Introduction

Early childhood development is influenced by the interaction of genetic, environmental, and socioeconomic factors. The risk factors affecting early childhood development include biological (e.g., stunting, infections, low birth weight, and preterm birth), environmental (i.e., undernutrition, tobacco smoking and chemical exposure, and stress), and social (e.g., poverty and air pollution).1,2,3 The Developmental Origins of Health and Disease (DOHaD) concept has proposed that several environmental factors in fetal stage and infancy are associated with a higher risk of non-communicable diseases among offsprings.4,5 A poor environment early in life has long-term physiological and epigenetic effects on brain development and maturation.6 Our previous study showed that active smoking during pregnancy was associated with altered DNA methylation and attention-deficit/hyperactivity disorder symptoms among preschool children.7

Malnutrition is defined as undernutrition, overnutrition, or energy or nutrient intake imbalance. The nutrients involved in maternal malnutrition include calories, proteins, carbohydrates, fats, vitamins, and minerals. Maternal malnutrition during pregnancy can adversely affect fetal growth and development. The effects include intrauterine growth restriction, low birth weight, preterm birth, and birth defects (e.g., neural tube defects and iodine deficiency).8,9 In addition, various animal and epidemiological studies have shown that maternal malnutrition during pregnancy increases the risk of obesity, diabetes, coronary heart disease, and hypertension among offsprings.10,11 Notably, low and high maternal protein intakes have an adverse effect on the growth of not only fetuses but also children.12 However, recent studies have revealed that protein intake is not associated with child growth and metabolic health.13,14 Thus, its effects on children are controversial.

Animal studies have shown that maternal protein restriction during pregnancy and lactation causes abnormal brain development among offsprings.15,16 Moreover, Gould et al. revealed that maternal protein restriction during preimplantation alone can adversely affect neural cell differentiation and short-term memory among offsprings.17 In addition, in a model excluding the effects of pre-pregnancy egg formation and post-pregnancy lactation caused by low protein (LP) intake, Furuse et al. showed that maternal protein restriction after implantation causes gene expression, epigenetic abnormalities in the brain, and behavioral abnormalities among offsprings.18 However, no study has validated the association between LP intake among pregnant mothers according to birth cohort and child developmental delay. Thus, this study aimed to assess the correlation between low maternal protein intake and child developmental delay at age 3 years using data obtained from a large cohort.

Methods

Study design and population

The Japan Environment and Children’s Study (JECS) is a national birth cohort study planned by the JECS Working Group. The detailed protocols have been published elsewhere.19,20 Regional centers across Japan participate in the JECS, and women in early pregnancy and their partners who lived in the area around the regional center were recruited and followed up by the Center. Participants were recruited between January 2011 and March 2014. Children who participated will be monitored until the age of 13. The JECS protocol was reviewed and approved by the Ministry of the Environment’s Institutional Review Board on Epidemiological Studies and the Ethics Committees of all participating institutions (ethical approval number: 100910001). The JECS was conducted in accordance with the Declaration of Helsinki, and written informed consent was obtained.

We used the dataset jecs-ta-20190930-qsn, which was released in October 2019. In total, there were 104,062 records. However, we excluded cases of abortion, stillbirth, and multiple pregnancies as well as incomplete data obtained using the Food Frequency Questionnaire (FFQ), a verified self-administered dietary questionnaire, during recruitment and the Japanese version of the Ages and Stages Questionnaire, Third Edition (J-ASQ-3) at age 3 years. Finally, 77,237 mother–child pairs were analyzed in the current study (Fig. 1).

Fig. 1
figure 1

Flow chart of the participants.

Full size image

Dietary information

Dietary information was obtained using FFQ, a validated self-administered dietary questionnaire evaluated in previous studies.21 The FFQ contains 172 food and beverage items, 9 frequency categories, and 3 portion sizes. Respondents were asked about their dietary status in early pregnancy (FFQ1, which was used to assess information from the previous year before conception) and mid-pregnancy (FFQ2, which was used to obtain data from conception to the answering date). The food and nutrient intakes were calculated using the FFQ items. Protein, fat, and carbohydrate (PFC) proportion was calculated as a percentage of energy intakes. We categorized the participants into three groups based on the following protein-to-total energy intake ratios obtained using the FFQ1: <9.39% (more than 2 standard deviation [SD] below the mean), 9.39–<13%, and ≤13%. The participants were grouped as follows in terms of protein intake during early pregnancy: ≥13%, the normal protein (NP) intake group; 9.39–<13%, the LP intake group; and <9.39%, the severely low protein (SLP) intake group. Similarly, we categorized the participants based on the protein-to-total energy ratios obtained using the FFQ2: <9.36% (more than 2 SD below the mean), 9.36–<13%, and ≤13%. The participants were grouped as follows in terms of protein intake during mid-pregnancy: ≥13%, the NP intake group; 9.36–<13%, the LP intake group; and <9.36%, the SLP intake group.

The individual food taken (gram per day) included cereals, vegetables, fish and shellfish, meats, fruits, potatoes and starches, sugar and sweets, beans, nuts and seeds, pickles, green and yellow vegetables, other types of vegetables, mushrooms, algae, eggs, milk, oils and fats, seasonings and spices, confectionery, alcoholic drinks, soft drinks, water, and soup. The food intakes were converted per 1000 kcal of energy intake.

The frequency of breakfast intake was extracted from a self-reported questionnaire in early pregnancy. Each mother answered about how often she ate breakfast by choosing frequency from either less than once a month, 1–3 times a month, 1–2 times a week, 3–4 times a week, 5–6 times a week, or every day. We have recategorized these data into three categories: less than 3 times a month, 1–4 times a week, 5 times a week or more.

Outcome definitions

Children’s neurodevelopment at the age of 3 years was assessed using the J-ASQ-3, which was included in the questionnaire sent when the child reached age 3 years (35–36 months). The ASQ-3 is a developmental screening questionnaire completed by parents/caregivers. ASQ-3 is a set of questionnaires about children’s development from 2 to 66 months which can be self-administered by parents/caregivers. It is a Japanese translation of ASQ-3, a tool for screening children with developmental delays.22 The J-ASQ-3 includes 30 questions across five domains: communication, gross motor, fine motor, problem-solving, and personal–social skills. A respondent chooses yes, sometimes, or no for each question and scored 10 for yes, 5 for sometimes, and 0 for no. The total scores for each domain were calculated. Mezawa et al. verified the J-ASQ-3 and reported a cutoff score (more than 2 SD below the mean) for developmental delay.22 The cutoff scores for each domain among children aged 3 years were as follows: communication = 29.95, gross motor = 39.26, fine motor = 27.91, problem-solving = 30.03, and personal–social = 29.89.

Covariates

We selected potential confounders associated with child development during pregnancy and early infancy based on previous studies.23,24,25 Information on maternal smoking and drinking during pregnancy, annual household income (million yen), maternal age at birth, paternal and maternal educational level, and older siblings was collected from mothers via self-administered questionnaires during the first and second/third trimesters of pregnancy. Child sex, pre-pregnancy body mass index (BMI), birth weight, and gestational week at birth were extracted from medical record transcripts. Information on childcare facility attendance and breastfeeding was collected from the self-questionnaire administered to the mothers when the child was age 1 year. Maternal age at birth was classified as <25 years, 25–29 years, 30–34 years, and ≥35 years. Preterm birth was defined as delivery before 37 weeks of gestation. Pre-pregnancy BMI was divided into three groups: underweight (<18.5 kg/m2), normal (18.5–24.9 kg/m2), and overweight (≥25 kg/m2).

Statistical analyses

The association between the proportion of protein intake before and during pregnancy (FFQ1) and child development delay at age 3 years was evaluated via a logistic regression analysis to estimate the crude odds ratio (cOR) and adjusted OR (aOR) after adding all the covariates and 95% confidence interval (CI). Then, the following three models were used: model 1, crude model; model 2, adjusted for maternal smoking and drinking during pregnancy, maternal age at birth, pre-pregnancy BMI, maternal and paternal education level, annual household income, older siblings, child sex, birth weight, gestational age at birth, attends a childcare facility at age 1 year, and breastfeeding until age 1 year; and model 3, further adjusted based on the proportion of protein intake in mid-pregnancy (FFQ2) to clarify the effects of protein intake in early pregnancy (FFQ1). A similar analysis of the proportion of protein intake in mid-pregnancy (FFQ2) was performed. In this case, model 3 used the proportion of protein intake obtained using the FFQ1. Missing values were excluded from the multivariate analysis. Differences in the intake of total energy, three major nutrients, and food group among the three groups were evaluated using a one-way analysis of variance. All statistical analyses were performed using the Statistical Package for the Social Sciences software version 27. A p value of 0.05 (two-sided) was considered statistically significant.

Results

Table 1 shows the characteristics of the participants. Based on the FFQ1, the protein-to-total energy intake ratios of 1389, 29,251, and 46,597 participants were <9.39% (−2 SD), 9.39–<13%, and ≤13%, respectively. According to the FFQ2, the protein-to-total energy intake ratio of 1492, 28,899, and 46,846 were <9.36%, 9.36–<13%, and 46,846 for ≥13%. The numbers of children aged 3 years who had scores below the cutoff for each domain in the J-ASQ-3 were as follows: 3144 (4.1%) for communication skill, 3570 (4.6%) for gross motor skill, 5927 (7.7%) for fine motor skill, 5717 (7.4%) for problem-solving skill, and 2618 (3.4%) for personal–social skill.

Table 1 Characteristics of the study population.
Full size table

Table 2 depicts the results of the univariate and multivariate logistic regression analyses of the association between the proportion of protein intake before and during early pregnancy (FFQ1) and child developmental delay at age 3 years. For communication skill, SLP intake was significantly associated with a higher risk of child developmental delay in model 1 (OR 1.54; 95% CI, 1.23–1.93) and model 2 (OR 1.38; 95% CI, 1.04–1.83). For fine motor skill, there was a significant correlation between SLP intake and a higher risk of child developmental delay in all models ([model 1: OR, 1.52; 95% CI, 1.28–1.81], [model 2: OR, 1.46; 95% CI, 1.18–1.80], and [model 3: OR, 1.43; 95% CI, 1.15–1.78]). For problem-solving skill, LP intake was significantly associated with a higher risk of child developmental delay in model 1 (OR 1.06; 95% CI, 1.00–1.12) and model 2 (OR 1.08; 95% CI, 1.01–1.15). Moreover, there was a significant correlation between SLP intake and a higher risk of child developmental delay in all models ([model 1: OR, 1.46; 95% CI, 1.22–1.74], [model 2: OR, 1.45; 95% CI, 1.17–1.80], and [model 3: OR, 1.40; 95% CI, 1.12–1.75]).

Table 2 Association between protein intake in early pregnancy (FFQ1) and development delay at age 3 years.
Full size table

Table 3 shows the results of the univariate and multivariate logistic regression analyses of the association between the proportion of protein intake in mid-pregnancy (FFQ2) and development delay at age 3 years. For communication skill, SLP intake was significantly correlated with a higher risk of child developmental delay in model 1 (OR 1.33; 95% CI, 1.06–1.67) and model 2 (OR 1.36; 95% CI, 1.03–1.80). For fine motor skill, there was a significant association between LP and a higher risk of child developmental delay in all models ([model 1: OR, 1.12; 95% CI, 1.06–1.19], [model 2: OR, 1.09; 95% CI, 1.02–1.16], and [model 3: OR, 1.07; 95% CI, 1.00–1.14]).

Table 3 Association between protein intake during mid-pregnancy (FFQ2) and development delay at age 3 years.
Full size table

We focused on the proportion of LP intake based on the FFQ1 and compared nutrient and food intakes. Table 4 shows the difference between the total energy and three major nutrient intakes among the three groups. In early pregnancy, the LP and SLP groups had a significantly lower total caloric intake than the NP group. In addition, the protein and lipid intakes were low. Meanwhile, the rate of carbohydrate intake was high.

Table 4 Difference in the total caloric intake and that of three major nutrients among the three groups.
Full size table

Table 5 shows the intakes per 1000 kcal of each food group among the three groups based on the FFQ1. The intakes of 17 food groups (potatoes and starches, beans, nuts and seeds, vegetables, pickles, green and yellow vegetables, other vegetables, mushrooms, algae, fish and shellfish, meats, eggs, milk, oils and fats, seasonings and spices, water, and soup) in the LP and SLP groups significantly decreased. Meanwhile, the intake of cereals, sugar and sweetener, fruits, confectionery, alcoholic drinks, and soft drinks significantly increased.

Table 5 Difference in the intake of each food group among the three groups based on the FFQ1.
Full size table

Figure 2 shows the frequency of eating breakfast among the three groups as assessed using the FFQ1. The proportions of mothers who ate breakfast five or more times a week were as follows: 84.4%, 77.9%, and 59.8% for the NP, LP, and SLP groups, respectively.

Fig. 2: Comparison of breakfast intake frequency before and in early pregnancy among the three groups.
figure 2

NP Normal protein, LP Low protein, SLP Severely low protein.

Full size image

Discussion

We found that SLP intake before and during early pregnancy was associated with a significantly higher risk of impaired communication, fine motor and problem-solving skills at age 3 years. Mothers with SLP intake before or during early pregnancy had low food intakes of several food groups such as fish and shellfish, meats, and vegetables. However, they had high intakes of cereals, confectionery, and alcoholic and soft drinks. In addition, they were more likely to skip breakfast.

According to the Japanese Dietary Intake Standards revised in 2020, the estimated average protein requirement for adult women is 50 g per day, with a protein proportion of 13–20% of the PFC ratio.26 Thus, pregnant women should take an additional 5 g of protein in the second trimester and 25 g of protein in the third trimester. Animal studies have reported that severe protein restriction during pregnancy (from the usual 18–20% to 5–10%) is associated with neurodevelopmental abnormalities in offspring.17,18,27 In humans, the protein ratio is not as severely low as in experimental animals; however, we aimed to identify the mothers with SLP intake ratios in the Japanese population and clarify the effects on their child’s development. In this study, the cutoff value was set at >2 SD below the mean to extract mothers with severely lower protein intake (SLP) during pregnancy. Since the number of mothers with a protein ratio of 20% (n = 545 in early pregnancy and n = 562 in the mid-pregnancy) was small, mothers with a protein ratio of ≥13% were defined as NP. Therefore, we defined the NP intake group as ≥13%, the LP intake group as mean −2 SD to 13% and the SLP intake group as mean <−2 SD.

In Japan, the birth weight is rapidly decreasing, and one of the causes is weight loss due to diet among pregnant mothers.28,29 Low birth weight and preterm birth are associated with a high risk of child developmental delay.1,2 Moreover, this study confirmed this association (data not shown). Meanwhile, LP intake during pregnancy can adversely affect fetal growth and development,30 and its association with low birth weight and preterm birth is controversial.31,32,33 Our results showed that LP intake in early and mid-pregnancy was not significantly associated with a higher risk of low birth weight, although birth weight was slightly low. The high risk of preterm birth was significantly associated with SLP intake in mid-pregnancy (data not shown). Maternal undernutrition is more detrimental in early than in late pregnancy.34 Animal studies have shown that protein restriction in early pregnancy causes brain development and behavioral disorders in pups.17,35 Our results also show that SLP intake in early pregnancy has a greater effect on child development than SLP intake in mid-pregnancy. In addition to preterm birth and low birth weight, this study adjusted for factors affecting child development such as maternal smoking and drinking during pregnancy, maternal age, pre-pregnancy BMI, education background of parents, annual household income, older siblings, child sex, attends a childcare facility, and breastfeeding until age 1 year.1,2,3 We showed a significant association even after adjusting for these confounding factors (model 2). We adjusted the protein intake status at another time in addition to Model 2 to clarify whether the effects of SLP intake are in early or mid-pregnancy (Model 3). As a result, SLP intake during early pregnancy is significantly associated with an increased risk of delayed child development. However, a moderate correlation was observed between protein proportions of FFQ1 and FFQ2 (r = 0.483, p < 0.01), suggesting that Model 3 results may be over-adjusted.

Our results showed that the energy intake is lower in the LP proportion group. In addition, since the proportion of carbohydrate-to-energy intake is high in the LP proportion group, insufficient protein, and fatty acid intake leads to low energy intake. Recently, not only LP intake but also high protein intake was found to be associated with a risk of growth retardation in childhood.36,37 The research also examined the effect of protein intake above 20%. Results showed no significant association between protein intake and a higher risk of child developmental delay at age 3 years. Our results showed that women with LP proportion had an adequate or excessive intake of sweets and alcohol. In addition, the percentage of women who skipped breakfast was high. Women who skip breakfast during pregnancy have low blood levels of EPA, DHA, and β-carotene, and low urinary levels of urea nitrogen and potassium, which can affect fetal growth and development.38 The group with an LP proportion was likely to be younger, thereby supporting the diet preference and nutritional bias of young pregnant women, which has become a problem in Japan in recent years.29 Therefore, LP intake due to an inappropriate dietary environment among young women before and during early pregnancy can adversely affect child development.

Using a large birth cohort, we showed for the first time that SLP intake during early pregnancy was associated with a higher risk of child developmental delay. Several animal studies have shown that maternal protein restriction in early pregnancy causes behavioral abnormalities such as less fear/anxiety-like behavior and high impulsiveness and exploration in pups.39,40,41 The involvement of epigenetics, particularly DNA methylation, has been considered a molecular mechanism linking maternal LP intake in early pregnancy and abnormal neural development in pups.17,18 Maternal protein restriction results in certain amino acid deficiencies, referred to as amino acid imbalance, in both the mother and the fetus. Amino acids function as not only protein constituents but also essential precursors for the synthesis of various molecules such as hormones, neurotransmitters, nitric oxide, glutathione, carnitine, and polyamines.42 Amino acid imbalance can affect not only the risk of intrauterine growth restriction and preterm birth but also postnatal development.11,43 Hence, future cohort studies should validate the molecular mechanism linking amino acid imbalance in early pregnancy and the risk of child developmental delay.

The current study had several strengths. That is, a large sample size was included, which regulated several potential confounders and improved estimation accuracy. In addition, nutritional status between early and mid-pregnancy was analyzed. Our results provided evidence regarding the importance of reinforcing the DOHaD concept, which showed that malnutrition in early pregnancy is associated with a higher risk of neurodevelopmental impairment. Nevertheless, this study also had several limitations. First, because the survey used self-administered questionnaires, recall bias and personal understanding of the questions could have affected the responses. We require that the completed questionnaire be returned within 2 weeks after their arrival. However, some of them may not be returned within the deadline; therefore, a misclassification is possible, especially in ASQ at the age of 3 years. In addition, the validity of the FFQ in Japan has been well verified.44 However, its validity in pregnant women is unclear. Second, the FFQ1 was used to collect information about maternal diet and nutrition over the previous year. Moreover, it is expected that a woman will not change her dietary habits until she gets pregnant. However, the responses might reflect pre-pregnancy status, not early pregnancy. Finally, regarding the nutritional environment after birth, breastfeeding was adjusted. However, the content of baby food and the nutritional environment in early childhood were not considered.

Conclusion

SLP intake caused by inadequate diet during early pregnancy was associated with a higher risk of child developmental delay at age 3 years. Since nutritional imbalance in early pregnancy affects not only fetal growth but also postnatal neurodevelopment, nutritional management before pregnancy is considered important.