Association Between Periconceptional Diet Quality and Hypertensive Disorders of Pregnancy: The Japan Environment and Children's Study

Abstract

Background

Little is known about the relationship of healthy diets, which are widely recommended to prevent diseases in general populations, with the risk of hypertensive disorders of pregnancy (HDP), particular among non‐Western populations with different dietary habits. We aimed to investigate the association between periconceptional diet quality and the risk of HDP among pregnant Japanese women.

Methods and Results

Dietary intake over 1 year before the first trimester of pregnancy was assessed using a validated, self‐administered food frequency questionnaire among 81 113 pregnant Japanese women who participated in a prospective cohort of the Japan Environment and Children's Study. Overall diet quality was assessed by the Balanced Diet Score (BDS) based on adherence to the country‐specific dietary guidelines and the Dietary Approaches to Stop Hypertension (DASH) score. Cases of HDP were identified by medical record transcription. The association between diet quality and HDP risk was examined using Bayesian logistic regression models with monotonic effects. We identified 2383 (2.9%) cases of HDP. A higher BDS was associated with a lower risk of HDP. When comparing the highest with the lowest quintile of the BDS, the adjusted odds ratio (aOR) of HDP was 0.83 (95% credible interval [CrI], 0.73–0.94). The DASH score and HDP risk were inversely associated in a monotonic dose–response manner (aOR per 1‐quintile increase in the DASH score, 0.92 [95% CrI, 0.89–0.95]).

Conclusions

A high‐quality diet, which is recommended for disease prevention in general populations, before conception may also reduce the risk of HDP among pregnant Japanese women.

Nonstandard Abbreviations and Acronyms

BDS

Balanced Diet Score

DASH

Dietary Approaches to Stop Hypertension

FFQ

food frequency questionnaire

HDP

hypertensive disorders of pregnancy

JECS

Japan Environment and Children's Study

Clinical Perspective.

What Is New?

• This study demonstrates that the beneficial effect of healthy diets, recommended by the Dietary Approaches to Stop Hypertension widely used for general populations in Western countries as well as country‐specific dietary guidelines, may also help to prevent hypertensive disorders of pregnancy in pregnant women with different dietary habits.

• Pregnant women with a higher quality diet based on both the country‐specific Balanced Diet Score and Dietary Approaches to Stop Hypertension score had a lower risk of hypertensive disorders of pregnancy.

What Are the Clinical Implications?

• Improving overall diet quality before conception may be an efficient and low‐cost strategy to reduce hypertensive disorders of pregnancy and their induced health consequences in both mother and child.

Hypertensive disorders of pregnancy (HDP), which affects 5% to 10% of all pregnancies, ¹ remains one of the leading causes of maternal and fetal morbidity and mortality worldwide. ² , ³ Furthermore, HDP is associated with an increased risk of future cardiovascular disease for both mothers and offspring. ⁴ Such lifelong and intergenerational adverse health consequences emphasize the need to identify modifiable risk factors and efficient strategies that may help to prevent HDP.

Diet is a modifiable factor, and the importance of optimal diet before and during pregnancy is well known. Although associations between diet and HDP remain less clear, various nutritional factors, such as calcium, magnesium, vitamins C and E, folate, and polyunsaturated fatty acids, have been suggested to be directly related to the inflammatory response, oxidative stress, vascular endothelial dysfunction, and lipid metabolism in the pathogenesis of HDP. ⁵ , ⁶ Some studies suggest that a healthy dietary pattern, such as a Mediterranean diet rich in fruits, vegetables, nuts and seeds, fish, olive oil, and cereals, may protect against HDP, ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ whereas a Western‐style dietary pattern with high consumption of processed meat and sugar‐sweetened beverages is associated with a higher risk of HDP. ⁷ , ⁸ Compositionally similar to the Mediterranean diet, great attention has focused on the Dietary Approaches to Stop Hypertension (DASH) diet, which was originally developed in the United States with the aim of treating and managing hypertension in general populations, ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ and its relationship with a reduced risk of HDP among pregnant women; however, the results are inconsistent. ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ Almost all observational studies were conducted in Western countries. ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² Furthermore, the majority of studies focused on the role of diet during pregnancy. ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ Given that early pregnancy is a critical period for placental development and cardiovascular adaptation related to blood pressure development, ²⁵ , ²⁶ additional studies focusing on the effect of diet before conception on the risk of HDP are needed.

Using data from a large prospective cohort of pregnant Japanese women, the aim of this study was to examine the association between periconceptional diet quality and the risk of HDP. To assess overall diet quality in relation to HDP risk, we applied the Balanced Diet Score (BDS) based on adherence to the Japanese Food Guide Spinning Top as the country‐specific dietary guidelines developed for Japanese populations, ²⁷ , ²⁸ , ²⁹ in addition to the DASH score. ¹⁷ The Japanese Food Guide Spinning Top was developed in 2005 by the Japanese Ministry of Health, Labour and Welfare and the Ministry of Agriculture, Forestry and Fisheries to provide recommendations about food selection and quantities for a healthy diet that can be easily adopted by the public, while taking into account the typical style of Japanese meals (ie, combination of a staple food, a main dish, and side dishes). ²⁷ , ²⁸ , ²⁹ , ³⁰ Higher adherence to the Japanese Food Guide Spinning Top, consisting of grain dishes, vegetable dishes, fish and meat dishes, milk, and fruit as well as moderate intake of snacks and alcoholic beverages, is effective to lower future mortality in middle‐aged women, ³¹ , ³² the risk of delivering a low birth weight infant, ³³ and waist circumference and low‐density lipoprotein‐cholesterol concentrations in young women. ³⁴ Our hypothesis in the present study was that a higher quality of overall diet before pregnancy would be associated with a reduced risk of HDP, regardless of the type of dietary quality indices used.

Methods

Data are unsuitable for public deposition due to ethical restrictions and the legal framework of Japan. The Act on the Protection of Personal Information (Act No. 57 of May 30, 2003, amendment on September 9, 2015) prohibits public deposition of data containing personal information. Ethical Guidelines for Medical and Health Research Involving Human Subjects enforced by the Japan Ministry of Education, Culture, Sports, Science and Technology and the Ministry of Health, Labour and Welfare also restrict the open sharing of epidemiologic data. All inquiries about access to data should be sent to the JECS Programme Office, National Institute for Environmental Studies at jecs-en@nies.go.jp.

Study Procedure and Subjects

The present study was based on the JECS (Japan Environment and Children's Study), an ongoing nationwide prospective birth cohort study. Details of the rationale, study design, protocol, and profile of the study participants have been published elsewhere. ³⁵ , ³⁶ Briefly, expectant mothers in early pregnancy were registered at 15 regional centers across Japan between January 2011 and March 2014. Women who agreed to participate were asked to complete questionnaires twice during the first and second/third trimesters. Information on complications related to pregnancy was separately collected from the medical record transcriptions. The JECS protocol was in accordance with the guidelines laid down in the Declaration of Helsinki and 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. Written informed consent was obtained from all participants.

The present study used jecs‐ta‐20190930 and jecs‐qa‐20210401, which were released in October 2019 and April 2021, respectively, and contain data on 103 057 pregnancies of 97 410 unique pregnant women (Figure). The study subjects were restricted to 93 600 singleton pregnant women by excluding women with multiple gestation (n=948) and a history of cardiovascular disease (n=456), type 1 or type 2 diabetes (n=201), hypertension (n=445), or renal disease (n=1833). Of these, we excluded 9969 women who did not have dietary data with plausible self‐reported energy intake (<500 or >4000 kcal/day) and information on physical activity to calculate the BDS. After further exclusion of 2518 women without data on the incidence of obstetric complications, the final analysis comprised 81 113 pregnant women.

Figure 1. Flow chart of study subject selection.

FFQ indicates food frequency questionnaire.

Dietary Assessment

Dietary intake over 1 year before the baseline survey during the first trimester of pregnancy (median [interquartile range]: 15.3 [12.4–19.1] weeks of gestation) was assessed using a validated, self‐administered food frequency questionnaire (FFQ). ³⁷ , ³⁸ Estimates of the daily intake of food and beverage items, and of energy and nutrients, were calculated using an ad hoc computer algorithm for the FFQ that was based on the Standard Tables of Food Composition in Japan 2010. ³⁹ The relative validity of the FFQ was determined by comparing intakes assessed using 12 days weighed food records and multiple sets of 24 hours urinary excretion data, as described previously. ³⁷ , ³⁸

Calculation of Diet Quality Scores

Using dietary information obtained by the FFQ, overall diet quality during the periconceptional period was assessed by the BDS and DASH score. ¹⁷ , ³³ , ⁴⁰ , ⁴¹ The calculation methods and components of the 2 diet quality scores are summarized in Tables S1 and S2.

The BDS is based on the Japanese Food Guide Spinning Top. ²⁷ , ²⁸ , ²⁹ It comprises 7 components, including grain dishes, vegetable dishes, fish and meat dishes, milk, fruits, snacks, confection, and beverages, and the amount of sodium from seasonings. ³³ , ⁴⁰ , ⁴¹ The 7 components were scored based on an energy density of 2100 kcal for women with at least a moderate physical activity level (or 1800 kcal for those with a low physical activity level), except for snacks, confection, and beverages, which was described as total energy content. Women who exceeded the recommended lower cutoff value received a score of 10, and those who consumed below the limits for the 5 dish categories or exceeded the recommended values for snacks, confection and beverages and sodium from seasonings received a score between 0 and 10, which was calculated proportionally (Table S1). Component scores were summed to obtain the total BDS, which ranged from 0 to 70, with a higher score indicating a better quality of overall diet.

The DASH score developed by Fung et al. consists of 8 components (7 food groups and 1 nutrient). ¹⁷ We modified the sweetened beverages group to better account for the multiple food sources of added sugar in the Japanese diet and included soft drinks, confectionaries, and sugar. ⁴¹ The scoring system is based on quintile rankings of energy‐adjusted intake and ranges from 1 point (lowest quintile) to 5 points (highest quintile) for fruits, vegetables (excluding potatoes), nuts and legumes, whole grains, and low‐fat dairy components (Table S2). For red and processed meat, sugar‐sweetened beverages, and sodium, scores ranged from 1 (highest quintile) to 5 (lowest quintile). Component scores were summed to obtain a total score ranging from 8 to 40, with a higher score reflecting better adherence to the DASH diet.

Hypertensive Disorders of Pregnancy

The primary outcome of the present study was the development of HDP, which was defined as hypertension (blood pressure ≥140/90 mm Hg), with or without proteinuria (≥300 mg/24 hours), that emerged after 20 weeks of gestation but resolved up to 12 weeks postpartum, or as eclampsia in accordance with the clinical guidelines of the Japan Society of Obstetrics and Gynecology and Japan Society for the Study of Hypertension in Pregnancy. ⁴² We obtained information on HDP from medical institutions through a questionnaire completed by obstetricians or other medical staff using medical record transcripts. This questionnaire only revealed whether women were diagnosed with HDP, which was coded as “Yes” or “No.” Information on the specific HDP (ie, gestational hypertension, preeclampsia, or superimposed preeclampsia and eclampsia) was not obtained in the JECS.

Assessment of Lifestyle Variables and Covariates

We collected information regarding age at baseline, smoking habits during early pregnancy, and educational attainment using self‐administered questionnaires completed at enrollment and during the second or third trimester. The physical activity level before pregnancy was assessed by a self‐administered, Japanese short version of the International Physical Activity Questionnaire ⁴³ , ⁴⁴ and was categorized as low, moderate, or high. Prepregnancy body weight and height were transcribed from medical records. Prepregnancy body mass index (kg/m²) was calculated by dividing prepregnancy body weight (kg) by the square of height (m²).

Statistical Analysis

Descriptive data are presented as mean±SD for continuous variables and percentages of subjects for categorical variables. Pearson's correlation coefficient was used to examine the strength and direction of the relationship between the BDS and DASH score. To examine the associations between diet quality and the risk of HDP, logistic regression was fit under a Bayesian paradigm. In general, the Bayesian logistic regression approach allows incorporation of prior knowledge or beliefs about the relationships between model parameters and provides distributions for risk predictions rather than point estimates of disease outcomes, thus improving understanding of predictive uncertainty. ⁴⁵ The diet quality scores were categorized at fifth points based on their distributions. A number of potential confounders defined by a directed acyclic graph based on scientific considerations and previous literature (Figure S1) were considered for adjustment, including age at baseline (continuous), prepregnancy body mass index (continuous), educational attainment (13, 13–14, or ≥15 years), parity (0 or ≥1), history of HDP (yes or no), history of gestational diabetes (yes or no), smoking status during early pregnancy (never smoked, ex‐smoker, or smoked), alcohol drinking during early pregnancy (never drunk, ex‐drinker, or drunk during early pregnancy), and physical activity level (low, moderate, or high). After checking the ordinality of covariates by examining the relation with outcome, the diet quality scores as well as the other ordinal covariates (ie, educational attainment, smoking status, alcohol drinking, and physical activity level) were entered as ordinal variables into the Bayesian logistic regression model with monotonic effects using the ‘brms’ package in R. ⁴⁶ , ⁴⁷ A detailed description of the modeling approach including the model equation and R code can be found in Data S1. This monotonic modeling assumes the effect is consistently negative or positive across the full range of an ordinal variable but allows the size of changes to vary across ordinal categories by a substantial amount. ⁴⁷ This approach results in 2 values for the variable. One is a regression coefficient of the average increase of the dependent variable when the ordinal predictor is increased from any of the ordered categories to the adjacent category (ie, effect per 1‐quintile increase). The other approach is a set of simplex parameters that describes the expected difference between 2 adjacent categories as a proportion of the difference between the lowest and highest categories. In other words, the simplex parameters indicate for which change between adjacent categories the expected change in the dependent variable would be highest. In the present study, the priors were chosen to be weakly informative priors due to a lack of consistent evidence on the potential distribution of the parameters based on a systematic review and meta‐analysis. We therefore assigned a Cauchy distribution with center 0 and scale 2.5 (Cauchy [0, 2.5]), as a default weakly informative prior, recommended by Gelman et al. ⁴⁸ The diet quality scores as well as the other ordinal variables of covariates (ie, educational attainment, smoking status, alcohol drinking, and physical activity level) were included as monotonic effects, and a Dirichlet distribution with a constant alpha of 2 was used for all simplex parameters. ⁴⁹ To ensure convergence of the algorithm, we ran 4 chains with a total of 5000 iterations, discarding the first 2000 iterations as burn‐in, leaving 3000 iterations as posterior samples. We assessed convergence through visual inspection such as trace plots and the Rhat statistic. The posterior distribution of the median odds ratio (OR) with probability of direction, 95% credible interval (CrI) using a high‐density interval, and percentage in region of practical equivalence for developing HDP were estimated according to the quintile of the diet quality scores, with the lowest category used as the reference. In the present study, percentage in region of practical equivalence was defined as the proportion of the posterior distribution of adjusted ORs within the [0.97–1.03] range, and then probability of direction >99% and percentage in region of practical equivalence <2.5% were considered as existing and significant.

To test the robustness of our results, we conducted several sensitivity analyses. First, we imputed missing values for covariates (5.1%) using the ‘miceRanger’ package in R and then compared the estimates of associations from complete cases (n=76 985) and imputed data (n=81 113). Second, we conducted analyses using an unordered categorical model in which the quintile of diet quality scores and other ordinal variables were treated as categorical variables and then compared fitness between the monotonic and categorical models. For the sensitivity of priors, we used a weakly informative prior with a normal distribution of parameters [Normal(mean=0, SD=0.35)] for diet quality scores, operating under the assumption that there was no effect, denoted mathematically as OR=1. Based on previous studies regarding diet quality and HDP risk, ⁷ , ¹⁰ , ¹¹ , ¹² , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ the SD was selected to encompass 95% of the probability between an OR of 0.5 and 2, resulting in an SD of 0.35. The prior for the simplex parameter was also changed to the Dirichlet distribution with a constant alpha of 1. The impacts of different priors on posterior distributions of key model parameters (ie, BDS and DASH score) and final model estimates were examined. Then, we compared the estimates of associations and fitness of 5 models (4 monotonic models with different priors in combinations and the categorical model).

Statistical analyses were performed using SAS statistical software, version 9.4 (SAS Institute, Inc, Chicago, IL, USA) and the R statistical package, version 4.3.0 (The R Foundation, Vienna, Austria). The Strengthening the Reporting of Observational Studies in Epidemiology checklist for cohort studies was used to ensure appropriate reporting of study methods and findings.

Results

Subject Characteristics

The basic characteristics of the subjects are shown in Table 1. Our analytical sample of 81 113 pregnant women, in comparison with those who were excluded from the analysis due to missing or incomplete values on exposure and outcome variables (n=12 487), were more likely to be better educated, never smokers, and primiparous and to have lower energy intake but less likely to engage in physical activity. The values for the other variables were comparable between the 2 groups. Of 81 113 pregnant women in the final analysis, 2383 (2.9%) women developed HDP. The mean values (SD) of the BDS and DASH score during the periconceptional period were 47.1 (7.7) and 23.8 (4.5), respectively, with wide ranges of total scores, which enabled us to detect meaningful differences (Table S3). Pearson's correlation coefficient between the BDS and DASH score was 0.59. When the pregnant women were categorized into quintiles according to their diet quality scores (Tables S4 and S5), those with higher scores tended to be older, to have higher educational attainment, and to have smoked less during early pregnancy. Whereas women with higher BDS were more likely to be multiparous and to have a history of HDP and less likely to have drunk alcohol during early pregnancy, those with higher DASH score were more likely to have consumed alcohol during early pregnancy. Intake patterns of nutrients were very consistent between the BDS and DASH score. Women in the higher quintiles of the BDS and DASH score tended to have lower intakes of total fat, saturated fatty acids, and monounsaturated fatty acids and higher intakes of protein, carbohydrate, dietary fiber, and all vitamins and minerals examined, with some exceptions including polyunsaturated fatty acids and sodium.

Table 1.

Basic Characteristics of 81 113 Pregnant Japanese Women Who Participated in the Japan Environment and Children's Study

	Value* , †
Maternal age at baseline, y	30.8 (5.0)
Prepregnancy BMI, kg/m2	21.2 (3.3)
Educational attainment, %
<13 y	35.4
13–14 y	42.2
≥15 y	22.4
Parity, %
0	42.4
≥1	57.6
History of gestational diabetes, %	0.70
History of HDP, %	1.88
Smoking status during early pregnancy, %
Never smoked	59.1
Ex‐smokers who quit before pregnancy	36.4
Smokers during early pregnancy	4.5
Alcohol drinking during early pregnancy, %
Never drunk	34.6
Ex‐drinkers who quit before pregnancy	55.2
Drinkers during early pregnancy	10.1
Level of physical activity, %
Low	49.7
Moderate	31.7
High	18.6
Dietary intake
Total energy, kcal/d	1772 (587)
Protein, % of energy	13.5 (2.0)
Fat, % of energy	29.5 (6.5)
Saturated fatty acids, % of energy	9.3 (2.7)
Monounsaturated fatty acids, % of energy	10.9 (2.8)
Polyunsaturated fatty acids, % of energy	6.0 (1.4)
Carbohydrate, % of energy	55.3 (7.8)
Dietary fiber, g/1000 kcal	6.2 (2.0)
Vitamin A, μg/1000 kcal	278 (203)
Vitamin C, mg/1000 kcal	52 (27)
Vitamin E, mg/1000 kcal	3.7 (1.1)
Folate, μg/1000 kcal	152 (54)
Sodium, mg/1000 kcal	1735 (511)
Potassium, mg/1000 kcal	1264 (328)
Calcium, mg/1000 kcal	289 (140)
Magnesium, mg/1000 kcal	132 (30)
Iron, mg/1000 kcal	3.9 (0.9)
Diet quality scores
Balanced Diet Score based on the Japanese Food Guide Spinning Top	47.1 (7.7)
Dietary Approaches to Stop Hypertension score	23.8 (4.5)
HDP, %	2.9

BMI indicates body mass index; and HDP, hypertensive disorders of pregnancy.

^*Values are mean±SD for continuous variables and percentages for categorical variables.

^† The number of subjects with missing values is 17 for age at baseline, 30 for prepregnancy BMI, 1462 for educational attainment, 1952 for parity, 515 for smoking status during early pregnancy, and 267 for alcohol drinking during early pregnancy.

Associations Between Diet Quality Scores Before Pregnancy and the Risk of HDP

As show in Table 2, both the BDS and DASH score were inversely associated with the risk of HDP (OR per 1‐quintile increase in the BDS, 0.95 [95% CrI, 0.93–0.98] and OR per 1‐quintile increase in the DASH score, 0.92 [95% CrI, 0.89–0.95]). According to the cumulative sum of the simplex parameters (ie, cumulative proportion of the total effect), the HDP risk decreased with an almost equal size across categories of the BDS. Compared with women in the lowest quintile of the BDS, women in the highest quintile had 17% (95% CrI, 6%–27%) lower odds of HDP. The risk of HDP decreased monotonically and sharply up to the first 3 quintiles of the DASH score, accounting for 61% of the total effect, and then the decrease slowed down.

Table 2.

Associations of Periconceptual Diet Quality, Evaluated by the BDS and DASH Score, With the Risk of HDP Among 81 113 Pregnant Japanese Women from the Japan Environment and Children's Study

	Prevalence of HDP, %	Cumulative sum of simplex parameters, %*	OR (95% CrI)† , ‡	PD, %	% in ROPE
BDS
Quintile 1 (13.1–40.5)	3.3	…	Reference
Quintile 2 (40.6–45.5)	3.1	29	0.95 (0.87–0.99)	99.9	21.4
Quintile 3 (45.6–49.6)	2.9	57	0.90 (0.82–0.97)	99.9	2.4
Quintile 4 (49.7–53.8)	2.7	79	0.86 (0.78–0.95)	99.9	0.7
Quintile 5 (53.9–70.0)	2.7	100	0.83 (0.73–0.94)	99.9	0.5
DASH score
Quintile 1 (8.0–19.0)	3.5	…	Reference
Quintile 2 (20.0–22.0)	3.1	32	0.90 (0.81–0.97)	100	2.5
Quintile 3 (23.0–25.0)	2.8	61	0.81 (0.73–0.90)	100	0
Quintile 4 (26.0–27.0)	2.7	82	0.76 (0.68–0.85)	100	0
Quintile 5 (28.0–40.0)	2.6	100	0.72 (0.63–0.81)	100	0

BDS indicates Balanced Diet Score; CrI, credible interval; DASH, Dietary Approaches to Stop Hypertension; HDP, hypertension disorders of pregnancy; OR, odds ratio; PD, probability of direction; and ROPE, region of practical equivalence.

^*Values are the cumulative sum of simplex parameters between the specified quintile and the lowest quintile (reference). Simplex parameters, which sum up to 100%, represent the expected difference between two adjacent categories as a proportion of the difference between the lowest and highest categories. Simplex parameters indicate for which change between adjacent categories the expected change in the dependent variable would be highest.

^† All values are the odds ratio and 95% credible interval, estimated using Bayesian logistic regression analysis.

^‡ Adjustment was made for age at baseline (continuous), prepregnancy body mass index (continuous), educational attainment (<13, 13–14, or ≥15 years), parity (0 or ≥1), history of gestational diabetes (yes or no), history of HDP (yes or no), smoking status during pregnancy (never smoked, ex‐smoker, or smoked during early pregnancy), alcohol drinking during early pregnancy (never drunk, ex‐drinker, or drunk during early pregnancy), and physical activity level (low, moderate, or high).

Results were comparable in sensitivity analyses using complete case data (Table S6). Sensitivity analyses using different priors yielded similar posterior distributions (Figure S2) and estimates of association (Table S7), suggesting the robustness of the final model estimates. In addition, the monotonic model used in the main analysis had the highest value of expected log posterior predictive density (Table S8), indicating that it had a better fit than the other models used in sensitivity analyses.

Discussion

In this large nationwide prospective cohort of pregnant Japanese women, a high‐quality diet during the periconceptional period, evaluated by the BDS based on the Japanese Food Guide Spinning Top, was inversely associated with the risk of HDP. We also observed a clear monotonic dose–response association between adherence to the DASH‐style diet, which was originally developed in the United States for treatment and management of hypertension, and a reduced risk of HDP. Our findings suggest that healthy diets, which are widely recommended to prevent diseases and improve hypertension in general populations, may also help to prevent the development of HDP among pregnant Japanese women.

Although both the BDS and DASH score were inversely associated with the risk of HDP, the inverse dose–response association of the DASH score with the HDP risk was much clearer and stronger. This may be partly because the BDS does not strictly account for the types of grains (ie, refined and whole grains) and dairy products (ie, low fat and full fat), which are associated with lipid profiles and chronic diseases including hypertension. These foods are less commonly consumed in Japan and therefore are not specifically advocated in the Japanese Food Guide. ²⁷ , ²⁸ , ²⁹ A previous study of the Japanese population modified the DASH score by excluding whole grains and low‐fat dairy products as score components because of the large number of nonconsumers and reported that the expected inverse association with blood pressure was not observed. ³⁰ When we excluded these 2 components from the total score, the inverse association between the DASH score and the risk of HDP was slightly attenuated (OR per 1‐quintile increase in the DASH score, 0.96 [95% CrI, 0.93–0.99]), indicating that whole grains and low‐fat dairy products might be important components of the Japanese diet to prevent HDP. Additionally, our findings are consistent with previous intervention studies that showed the beneficial effect of the DASH diet on blood pressure reduction in the general Japanese population. ⁵⁰ , ⁵¹ The present study demonstrated the beneficial effect of healthy diets, characterized by a greater intake of vegetables, fruits, whole grains, legumes, fish, and low‐fat dairy products and a low intake of meat and sugar‐sweetened beverages, ⁴¹ on the risk of developing HDP. Our finding suggests that healthy diets recommended by disease‐specific guidelines widely used in Western countries as well as the country‐specific dietary guidelines may help to prevent HDP in pregnant Japanese women.

Little is known about the influence of adherence to the DASH diet before and during pregnancy on hypertensive disorders, and the results are inconsistent. ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ A study of an American cohort of 11 535 women showed that greater prepregnancy adherence to the DASH diet was associated with a lower risk of preeclampsia but not gestational hypertension. ¹⁸ A similar beneficial effect of the DASH diet during pregnancy on preeclampsia was observed in a Chinese case–control study ²³ and a randomized controlled clinical trial of pregnant Chinese women diagnosed with gestational hypertension and chronic hypertension. ²⁴ By contrast, the Generation R study of 3414 pregnant women showed the association of a higher DASH score assessed at early pregnancy with a reduction in midpregnancy diastolic blood pressure but not with HDP. ²¹ Furthermore, the Project Viva cohort of 1760 women ²² and the Danish National Birth Cohort of 66 651 women ¹⁹ did not find associations. The DASH score does not measure adherence to the DASH diet per se, and the scoring system of the DASH score depends on the distribution of the population examined, making it difficult to compare the degree of adherence to the DASH diet and its effects between different populations. Furthermore, the inconsistent findings across studies may be partly due to differences in study populations (eg, pregnant women with or without disorders), ethnicity, study design, sample size, timing of dietary assessment (ie, before or during pregnancy), diagnosis of HDP (ie, self‐reported or physician diagnosed), subtypes of HDP, severity of HDP, and timing of HDP onset. Although we provided evidence of the beneficial effect of the DASH diet in the periconceptional period on the risk of HDP, further studies are needed to draw clear conclusions about the role of the DASH diet in pregnant women.

Although the cause of HDP remains unclear because it is a hyper‐heterogenous syndrome with a multifactorial nature, HDP is linked to endothelial dysfunction, inflammation, oxidative stress, insulin resistance, and dyslipidaemia. ⁵² Diets with higher BDS and DASH score are rich in antioxidant nutrients (ie, vitamins A, E, and C), vitamin D, and minerals (zinc, calcium, iron, and magnesium) (Tables S4 and S5). ⁴¹ Antioxidant nutrients prevent development of HDP by enhancing the antioxidant capacity during pregnancy, inhibiting NAD(P)H oxidase activation, and preventing an exaggerated inflammatory response. ⁵² , ⁵³ Maternal nutritional deficiencies and a nutritional imbalance in early pregnancy are related to poor placental development and blood pressure regulation. In addition, dietary fiber from vegetables, fruits, grains, and pulses also has beneficial effects on glucose tolerance and lipid profiles that increase the risk of preeclampsia. ⁶ , ⁵³ Therefore, the beneficial effects of healthy diets on HDP may result from the additive and synergistic effects of nutrients that protect against oxidative stress, the inflammatory response, and endothelial cell dysfunction during the pathogenesis of HDP.

The main strengths of the present study include its well‐designed prospective nature, assessment of diet during the periconception period, use of medical record transcription to obtain information about HDP based on a unified definition in clinical practice, consideration of a wide range of potential confounders present before and during pregnancy, use of multiple imputation to reduce bias due to missing data on covariates, and investigation of a general population of pregnant Japanese women with a large sample size, which allowed us to estimate ORs with greater statistical power than previous studies that investigated diet quality and development of HDP. However, our study also has several limitations. First, we used self‐reported dietary information assessing a long‐term period (ie, the previous year). Incompleteness of the assessment due to poor recall (memory), inaccurate portion size estimation, and dietary change increases the chance of nondifferential misclassification and may attenuate rather than enhance associations. Misreporting of dietary intake, which is associated with self‐reported dietary assessment, was minimized by excluding subjects with biologically implausible energy intakes and using energy‐adjusted dietary variables. In addition, we conducted the same analysis among 76 846 pregnant women who did not intentionally change their diet within 1 year, and the results did not change materially. Second, the FFQ used in the current study was not designed specifically to evaluate the overall diet quality score based on the predefined diet quality indices. Although we used a validated FFQ, its reliability for estimating the intake of specific food and nutrients, such as whole grains, low‐fat dairy products, and sodium, which are components of diet quality indices, is not well confirmed among pregnant women. Additionally, each dietary component of the BDS and DASH score was allocated the same weight. It is implausible that all dietary components have the same impact on the HDP risk, but it is currently unknown how to weight the relative contributions of components to the total scores with little guidance available. Third, we could not classify HDP phenotypes, such as gestational hypertension, preeclampsia, and other specific conditions, due to the lack of detailed information in the JECS. Therefore, comparison of the effect of diet quality on each subtype of HDP between this study and other studies is limited. Given the relatively low rate of HDP (2.9%) in comparison with previous studies, ¹ even though HDP cases were diagnosed based on the clinical guidelines, the incompleteness of ascertainment of HDP in the present study cannot be ruled out. Finally, although we controlled for a series of potential confounders, we cannot completely exclude the possibility of unmeasured (eg, health conscious and family history of hypertension, diabetes, and HDP) and residual confounding factors due to the nature of observational studies.

Conclusions

A higher quality periconceptional diet evaluated with 2 predefined diet quality indices (the BDS and DASH score) was consistently associated with a lower risk of HDP. Although the evidence is limited and the cause has not been elucidated, improving overall diet quality before pregnancy could be an efficient and low‐cost strategy to reduce HDP and HDP‐induced health consequences in both mother and child. Further studies with more rigorous assessment methods of diet and HDP are needed to confirm the present findings and to draw firm conclusions about this relationship in various populations with different food cultures.

Appendix

Members of the JECS Group as of 2023: Michihiro Kamijima (principal investigator, Nagoya City University, Nagoya, Japan), Shin Yamazaki (National Institute for Environmental Studies, Tsukuba, Japan), Yukihiro Ohya (National Center for Child Health and Development, Tokyo, Japan), Reiko Kishi (Hokkaido University, Sapporo, Japan), Nobuo Yaegashi (Tohoku University, Sendai, Japan), Koichi Hashimoto (Fukushima Medical University, Fukushima, Japan), Chisato Mori (Chiba University, Chiba, Japan), Shuichi Ito (Yokohama City University, Yokohama, Japan), Zentaro Yamagata (University of Yamanashi, Chuo, Japan), Hidekuni Inadera (University of Toyama, Toyama, Japan), Takeo Nakayama (Kyoto University, Kyoto, Japan), Tomotaka Sobue (Osaka University, Suita, Japan), Masayuki Shima (Hyogo Medical University, Nishinomiya, Japan), Seiji Kageyama (Tottori University, Yonago, Japan), Narufumi Suganuma (Kochi University, Nankoku, Japan), Shoichi Ohga (Kyushu University, Fukuoka, Japan), and Takahiko Katoh (Kumamoto University, Kumamoto, Japan).

Sources of Funding

The Japan Environment and Children's Study was funded by the Ministry of the Environment, Japan. The findings and conclusions of this study are the sole responsibility of the authors and do not represent the official views of the Japanese government.

Disclosures

None.

Supporting information

Data S1