artificially sweetened products were significantly associated with increased birth weight (resp. (β
= 0.047 [95% CI 0.004 to 0.089, p
= 0.03]) and (β
= 0.050 [95% CI 0.009 to 0.093, p
= 0.02])) (Table 3
). For analysis with macronutrients from food groups vegetables and eggs, we did not find any significant results.
The aim of this study was to investigate the association of maternal diet based on specific food group intake during the preconception period with the birth weight of the offspring in a linked birth cohort of a representative group of women in the north of The Netherlands [4
]. We showed that after correction for confounders, including maternal BMI and total energy intake, specifically higher intake of food groups “artificially sweetened products” and “vegetables” were associated with increased birth weight, and higher intake of food groups “eggs” was associated with decreased birth weight. In a previous study, we examined the association between preconception macronutrient intake and birth weight. The advantage of representing diet as specific compounds or group of compounds is that such information can be directly related to our dietary intake behavior. However, this approach does not provide any insights into what specific foods may contribute to the associations found.
Our finding that the food group artificially sweetened products was positively associated with birth weight is intriguing given recent studies carried out by researchers at Israel’s Weizmann Institute. They examined three artificial sweeteners (saccharin, sucralose, and aspartame) which are often incorporated into low-caloric snacks and beverages [17
]. The research showed that all three sweeteners may induce metabolic changes such as glucose intolerance which is associated with an increased risk of diabetes mellitus and obesity. More specifically, they analyzed the possible associations between consumption of artificial sweeteners, microbiome composition, and metabolic outcomes in 381 of non-diabetic individuals (44% males and 56% females), and showed that intake of artificial sweeteners was not only associated with various clinical parameters such as BMI, blood pressure, HbA1 C%, and fasting glucose levels, but also with the presence of certain microbiota taxa [18
]. These data suggest that artificial sweeteners and their associated microbiomes may play a crucial role in the regulation of glucose metabolism in humans. Artificial sweeteners stimulate intestinal sugar absorption [19
], cause disruption of the ability of sweet taste to signal caloric consequences [20
], they cause an increase in appetite [22
], and finally, they elicit impaired glycemic or insulin responses [23
The fact that we show a positive association between preconception intake of artificially sweetened products and birth weight may suggest that artificial sweeteners induce similar metabolic changes during preconception that may affect the offspring’s birth weight. During normal pregnancy, changes in insulin sensitivity are a normal physiological phenomenon that provides increased nutrient supply to the fetus. However, if insulin resistance develops and the beta-cell compensation of the pancreas is inadequate in secreting sufficient insulin to maintain normal glycemia in the mother, gestational diabetes mellitus may occur [24
]. It would be interesting to further explore if and how the intake of products with artificial sweeteners may contribute to the pathophysiological mechanisms that also drive the association between gestational diabetes mellitus and obesity and increased birth weight [25
We showed that the intake of artificially sweetened products increased over stratified groups of BMI; however, we adjusted for BMI via linear regression analyses, so our findings cannot fully be explained by the contribution of BMI. Unfortunately, we do not have information on gestational weight gain nor on glucose metabolism or occurrence of (gestational) diabetes mellitus available in this population, which could potentially confound the association with birth weight. However, the prevalence of diabetes mellitus (type 1 and 2) in the complete Lifelines Cohort Study is low, e.g., 4% [26
], approximately 5% lower than reported by the WHO for the Netherlands [27
]. We assume that the prevalence within the complete Lifelines Cohort Study is approximately the same as within our cohort of Lifelines participants, and therefore too low to have any major effect on the association found in this study [28
]. In addition, as described in our previous paper, our cohort is relatively healthy in terms of BMI [4
], one of the major risk factors for the development of gestational diabetes mellitus [29
Within our study, the food group “artificially sweetened products” consists of two main food items. The first is “light soft drinks/lemonade without sugar,” and the second food item is “dairy drinks without sugar/with artificial sweeteners.” Given the fact that we found that it was (animal) protein within this food group that was positively associated with birth weight, this suggests that it is probably the food item ‘dairy drinks without sugar/with artificial sweeteners’ that is driving the positive association with birth weight as the other food item does not contain any protein. These results are in line with results from the Generation R study, a population-based prospective cohort study in Rotterdam, The Netherlands, who showed that higher intake of protein, especially animal protein, in early postnatal life was associated with a greater height, weight, and BMI in childhood up to 9 years of age in the offspring [30
Besides “artificially sweetened products,” we also found an association between food groups “vegetables” and “eggs” with birth weight. Results from the Danish National Birth Cohort showed that fruit and vegetable consumption in pregnancy was positively associated with birth weight in well-nourished Danish women, especially among lean women [31
]. The association between intake of vegetables and birth weight could potentially be due to the higher vitamins and minerals content of these foods, e.g., vitamin C and folic acid. In a study from Portsmouth, an increased birth weight was associated with vitamin C intake in early pregnancy [32
]. Given the fact that the group of women in our cohort represents relatively healthy women with 60% of the women having a normal BMI according to the WHO classification [4
], the association in our study may also be, partially, explained by improved micronutrient intake. As we did not have any information on micronutrient intake, we have not been able to further investigate this.
Regarding the association found between food groups, eggs, and birth weight, there is less scientific evidence to support this finding.
For many food groups (e.g., fruits, whole-grain products, red and processed meat) and overall diet quality (LLDS) we found a clear association with maternal BMI but most were not significantly associated with birth weight. More specifically we showed that mean diet quality (LLDS) slightly decreased with increasing BMI [8
]. We also showed that on the one hand, sugar containing product intake decreased with increase of BMI, while on the other intake of artificially sweetened products increased. This shift in product preference with BMI has been described previously by several large-scale studies, including the National Health and Nutrition Examination Survey [34
]. This supports our findings, shows our cohort to be representative, and suggests that similar findings can already be identified in much smaller but focused cohort studies like the Perined-Lifelines linked birth cohort.
As maternal BMI is a stronger predictor for birth weight than dietary intake [4
], it is potentially the diet quality of the mother that can support a healthy pregnancy, but it is not something that is likely to have an acute effect on the offspring. Obviously, the diet may have already impacted maternal BMI, which is subsequently translated to the birth weight of their offspring. In this sense, maternal BMI may simply be the mediator between preconception dietary intake and birth weight.
Within our analysis we have not corrected for multiple testing given the more exploratory background of the analysis. Our results are indicating a direction of association between preconception dietary intake and birth weight. Further research should be performed within large cohorts investigating the association between dietary intake and birth weight to be able to draw more firm conclusions.