Índice de alimentação saudável para gestantes : adaptação para uso em gestantes brasileiras Healthy eating index for pregnancy : adaptation for use in pregnant women in Brazil

MÉTODOS: Gestantes entre a 16a e 36a semana de gestação (n = 712) foram arroladas em unidades básicas de saúde em Porto Alegre e Bento Gonçalves, RS, em 2010. Com base no índice americano Alternate Healthy Eating Index for Pregnancy (AHEI-P) foi criado o Índice de Alimentação Saudável para Gestantes Brasileiras (HEIP-B). Foram aplicados o questionário de frequência alimentar e o questionário sociodemográfi co. Foi utilizada a análise de componentes principais focada para avaliar a relação entre os índices e os nutrientes relevantes à gestação.

Pregnancy is a period during which nutritional necessities increase, and appropriate nutrition is essential to the health of mother and baby.Pregnant women should consume specifi c varieties and quantities of foods, bearing in mind dietary guidelines and cultural eating habits, in order to meet energy and nutrition needs and to follow weight gain recommendations.a From a public health perspective, previous diagnosis and assessment of subsequent food intake in order to determine the prevalence of appropriate consumption is as important as correct guidance.The study of dietary intake is a complex process, as nutrition involves biological, socio-economic, cultural and symbolic dimensions. 2 One way of assessing nutrition patterns is through a priori methods, using existing nutritional knowledge.For example, indexes which evaluate individuals' consumption compared to what is recommended are created using specifi c dietary guidelines. 9,11e such index, called the Healthy Eating Index (HEI) was produced by the US Department of Agriculture in order to assess the quality of the Americans' diet.It includes ten components: fi ve refer to food groups from the American Dietary Guide ("Grains", "Vegetables", "Fruit", "Dairy" and "Meat"), four to nutrients ("total fat", "saturated fat", "cholesterol" and "sodium"), and one that measures variety in the diet.Each component scores between 0 and 10. 10 There are few studies on specifi c dietary standards in pregnancy, especially with dietary indexes.Rifas-Shiman et al, 25 used the Alternate Healthy Eating Index (AHEI) to create a healthy eating index adapted for pregnancy: Alternate Healthy Eating Index for Pregnancy (AHEI-P).This index, created to assess diet quality, has nine components, each of which is worth ten points: vegetables; fruit; two ratios ("white meat/ red meat" and "polyunsaturated/saturated"); fi ber; trans fats; folate, calcium and iron. 25national study carried out by Maeda b applied the Healthy Eating Index adapted for pregnancy.In spite of the name used, this index is not adapted from a pregnancy specifi c instrument but rather from the HEI 10 developed to evaluate the American population's diet.
The present study aimed to assess the overall diet quality in a sample of pregnant Brazilian women based on one simple, objective parameter.This article refers to data collected in stage II of ECCAGe, in which 785 pregnant women, aged 13-42, were selected consecutively in the waiting room of prenatal appointments in the cities of Bento Gonçalves (eight Primary Care Units and one Maternal-Infant Referral Center) and Porto Alegre (seven PCU belonging to the School Health Center Murialdo and three belonging to the Municipal Health Department).Of the pregnant women selected, 59 refused to participate (7.5%) and there were nine losses to follow up (1.2%), making a total of 712 pregnant women in the fi nal sample.
Women completed a sociodemographic questionnaire.Clinical data and data referring to the women's height and weight (barefoot and lightly clothed) were collected according to the Ministry of Health.c The semi-qualitative Food Frequency Questionnaire (FFQ), previously validated for pregnant women, 6 was used to collect data on diet during pregnancy.Coeffi cients of correlation, adjusted for energy ranging from 0.10 (vitamin E) to 0.50 (vitamin C), were presented in the validation of this tool.The FFQ has eight options for frequency of consumption, which range from "more than three times a day" to "never or almost never".The list of foods was composed of 88 items.In order to assess the quantity consumed, these food items were presented in standardized portions either in serving sizes or units of the food.The interviewee reported her usual consumption in multiples of these portions.
In order to determine the quantity of the product in a standard portion in grams (g) or milliliters (ml), i.e., the portion corresponding to the serving size pre-established in the FFQ, the Table for Assessing Food Consumption in Portion Sizes was used. 24Foods which were not included in this table were weighed or measured in order establish a standard value in grams (g) or milligrams (mg) or micrograms (mcg).
For each food item, the value of micronutrients and fi ber was calculated in g, mg or micrograms (mcg).For macronutrients, the value was calculated in g and the percentage of the recommended Daily Allowance (%RDA).
The Brazilian Table of Food Composition (TACO) d was used as the reference to obtain the foods' nutritional composition.A second table, the Table of Food Composition: support for dietary decisions, 22 was consulted when any of the food items shown in the FFQ or nutrients being investigated were not included in the TACO.d A third option was to consult the food's label.Some items in the FFQ had more than one subtype in the composition tables, such as the banana which, in Pinheiro's table, 24 had fi ve different types listed.The mean of these subtypes was calculated in such cases in order to obtain a value in g or ml which took into account the variability of the food in question.e Estimates for the consumption of each nutrient were consolidated based on the calculation: Total of the nutrient of a food = nº portions x frequency of consumption/day x nutrient in the portion.e The food items were separated into groups according to their category in the dietary guide for the Brazilian population.f The number of portions consumed of each food was calculated according to weight in grams or volume of the foods present in the groups of the Brazilian guide.f The mean of the subtypes of a specifi c food was used to fi nd the most representative value for the food in question.For example, for the food item "cake".There were four subtypes in the Brazilian guide: f banana cake, carrot cake, chocolate cake and coconut cake.A syntax g for constructing the variables necessary for calculating the indexes of health eating was created in the SPSS v.16 program.
The Brazilian dietary guide establishes portions based on a diet of 2000 kcal.f Adapting the number of recommended portions for each group included in the indexes to the need of pregnant women was based on assuming an increase of 300kcal/day during pregnancy. 12(Table 1).To that end, we used a simple rule of three.
The AHEI-P 25 was based on the AHEI, excluding "alcohol", as it is not recommended for pregnant women, and "nuts and soya protein", due to allergic reactions (tofu and soya were included in the "vegetable" component).Three components were added to refl ect the intake of nutrients which are particularly important during pregnancy: folate, iron and calcium. 25e AHEI-P includes nine components: two food groups ("vegetables" and "fruit"), two ratios ("white meat/red meat" and "polyunsaturated/saturated") and fi ve nutrients ("fi ber", "trans fats", "calcium", "folate" and "iron").Each of these components has a score varying between 0 and 10 (Table 1). 25e Brazilian population dietary guide f was used in the creation of the HEIP-B in order to establish the recommended number of portions for each food group.The DRI [13][14][15] were used to establish recommendations for nutrients.Minimum and maximum portions determined by the Brazilian guide f were adapted to the additional calorie needs of pregnant women, as mentioned above.In addition, another component, present in the Brazilian guide, was included: "beans and other protein-rich vegetables".The AHEI-P ratios for "white meat/red meat" and "polyunsaturated/saturated" were used, as there were no Brazilian references which documented limits.
The scores for each component were calculated (equations 1 and 2) according to the criteria shown in Table 1 and intermediate values were calculated proportionally: g The only item for which this formula could not be used was trans fats, as the higher the consumption of foods which contained them, the worse the score.Therefore, the formula used was: In both cases, = Quantity of the component ingested x.For the HEIP-B, the total sum of components was divided by 10 and multiplied by 9, to enable comparison with the AHEI-P.

Min
AHEI-P and HEIP-B were categorized similarly to Basiotis et al, 1 with the aim of visualizing distribution patterns in pregnant women's diets.The cut off points to classify the quality of the diet established by these authors may be interpreted as follows: out of a total score of 100 points, a score which classifi es a diet as "good quality" (total score > 80) indicates that at least 80% of the dietary necessities included in the index are being achieved.Similarly, for a total score of 90 points the following cut off points were established: Poor Quality (< 45 points), Needs Improvement (45 to 72 points) and High quality (72 points). 1 Socio-demographic and anthropometric data were described as means, standard deviations or absolute and relative frequencies.Quartiles and percentages for those pregnant women who scored maximum and minimum scores were used to describe the two dietary indexes.
Spearman's correlation coeffi cient was used to measure the linear association between the indexes and nutrients investigated, due to the asymmetry of the data.Focused principal component analysis (FPCA) was used with Spearman's correlation 5 in order to obtain graphic representations of each of the indexes with some of the nutrients.This technique enables the description and understanding of the relationships between differing variables, particularly response variables, in the case of this study, the two indexes.The results are shown graphically and the variable of interest is in the center.Circles with a smaller radius indicate closer correlation.The 2nd circle from the outside defi nes correlation coeffi cients statistically signifi cant at the 5% level. 3his analysis was carried out using the R program (R Development Core Team, 2008) using the psy library. 5he other analyses were carried out with SPSS v. g Syntax for creating an index of healthy eating for pregnant women, according to the groups from the dietary guide of the Brazilian population.Available from: www.mat.ufrgs.br/~camey/HEIP_B/HEIsyears old (SD = 6.4) and a mean of 7.6 years of schooling (SD = 2.7).The majority of the women (78.5%) were married or living with their partner and the mean household income was 1.1 minimum wages (SD = 0.7).The mean gestational age was 24.5 weeks (SD = 5.8) Gonçalves (Protocol nº 2006561/2006).The interviews were carried out in a designated area, after a consent form had been signed.

RESULTS
The 712 pregnant women had a mean age of 24.6  The median, the 1 st and 3 rd quartiles and the percentage of the pregnant women who scored maximum and minimum scores in the indexes and their components are shown in Table 2.
The medians and interquartile intervals (with the exception of "calcium"), and the values of total scores were similar when AHEI-P and HEIP-B were compared.The median for the "calcium" component differed only for pregnant women aged 19 and over (9.1 for AHEI-P and 10 for HEIP-B).The absence of differences between medians in the other components occurred due to the majority of maximum and minimum values attributed being the same.
Based on the fi nal scores of the indexes, a dietary classifi cation was performed which varied according to the dietary index: for AHEI-P and HEIP-B, 65.4% (n = 465) and 62.6% (n = 446) respectively, were in the Needs Improvement category.HEIP-B showed signifi cant correlation with the nutrients under investigation: lower for vitamin B12 (r = 0.26, < 0.001) and higher for folate and potassium (r = 0.78, < 0.001).HEIP-B showed the following correlations for nutrients specifi cally recommended for pregnancy: folate (r = 0.78; p < 0.001), calcium (r = 0.63; p < 0.001) and iron (r = 0.68; p < 0.001).
Nutrients relevant to pregnancy (Figure ) showed similar patterns in AHEI-P (left) and HEIP-B (right), with positive correlations above 0.4, except for vitamin B12.There was weak correlation of two nutrients in HEIP-B, but which had high correlation between themselves (vitamins A and B12).A strong correlation was observed between two nutrients (folate and fi ber) and the index, close to the line closest to the center (r = 0.8).All of the correlations were signifi cant, once the variables were located within the circle derived by the line between correlations 0 and 0.2 (Figure).

DISCUSSION
HEIP-B showed good correlations with the nutrients being investigated during pregnancy and 62.6% of the pregnant women presented diets that "needs improvement".
National studies using dietary indexes to assess dietary quality have found that the majority of the population studied "need to improve" their diet, 19,b although cut off points for this classifi cation vary between studies.
One of the reasons for adapting the AHEI-P to the national dietary guide f is that there are differences, in terms of portions of the food groups, between the American and the Brazilian guide.Moreover, the "beans and other protein-rich vegetables" group was included in the "vegetable" group in the American guide, whereas the Brazilian guide f contained specifi c recommendations.In a national study, Mota et al adapted the Health Eating Index to the Brazilian population and included the group of legumes in the same way. 19cluding the "beans" group, a typically Brazilian food, and adopting the portions of the Brazilian guide to the extra 300 kcals needed for pregnancy may have increased the fi nal HEIP-B score.
The number of portions recommended for each group was rounded up to facilitate clinical practice.There was diffi culty with the "beans" group, since the number of portions obtained (1.2 portions) made it diffi cult to apply it in practice, limiting the recommendation to the public.
There are no nutritional and weight gain recommendations specifi c to Brazilian pregnant women.National authors 27 working with pregnant women use, for example, the Institute of Medicine 16 recommendations to estimate weight gain during pregnancy and to quantify the needed nutrients.
In this study, the FPCA technique was chosen to assess correlation between nutrients investigated during pregnancy and the dietary indexes because it has the advantage of showing correlations graphically and simultaneously.Based on this, we can state that these nutrients basically belong to one single cluster, i.e. the variability of each nutrient is captured in a similar way by the AHEI-P and the HEIP-B.Moreover, the HEIP-B showed moderate correlations (r > 0.4) with all nutrients, except vitamin B12.The reason this vitamin does not show a good correlation with the index is that few pregnant women consume less than the recommended quantity.Therefore, irrespective of the HEIP-B score, consumption of this vitamin was high.
In a Canadian study carried out by Pick et al, 23 the quality of pregnant women's diet was assessed using a dietary index.The authors examined intake of nutrients of interest during pregnancy in the women who had what was considered a "good" index (HEI > 80), and observed that 4% reached the recommended level of folate and none consumed the recommended amount of iron.The authors concluded that the index was useful for analyzing whether dietary guidelines are followed, but that it did not capture the uniqueness of the pregnant women's diet.The index proposed in the study by Pick et al 23 did not include any pregnancy-specifi c nutrient.
A limitation of these indexes is that that maximum score for each food group is based only on the minimum number of portions to be consumed, i.e., there is no "penalty" for excessive consumption.One of the reasons for not including penalization of high consumption in the HEIP-B is that there is no consensus on limits for the number of recommended portions in each food groups.However, as excessive weight gain during pregnancy may affect the health of mother and baby, 16,21 further discussion in order to establish maximum limits is necessary.
Recommended intakes adapted to the energy needs of each individual were not used in constructing the HEIP-B, which may be another limitation.Instead we opted for a single recommendation, as in the Brazilian guide.f However, calculating recommendations based on energy density may have led to more precise results.
In a revision of dietary indexes, Waijers et al 28 indicated the limitations of using such tools to assess dietary quality: they reported that the indexes may not deal adequately with correlation in intake of various dietary factors and with the existence of interactions between nutrients. 28Other authors describe how, although limitations exist, the indexes manage to assess overall diet, characterizing dietary patterns, in contrast to the analysis of nutrients in isolation. 4,7,17,18other limitation of the study was using the FFQ as an assessment method of dietary consumption, as it contains a limited number of foods in comparison with dietary history. 28Moreover, as it is a semi-quantitative questionnaire, the FFQ used may have contributed to overestimating the quantities of each of the investigated food items consumed. 6The validation of the FFQ used was for nutrients and not food groups, which may have had some infl uence in the construction of the dietary index.However, some authors have used the FFQ to calculate this type of index. 8,26 spite of these limitations, dietary indexes are instruments of recognized usefulness in assessing dietary quality, especially in the primary health care fi eld.These instruments should be adapted to national guidelines and to the population under study, so that cultural realities are considered.When comparing these indexes, dietary quality was found to be better when assessed by the HEIP-B.
In this study, the dietary quality of the majority of pregnant women was classifi ed as within the "needs improvement" cut off point, demonstrating the need to work more specifi cally in the dietary education of this group.
Figure.Correlation of dietary indices with nutrients relevant to pregnancy using focused principal component analysis.Bento Gonçalves and Porto Alegre, Southern Brazil, 2007.(n = 712)

Table 1 .
Components of the AHEI-P and HEIP-B, with scoring criteria.Bento Gonçalves and Porto Alegre, Southern Brazil, 2007.(n= 712)Number of portions was obtained by adding the number of portions of all foods in that group.

Table 2 .
Descriptive measures for each component and total scores for the AHEI-P and HEIP-B.Bento Gonçalves and Porto Alegre, Southern Brazil, 2007.(n = 712) AHEI-P: Alternate Healthy Eating Index for Pregnancy HEIP-B: Healthy Eating Index for Pregnancy -adapted to Brazil