Diet significantly contributes to human health [1], and during pregnancy, dietary energy and nutrient requirements, specifically micronutrient requirements, are generally greater to support increased maternal metabolism and blood volume, red cell mass expansion, and to secure the delivery of nutrients to the fetus [2]. Poor nutrition during pregnancy is indicative of greater short- and long-term health risks to both mother and child [34]. Potential health risks to women and their off-spring necessitate continuous monitoring of the nutritional statuses and dietary intakes of pregnant women. Moreover, because a healthy diet reduces the risk of short- and long-term adverse pregnancy outcomes [5], the identification of the optimal diets during pregnancy is an important public health issue.

Dietary intakes during pregnancy are commonly assessed using intakes of energy, macronutrients, micronutrients, and food groups [67] and maternal diet imbalance suggests poor nutritional status. In recent years, the focus of dietary studies has shifted from evaluations of single nutrients toward evaluations of dietary patterns, which describe combinations of commonly consumed foods [8]. This approach allows diets to be described holistically and the roles played by social and economic factors with respect to compliance with dietary patterns to be elucidated [9]. Dietary patterns may also help capture the complexities of diet, which are often inaccessible by nutrient analysis. The latter point is relevant because nutritional health outcomes are often the result of multiple synergies between nutrients and foods rather than the summed effects of individual components [10].

Previous studies have shown that nutrition prior to and during pregnancy may critically effect maternal health and subsequent fetal development [111213]. A dietary pattern high in fruits and vegetables is associated with a reduced risk of preterm delivery [14151617], whereas dietary patterns characterized by high energy, saturated and trans fats, refined sugar, and sodium are associated with higher risks of preterm delivery [18], low birth weight [19], and small-for-gestational-age (SGA) infants [20]. Maternal diet not only affects fetal development but also infant risk of chronic disease in later life [212223]. Mediterranean diets and diets high in fruits and vegetables have been found to protect allergies in children [2425], while maternal diets high in white bread, red and processed meat, French fries, fried chicken, and vitamin C rich drinks have been reported to increase the risks of childhood overweightedness and obesity significantly at 3 years of age [16].

Although several approaches have been used to identify DPs among pregnant women, the posterior approach derived from principal component analysis is the most commonly used [26272829]. However, most studies on the topic have been conducted in Western populations, and due to diet and lifestyle behavior differences, the dietary pattern exhibited by pregnant Malaysian women is certain to differ. In addition, the majority of nutritional studies focused on one of the phases of pregnancy, such as pre-pregnancy [3031] or mid-to-late pregnancy [3233], and as maternal diets appear to change during pregnancy. Furthermore, we should understand dietary changes during this period, as dietary changes during pregnancy appear to reflect women's efforts to balance physiological changes in an effort to secure a healthy outcome. However, such information is unavailable for pregnant Malaysian women. In addition, understanding the characteristics of pregnant women with different dietary patterns is important for devising appropriate strategies to promote healthy eating. Thus, the present study was undertaken to determine dietary patterns before and during pregnancy and to document the consistencies of and deviations from dietary patterns over time and the maternal factors associated with observed patterns.

The characteristics of the 737 study subjects are summarized in Table 1. Mean age at study entry was 30.34 ± 4.55 years. Most women were Malays (89.3%), nulliparous (35.8%), had completed secondary education (44.6%), were employed (69.7%), and were from low-income households (60.8%). A small percentage of women reported having GDM (8.1%) or pre-eclampsia (2.7%) during a previous pregnancy. More than one-fifth had a family history of diabetes mellitus (DM) (27.0%) or hypertension (23.1%). About 40.4% had an early waist circumference measurement of ≥ 80 cm, and 73.9% were overweight (33.6%) or obese (40.3%). More than half (54.1%) reported a normal body mass index (BMI 18.50-24.99 kg/m²) before pregnancy.

The factor loadings of DPs before and during pregnancy (first, second, and third trimester) are provided in Table 2. Three DPs were identified at each time point: DP 1–3 for pre-pregnancy, DP 4–6 for the first trimester, DP 7–9 for the second trimester, and DP 10–12 for the third trimester. Pre-pregnancy, DP 1 was the most dominant food pattern and accounted for 13.5% of total dietary intake variation. During pregnancy, DP 6, DP 9 and DP 12 explained the largest percentage of total variance, with total variances of 14.4%, 15.4% and 13.5% for first, second, and third trimesters, respectively. DP 1 and DP 4 appeared to be a more prudent diet and were characterized by higher intakes of other vegetables, nuts, seeds & legumes, green leafy vegetables, eggs, fruits, and milk & dairy products. DP 7 in the second trimester became more plant- and staple food-based. However, in the third trimester, DP 10 was characterized by additional food sources choice rich in protein, such as poultry & meat; nonetheless, this pattern was still considered healthy based on overall vegetable and fruit intakes. DP 2, 5, 8, and 11 had greater loadings of condiments & spices, sugar, spread & creamer, though DP 2 had additional sweet foods, DP 5 and DP 8 had additional oils & fats, and DP 11 had additional tea & coffee. DP 3 and DP 6 were characterized by high protein (poultry, meat, processed, dairy, eggs, and fish), sugars (mainly as beverages and sweet foods), and energy (bread, cereal & cereal products, rice, noodles & pasta). A similar pattern (DP 9) was found in the second trimester with additional fruits. However, in the third trimester, DP 12 had greater loadings of energy (bread, cereal & cereal products, rice, noodles & pasta), sugars (mainly as beverages, and sweet foods), and good protein sources (eggs and nuts, seeds & legumes). Sensitivity analyses were performed on 487 pregnant women with complete data, and findings concurred with main findings (Supplementary Table 1).

Table 3 shows the associations between women's characteristics and DPs. Pre-pregnancy, women were more likely to have low adherence (LA) for DP 1, but less likely to have LA for DP 2. No significant association was found between maternal characteristics and DP during the first trimester, but during the second trimester, were more likely to have high adherence (HA) for DP8, and in the third trimester, were more likely to have LA for DP 10 and HA for DP11. Whereas, women with a higher waist circumference (WC) at first prenatal visit were more likely to have LA for DP 11. In addition, women with higher education were more likely to have LA for DP10.

The present study identified three distinct DPs in Malaysian pregnant women with specific adherence patterns over time and their dependence on maternal characteristics. Pre-pregnancy DP 1 was dominated by plant-based foods, eggs, milk and dairy products, and most resembled the DP commonly identified as “Prudent/Healthy” (vegetables, fruits, fish, legumes, whole grains, and dairy products) reported in Western studies [1333474849]. DP 4, DP 7 and DP 10 during pregnancy remained somewhat plant-based with the addition of an additional staple-food in the second trimester and protein (poultry & meat) in the third trimester. In a Korea study conducted on a general adult population, a healthy dietary pattern was characterized by high intakes of vegetables, fruits, whole grains, legumes, and seafood and low intakes of sweetened foods, refined grains, and processed meat [50]. However, during pregnancy, a healthy diet should be characterized as a balanced diet that includes a combination of all food groups to ensure that increased nutritional needs are met during pregnancy [51]. Thus, although rice, noodles & pasta, and poultry & meat were included in DP 7 and DP 10, these patterns were still considered as healthy. A similar pattern was reported in studies conducted in China (‘vegetarian/vegetable pattern’) and in the Growing Up in Singapore Towards Healthy Outcomes (GUSTO) study (‘vegetable, fruit and white rice’). The patterns reported by these studies were characterized by high loadings of plant-based foods such as root vegetables, beans, leafy vegetables, fruits, and legumes [173233]. Dietary patterns weighted toward plant-based foods (especially fruits), vegetables, whole grains, and lean meat throughout pregnancy have been shown to benefit both maternal and perinatal health [21333]. He et al. [33] concluded vegetables and fruits were important components of protective dietary patterns and that a ‘vegetable dietary pattern’ was significantly associated with a lower risk of GDM [33]. In addition, Chia et al. [17] reported this dietary pattern was significantly associated with preterm birth and larger birth size [17].

DP2 and DP3 during pre-pregnancy, DP 5 and DP 6 during first trimester, DP 8 and DP 9 during second trimester, and DP 11 and DP 12 in third trimester were less healthy DPs characterized by either high intakes of fats and energy or protein and sugar. DP 3 changed from pre-pregnancy to pregnancy (DP 6, DP 9, and DP 12) to account for the largest proportion of total variance. Plausibly, preference for these energy dense foods is due to increased appetite during pregnancy. De-Graft Aikins [52] stated that Ghanaian women reported increased appetite during pregnancy and reported a craving for starch and protein rich foods [52]. In addition, many pregnant women believe that the consumption of high protein foods, such as meat, cheese, and milk, is important for infant growth and development [53]. Based on current knowledge, excessively low protein intake during pregnancy is associated with negative effects in terms of weight and length at birth. However, excessively high protein could also negatively impact fetal development [54]. Although fish & seafood and eggs contain lower fat levels than meats, the methods used to prepare them may substantially increase fat contents. In Malaysia, fish & seafood and eggs are usually deep-fried, grilled with oil, or with added oil and/or fat (e.g. curry, fried chili, coconut milk-based gravy, or sauce) [37]. In other words, increased fish & seafood and egg intake may be associated with energy and fat intake increases.

DP 2 (pre-pregnancy), DP 5 (first trimester), DP 8 (second trimester), and DP 11 (third trimester) were much more consistent than other DP over time. This pattern was characterized by high loadings of condiments & spices (at all time points), sugar, spread & creamer (all time points), and lower loadings of oils & fats (1^st and 2^nd trimester) and tea & coffee (3^rd trimester). These DPs appear to be unique to the Malay population and are markedly different from those reported in Western populations [13334749]. This pattern appears to be similar to the ‘less-healthy’ pattern reported in pregnant women in the Universiti Sains Malaysia (USM) Birth Cohort Study [55]. The Malaysian Adult Nutrition Survey (MANS) also showed the majority of Malaysians consume tea (47%) and coffee (28%) at least once a day and habitually add sugar (four teaspoons per day) and creamer (three teaspoons per day) to beverages [37]. The condiments and spices included in this pattern were soy sauce (salty soy sauce or sweet soy sauce), oyster sauce, chili paste, and chili sauces, all of which are frequently used during food preparation. This finding illustrates specific dietary pattern differences between Asian and Western populations.

Although the present study showed Malays were more likely to have low adherence (LA) for DP 1 before pregnancy and DP 10 during the 3^rd trimester, average daily intakes of fruits and vegetables based on 24-hour dietary recall fell much below recommended levels. This finding is similar to that of the MANS, which reported that about 75.5% of Malaysian women aged 18 to 59 years did not meet the dietary guideline for fruit and vegetable intakes (3 to 5 servings per day) [37]. We also found that women with at least a tertiary education were significantly more likely to have LA for DP 10, but only in the third trimester. Previous studies showed that education level importantly enhances nutritional knowledge [5657], and thus, it is possible that these women increased their energy and protein intakes in the 3^rd trimester to support fetal growth and development. However, whether this observation indicates that as pregnancy progresses, women with higher education become more aware of the relationship between diet and fetal growth and subsequently adopt a more diverse dietary pattern, requires further investigation.

We also found significant ethnic differences in DP 2 (condiments, spices, sugar, spread and creamer) during pre-pregnancy, DP 8 during the second trimester, and DP 11 during the third trimester, which probably reflect differences in food culture among ethnic groups. Malays were less likely to have LA before pregnancy but tended toward higher adherence (HA) to DP 8 and DP 11 than non-Malays. The main characteristic of Malay cuisine is the generous use of condiments and spices in dishes (either animal or plant-based). In addition, tea and coffee are served daily for breakfast and snacks and sugar is usually added to beverages. These findings highlight the importance of understanding dietary patterns of specific ethnic groups when designing targeted health promotion interventions.

In the third trimester, women with a higher WC at first prenatal visit were found to be more likely to have LA for DP 11 (condiments, spices, sugar, spread and creamer). About 73.9% of women with a WC of ≥ 80 cm at first prenatal visit were overweight (33.6%) or obese (40.3%) at study entry. Within the Malaysian health care setting, it is plausible that overweight or obese women received diet counseling from a health professional during pregnancy to reduce amounts of sugary foods, oil, and fats in diet to prevent excessive gestational weight gain (GWG) and pregnancy complications, such as gestational diabetes mellitus (GDM). Furthermore, previous studies have shown taste preferences may differ according to weight status, in that those who are overweight or obese appear show greater preference for energy-dense foods [5859].

This study has several limitations that warrant consideration. First, the study cohort was not representative of all pregnant Malaysian women. Most of the women enrolled were Malays, had a secondary education and lower, and were from low- or middle-income households. As, dietary patterns can vary by sex, socioeconomic status, ethnicity, and culture, we suggest the results of the study be compared with those of other pregnant Malaysian women representing various ethnic and socioeconomic groups. Second, the factor analysis and statistical methods used to define dietary patterns, for example, the consolidation of food items into food groups, the number of food groups included, and the number of factors extracted, were subjective. Sensitivity analysis was performed to assess robustness of results. Furthermore, all statistical analyses were checked for violation of the assumptions underlying those analyses (e.g., factor analysis) in order to minimize the potential bias. Third, use of the SFFQ for dietary assessments introduced the possibilities of recall bias and misreporting. However, the use of well-trained interviewers for data collection probably reduced these errors. Fourth, although the borderline significances of our findings caution that interpretations be approached carefully. Fifth we did not assess food preferences, food taboos, beliefs, or practices during pregnancy, or the social contexts of meal consumption. As these are important determinants of dietary intake, such information could explain adherence to eating patterns during pregnancy. Nevertheless, despite these limitations, we believe our findings provide important insights of the DPs of Malaysian women from pre-pregnancy and throughout pregnancy.

In conclusion, the DPs study differed substantially from those reported in Western populations. Furthermore, the different DPs observed before and during pregnancy, reflect clear changes in eating and drinking habits, although some healthy dietary adaptations did not last until the end of pregnancy. In addition, ethnicity, education, and WC in early pregnancy were significantly associated with specific DPs. These findings indicate the importance of understanding DPs when designing interventions aimed at promoting healthy diets for pregnant women. Such interventions should also consider the factors associated with the dietary intakes of pregnant women so that interventions can be tailored to meet the specific needs of women prior to and during pregnancy. Further studies are needed to determine the associations between DPs and pregnancy outcomes, child development, and later health.