Learn more: PMC Disclaimer | PMC Copyright Notice
Methodological aspects of the study of dietary patterns during pregnancy and maternal and infant health outcomes. A systematic review
Abstract
The objective of the present study was to systematically review the literature exploring the associations between different dietary patterns obtained from Food Frequency Questionnaires during pregnancy and the development of health‐related maternal and infant outcomes in the Framework of the Eurreca Network of Excellence. A systematic search was conducted on Pubmed for literature published up to September 2009. The search strategy resulted in an initial amount of 2048 articles. After applying the selection criteria, seven studies were finally identified. Five articles were based on prospective cohort studies and the other two were case‐control studies. The methods used to elaborate the dietary pattern could be classified as hypothesis‐oriented (three studies) or empirically‐derived (four studies). The different food frequency questionnaires used for diet assessment were self‐administered, semi‐quantitative and had been previously validated, but just four studies employed questionnaires validated specifically for their use in a pregnant population. The divergent methods used to assess the dietary patterns make it difficult to compare results. However, some resulting recommendations can be applied when dietary patterns during pregnancy are analyzed: to employ a validated food frequency questionnaire designed for use in pregnancy, to consider the special role exerted by mineral and vitamin supplements in this particular population group, to adequately select the time in which dietary data is collected, to adjust the results for life‐style and educational characteristics, and in the case of hypothesis‐oriented dietary patterns, to correctly choose the components comprising the score.
Introduction
Several studies have analyzed the role of maternal diet during pregnancy in maternal and child health related conditions such as gestational diabetes, pre‐eclampsia, fetal growth, pre‐term birth, post‐partum depression, birthweight, neuro‐psychiatric, allergic, dermatological and respiratory diseases (Klemmensen et al. 2009; Miyake et al. 2009; Strøm et al. 2009).
Most of the studies published until now have evaluated the role of different nutrients such as omega‐3 fatty acids, vitamins and minerals or several food items like fish, fruits or meat in these pathologies and health outcomes. However, foods are not consumed in isolation. There can exist antagonistic and synergistic effects between nutrients and food items. An overall dietary pattern (DP) might have a greater effect on health than a single food or nutrient (Jacques & Tucker 2001) and may explain more completely the pathogenesis of nutrition‐related diseases or the health benefits derived from diet. As such, since each nutrient or food item is part of a larger pattern consisting of many nutrients and foods, interest has drifted from the study of single nutrients to analysis that identifies DP of consumed foods, which can be used to describe the overall diet (Hu 2002; Kant 2004).
Methods most often used to characterize diet and ascertain DP have been based either on a priori knowledge (Hypothesis‐oriented DP) or derived from food frequency questionnaires using data‐driven dimension reduction techniques, such as principal components analysis (PCA) (empirically‐derived DP) (Sánchez Villegas & Serra Majem 2005; Román‐Viñas et al. 2009). The aim of this study was to systematically review and analyse those studies exploring the association between different DP identified during pregnancy and the development of health‐related maternal and infant outcomes using both methodological approximations.
Key messages
- •Food‐frequency questionnaires used to asses diet during pregnancy are generally self‐administered, semi‐quantitative and have been previously validated.
- •Divergent methods are used to assess dietary patterns so comparing results is complicated.
- •When a questionnaire is designed to assess diet during pregnancy, it is important to consider the role of vitamins and minerals supplements in order to correctly select the time covered by the questionnaire and the time in which dietary data is collected, and to adjust the results for lifestyle and educational characteristics of the sample.
Materials and methods
A Pubmed search was conducted up to September 2009. We wanted to identify studies that utilized Food Frequency Questionnaires (FFQ) and linked DPs during pregnancy to health outcomes for the mother and the infant. Keywords as well as Medical Subject Headings (MeSH) terms were used for this purpose. The search strategy applied was as follows:
[‘pregnancy’(mesh terms) or ‘pregnant women’(mesh terms) or ‘Maternal Nutritional Physiological Phenomena’(Mesh) or ‘infant’(Mesh)] and [‘diet’(mesh terms) or ‘diet’(all fields)] and [‘questionnaires’(mesh) or ‘questionnaire’ (all fields) or ‘diet records’(mesh) or ‘nutrition assessment’ (mesh) or ‘dietary pattern’(all fields)]
Selected studies had to comply with the selection criteria detailed in Table 1. We selected only studies that collected information regarding food intake though food frequency questionnaires because this is the most appropriate method to assess average intake over an extended period of time in epidemiologic studies (Willett 1998).
Table 1
Criteria for the selection of relevant papers
| Inclusion criteria |
| • Studies based on dietary patterns |
| • Studies relating diet during pregnancy (including the periconceptional period) with health outcomes in the mother and the infant |
| Exclusion Criteria |
| • Studies in languages other than English |
| • Descriptive Studies |
| • Articles comparing socioeconomic or ethnic features of populations |
| • Studies evaluating nutrients or foods individually |
| • Studies using other questionnaires than FFQ |
| • Studies based on minority population groups |
| • Articles assessing pre‐pregnancy diet |
| • Studies considering health outcomes in offspring beyond the second year of age |
Results
The Pubmed search strategy yielded a total amount of 2048 articles. After applying the selection criteria by screening titles and abstracts, 37 articles were selected. This task was simultaneously completed by two different reviewers (A.S.V., N.B.) and individual results were compared; any difference between selection results was discussed for final inclusion. Further examination of full papers (again doubled reviewed) permitted a final selection of seven articles. Thirty of the articles did not fulfill the inclusion criteria.
Table 2 shows the main characteristics of the selected articles. The number of participants in the studies varied from 131 (Vujkovic et al. 2009) to 44 612 (Knudsen et al. 2008). Two of the publications showed results from Norway (Norwegian Mother and Child Cohort Study) (Haugen et al. 2008; Brantsaeter et al. 2009), two from Denmark (Danish National Birth Cohort) (Knudsen et al. 2008; Mikkelsen et al. 2008), two from the Netherlands (2007, 2009) and one study from the USA (Rifas‐Shiman et al. 2009).
Table 2
Characteristics of included studies
| Author / year | Country / Sample characteristics | Study design/Study name (for cohort studies) | Questionnaire Characteristics | Period assessed by FFQ | Studied Health outcome | Method for Dietary pattern Assessment | Dietary patterns label | Results / Conclusions |
|---|---|---|---|---|---|---|---|---|
| Brantsaeter et al. 2009 | Norway n = 23 423 women Nulliparous | Prospective cohort study Norwegian Mother and Child Cohort S(MoBa) ‐ | Semi‐quantitative FFQ, 255 items Validated Administered around week 22 of gestation | First 4–5 months of pregnancy | Risk of pre‐eclampsia during pregnancy | Empirically‐derived: Principal Component Analysis (PCA) | • Vegetable • Processed food • Potato & fish • Cakes & sweets | Estimated relative risks of pre‐eclampsia: DP OR tertile 3 vs. tertile 1 (95%CI) Vegetable: 0.72 (0.73–0.97) Processed food: 1.21 (1.03–1.41) Potato&fish: 1.00 (0.84–1.18) Cakes sweets: 0.90 (0.76–1.06) |
| Haugen et al. 2008 | Norway n = 26 563 women Non‐smoking BMI: 19–32 Age: 21–38 Singleton pregnancy | Prospective cohort study Norwegian Mother and Child Cohort S(MoBa) | FFQ (presumably same features as Brantsaeter et al. 2009, but not stated) Validated Administered at week 18–22 of gestation | First 4–5 months of pregnancy | Early pre‐term delivery Pre‐term delivery Late pre‐term delivery | Hypothesis oriented | • Mediterranean diet (5 criteria) • Mediterranean diet (1–4 criteria) • Non‐Mediterranean diet (0 criteria) | Comparison of MD (5 criteria) vs. non‐MD consuming women: Pre‐term OR (95% CI): 0.73 (0.32–1.68) Early Pre‐term OR (95% CI): 0.93 (0.16–5.37) No significant association between MD and the frequency of pre‐term births |
| Vujkovic et al. 2009 | Netherlands Cases = 50 mothers Controls = 81 mothers Dutch Caucasian women | Case‐control study Exclusion criteria: Pregnancy at study moment Consanguinity Family relation between cases and controls Maternal diabetes mellitus | Self‐administered / Semi‐quantitative FFQ, 200 items Validated Administered at the moment of 14 months after child birth | From 3 months before pregnancy until Administration of the FFQ | Spina bifida in the offspring | Empirically‐derived: • Principal Component Analysis (PCA) • Reduced Rank Regression (RRR) | • Mediterranean diet | Low vs. high percentile of DP associated with risk of the outcome: PCA Adjusted OR (95%CI) <p25, n(%) 2.3 (1.2–6.01) ≥p25, n(%) 1.0 (reference) RRR Adjusted OR (95%CI) <p25, n(%) 3.5 (1.5–7.9) ≥p25, n(%) 1.0 (reference) |
| Vujkovic et al. 2007 | Netherlands Cases = 203 mothers Controls = 178 mothers Dutch European women | Case‐control study Exclusion criteria: Pregnancy Breastfeeding, Folic acid‐containing supplement use Nausea starting after the first week of pregnancy | Self‐administered / Semi‐quantitative FFQ, 102 items Validated Administered 14 months after the birth of index child | 24 months before administration | Cleft lip and/or cleft palate in the offspring | Empirically‐derived: Principal Component Analysis (PCA) | • Western dietary pattern • Prudent pattern | DP associated with risk of the outcome: Western Adjusted OR (95%CI) Tertile 1 1.0 (reference) Tertile 2 1.2 (0.7–2.1) Tertile 3 1.7 (1.0–3.0) Prudent Adjusted OR (95%CI) Tertile 1 1.0 (reference) Tertile 2 0.8 (0.5–1.4) Tertile 3 1.3 (0.8–1.8) |
| Mikkelsen et al. 2008 | Denmark n = 35 530 women Non‐smoking BMI: 19‐32 Age: 21‐38 Singleton pregnancy | Prospective cohort study Danish National Birth Cohort (DNBC) † | Self‐administered / Semi‐quantitative FFQ, 360 items Validated Administered at gestation week 25 | Previous 4 weeks | Early pre‐term delivery Pre‐term delivery Late pre‐term delivery | Hypothesis oriented | ‐Mediterranean diet (5 criteria) ‐Mediterranean diet (1‐4 criteria) ‐Non‐Mediterranean diet (0 criteria) | Comparative of MD (5 criteria) vs. non‐MD consuming women: Pre‐term OR (95% CI): 0.61 (0.65–1.05) Early Pre‐term OR (95% CI): 0.28 (0.11–0.76) Risk of early pre‐term delivery: 72% lower (Statistically significant) |
| Knudsen et al. 2008 | Denmark n = 44 612 pairs (mothers and child) Live born Full‐term Singleton pregnancy | Prospective cohort study Danish National Birth Cohort (DNBC) | Self‐completed / Semi‐quantitative FFQ, 360 items Validated Administered at week 25 of gestation | Previous 4 weeks | Fetal Growth | Empirically‐derived: Principal Component Analysis (PCA) | ‐Factor 1 (Class Western diet) ‐Factor 2 (Class Health conscious) ‐An additional class was described as Intermediate between factor 1 and 2 | Comparative of SGA across the three classes: Class OR (95%CI) Factor 1 1.00 (reference) Intermediate 0.68 (0.55–0.84) Factor 2 0.74 (0.64–0.86) |
| Rifas‐Shiman et al. 2009 | USA n = 1777 women Able to answer in English No plans to move out before giving birth <22 gestational weeks Singleton pregnancy | Prospective cohort study Project Viva | 2 Self‐administered, semi‐quantitative FFQ, 166 items Validated 1st FFQ administered at 1st prenatal visit 2nd FFQ administered at 26–28 weeks of gestation | 1st: since women last menstrual period until FFQ completion date 2nd: 3 previous months | Pre‐eclampsia Birthweight for gestational age Pregnancy weight gain Glucose status | Hypothesis oriented: Alternate Healthy Eating Index modified for pregnancy (AHEI‐P) | ‐Alternate Healthy Eating modified for pregnancy (AHEI‐P) |
Pre‐eclampsia: First trimester AHEI‐P OR 0.96 (95% CI 0.84–1.10) Second trimester AHEI‐P OR 0.87 (95% CI 0.76–1.00) No conclusive differences for birthweight for gestational age, pregnancy weight gain and glucose status |
Study design and sample characteristics
Five of the articles were based on prospective cohort studies (Haugen et al. 2008; Knudsen et al. 2008; Mikkelsen et al. 2008; Brantsaeter et al. 2009; Rifas‐Shiman et al. 2009) and the other two were case‐control studies (2007, 2009). All studies were published in the last 2 years.
Cohort studies
Brantsaeter et al. selected nulliparous women from the Norwegian Mother and Child Cohort Study (MoBa) in order to investigate DPs and the risk of pre‐eclampsia. Mikkelsen et al. and Haugen et al. were parallel studies performed collaboratively but using different cohorts, the Danish National Birth Cohort (DNBC) and MoBa, respectively; selected pregnant women had to be non‐smoking, have a BMI between 19 and 32, aged 21 to 38 years when giving birth and having a singleton pregnancy. Assessing the association between Mediterranean diet and reduced risk of pre‐term birth was the aim of both studies. Knudsen et al. included the highest number of subjects of all the reviewed studies (44 612), using the DNBC data for selecting women with live born, full‐term singleton deliveries; this research evaluated possible associations between maternal diet during pregnancy and fetal growth. Rifas‐Shiman et al. recruited participants into the Project Viva cohort study, as a secondary purpose of their research, to examine associations of diet quality with pregnancy outcomes (which included pre‐eclampsia, among others). Women selected for this study had to be able to answer questions in English, be resident in the study area before delivery and their initial prenatal clinical appointment had to have occurred before gestational week 22.
Case‐control studies
Vujkovic et al. recruited 381 (203 cases and 178 controls) and 131 (50 cases and 81 controls) mothers, respectively, in two studies published in 2007 and 2009. The study published in 2007 tried to identify the associations of maternal DP with risk of cleft lip, cleft palate, or both in the offspring. The study published in 2009 focused on evaluating the role of maternal DP in relation to the risk of spina bifida in the newborn. Detailed information about the sample and exclusion criteria are shown in Table 2.
Questionnaire characteristics
All the food frequency questionnaires used by the seven studies comprised in this review were self‐administered, semi‐quantitative and had been previously validated. The number of items included in the questionnaires varied from 102 to 360. Haugen et al. did not state explicit information about the questionnaire characteristics but as the data is extracted from the same cohort as Brantsaeter et al., it was assumed to have the same features.
Period assessed by the FFQ
Although all the questionnaires assessed dietary intake during pregnancy, high variability was found regarding the moment of their administration and the dietary period they covered. Brantsaeter et al. and Haugen et al. used the MoBa FFQ, which was administered around week 22 assessing the maternal diet during the first 4–5 months of pregnancy while Vujkovic et al. administered an FFQ 14 months after delivery to describe mother's diet since the preconception period. On the other hand, questionnaires from the two studies based on the DNBC were administered at week 25 covering the diet during the previous 4 weeks. Finally, two questionnaires had to be completed during the first and second trimester of pregnancy in the Project Viva cohort study.
Validation
Although the FFQs used in the different studies had been previously validated, not all of them had been specifically validated for collecting data from diet during pregnancy. The two case‐control studies used the same questionnaire developed for the Dutch cohorts of the European Prospective Investigation into Cancer and Nutrition study. Likewise, Rifas‐Shiman et al. applied the Willett FFQ extensively validated for use in the Nurses' Health Study and other large cohort studies but slightly modified for maternal diet assessment. Only the intake of nutrients relevant for pregnancy (folate, protein, retinol and n‐3 fatty acids) and fruit and vegetable intake were validated in the Danish National Birth Cohort FFQ (Mikkelsen et al. 2007).
Supplements
Intake of supplemental minerals and vitamins were considered in some of the studies. Vujkovic et al. (two studies) and Brantsaeter et al. included supplements among the confounders that were taken into account for adjusting their results, while Knudsen et al., Mikkelsen et al., Haugen et al. and Rifas‐Shiman et al. did not consider them.
Development and definition of the DPs
The studies were classified into two categories according to the method used to establish the DP, resulting in three studies with a hypothesis‐oriented assessment of the DP and four studies with an empirically‐derived diet assessment.
Hypothesis‐oriented DPs
The construction of the a priori or theoretical approach (Hypothesis‐oriented DP) consists of the definition of certain scores or indices based on dietary recommendations (Kant 2004; Waijers et al. 2007; Román‐Viñas et al. 2009).
Out of the three studies, two were based on the Mediterranean diet (Haugen et al. 2008; Mikkelsen et al. 2008) and another on alternate healthy eating (Rifas‐Shiman et al. 2009). The criteria used to define Mediterranean diet were: consumption of fish twice a week or more, consumption of vegetables/fruit five times a day or more, use of olive or rape seed oil for cooking and seasoning, meat consumption not exceeding two servings per week and not more than two cups of coffee per day. Alternate healthy eating modified for pregnancy (AHEI‐P) is a 90‐point scale with nine components: vegetables, fruit, ratio of white to red meat, fibre, trans fatty acids, polyunsaturated to saturated fatty acid ratio, folate, calcium and iron from foods (alcohol, nuts and soy proteins were excluded for the assessment of the index because alcohol exerts a detrimental effect during pregnancy and nuts may be avoided during pregnancy due to possible allergen sensitization).
Empirically‐derived DPs
The most common technique used to derive patterns was the PCA, used in all the studies. One study also applied a reduced rank regression (Vujkovic et al. 2009).
Labels for these derived patterns were: ‘Western’ and ‘Prudent’ (Vujkovic et al. 2007), ‘Western’ and ‘Health Conscious’ (Knudsen et al. 2008), ‘Vegetable’, ‘Potato and fish’, ‘Processed food’, ‘Cakes and sweets’ (Brantsaeter et al. 2009) and ‘Mediterranean’ (Vujkovic et al. 2009). Table 3 shows the main food groups included in each DP.
Table 3
Main food groups or food items included in the empirically‐derived (PCA) Dietary Patterns †
| Study | Knudsen et al. 2008 | Vujkovic et al. 2007 | Vujkovic et al., 2009 | Brantsaeter et al. 2009 | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Western † | Health conscious | Western ‡ | Prudent ‡ | Mediterranean | Mediterranean § | Vegetable | Processed food | Potato and fish | Cakes and sweets | |
| Positive | Red and processed meats | Vegetables | Organ meat | Fish | Vegetables | Vegetable oil | Onion, garlic | Processed meat | Cooked potatoes | Cakes |
| Animal fat | Fruit | Red meat | Garlic | Vegetables oil | Fruits | Vegetables | White bread | Processed fish | Waffles and pancakes | |
| Margarine | Fish | Processed meat | Nuts | Fruit | Dairy products | Mushrooms | French fries | Meat and fish spread | Ice cream | |
| Refine grains | Water | Pizza | Vegetables | Fish | Cereal products | Olive and cooking oil | Ketchup/mayonnaise | Lean fish | Sweets and chocolate | |
| Eggs | Vegetable fats | Legumes | Legumes | Fruits | Salty snacks | Margarine | Salty snacks | |||
| Potatoes | Poultry | Potatoes | Cereal products | Rice | Sweets | Rice pudding | ||||
| Snacks | French fries | Chicken | Sugar‐sweetened drinks | Jam and honey | ||||||
| Sweets | Condiments | Water | Pizza | |||||||
| High‐fat dairy | Mayonnaise | |||||||||
| Negative | Fruits | Meat | Sauces and condiments | Blue and white fish | Chicken | |||||
| Potatoes | Potatoes | High‐grain cereals | ||||||||
| Sugars | ||||||||||
Health outcomes
Two studies analyzed the risk of pre‐eclampsia associated with different DP (Brantsaeter et al. 2009; Rifas‐Shiman et al. 2009). Maternal glucose status and weight gain during pregnancy was only ascertained by Project Viva (Rifas‐Shiman et al. 2009). The risk of pre‐term birth associated to DP was analyzed in two studies (Haugen et al. 2008; Mikkelsen et al. 2008) and two others evaluated the role of DP on birthweight (Knudsen et al. 2008; Rifas‐Shiman et al. 2009). Finally, one study associated maternal diet with the risk of spina bifida in offspring (Vujkovic et al. 2009) and another with the risk of developing a cleft lip and palate (Vujkovic et al. 2007).
In the study of Brantsaeter et al. women with high scores for the ‘Vegetable’ pattern had lower risk of pre‐eclampsia whereas women with high scores for the pattern labelled as ‘Processed food’ were at increased risk.
Mikkelsen et al. studied the association of maternal intake of a Mediterranean‐type diet (MD) with the risk of pre‐term birth. A reduction in early pre‐term birth (before 35th week) of more than 70% was observed for women showing all the MD criteria as compared with those who met none of the criteria. Haugen et al. also investigated whether the MD may be associated with a lower risk of pre‐term birth using a similar methodology and definition for the outcome (parallel studies). However, these authors failed to find a reduction in risk.
In 2009, Vujkovic et al. assessed the relationship of maternal DP with the risk of spina bifida. The authors observed a significantly increased risk of spina bifida among offspring from mothers whose consumption did not follow a Mediterranean DP (OR: 3.5; 95% CI: 1.5–8.2).
Knudsen et al. examined possible associations between maternal diet during pregnancy and fetal growth. A significant inverse relationship was observed for low birthweight (birthweight below the 2.5th percentile for gestational age and sex) for women in the ‘Health Conscious’ DP class as compared with women in the ‘Western’ DP category.
Vujkovic et al. (2007) aimed to identify the relationship between DP and cleft lip or cleft palate, or both, in newborns. The ‘Western’ DP was associated with a marginally higher risk of the outcome (OR: 1.7; 95% CI: 1.0–3.0). A lack of effect was reported for the ‘Prudent’ DP.
Finally, Rifas‐Shiman et al. studied the association between maternal diet quality and the risk of pre‐eclampsia and low birthweight, as well as glucose status and weight gain during pregnancy. An increment of 5 points in the AHEI‐P score was associated with a slightly lower risk of developing pre‐eclampsia (OR: 0.87; 95% CI: 0.76–1.00), but only during the second trimester of pregnancy. AHEI‐P score was not associated with pregnancy weight gain, glucose status or birthweight.
Discussion
Epidemiological evidence ascertaining the association between different DP during pregnancy and maternal and infant health related outcomes is limited and conflicting. Only seven studies have analyzed these potential relationships. Moreover, health outcomes were diverse in these analyses. In addition, the divergent methods used to assess the DP make it difficult to compare study results. For example, Brantsaeter et al. (2009) found a protective effect on pre‐eclampsia risk of the ‘Vegetable’ DP extracted by a principal component analysis, whereas Rifas‐Shiman et al. (2009) only found a marginally effect on this outcome for the AHEI‐P (OR 0.87, CI 95% 0.76–1.00). Although both patterns shared certain characteristics such as a high intake of vegetables, fruits and white meat, other components of the diet (foods or nutrients) were included in one but not in the other DP. In addition, the first is an empirically derived pattern whereas the second is a hypothesis‐oriented DP.
Hypothesis‐oriented DPs
A 5‐point increment in the adherence to the AHEI‐P during the second trimester of pregnancy was associated with lower risk of pre‐eclampsia and with a reduction in blood glucose levels in Project Viva (Rifas‐Shiman et al. 2009). However, the research failed to find a relationship between this increment and pregnancy weight gain, birthweight or the presence of gestational diabetes or impaired glucose tolerance. The lack of association with pregnancy weight gain could be explained by the fact that the AHEI‐P was not designed to assess total energy intake.
The main limitation in the use of a hypothesis‐oriented DP is that the constructed scores or indices are based on food guidelines or cut‐off points that may be applicable for certain populations, but not for others (Román‐Viñas et al. 2009). Thus, some dietary indices have to be modified for their use in special populations such as pregnant women. This was the case for the AHEI‐P, a modified version of the AHEI (Bach et al. 2006). Alcohol and nut intakes are not considered in the AHEI‐P because of the detrimental effect of alcohol intake during pregnancy and the advice to avoid the consumption of nuts by the pregnant women due to their allergenic properties (Frank et al. 1999). However, the AHEI‐P reflects the intake of nutrients that are particularly important during pregnancy such as folate, iron and calcium. Moreover, for each nutrient or food included in the pattern, most of the hypothesis‐oriented DPs define a cut‐off point using the intake distribution for the entire sample (median for example). Thus, if the consumption of specific components of the DP is low in the study population, even small intakes could be considered above the median. The AHEI‐P does not have this problem because the contribution of each component to the score is based on the adequacy of its intake.
Hypothesis‐oriented DPs have been used to study the association between diet during pregnancy and pre‐term birth as well. In this case, two different cohort studies defined a DP labelled ‘Mediterranean Diet’. This pattern was characterized by a high consumption of vegetables, fruits, fish, olive or seed oil and low consumption of red meat and coffee. However, the definition of the pattern, based on that used by Kourhy et al. (2005), included coffee intake, considered olive and seed oil in the same category and did not take into consideration the consumption of legumes, nuts, cereals or whole fat dairy products. These are included in other internationally accepted definitions of a Mediterranean DP (Bach et al. 2006). Furthermore, the results obtained for these studies were contradictory. In spite of the beneficial effect found for a high Mediterranean DP score by the Danish cohort study, the Norwegian Mother and Child Cohort Study did not find any association with this pattern. It is difficult to explain the difference in the results when both studies had a similar prospective design, the size and characteristics of the sample excluding women smokers and those with abnormal values for energy intake and BMI were comparable. Both collected dietary data during the mid‐pregnancy period (around 22–25 weeks of gestation) using a self‐administered and validated food frequency questionnaire with supplement information, and applied the same definition to assess the Mediterranean DP and the outcome, controlling for similar confounders such as socio‐economic status or parity.
It is important to point out that both studies were conducted amongst Northern European populations where the consumption of a Mediterranean DP or of some of its components is scarce and limited to individuals with high education levels (Petkeviciene et al. 2007; Prättäläet al. 2009). As such, although the authors adjusted for socio‐economic status we can not exclude the possibility of some residual confounding where diet may be a marker of a healthier life style with less total energy intake and weight gain during pregnancy and more physical activity and supplement use.
Empirically derived DPs
Two case–control studies used statistical techniques to derive DP and to analyze their effect on the risk of developing a cleft lip, cleft palate or spina bifida. Both collected dietary intake data 14 months after delivery. The authors argue about the validity of the dietary data since nutritional habits seem to be constant during life except for periods of dieting, pregnancy, breastfeeding and extreme growth (Willett 1998). They suggest that the assessment of diet 14 months after delivery would reflect dietary habits in the periconceptional period. This period is critical in the aetiology and development of the studied diseases (Cetin et al. 2009). One of the main problems in nutritional epidemiology for establishing diet‐disease relationships is to assess the relevant period of exposure and to ascertain dietary intake adequately. The first task is particularly difficult during pregnancy and the second one is even more critical when the epidemiological design is a case‐control study with small numbers of subjects. Thus, the results derived from these studies should be cautiously interpreted.
Two major empirically‐derived DPs and their relationship with health outcomes in adult population have been extensively described (Fung et al. 2005; Varraso et al. 2007). The Prudent DP is characterized by a high consumption of fruits, vegetables, whole‐grains, fish and poultry whereas the Western DP includes red and processed meat, french fries, whole fat dairy and sweets. Similar results were obtained in this review when pregnant women were studied although with minor differences in composition and labelling (Table 3). The results showed a beneficial effect for the ‘Vegetable’, ‘Mediterranean or ‘Health conscious’ DP and a detrimental effect for the ‘Western’ or ‘Processed foods’ DP on maternal diseases (risk of pre‐eclampsia) and infant outcomes and conditions (risk of spina bifida or cleft lip, birthweight). Nevertheless, we have to take into account that having a healthier DP is also associated with having a healthier life‐style (more physical activity, use of vitamins and minerals and lower weight gain during pregnancy) and with having a higher educational level. Hence, all these possible confounders should be considered in the analyses.
One of the most important disadvantages of the empirically‐derived DP is that several subjective decisions from the investigators are necessary to determined the inclusion or exclusion of certain food groups in the construction of the patterns (Román‐Viñas et al. 2009).
Limitations in the use of DPs during pregnancy
Some of the most important limitations are those derived from the use of DPs in general (Bach et al. 2006). All the components included in the DP are considered of similar relevance when the DP is elaborated independent of their actual contribution to the aetiology of the disease or condition. In addition, some scores establish a linear relationship between a component intake and disease risk when some associations are known not to have a dose‐response trait (Serra‐Majem et al. 2004) such as that based on alcohol intake and cardiovascular risk (Panagiotakos et al. 2007).
Most of the studies included in this review used validated food‐frequency questionnaires to assess the components (foods or nutrients) of the DP. However, their use among pregnant women had not been evaluated in some cases (2007, 2009; Rifas‐Shiman et al. 2009). Moreover, only a few validation studies have evaluated DPs (Román‐Viñas et al. 2009) and none of them was based on pregnant women.
Moreover, supplement use during pregnancy is a key point in the study of the association between maternal diet and offspring outcomes (Shah & Ohlsson 2009). Therefore, the use of vitamin and mineral supplements during pregnancy should be considered in the analyses either including its use as a component of the index, or incorporating its composition into the calculation of total nutrient intake or adjusting for the variable in the multivariable models.
A potential problem in analyses involving pregnant women is, first, to adequately select the period in which the FFQ is administered. Second, it is crucial to establish the period of time covered by the FFQ. This review aimed to include articles where dietary exposure was ascertained during pregnancy; however, two of the studies assessed diet 14 months after the pregnancy (2007, 2009). Although the aim of these authors was to ascertain the diet followed by the pregnant women during their periconception period, it could be possible that the presence of diseases in their offspring had led to changes in mother diets after delivery, especially in the case of the Spina Bifida that has been associated to diet.
Moreover, since maternal nutritional status before pregnancy could be associated with mother and infant health outcomes (Moore & Davies 2005; Zhang et al. 2006), it would be of interest to also analyze the effects of pre‐gravid diet.
In conclusion, some recommendations should be taken into account when DP during pregnancy is analyzed. First, it is important to use a validated food frequency questionnaire designed to assess food and nutrient intake or to derive DPs during pregnancy. It is also critical to consider the special role exerted by vitamin and mineral supplements in this special population and include this information in the DP or in the analysis as a separate variable. Moreover, special attention should be paid to when dietary data is collected (the most relevant period for the outcome of interest), and to adjust the results for lifestyle characteristics and educational levels of the participants so as to avoid confounding. Finally, it is important to appropriately label the constructed DP. In the case of the hypothesis‐oriented DP, this may be specifically designed for the aim of the study or adapted to pregnancy features. The key factor is to correctly choose the components comprising the score and the expected relationship with the disease outcome of interest (which may not necessarily be linear).
Source of funding
The work reported here has been carried out within the EURRECA Network of Excellence (http://www.eurreca.org) that is financially supported by the Commission of the European Communities, specific Research Technology and Development (RTD) Programme Quality of Life and Management of Living Resources, within the Sixth Framework Programme, contract no. 036196. This report does not necessarily reflect the Commission's views or its future policy in this area.
Conflict of interest
The authors declare no conflict of interest.
References
- Bach A., Serra‐Majem L., Carrasco J.L., Roman B., Ngo J., Bertomeu I., et al (2006) The use of indexes evaluating the adherence to the Mediterranean diet in epidemiological studies: a review. Public Health Nutrition 9, 132–146. [PubMed] [Google Scholar]
- Brantsaeter A.L., Haugen M., Samuelsen S.O., Torjusen H., Trogstad L., Alexander J., et al (2009) A dietary pattern characterized by high intake of vegetables, fruits, and vegetable oils is associated with reduced risk of preeclampsia in nulliparous pregnant Norwegian women. The Journal of Nutrition 139, 1162–1168. [PMC free article] [PubMed] [Google Scholar]
- Cetin I., Berti C. & Calabrese S. (2010) Role of micronutrients in the periconceptional period. Human Reproduction Update 16, 80–95. [PubMed] [Google Scholar]
- Frank L., Marian A., Visser M., Weinberg E. & Potter P.C. (1999) Exposure to peanuts in utero and in infancy and the development of sensitization to peanut allergens in young children. Pediatric Allergy and Immunology 10, 27–32. [PubMed] [Google Scholar]
- Fung T.T., Hu F.B., Holmes M.D., Rosner B.A., Hunter D.J., Colditz G.A., et al (2005) Dietary patterns and the risk of postmenopausal breast cancer. International Journal of Cancer 116, 116–121. [PubMed] [Google Scholar]
- Haugen M., Meltzer H.M., Brantsaeter A.L., et al (2008) Mediterranean‐type diet and risk of preterm birth among women in the Norwegian Mother and Child Cohort Study (MoBa): a prospective cohort study. Acta Obstetricia et Gynecologica Scandinavica 87, 319–324. [PubMed] [Google Scholar]
- Hu F. (2002) Dietary pattern analysis: a new direction in nutricional epidemiology. Current Opinion in Lipidology 13, 3–9. [PubMed] [Google Scholar]
- Jacques P.F. & Tucker K.L. (2001) Are dietary patterns useful for understanding the role of diet in chronic disease? The American Journal of Clinical Nutrition 73, 1–2. [PubMed] [Google Scholar]
- Kant A. (2004) Dietary patterns and health outcomes. Journal of the American Dental Association 104, 615–636. [PubMed] [Google Scholar]
- Klemmensen A., Tabor A., Østerdal M.L., et al (2009) Intake of vitamin C and E in pregnancy and risk of pre‐eclampsia: prospective study among 57 346 women. BJOG 116, 964–974. [PubMed] [Google Scholar]
- Knudsen V.K., Orozova‐Bekkevold I.M., Mikkelsen T.B., Wolff S. & Olsen S.F. (2008) Major dietary patterns in pregnancy and fetal growth. European Journal of Clinical Nutrition 62, 463–470. [PubMed] [Google Scholar]
- Kourhy J., Haugen G., Tonstad S., Froslie K.F. & Henriksen T. (2005) Effect of a cholesterol‐lowering diet on maternal, cord, and neonatal lipids, and pregnancy outcome: a randomized clinical trial. American Journal of Obstetrics and Gynecology 193, 1292–1301. [PubMed] [Google Scholar]
- Magnus P., Irgens L.M., Haug K., Nystad W., Skjaerven R. & Stoltenberg C. (2006) Cohort profile: the Norwegian Mother and Child Cohort Study (MoBa). International Journal of Epidemiology 35, 1146–1150. [PubMed] [Google Scholar]
- Mikkelsen T.B., Olsen S.F., Rasmussen S.E. & Osler M. (2007) Relative validity of fruit and vegetable intake estimated by the food frequency questionnaire used in the Danish National Birth Cohort. Scandinavian Journal of Public Health 35, 172–179. [PubMed] [Google Scholar]
- Mikkelsen T.B., Osterdal M.L., Knudsen V.K., et al (2008) Association between a Mediterranean‐type diet and risk of preterm birth among Danish women: a prospective cohort study. Acta Obstetricia et Gynecologica Scandinavica 87, 325–330. [PubMed] [Google Scholar]
- Miyake Y., Sasaki S., Tanaka K., Ohfuji S. & Hirota Y. (2009) Maternal fat consumption during pregnancy and risk of wheeze and eczema in Japanese infants aged 16–24 months: the Osaka maternal and child health study. Thorax 64, 815–821. [PubMed] [Google Scholar]
- Moore V.M. & Davies M.J. (2005) Diet during pregnancy, neonatal outcomes and later health. Reproduction, Fertility, and Development 17, 341–348. [PubMed] [Google Scholar]
- Olsen J., Melbye M., Olsen S.F., Sørensen T.I., Aaby P., Andersen A.M. et al (2001) The Danish National Birth Cohort – its background, structure and aim. Scandinavian Journal of Public Health 29, 300–307. [PubMed] [Google Scholar]
- Panagiotakos D.B., Kourlaba G., Zeimbekis A., Toutouzas P. & Polychronopoulos E. (2007) The J‐shape association of alcohol consumption on blood pressure levels, in elderly people from Mediterranean Islands (MEDIS epidemiological study). Journal of Human Hypertension 21, 585–587. [PubMed] [Google Scholar]
- Petkeviciene J., Klumbiene J., Prättälä R., Paalanen L., Pudule I. & Kasmel A. (2007) Educational variations in the consumption of foods containing fat in Finland and the Baltic countries. Public Health Nutrition 10, 518–523. [PubMed] [Google Scholar]
- Prättälä R., Hakala S., Roskam A.J., et al (2009) Association between educational level and vegetable use in nine European countries. Public Health Nutrition 12, 2174–2182. [PubMed] [Google Scholar]
- Rifas‐Shiman S.L., Rich‐Edwards J.W., Kleinman K.P., Oken E. & Gillman M.W. (2009) Dietary quality during pregnancy varies by maternal characteristics in Project Viva: a US cohort. Journal of the American Dietetic Association 109, 1004–1111. [PMC free article] [PubMed] [Google Scholar]
- Román‐Viñas B., Ribas Barba L., Ngo J., Martínez‐González M.A., Wijnhoven T.M. & Serra‐Majem L. (2009) Validity of dietary patterns to assess nutrient intake adequacy. The British Journal of Nutrition 101 (Suppl. 2), S12–S20. [PubMed] [Google Scholar]
- Sánchez Villegas A. & Serra Majem L. (2005) Epidemiología nutricional In: Tomo III: Nutrición Humana en el estado de salud (eds Gil Hernández A., Camero E., Culebras J.M., González J. & Leon M.), pp 599–622. Grupo Acción Médica: Madrid. [Google Scholar]
- Serra‐Majem L., Trichopoulou A., Ngo J., Cervera P., Garcia Alvarez A., La Vecchia C. et al (2004) Does the definition of the Mediterranean diet need to be updated? Public Health Nutrition 7, 927–929. [PubMed] [Google Scholar]
- Shah P.S. & Ohlsson A. (2009) Knowledge synthesis group on determinants of low birth weight and preterm births. Effects of prenatal multimicronutrient supplementation on pregnancy outcomes: a meta‐analysis. Canadian Medical Association Journal 180, E99–E108. [PMC free article] [PubMed] [Google Scholar]
- Strøm M., Mortensen E.L., Halldorsson T.I., Thorsdottir I. & Olsen S.F. (2009) Fish and long‐chain n‐3 polyunsaturated fatty acid intakes during pregnancy and risk of postpartum depression: a prospective study based on a large national birth cohort. The American Journal of Clinical Nutrition 90, 149–155. [PubMed] [Google Scholar]
- Varraso R., Fung T.T., Barr R.G., Hu F.B., Willett W. & Camargo C.A. Jr (2007) Prospective study of dietary patterns and chronic obstructive pulmonary disease among US women. The American Journal of Clinical Nutrition 86, 488–495. [PMC free article] [PubMed] [Google Scholar]
- Vujkovic M., Ocke M.C., Van Der Spek P.J., Yazdanpanah N., Steegers E.A. & Steegers‐Theunissen R.P. (2007) Maternal Western dietary patterns and the risk of developing a cleft lip with or without a cleft palate. Obstetrics and Gynecology 110, 378–384. [PubMed] [Google Scholar]
- Vujkovic M., Steegers E.A., Looman C.W., Ocké M.C., Van Der Spek P.J. & Steegers‐Theunissen R.P. (2009) The maternal Mediterranean dietary pattern is associated with a reduced risk of spina bifida in the offspring. BJOG 116, 408–415. [PubMed] [Google Scholar]
- Waijers P.M., Feskens E.J. & Ocke M.C. (2007) A critical review of predefined diet quality scores. The British Journal of Nutrition 97, 219–231. [PubMed] [Google Scholar]
- Willett W.C. (1998) Nutritional Epidemiology, 2nd edn. Oxford University Press: New York. [Google Scholar]
- Zhang C., Schulze M.B., Solomon C.G. & Hu F.B. (2006) A prospective study of dietary patterns, meat intake and the risk of gestational diabetes mellitus. Diabetologia 49, 2604–2613. [PubMed] [Google Scholar]