- Email: [email protected]
Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54
YCLNU2805_proof ■ 29 April 2016 ■ 1/7
Clinical Nutrition xxx (2016) 1e7
Contents lists available at ScienceDirect
Clinical Nutrition journal homepage: http://www.elsevier.com/locate/clnu
Original article
Q6 Q5
Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern Tatiana Papazian a, b, Hala Hout b, Darine Sibai b, Nour Helou b, Hassan Younes b, c, Nada El Osta d, e, Lydia Rabbaa Khabbaz a, b, * a ^le de qualit Laboratoire de Pharmacologie, Pharmacie clinique et Contro e des M edicaments, Faculty of Pharmacy, Saint-Joseph University of Beirut, Beirut, Lebanon b Department of Nutrition, Faculty of Pharmacy, Saint-Joseph University of Beirut, Beirut, Lebanon c Department of Nutrition and Health Sciences, Institut Polytechnique LaSalle Beauvais, France d Department of Public Health, Faculty of Medicine, Saint-Joseph University, Beirut, Lebanon e Department of Prosthodontics, Faculty of Dental Medicine, Saint-Joseph University, Beirut, Lebanon
a r t i c l e i n f o
s u m m a r y
Article history: Received 12 October 2015 Accepted 8 April 2016
Background & aims: Accurate dietary assessment tools are required to ensure that maternal diet supplies all the nutrients needed for fetal development. However, no dietary method could accurately estimate food intake during gestation. Food Frequency Questionnaires (FFQ), frequently used in epidemiological studies, estimate long term nutritional status of the target population. However, it is recommended to create and validate a FFQ compatible with the dietary habits of the studied population, to avoid cultural and social discrepancies. This study aimed to develop and test the reproducibility and the validity of a semi-quantitative FFQ compatible with the diet of Mediterranean and Middle-Eastern population, in a sample of Lebanese pregnant women. Methods: 128 women participated in the validation study, while 38 took part in the reproducibility phase, which was repeated in a time frame of 21 days. The FFQ was validated against a 24 h dietary recall (DR). Results: The intra class correlation coefficient (ICC) ranged from 0.935 for calcium to 0.984 for vitamin D (p value < 0.001), indicating an excellent reproducibility. A satisfactory agreement between the two dietary tools was demonstrated using BlandeAltman plot and Spearman's and Pearson's correlations coefficients which varied between 0.294 for iron to 0.762 for caloric intake (p value < 0.01). Conclusions: The newly developed FFQ englobing Mediterranean food items was culture specific and assessed the nutrient intake of our population. Administering this tool in future researches will help monitor the nutritional status of pregnant women, aiming at improving maternal and newborn health. © 2016 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism.
Keywords: Maternal nutritional status Food frequency questionnaire 24 h dietary recall Validity Reproducibility
Q2
1. Introduction Adequate food intake during pregnancy is a key factor affecting the health of both the mother and the infant. Along the
Q1
* Corresponding author. Laboratoire de Pharmacologie, Pharmacie clinique et ^le de qualite des Me dicaments, Faculte de pharmacie, Campus des sciences Contro dicales (CSM), Rue de Damas, Universite Saint-Joseph, B.P. 11-5076, Riad el Solh, me Beirut 1107 2180, Lebanon. Tel.: þ961 3 224 449, þ961 1 4210 000x6792; fax: þ961 1 421 022. E-mail address: [email protected] (L.R. Khabbaz).
physiological changes of her body, a pregnant woman faces the challenges of preventing and correcting any existing nutrient deficiency during both the periconceptional phase and pregnancy. This adequacy can only be achieved by a well-balanced diet providing all the necessary nutrients and energy needed for her and the fetus [1]. Strong scientific evidences confirm the impact of maternal diet on neonatal outcomes [2] and the occurrence of growth problems and chronic diseases during childhood [3]. Despite the large number of studies investigating the role of nutrition in pregnancy and neonatal outcome, few valid
http://dx.doi.org/10.1016/j.clnu.2016.04.015 0261-5614/© 2016 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism.
Please cite this article in press as: Papazian T, et al., Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.04.015
55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
YCLNU2805_proof ■ 29 April 2016 ■ 2/7
2
T. Papazian et al. / Clinical Nutrition xxx (2016) 1e7
questionnaires to assess nutritional status during gestation exist. Hence, to monitor the dietary profile of a selected population, researchers often rely on different tools, each one having its strengths and limitations [4]. Food frequency questionnaires (FFQs) have been designed and used in many epidemiological studies, to assess the nutrient intake of different age groups, healthy or suffering from a disease state or pregnant women [4]. This questionnaire, ideally conducted by a trained dietitian, permits to evaluate the dietary intake of the respondents by asking them to report the frequency of their usual consumption of common food items, from different food groups, over a specified period of time (daily, weekly, or monthly). In addition, depending on the purpose of the study, portion sizes of the components present in the questionnaire can also be specified. In this case, the FFQ is known as a semi quantitative FFQ and permits to capture the usual intake with more precisions. However, this method cannot predict alone the exact composition of nutrients, since it reflects more the habitual diet over a certain period of time [5] and should be matched with dietary records such as recalls, which are more representative and precise [4]. Therefore, in any epidemiological study, a dietary recall should accompany the FFQ to assess the nutritional intake. Since dietary habits are under the influence of the participants' cultural background [6,7] and because pregnancy is a critical window in a woman's life [1], it is of great interest to develop a valid tool, compatible with the diet of Mediterranean and Middleeastern population and able to estimate the intake of nutrients in this target population. Lebanon is a small country on the Eastern shore of the Mediterranean Sea, having dietary food ingredients highly representative of the traditional Mediterranean diet. Lebanese cuisine is composed mainly of cereals, legumes and vegetables, together with fish, meat or poultry and stuffed or blended most often with olive or sesame oil and herbs, to end up with dishes commonly known as “mezze” and “stews”. However, during the last decades, food habits of people living around the Mediterranean basin underwent some mutations, from plant-based traditional Mediterranean meals composed mainly of nonrefined cereals, nuts, seeds, vegetables, fruits, olive oil and vegetables-based protein meals to animal-based westernized food patterns, low in fiber, rich in animal proteins and saturated fats [8,9]. The adherence to the Mediterranean diet and the effect of recent culinary mutations on the metabolic syndrome was studied on some population groups, such as Lebanese adults, however the tool used to assess the nutritional status was a 78 item qualitative FFQ of only individual food ingredients, lacking composite dishes and hence not truly representing the local food items [10]. Hence, the aim of this research was the development of a 157 item semi-quantitative FFQ, containing Mediterranean and Middle Eastern food items, designed to capture usual food intake of Lebanese pregnant women, to assess their nutritional status with more precisions and to help researchers in the future to use it in other population subgroups living in the Mediterranean region. The developed tool was tested for reproducibility and validity. 2. Methodology 2.1. Study design This was a cross-sectional observational study, conducted as a part of a large longitudinal one, aiming to assess the nutritional status of women during pregnancy on neonatal outcomes and on the biochemical measures in both the mother and the offspring.
2.2. Study population The total number of participants was 128 Lebanese women (103 pregnant and 25 included at day 1 post-partum) aged between 18 and 40 years old, healthy, not suffering from gestational diabetes, preeclampsia, or any other chronic disease affecting their nutritional status. Women with multiple gestations were excluded from participation. Participants were recruited, independently of their weeks of gestation, in private clinics in different geographic districts in the capital Beirut, the regions of Mount Lebanon, the North and the South. The sample size was determined based on the recommendations of professionals in this field confirming that 100 individuals are required to assess the agreement between tools used to evaluate dietary intakes [4,5]. Field work was conducted between January and March 2015 by trained dietitians, by face to face interviews. The study protocol was approved by the Institutional Review Board of Saint-Joseph University at Beirut Lebanon, ^ tel-Dieu Hospital Ethic Committee (CE 624/FP 49) and the the Ho participating gynecologists. All subjects gave their written consent prior to the study. 2.3. Development of the study material A panel composed of researchers and nutrition instructors developed the questionnaire and held multiple meetings for reviewing and making the necessary adjustments on the final consensual version. At the end of the adaptation, thirty Lebanese women, aged between 18 and 40 years, tested the pre-final version of the study material. The meaning, comprehensibility and acceptability were studied by means of individual interviews. It was divided into 3 parts. The first one covered sociodemographic characteristics and personal information related to gestational complications, appetite modifications, gastrointestinal problems, smoking and food habits. The indicators used in the first part of the questionnaire dealing with the socioeconomic status were the crowding index (deducted by the interviewer as the total number of co-residents per household, excluding the newborn infant, divided by the total number of rooms, excluding the kitchen and bathrooms), the educational level, and the work status. A crowding index more than one, suggested a household with restricted economic resources. In this study, education was measured as a categorical variable and participants had to choose between six categories (primary, intermediate, high school, technical, a bachelor or a master degree). Both the educational level and the crowding index are recommended to be used in epidemiological studies to classify the economic status of the studied population [11]. Personal income was omitted during the statistical analyses, since a large proportion of women declined to confess it during the face to face interview. Details on gestational weight gain, weeks of gestation, and use of vitamin supplements were copied directly from the medical files, to avoid ending-up with irrelevant data. Body mass index (BMI) was calculated as the ratio of pre-gestational weight (kg) to square of height (m), relying on the anthropometric measures (height, weight) present in the medical file. It was then categorized according to the WHO cut-off points (underweight <18.5, normal 18.5e24.9, overweight 25e29.9 and obese >30) [12]. The third part of the questionnaire dealt with the development of a semi-quantitative 157 items FFQ containing Mediterranean foods and Middle Eastern ingredients. It was initially developed in a pilot project, which consisted of recalling the intake of 50 females aged between 18 and 40 years and by compiling all the food items mentioned in their recall. In a second step, the research team reviewed all the food items and added the missing ones, due to
Please cite this article in press as: Papazian T, et al., Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.04.015
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
YCLNU2805_proof ■ 29 April 2016 ■ 3/7
T. Papazian et al. / Clinical Nutrition xxx (2016) 1e7
seasonal variability. To avoid underreporting of items from each food group, open-frequency categories were used in decreasing order: daily, weekly, monthly, all over the gestation and never. In addition, to help the women quantify the exact amount and portion of foods consumed, standard measuring cups and spoons, plastic food models and local food photos in frequently consumed sizes (provided by www.numed.com) were used during the interview. It was further subdivided into 12 categories: Breads and cereals (12 items), Rice, pasta, potato and legumes (14 items), Milk and dairy products (9 items), Fruits and fruit juices (10 items, subdivided by fruits having same portion sizes or from same family), Vegetables (12 items, again subdivided), Meat, poultry, fish, eggs and ham (25 items), Nuts and condiments (16 items), Sugar based snacks, desserts and jams (23 items), Bakery products (11 items), Salty snacks (3 items), Oils and fats (9 items), and Beverages (13 items). These subdivisions ended up with a total of 157 food items mentioned in the FFQ. The weight in grams of each food was multiplied by its frequency of consumption, and divided for example by 7, if it was consumed just once a week. Participants' responses were then converted into average daily intake, in grams. Finally, the last part was a 24 h dietary recall used as our reference method of choice for validating the FFQ. It was filled by a trained dietitian, with the same techniques described before. The participant had to recall all food items and drinks with maximum details possible, consumed the day before, from morning till evening. All interviews during data collection were conducted from Tuesday till Saturday, to avoid recalling the intake of a participant during a weekend, since people usually modify their eating patterns at the end of the week. The FFQ was administered before the 24 h recall as described elsewhere by the same interviewer, the day of the enrollment [13]. We collected detailed information on supplement use (date of first use, time, frequency, dose and brand names), however the contribution of micronutrients from dietary supplements to the recall was not included in the analysis, because of inter-individual variation of absorption of those supplements and since our focus was mainly on nutrients provided from the daily diet. 2.4. Nutrient calculation The Nutrilog software (version 2.30) was used to analyze the food records of the dietary recall and the FFQ, using databases from US Department of Agriculture [14]. For local and traditional Lebanese dishes, nutrient values were obtained from the chemical analyses of standardized recipes derived from the databases of the American University of Beirut [15]. However, those analyses were done in 1970s, and many new local food items were lacking in that unique national reference. Hence, for each new Lebanese dish encountered in the recalls and missing in Nutrilog, a standardized recipe was cooked, ingredients split to a manual analysis and then added to the software as extra recipes. 3. Statistical analysis Means and standard deviations were computed for continuous data (age, number of meals and nutritional status) and frequencies for categorical variables (place of residence, education level, occupation, crowding index, BMI, presence of digestive disorders and complications, smoking, taking supplements …). KolmogoroveSmirnov tests were used to assess the normal distribution of energy and nutrients. Since not all micronutrients were normally distributed, nonparametric tests were performed for comparison between the 24 h dietary recall and the FFQ. Reproducibility study: Means and standard deviations for nutrient intakes were calculated for both food frequency
3
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 4. Results 91 92 4.1. General characteristics of the population 93 94 The baseline characteristics of the participants are shown in 95 Table 1. The mean age of the participants was 29 years (±5.13 SD), 96 with a mean pre-pregnancy weight of 60.67 kg and a height of 97 163.12 cm. The majority of our participants' (75.8%) had a body 98 mass index (BMI) between 18.5 and 24.9 kg/m2. 72.7% had achieved 99 higher education and only 9% had a crowding index >1. This reflects 100 the relatively high educational and economical level of our sample. 101 The distribution was as follows: 31 women in their first trimester of 102 pregnancy, 39 in the second, 33 in the last and 25 who were 103 recruited at delivery. 104 105 4.2. Reproducibility of the food-frequency questionnaire 106 107 Thirty eight women participated in the reproducibility study. It 108 consisted in administrating the questionnaires twice under the 109 same conditions. The mean daily intake of energy, macro and 110 micronutrients derived from the FFQ1 and the FFQ2, measured 3 111 weeks apart, are presented in the table below. Highly significant 112 intra-class correlation coefficients were generated by all the stud113 ied nutrients, ranging from 0.935 for calcium to 0.984 for vitamin D, with a mean ICC of 0.962 for all nutrients (p < 0.001) (see Table 2). Q3 114 115 116 4.3. Validity of the food-frequency questionnaire 117 118 The estimated intake of energy, macronutrients and most 119 essential micronutrients in pregnancy were used to assess the ac120 curacy of the FFQ and to measure its agreement with the chosen 121 gold standard, the 24 h dietary recall. Details showing median daily 122 intake of nutrients and the correlations between the two methods 123 are presented in Table 3. The FFQ produced significantly higher 124 intakes reaching respectively a relative difference of 18.9%, 29.8%, 125 32.9% and 30.4% for caloric intake, iron, cholesterol, and folic acid as 126 compared to the 24 h dietary recall (average relative difference 127 16.98%). The contribution of macronutrients to the total energy 128 intake was almost similar in both tools. Spearman correlation co129 efficients ranged between 0.294 for iron and 0.762 for caloric 130 intake, with a highly significant p value.
questionnaires (FFQ1 and FFQ2). Intra class correlation coefficients (ICC) were used between the first and the second questionnaires, both administrated under the same conditions by the same interviewer, three weeks apart. This coefficient is known to measure the fraction of total variation due to between-individual variability, with a high value indicating a low within-person variation. Validation study: Crude nutrient intakes estimated from average FFQ (1 and 2) were compared with the average 24 h dietary recall (1 and 2) using Wilcoxon signed rank test (paired data). Spearman correlation coefficients were calculated to measure the strength of the linear association between the FFQ estimates and the 24 h dietary recall estimates. BlandeAltman analysis was carried out to assess the extent of agreement between absolute estimates from the FFQ and the reference tool (the 24 h dietary recall). The differences between the two methods were plotted against the mean of the two dietary measures, with a 95% calculated as limits of agreement (LOA). The plot of the difference shows the relationship between the measurement errors, the true value and to what extent the two tools agree. This was performed for all nutrients [16]. All statistical analyses were performed using the SPSS statistical software package version 20. Results were considered significant at p < 0.05.
Please cite this article in press as: Papazian T, et al., Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.04.015
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
YCLNU2805_proof ■ 29 April 2016 ■ 4/7
4
T. Papazian et al. / Clinical Nutrition xxx (2016) 1e7
Table 1 Sociodemographic characteristics of Lebanese pregnant women at baseline (N ¼ 128). Characteristic
Mean
SD
Age (years) Height (cm) Pre-gravid weight (kg) Pre-gravid BMI (kg/m2)
29.5 163.1 60.7 22.8
5.1 5.7 9.1 3.4
N
%
40.0 56.0 15.0 5.0 11.0 1.0
31.3 43.8 11.7 3.9 8.6 0.8
3.0 10.0 13.0 9.0 75.0 18.0
2.3 7.8 10.2 7.0 58.6 14.1
55.0 56.0 9.0 2.0 3.0 2.0 1.0
43.0 43.8 7.0 1.6 2.3 1.6 0.8
119.0 9.0
93.0 7.0
31.0 39.0 33.0 25.0
24.2 30.5 25.8 19.5
7.0 97.0 19.0 5.0
5.5 75.8 14.8 3.9
Place of Residence Beirut Mount Lebanon Southern Lebanon Northern Lebanon Nabatiyeh Bekaa Educational level Elementary Intermediate Secondary Technical BA/Bsc Master Profession Housewife Employed Entrepreneur Unemployed Liberal Trainee Student Crowding index 1 >1 Trimester of gestation First trimester Second trimester Third trimester At delivery Pre-gravid BMI (kg/m2) <18.5 (underweight) 18.5e24.9 (normal) 25e29.9 (overweight) 30 (obese)
Good Spearman correlations (r 0.5) were found for energy, macronutrients, calcium, vitamin D and sodium. Fair correlations (0.3 < r < 0.49) were found for sugars, fibers, cholesterol, and folic acid. A weak correlation was found for iron (r < 0.3). After
adjustments for energy, good correlations were observed for fats, calcium and sodium and fair correlations were observed for all other nutrients. The average correlations increased after energy adjustment for intakes of nutrients (0.44). An average of 77.26% of pregnant women, ranging from 62% to 88% were classified into the correct quartile, while less than 10% of them were misclassified. 4.4. BlandeAltman analyses Results of the BlandeAltman analyses for proteins and folic acid intake are shown respectively in Figs. 1 and 2. They were selected because of their crucial role during fetal development. Plots of other nutrients resembled those presented below (plots not shown). As mentioned earlier, the difference in intake between the FFQ and the dietary recall is plotted on the Y axis, and the mean intake from the two tools is presented on the X axis. Most data points were clustered around the line zero or the mean difference line across the range of intakes. Few data points fell above the upper LOA. 5. Discussion Valid questionnaires to estimate nutrient intake in the diet of Mediterranean and Middle Eastern population and particularly during pregnancy are lacking. This is to our knowledge the first validation study of a culturally adequate FFQ conducted among a group of Lebanese pregnant women. As in other validation studies, its single administration was enough to estimate the nutrient intake of this particular population [17]. The research team relied on the development and administration of this FFQ, because it is an appropriate tool, reflecting intakes over a prolonged period of time (such as the gestation), less expensive and easy to be administered and interpreted [17,18]. In addition, to assess its validity, literature reviews show that comparing it to a 24 h dietary recall is an optimal comparison method, commonly employed in epidemiological studies [4]. However, because this type of recall relies greatly on memory and is affected by day to day and seasonal variations, some authors suggest the collection of at least two 24 h recalls [13], spread along a specific time frame [19]. In the context of our project, we had two dietary recalls only from one third of our participants', because of
Table 2 Reproducibility study: Mean daily intake of energy and nutrients and the intra-class correlation coefficient for the comparison between FFQ1 and FFQ2 in Lebanese pregnant women (n ¼ 38).
Energy (kcal) Carbs (g) Carbs (%) Proteins (g) Proteins (%) Fats (g) Fats (%) Sugars (g) Fibers (g) Cholesterol (g) Iron (mg) Calcium (mg) Vitamin D (mg) Folic acid (mg) Sodium (mg)
FFQ 1
FFQ 2
Mean ± SD
Mean ± SD
2475 312 48.9 94 15 99 35.6 93 22 237 16.3 889 9.98 340 5335
± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
537 82 6.1 24 2.3 23 5.7 24 6.8 95 4.8 291 2.18 114 855
2533 307 48.7 95 15 102 36.4 93 23 232 16.4 884 10.03 341 5367
± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
494 84 5.3 21 2.1 23 5.1 21 6.3 91 4.5 261 2.23 106 740
ICC [95% CI]
p value
0.957 0.970 0.969 0.962 0.961 0.953 0.974 0.941 0.976 0.936 0.977 0.935 0.984 0.980 0.958
<0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001
% of FFQ 1 103 98 99 101 100 103 102 100 105 98 101 99 101 101 101
[0.917e0.978] [0.943e0.985] [0.939e0.984] [0.927e0.980] [0.924e0.980] [0.910e0.976] [0.950e0.987] [0.886e0.969] [0.953e0.987] [0.877e0.967] [0.956e0.988] [0.875e0.966] [0.969e0.992] [0.962e0.990] [0.919e0.978]
FFQ1: food frequency questionnaire during 1st administration; FFQ2: food frequency questionnaire during 2nd administration; ICC: Intra-class correlation coefficient; CI: Confidence interval (p < 0.05).
Please cite this article in press as: Papazian T, et al., Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.04.015
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
YCLNU2805_proof ■ 29 April 2016 ■ 5/7
T. Papazian et al. / Clinical Nutrition xxx (2016) 1e7
5
Table 3 Validation study: mean daily intake of nutrients, relative difference, Spearman correlation coefficient, and cross-classification for the comparison between FFQ and 24 h recall in Lebanese pregnant women (n ¼ 128).
Energy (kcal) Carbs (g) Carbs (%) Proteins (g) Proteins (%) Fats (g) Fats (%) Sugars (g) Fibers (g) Cholesterol (g) Iron (mg) Calcium (mg) Vitamin D (mg) Folic acid (mg) Sodium (mg)
p valuea
24 h recall
FFQ
Mean ± SD
Mean ± SD
2003 255.5 50.2 75.5 14.8 85.5 35 67 17.9 149 14.1 799 8.9 270 4818
± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
431 59 4.4 18 2.2 67 4 28 7.3 103 5.6 384 2.5 133 878
2381 299.5 51.1 87.5 14.6 95 35.7 87 23.1 198 18.3 881 10.6 352 5184
± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
565 78 15.3 23 2.1 25 5 27 11.7 84 30 311 10.9 246 930
<0.001 <0.001 0.489 <0.001 0.378 <0.001 0.102 <0.001 <0.001 <0.001 <0.001 0.004 0.001 0.001 <0.001
Relative difference (%)b
Spearman correlation coefficient Unadjusted
Energy adjusted
18.9 17.2 1.8 15.9 1.3 11.1 2 29.9 29.1 32.9 29.8 10.3 19.1 30.4 7.6
0.762** 0.757** 0.563** 0.630** 0.544** 0.642** 0.517** 0.389* 0.375* 0.361* 0.294* 0.602** 0.603** 0.445** 0.656**
0.762** 0.414* 0.563** 0.402** 0.538** 0.502** 0.425** 0.318* 0.475** 0.312* 0.349* 0.518** 0.469** 0.404** 0.562**
Percent agreement (same & adjacent quartile)
Extreme disagreement
73% 88% 84% 75% 77% 79% 87% 73% 72% 78% 74% 73% 87% 62% 77%
0% 2% 1% 0% 2% 1% 1% 5% 3% 5% 6% 0% 1% 4% 5%
*p value < 0.05; **p value < 0.01. a Wilcoxon signed rank test to test for differences between FFQ and 24 h dietary recall. b Relative difference ¼ [(FFQ e 24 h recall)/24 h recall]*100.
Fig. 1. BlandeAltman plot showing agreement between the average FFQ and the average of 24 h recall for protein intake. The solid line represents the mean difference in absolute intake between the two dietary assessment methods, while the dashed lines represent the limits of agreement (±2SD).
difficulties related to the respondents' cooperation. However, a special attention was taken, to fill both tools during their enrollment, unlike the methodology in other validation studies where the completion of both instruments differed in the time frame [20]. Sample size and participants' age range were similar to those in other FFQ validation studies [19,21]. However, no direct comparisons can be conducted with other FFQ validation studies, since our FFQ is the first tool assessing Middle Eastern foods' intake by women during pregnancy. In addition, previous studies testing the validity of the FFQ for pregnant women were restricted to food intake at a specific trimester of gestation: the second [21] or the third trimester of pregnancy [22,23]. Our study aimed to include women at different gestational time frame (31%, 39% and 33% respectively for the first, second and the third trimesters), with no restrictions to a particular trimester. However, during the interpretation of our questionnaire, one quarter of our participants admitted variations in appetite, nausea and vomiting during the first trimester, hence affecting
Fig. 2. BlandeAltman plot showing agreement between the average FFQ and the average of 24 h recall for folate intake. The solid line represents the mean difference in absolute intake between the two dietary assessment methods, while the dashed lines represent the limits of agreement (±2SD).
their food intake. Some studies advocate to incorporate participants', starting the fourteenth week of gestation, to avoid the negative impact of the hormonal changes on the dietary intake [24], however, we wouldn't recommend the exclusion of first trimester pregnant women in future studies, because this trimester is of crucial importance for the nutritional status, not only of the mother but of the baby as well. The selection of nutrients to be evaluated in the diets of pregnant women is similar to those analyzed in other studies [19,21]. In our study, we evaluated the contribution of macronutrients to the total energy intake, which is not mentioned in other publications. To determine the reproducibility of our FFQ, we relied on the intra-rater reliability of our tool which consists of administrating the same questionnaire to the same pregnant woman, by the same interviewer [4]. Although neither short nor long intervals are advised between the two collect, a special attention was given to have this second appointment, during a period not exceeding 21 days, to avoid hormonal fluctuations affecting appetite or the transition to a new trimester of gestation, hence influencing the nutritional needs. The mean intra class correlation coefficients (ICC)
Please cite this article in press as: Papazian T, et al., Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.04.015
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
YCLNU2805_proof ■ 29 April 2016 ■ 6/7
6
T. Papazian et al. / Clinical Nutrition xxx (2016) 1e7
obtained after both administrations implies the strong reproducibility of our tool, even though it was conducted to a restricted number of participants. The values of ICC were superior to those observed by Erkkola et al. (2001) for a similar study in Finnish pregnant women [22] and to those of Li et al. (2014) in China [25], but join the values published by Loy et al. (2011) in Malaysian pregnant women [23]. As in other validation studies, our FFQ overestimated food consumption and intakes of nutrients as compared with the 24 h recall [20,21]. Pinto et al. proposed that this overestimation is partly due to difficulties in accurately estimating the portion size consumed [17]. On the other hand, Iqbal et al. suggested that a FFQ containing a large number of food items may be the cause of this overestimation [26]. Moreover, according to Beaton, an overestimating FFQ is less of a concern than an underestimating one, since the latter can precipitate more wrong assessment and misclassification of individuals [27]. Spearman correlation coefficients obtained while comparing the two tools were 0.294 for iron, 0.762 for energy intake, and 0.361 for cholesterol. Similar results were achieved in other validation studies, assessing nutrient intake during pregnancy [5,18,23]. However, statistical analyses were highly significant for almost all studied nutrients compared to those published by Shatenstein et al. in 2011 [20], Mouratidou et al. in 2006 [13] and Loy et al., in 2011 [23]. Willet suggested that acceptable correlations for FFQ validations are between 0.5 and 0.7 [5]. However, these coefficients differed according to Hankin et al. who suggested considering <0.3 as a poor correlation, between 0.3 and 0.49 fair and >0.5 a good level of correlation [28]. Cross-classification permitted to classify the participants in the same or different categories by both tested tools [4]: more than three quarter of women fell into the same quartiles by both methods, which proves that the degree of misclassification was small (<10%), similar to other studies [17,23]. In the present study, the BlandeAltman scatter plots were used to provide a graphical interpretation of agreement between the FFQ and the 24 h dietary recall and to obtain further information that the correlation coefficient itself cannot provide. The plots indicated that, for some nutrients the difference between the FFQ and the 24 h dietary recall increased with increasing intake pointing to overestimates by the FFQ. This shows that participants who consumed higher amounts of food and energy reported more errors during diet assessments. These findings are consistent with the study conducted among pregnant women by Loy et al. [23] but differ than the findings noted by Mouratidou et al. [13] and Shatenstein et al. [20]. In addition, the LOA e defined as the bias (2SD) of the difference e were wide for some nutrients while others were narrow. The reason for this is unclear. New methods of dietary assessments are proposed by Boeing, such as the use of smart phones to translate food labels or pictures consumed by the patient to a server, for a direct nutrient analysis [29]. Another proposition is to rely on biochemical markers, which are more specific but often expensive, invasive, nutrient-selective and influenced by physiological changes [4]. In addition, those indicators are more related to recent nutrient status, unlike the FFQ which predicts more long-term exposure [30]. Therefore, valid FFQs and DRs remain valid tools that predict food intake and nutritional status in all population subgroups. The strengths of this study were the development of a FFQ containing foods culturally adapted to the Mediterranean area and the Arab world, the use of visual aids to achieve higher precision on portion size, the inclusion of the same participants in the reproducibility and the validity phases and the face to face interview administered by a trained dietitian. No one retrieved during the reproducibility phase and no missing data was reported which represent a high sense of motivation and commitment from both
parties. Finally, the use of different statistical analysis permitted to better highlight the reproducibility and the validity of our tool. However, some limitations were encountered in our study: bias related to memory, misreporting of portion sizes and cooking techniques, seasonal variations in some fruits and vegetables and the choice of a single recall as the ideal reference method. Multiple recalls spread along gestation insure a better estimate of nutritional status, however the research team had a limited time frame and could not reach a larger percentage of the respondents'. In addition, the computerized program englobing nutrient databases was not fully compatible with our food items, since policies related with folate enrichment of flour are not implemented by the Lebanese authorities. Therefore, the folate content of the diet of Lebanese pregnant women may be overestimated, since it's mostly derived from USDA food composition tables, where the enrichment of the flour by folic acid is a must. 6. Conclusion Monitoring nutritional status during pregnancy is crucial to implement prenatal nutritional strategies. However, dietary assessment techniques vary depending on the tools and the ways of data collection and analysis. The aim of this research was to develop a semi-quantitative FFQ, culturally convenient to Mediterranean and Middle Eastern countries and to study its reproducibility and validity among a sample of pregnant women. Thus, it has been proven that our newly developed tool is a valid and a reliable method, that provides a real assessment of nutritional intake during pregnancy and its future application on a larger number of participants will permit to assess the effect of dietary intake of pregnant Lebanese women on the neonatal outcomes in our prospective study. However, later work should aim in omitting some food items present in the FFQ to shorten its administration time and validating it in other population groups such as children, adolescents, adults and athletes. Authorship All authors reviewed and approved the submitted manuscript. T.P. conceived the study, supervised the recruitment of the participants, did the face to face interviews, conducted the nutrient analysis and prepared the manuscript. H.H contributed to the design of the study, did the face to face interviews, conducted the data analysis, the statistical interpretation and the drafting of the manuscript. D.S. contributed to the design of the study, did the face to face interviews and participated in the data analysis. N.H. helped in the design of the study material. N.O. carried out the statistical analysis and helped with the data interpretation. L.K. provided complete supervision, critical revision, data interpretation and correction of the manuscript. H.Y. participated in the study conception and design. All authors constituted the research team. Contributors All authors have approved the final article. Financial support This research received no specific grant from any funding agency, commercial or not-for-profit sectors. Conflict of interest The authors declare that they have no conflict of interest.
Please cite this article in press as: Papazian T, et al., Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.04.015
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
YCLNU2805_proof ■ 29 April 2016 ■ 7/7
T. Papazian et al. / Clinical Nutrition xxx (2016) 1e7
Acknowledgments The authors gratefully acknowledge the participants and the gynecologists.
References [1] Procter SB, Campbell CG. Position of the Academy of Nutrition and Dietetics: nutrition and lifestyle for a healthy pregnancy outcome. J Acad Nutr Diet 2014 Jul;114(7):1099e103. [2] Olson CM. Achieving a healthy weight gain during pregnancy. Annu Rev Nutr 2008;28:411e23. [3] Scientific Advisory Committee on Nutrition. The influence of maternal, fetal and child nutrition on the development of chronic disease in later life. 2011. Available at: http://www.sacn.gov.uk/pdfs/sacn_early_nutrition_final_report_ 20_6_11.pdf [accessed on June 2015]. [4] Cade J, Thompson R, Burley V, Warm D. Development, validation and utilisation of food-frequency questionnaires e a review. Public Health Nutr 2002 Aug;5(4):567e87. [5] Willett W. Nutritional epidemiology. 2nd ed. New York: Oxford University Press; 1998. [6] Barbieri P, Crivellenti LC, Nishimura RY, Sartorelli DS. Validation of a food frequency questionnaire to assess food group intake by pregnant women. J Hum Nutr Diet Off J Br Diet Assoc 2015 Jan;28(Suppl. 1):38e44. [7] Tayyem RF, Abu-Mweis SS, Bawadi HA, Agraib L, Bani-Hani K. Validation of a food frequency questionnaire to assess macronutrient and micronutrient intake among Jordanians. J Acad Nutr Diet 2014 Jul;114(7):1046e52. [8] Naja F, Nasreddine L, Itani L, Chamieh MC, Adra N, Sibai AM, et al. Dietary patterns and their association with obesity and sociodemographic factors in a national sample of Lebanese adults. Public Health Nutr 2011 Sep;14(9): 1570e8. [9] Bonaccio M, Iacoviello L, de Gaetano G, Moli-Sani Investigators. The Mediterranean diet: the reasons for a success. Thromb Res 2012 Mar;129(3): 401e4. [10] Issa C, Darmon N, Salameh P, Maillot M, Batal M, Lairon D. A Mediterranean diet pattern with low consumption of liquid sweets and refined cereals is negatively associated with adiposity in adults from rural Lebanon. Int J Obes 2005 2011 Feb;35(2):251e8. [11] Galobardes B, Shaw M, Lawlor DA, Lynch JW, Davey Smith G. Indicators of socioeconomic position (part 1). J Epidemiol Community Health 2006 Jan;60(1):7e12. [12] Report of a WHO Expert Committee Physical status: the use and interpretation of anthropometryWorld health organ tech rep ser, vol. 854; 1995. p. 1e452. [13] Mouratidou T, Ford F, Fraser RB. Validation of a food-frequency questionnaire for use in pregnancy. Public Health Nutr 2006 Jun;9(4):515e22.
7
[14] US Department of Agriculture: agricultural research Service, USDA National Nutrient Database for standard reference, release 27. Available from: http:// ndb.nal.usda.gov/ndb/search [accessed April 2015]. [15] Pellet P, Shadarevian S. Food composition table for use in the Middle East. 3rd ed. Lebanon: American University of Beirut; 1975. [16] Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet Lond Engl 1986 Feb 8;1(8476): 307e10. [17] Pinto E, Severo M, Correia S, dos Santos Silva I, Lopes C, Barros H. Validity and reproducibility of a semi-quantitative food frequency questionnaire for use among Portuguese pregnant women. Matern Child Nutr 2010 Apr;6(2): 105e19. [18] Baer HJ, Blum RE, Rockett HRH, Leppert J, Gardner JD, Suitor CW, et al. Use of a food frequency questionnaire in American Indian and Caucasian pregnant women: a validation study. BMC Public Health 2005;5:135. [19] Potosky AL, Block G, Hartman AM. The apparent validity of diet questionnaires is influenced by number of diet-record days used for comparison. J Am Diet Assoc 1990 Jun;90(6):810e3. [20] Shatenstein B, Xu H, Luo Z-C, Fraser W. Relative validity of a food frequency questionnaire for pregnant women. Can J Diet Pract Res 2011;72(2):60e9. Q4 [21] Mouratidou T, Ford F, Prountzou F, Fraser R. Dietary assessment of a population of pregnant women in Sheffield, UK. Br J Nutr 2006 Nov;96(5):929e35. €s a €nen L, Knip M, Virtanen SM. Val[22] Erkkola M, Karppinen M, Javanainen J, Ra idity and reproducibility of a food frequency questionnaire for pregnant Finnish women. Am J Epidemiol 2001 Sep 1;154(5):466e76. [23] Loy SL, Marhazlina M, Nor AY, Hamid JJM. Development, validity and reproducibility of a food frequency questionnaire in pregnancy for the Universiti Sains Malaysia birth cohort study. Malays J Nutr 2011 Apr;17(1):1e18. [24] Cheng Y, Yan H, Dibley MJ, Shen Y, Li Q, Zeng L. Validity and reproducibility of a semi-quantitative food frequency questionnaire for use among pregnant women in rural China. Asia Pac J Clin Nutr 2008;17(1):166e77. [25] Li M, Halldorsson TI, Bjerregaard AA, Che Y, Mao Y, Hu W, et al. Relative validity and reproducibility of a food frequency questionnaire used in pregnant women from a rural area of China. Acta Obstet Gynecol Scand 2014 Nov;93(11):1141e9. [26] Iqbal R, Ajayan K, Bharathi AV, Zhang X, Islam S, Soman CR, et al. Refinement and validation of an FFQ developed to estimate macro- and micronutrient intakes in a south Indian population. Public Health Nutr 2009 Jan;12(1):12e8. [27] Beaton G. Interpretation of results from diet history studies. L. Kohlmeier. London: Smith-Gordon; 1991. [28] Hankin JH, Wilkens LR, Kolonel LN, Yoshizawa CN. Validation of a quantitative diet history method in Hawaii. Am J Epidemiol 1991 Mar 15;133(6):616e28. [29] Boeing H. Nutritional epidemiology: new perspectives for understanding the diet-disease relationship? Eur J Clin Nutr 2013 May;67(5):424e9. ~ oz E-M, Gimenez-Monzo D, García-de-la-Hera M, [30] Vioque J, Navarrete-Mun Granado F, Young IS, et al. Reproducibility and validity of a food frequency questionnaire among pregnant women in a Mediterranean area. Nutr J 2013;12:26.
Please cite this article in press as: Papazian T, et al., Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.04.015
41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80
YCLNU2805_proof ■ 29 April 2016 ■ 1/7
Clinical Nutrition xxx (2016) 1e7
Contents lists available at ScienceDirect
Clinical Nutrition journal homepage: http://www.elsevier.com/locate/clnu
Original article
Q6 Q5
Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern Tatiana Papazian a, b, Hala Hout b, Darine Sibai b, Nour Helou b, Hassan Younes b, c, Nada El Osta d, e, Lydia Rabbaa Khabbaz a, b, * a ^le de qualit Laboratoire de Pharmacologie, Pharmacie clinique et Contro e des M edicaments, Faculty of Pharmacy, Saint-Joseph University of Beirut, Beirut, Lebanon b Department of Nutrition, Faculty of Pharmacy, Saint-Joseph University of Beirut, Beirut, Lebanon c Department of Nutrition and Health Sciences, Institut Polytechnique LaSalle Beauvais, France d Department of Public Health, Faculty of Medicine, Saint-Joseph University, Beirut, Lebanon e Department of Prosthodontics, Faculty of Dental Medicine, Saint-Joseph University, Beirut, Lebanon
a r t i c l e i n f o
s u m m a r y
Article history: Received 12 October 2015 Accepted 8 April 2016
Background & aims: Accurate dietary assessment tools are required to ensure that maternal diet supplies all the nutrients needed for fetal development. However, no dietary method could accurately estimate food intake during gestation. Food Frequency Questionnaires (FFQ), frequently used in epidemiological studies, estimate long term nutritional status of the target population. However, it is recommended to create and validate a FFQ compatible with the dietary habits of the studied population, to avoid cultural and social discrepancies. This study aimed to develop and test the reproducibility and the validity of a semi-quantitative FFQ compatible with the diet of Mediterranean and Middle-Eastern population, in a sample of Lebanese pregnant women. Methods: 128 women participated in the validation study, while 38 took part in the reproducibility phase, which was repeated in a time frame of 21 days. The FFQ was validated against a 24 h dietary recall (DR). Results: The intra class correlation coefficient (ICC) ranged from 0.935 for calcium to 0.984 for vitamin D (p value < 0.001), indicating an excellent reproducibility. A satisfactory agreement between the two dietary tools was demonstrated using BlandeAltman plot and Spearman's and Pearson's correlations coefficients which varied between 0.294 for iron to 0.762 for caloric intake (p value < 0.01). Conclusions: The newly developed FFQ englobing Mediterranean food items was culture specific and assessed the nutrient intake of our population. Administering this tool in future researches will help monitor the nutritional status of pregnant women, aiming at improving maternal and newborn health. © 2016 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism.
Keywords: Maternal nutritional status Food frequency questionnaire 24 h dietary recall Validity Reproducibility
Q2
1. Introduction Adequate food intake during pregnancy is a key factor affecting the health of both the mother and the infant. Along the
Q1
* Corresponding author. Laboratoire de Pharmacologie, Pharmacie clinique et ^le de qualite des Me dicaments, Faculte de pharmacie, Campus des sciences Contro dicales (CSM), Rue de Damas, Universite Saint-Joseph, B.P. 11-5076, Riad el Solh, me Beirut 1107 2180, Lebanon. Tel.: þ961 3 224 449, þ961 1 4210 000x6792; fax: þ961 1 421 022. E-mail address: [email protected] (L.R. Khabbaz).
physiological changes of her body, a pregnant woman faces the challenges of preventing and correcting any existing nutrient deficiency during both the periconceptional phase and pregnancy. This adequacy can only be achieved by a well-balanced diet providing all the necessary nutrients and energy needed for her and the fetus [1]. Strong scientific evidences confirm the impact of maternal diet on neonatal outcomes [2] and the occurrence of growth problems and chronic diseases during childhood [3]. Despite the large number of studies investigating the role of nutrition in pregnancy and neonatal outcome, few valid
http://dx.doi.org/10.1016/j.clnu.2016.04.015 0261-5614/© 2016 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism.
Please cite this article in press as: Papazian T, et al., Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.04.015
55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
YCLNU2805_proof ■ 29 April 2016 ■ 2/7
2
T. Papazian et al. / Clinical Nutrition xxx (2016) 1e7
questionnaires to assess nutritional status during gestation exist. Hence, to monitor the dietary profile of a selected population, researchers often rely on different tools, each one having its strengths and limitations [4]. Food frequency questionnaires (FFQs) have been designed and used in many epidemiological studies, to assess the nutrient intake of different age groups, healthy or suffering from a disease state or pregnant women [4]. This questionnaire, ideally conducted by a trained dietitian, permits to evaluate the dietary intake of the respondents by asking them to report the frequency of their usual consumption of common food items, from different food groups, over a specified period of time (daily, weekly, or monthly). In addition, depending on the purpose of the study, portion sizes of the components present in the questionnaire can also be specified. In this case, the FFQ is known as a semi quantitative FFQ and permits to capture the usual intake with more precisions. However, this method cannot predict alone the exact composition of nutrients, since it reflects more the habitual diet over a certain period of time [5] and should be matched with dietary records such as recalls, which are more representative and precise [4]. Therefore, in any epidemiological study, a dietary recall should accompany the FFQ to assess the nutritional intake. Since dietary habits are under the influence of the participants' cultural background [6,7] and because pregnancy is a critical window in a woman's life [1], it is of great interest to develop a valid tool, compatible with the diet of Mediterranean and Middleeastern population and able to estimate the intake of nutrients in this target population. Lebanon is a small country on the Eastern shore of the Mediterranean Sea, having dietary food ingredients highly representative of the traditional Mediterranean diet. Lebanese cuisine is composed mainly of cereals, legumes and vegetables, together with fish, meat or poultry and stuffed or blended most often with olive or sesame oil and herbs, to end up with dishes commonly known as “mezze” and “stews”. However, during the last decades, food habits of people living around the Mediterranean basin underwent some mutations, from plant-based traditional Mediterranean meals composed mainly of nonrefined cereals, nuts, seeds, vegetables, fruits, olive oil and vegetables-based protein meals to animal-based westernized food patterns, low in fiber, rich in animal proteins and saturated fats [8,9]. The adherence to the Mediterranean diet and the effect of recent culinary mutations on the metabolic syndrome was studied on some population groups, such as Lebanese adults, however the tool used to assess the nutritional status was a 78 item qualitative FFQ of only individual food ingredients, lacking composite dishes and hence not truly representing the local food items [10]. Hence, the aim of this research was the development of a 157 item semi-quantitative FFQ, containing Mediterranean and Middle Eastern food items, designed to capture usual food intake of Lebanese pregnant women, to assess their nutritional status with more precisions and to help researchers in the future to use it in other population subgroups living in the Mediterranean region. The developed tool was tested for reproducibility and validity. 2. Methodology 2.1. Study design This was a cross-sectional observational study, conducted as a part of a large longitudinal one, aiming to assess the nutritional status of women during pregnancy on neonatal outcomes and on the biochemical measures in both the mother and the offspring.
2.2. Study population The total number of participants was 128 Lebanese women (103 pregnant and 25 included at day 1 post-partum) aged between 18 and 40 years old, healthy, not suffering from gestational diabetes, preeclampsia, or any other chronic disease affecting their nutritional status. Women with multiple gestations were excluded from participation. Participants were recruited, independently of their weeks of gestation, in private clinics in different geographic districts in the capital Beirut, the regions of Mount Lebanon, the North and the South. The sample size was determined based on the recommendations of professionals in this field confirming that 100 individuals are required to assess the agreement between tools used to evaluate dietary intakes [4,5]. Field work was conducted between January and March 2015 by trained dietitians, by face to face interviews. The study protocol was approved by the Institutional Review Board of Saint-Joseph University at Beirut Lebanon, ^ tel-Dieu Hospital Ethic Committee (CE 624/FP 49) and the the Ho participating gynecologists. All subjects gave their written consent prior to the study. 2.3. Development of the study material A panel composed of researchers and nutrition instructors developed the questionnaire and held multiple meetings for reviewing and making the necessary adjustments on the final consensual version. At the end of the adaptation, thirty Lebanese women, aged between 18 and 40 years, tested the pre-final version of the study material. The meaning, comprehensibility and acceptability were studied by means of individual interviews. It was divided into 3 parts. The first one covered sociodemographic characteristics and personal information related to gestational complications, appetite modifications, gastrointestinal problems, smoking and food habits. The indicators used in the first part of the questionnaire dealing with the socioeconomic status were the crowding index (deducted by the interviewer as the total number of co-residents per household, excluding the newborn infant, divided by the total number of rooms, excluding the kitchen and bathrooms), the educational level, and the work status. A crowding index more than one, suggested a household with restricted economic resources. In this study, education was measured as a categorical variable and participants had to choose between six categories (primary, intermediate, high school, technical, a bachelor or a master degree). Both the educational level and the crowding index are recommended to be used in epidemiological studies to classify the economic status of the studied population [11]. Personal income was omitted during the statistical analyses, since a large proportion of women declined to confess it during the face to face interview. Details on gestational weight gain, weeks of gestation, and use of vitamin supplements were copied directly from the medical files, to avoid ending-up with irrelevant data. Body mass index (BMI) was calculated as the ratio of pre-gestational weight (kg) to square of height (m), relying on the anthropometric measures (height, weight) present in the medical file. It was then categorized according to the WHO cut-off points (underweight <18.5, normal 18.5e24.9, overweight 25e29.9 and obese >30) [12]. The third part of the questionnaire dealt with the development of a semi-quantitative 157 items FFQ containing Mediterranean foods and Middle Eastern ingredients. It was initially developed in a pilot project, which consisted of recalling the intake of 50 females aged between 18 and 40 years and by compiling all the food items mentioned in their recall. In a second step, the research team reviewed all the food items and added the missing ones, due to
Please cite this article in press as: Papazian T, et al., Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.04.015
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
YCLNU2805_proof ■ 29 April 2016 ■ 3/7
T. Papazian et al. / Clinical Nutrition xxx (2016) 1e7
seasonal variability. To avoid underreporting of items from each food group, open-frequency categories were used in decreasing order: daily, weekly, monthly, all over the gestation and never. In addition, to help the women quantify the exact amount and portion of foods consumed, standard measuring cups and spoons, plastic food models and local food photos in frequently consumed sizes (provided by www.numed.com) were used during the interview. It was further subdivided into 12 categories: Breads and cereals (12 items), Rice, pasta, potato and legumes (14 items), Milk and dairy products (9 items), Fruits and fruit juices (10 items, subdivided by fruits having same portion sizes or from same family), Vegetables (12 items, again subdivided), Meat, poultry, fish, eggs and ham (25 items), Nuts and condiments (16 items), Sugar based snacks, desserts and jams (23 items), Bakery products (11 items), Salty snacks (3 items), Oils and fats (9 items), and Beverages (13 items). These subdivisions ended up with a total of 157 food items mentioned in the FFQ. The weight in grams of each food was multiplied by its frequency of consumption, and divided for example by 7, if it was consumed just once a week. Participants' responses were then converted into average daily intake, in grams. Finally, the last part was a 24 h dietary recall used as our reference method of choice for validating the FFQ. It was filled by a trained dietitian, with the same techniques described before. The participant had to recall all food items and drinks with maximum details possible, consumed the day before, from morning till evening. All interviews during data collection were conducted from Tuesday till Saturday, to avoid recalling the intake of a participant during a weekend, since people usually modify their eating patterns at the end of the week. The FFQ was administered before the 24 h recall as described elsewhere by the same interviewer, the day of the enrollment [13]. We collected detailed information on supplement use (date of first use, time, frequency, dose and brand names), however the contribution of micronutrients from dietary supplements to the recall was not included in the analysis, because of inter-individual variation of absorption of those supplements and since our focus was mainly on nutrients provided from the daily diet. 2.4. Nutrient calculation The Nutrilog software (version 2.30) was used to analyze the food records of the dietary recall and the FFQ, using databases from US Department of Agriculture [14]. For local and traditional Lebanese dishes, nutrient values were obtained from the chemical analyses of standardized recipes derived from the databases of the American University of Beirut [15]. However, those analyses were done in 1970s, and many new local food items were lacking in that unique national reference. Hence, for each new Lebanese dish encountered in the recalls and missing in Nutrilog, a standardized recipe was cooked, ingredients split to a manual analysis and then added to the software as extra recipes. 3. Statistical analysis Means and standard deviations were computed for continuous data (age, number of meals and nutritional status) and frequencies for categorical variables (place of residence, education level, occupation, crowding index, BMI, presence of digestive disorders and complications, smoking, taking supplements …). KolmogoroveSmirnov tests were used to assess the normal distribution of energy and nutrients. Since not all micronutrients were normally distributed, nonparametric tests were performed for comparison between the 24 h dietary recall and the FFQ. Reproducibility study: Means and standard deviations for nutrient intakes were calculated for both food frequency
3
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 4. Results 91 92 4.1. General characteristics of the population 93 94 The baseline characteristics of the participants are shown in 95 Table 1. The mean age of the participants was 29 years (±5.13 SD), 96 with a mean pre-pregnancy weight of 60.67 kg and a height of 97 163.12 cm. The majority of our participants' (75.8%) had a body 98 mass index (BMI) between 18.5 and 24.9 kg/m2. 72.7% had achieved 99 higher education and only 9% had a crowding index >1. This reflects 100 the relatively high educational and economical level of our sample. 101 The distribution was as follows: 31 women in their first trimester of 102 pregnancy, 39 in the second, 33 in the last and 25 who were 103 recruited at delivery. 104 105 4.2. Reproducibility of the food-frequency questionnaire 106 107 Thirty eight women participated in the reproducibility study. It 108 consisted in administrating the questionnaires twice under the 109 same conditions. The mean daily intake of energy, macro and 110 micronutrients derived from the FFQ1 and the FFQ2, measured 3 111 weeks apart, are presented in the table below. Highly significant 112 intra-class correlation coefficients were generated by all the stud113 ied nutrients, ranging from 0.935 for calcium to 0.984 for vitamin D, with a mean ICC of 0.962 for all nutrients (p < 0.001) (see Table 2). Q3 114 115 116 4.3. Validity of the food-frequency questionnaire 117 118 The estimated intake of energy, macronutrients and most 119 essential micronutrients in pregnancy were used to assess the ac120 curacy of the FFQ and to measure its agreement with the chosen 121 gold standard, the 24 h dietary recall. Details showing median daily 122 intake of nutrients and the correlations between the two methods 123 are presented in Table 3. The FFQ produced significantly higher 124 intakes reaching respectively a relative difference of 18.9%, 29.8%, 125 32.9% and 30.4% for caloric intake, iron, cholesterol, and folic acid as 126 compared to the 24 h dietary recall (average relative difference 127 16.98%). The contribution of macronutrients to the total energy 128 intake was almost similar in both tools. Spearman correlation co129 efficients ranged between 0.294 for iron and 0.762 for caloric 130 intake, with a highly significant p value.
questionnaires (FFQ1 and FFQ2). Intra class correlation coefficients (ICC) were used between the first and the second questionnaires, both administrated under the same conditions by the same interviewer, three weeks apart. This coefficient is known to measure the fraction of total variation due to between-individual variability, with a high value indicating a low within-person variation. Validation study: Crude nutrient intakes estimated from average FFQ (1 and 2) were compared with the average 24 h dietary recall (1 and 2) using Wilcoxon signed rank test (paired data). Spearman correlation coefficients were calculated to measure the strength of the linear association between the FFQ estimates and the 24 h dietary recall estimates. BlandeAltman analysis was carried out to assess the extent of agreement between absolute estimates from the FFQ and the reference tool (the 24 h dietary recall). The differences between the two methods were plotted against the mean of the two dietary measures, with a 95% calculated as limits of agreement (LOA). The plot of the difference shows the relationship between the measurement errors, the true value and to what extent the two tools agree. This was performed for all nutrients [16]. All statistical analyses were performed using the SPSS statistical software package version 20. Results were considered significant at p < 0.05.
Please cite this article in press as: Papazian T, et al., Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.04.015
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
YCLNU2805_proof ■ 29 April 2016 ■ 4/7
4
T. Papazian et al. / Clinical Nutrition xxx (2016) 1e7
Table 1 Sociodemographic characteristics of Lebanese pregnant women at baseline (N ¼ 128). Characteristic
Mean
SD
Age (years) Height (cm) Pre-gravid weight (kg) Pre-gravid BMI (kg/m2)
29.5 163.1 60.7 22.8
5.1 5.7 9.1 3.4
N
%
40.0 56.0 15.0 5.0 11.0 1.0
31.3 43.8 11.7 3.9 8.6 0.8
3.0 10.0 13.0 9.0 75.0 18.0
2.3 7.8 10.2 7.0 58.6 14.1
55.0 56.0 9.0 2.0 3.0 2.0 1.0
43.0 43.8 7.0 1.6 2.3 1.6 0.8
119.0 9.0
93.0 7.0
31.0 39.0 33.0 25.0
24.2 30.5 25.8 19.5
7.0 97.0 19.0 5.0
5.5 75.8 14.8 3.9
Place of Residence Beirut Mount Lebanon Southern Lebanon Northern Lebanon Nabatiyeh Bekaa Educational level Elementary Intermediate Secondary Technical BA/Bsc Master Profession Housewife Employed Entrepreneur Unemployed Liberal Trainee Student Crowding index 1 >1 Trimester of gestation First trimester Second trimester Third trimester At delivery Pre-gravid BMI (kg/m2) <18.5 (underweight) 18.5e24.9 (normal) 25e29.9 (overweight) 30 (obese)
Good Spearman correlations (r 0.5) were found for energy, macronutrients, calcium, vitamin D and sodium. Fair correlations (0.3 < r < 0.49) were found for sugars, fibers, cholesterol, and folic acid. A weak correlation was found for iron (r < 0.3). After
adjustments for energy, good correlations were observed for fats, calcium and sodium and fair correlations were observed for all other nutrients. The average correlations increased after energy adjustment for intakes of nutrients (0.44). An average of 77.26% of pregnant women, ranging from 62% to 88% were classified into the correct quartile, while less than 10% of them were misclassified. 4.4. BlandeAltman analyses Results of the BlandeAltman analyses for proteins and folic acid intake are shown respectively in Figs. 1 and 2. They were selected because of their crucial role during fetal development. Plots of other nutrients resembled those presented below (plots not shown). As mentioned earlier, the difference in intake between the FFQ and the dietary recall is plotted on the Y axis, and the mean intake from the two tools is presented on the X axis. Most data points were clustered around the line zero or the mean difference line across the range of intakes. Few data points fell above the upper LOA. 5. Discussion Valid questionnaires to estimate nutrient intake in the diet of Mediterranean and Middle Eastern population and particularly during pregnancy are lacking. This is to our knowledge the first validation study of a culturally adequate FFQ conducted among a group of Lebanese pregnant women. As in other validation studies, its single administration was enough to estimate the nutrient intake of this particular population [17]. The research team relied on the development and administration of this FFQ, because it is an appropriate tool, reflecting intakes over a prolonged period of time (such as the gestation), less expensive and easy to be administered and interpreted [17,18]. In addition, to assess its validity, literature reviews show that comparing it to a 24 h dietary recall is an optimal comparison method, commonly employed in epidemiological studies [4]. However, because this type of recall relies greatly on memory and is affected by day to day and seasonal variations, some authors suggest the collection of at least two 24 h recalls [13], spread along a specific time frame [19]. In the context of our project, we had two dietary recalls only from one third of our participants', because of
Table 2 Reproducibility study: Mean daily intake of energy and nutrients and the intra-class correlation coefficient for the comparison between FFQ1 and FFQ2 in Lebanese pregnant women (n ¼ 38).
Energy (kcal) Carbs (g) Carbs (%) Proteins (g) Proteins (%) Fats (g) Fats (%) Sugars (g) Fibers (g) Cholesterol (g) Iron (mg) Calcium (mg) Vitamin D (mg) Folic acid (mg) Sodium (mg)
FFQ 1
FFQ 2
Mean ± SD
Mean ± SD
2475 312 48.9 94 15 99 35.6 93 22 237 16.3 889 9.98 340 5335
± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
537 82 6.1 24 2.3 23 5.7 24 6.8 95 4.8 291 2.18 114 855
2533 307 48.7 95 15 102 36.4 93 23 232 16.4 884 10.03 341 5367
± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
494 84 5.3 21 2.1 23 5.1 21 6.3 91 4.5 261 2.23 106 740
ICC [95% CI]
p value
0.957 0.970 0.969 0.962 0.961 0.953 0.974 0.941 0.976 0.936 0.977 0.935 0.984 0.980 0.958
<0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001
% of FFQ 1 103 98 99 101 100 103 102 100 105 98 101 99 101 101 101
[0.917e0.978] [0.943e0.985] [0.939e0.984] [0.927e0.980] [0.924e0.980] [0.910e0.976] [0.950e0.987] [0.886e0.969] [0.953e0.987] [0.877e0.967] [0.956e0.988] [0.875e0.966] [0.969e0.992] [0.962e0.990] [0.919e0.978]
FFQ1: food frequency questionnaire during 1st administration; FFQ2: food frequency questionnaire during 2nd administration; ICC: Intra-class correlation coefficient; CI: Confidence interval (p < 0.05).
Please cite this article in press as: Papazian T, et al., Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.04.015
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
YCLNU2805_proof ■ 29 April 2016 ■ 5/7
T. Papazian et al. / Clinical Nutrition xxx (2016) 1e7
5
Table 3 Validation study: mean daily intake of nutrients, relative difference, Spearman correlation coefficient, and cross-classification for the comparison between FFQ and 24 h recall in Lebanese pregnant women (n ¼ 128).
Energy (kcal) Carbs (g) Carbs (%) Proteins (g) Proteins (%) Fats (g) Fats (%) Sugars (g) Fibers (g) Cholesterol (g) Iron (mg) Calcium (mg) Vitamin D (mg) Folic acid (mg) Sodium (mg)
p valuea
24 h recall
FFQ
Mean ± SD
Mean ± SD
2003 255.5 50.2 75.5 14.8 85.5 35 67 17.9 149 14.1 799 8.9 270 4818
± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
431 59 4.4 18 2.2 67 4 28 7.3 103 5.6 384 2.5 133 878
2381 299.5 51.1 87.5 14.6 95 35.7 87 23.1 198 18.3 881 10.6 352 5184
± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
565 78 15.3 23 2.1 25 5 27 11.7 84 30 311 10.9 246 930
<0.001 <0.001 0.489 <0.001 0.378 <0.001 0.102 <0.001 <0.001 <0.001 <0.001 0.004 0.001 0.001 <0.001
Relative difference (%)b
Spearman correlation coefficient Unadjusted
Energy adjusted
18.9 17.2 1.8 15.9 1.3 11.1 2 29.9 29.1 32.9 29.8 10.3 19.1 30.4 7.6
0.762** 0.757** 0.563** 0.630** 0.544** 0.642** 0.517** 0.389* 0.375* 0.361* 0.294* 0.602** 0.603** 0.445** 0.656**
0.762** 0.414* 0.563** 0.402** 0.538** 0.502** 0.425** 0.318* 0.475** 0.312* 0.349* 0.518** 0.469** 0.404** 0.562**
Percent agreement (same & adjacent quartile)
Extreme disagreement
73% 88% 84% 75% 77% 79% 87% 73% 72% 78% 74% 73% 87% 62% 77%
0% 2% 1% 0% 2% 1% 1% 5% 3% 5% 6% 0% 1% 4% 5%
*p value < 0.05; **p value < 0.01. a Wilcoxon signed rank test to test for differences between FFQ and 24 h dietary recall. b Relative difference ¼ [(FFQ e 24 h recall)/24 h recall]*100.
Fig. 1. BlandeAltman plot showing agreement between the average FFQ and the average of 24 h recall for protein intake. The solid line represents the mean difference in absolute intake between the two dietary assessment methods, while the dashed lines represent the limits of agreement (±2SD).
difficulties related to the respondents' cooperation. However, a special attention was taken, to fill both tools during their enrollment, unlike the methodology in other validation studies where the completion of both instruments differed in the time frame [20]. Sample size and participants' age range were similar to those in other FFQ validation studies [19,21]. However, no direct comparisons can be conducted with other FFQ validation studies, since our FFQ is the first tool assessing Middle Eastern foods' intake by women during pregnancy. In addition, previous studies testing the validity of the FFQ for pregnant women were restricted to food intake at a specific trimester of gestation: the second [21] or the third trimester of pregnancy [22,23]. Our study aimed to include women at different gestational time frame (31%, 39% and 33% respectively for the first, second and the third trimesters), with no restrictions to a particular trimester. However, during the interpretation of our questionnaire, one quarter of our participants admitted variations in appetite, nausea and vomiting during the first trimester, hence affecting
Fig. 2. BlandeAltman plot showing agreement between the average FFQ and the average of 24 h recall for folate intake. The solid line represents the mean difference in absolute intake between the two dietary assessment methods, while the dashed lines represent the limits of agreement (±2SD).
their food intake. Some studies advocate to incorporate participants', starting the fourteenth week of gestation, to avoid the negative impact of the hormonal changes on the dietary intake [24], however, we wouldn't recommend the exclusion of first trimester pregnant women in future studies, because this trimester is of crucial importance for the nutritional status, not only of the mother but of the baby as well. The selection of nutrients to be evaluated in the diets of pregnant women is similar to those analyzed in other studies [19,21]. In our study, we evaluated the contribution of macronutrients to the total energy intake, which is not mentioned in other publications. To determine the reproducibility of our FFQ, we relied on the intra-rater reliability of our tool which consists of administrating the same questionnaire to the same pregnant woman, by the same interviewer [4]. Although neither short nor long intervals are advised between the two collect, a special attention was given to have this second appointment, during a period not exceeding 21 days, to avoid hormonal fluctuations affecting appetite or the transition to a new trimester of gestation, hence influencing the nutritional needs. The mean intra class correlation coefficients (ICC)
Please cite this article in press as: Papazian T, et al., Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.04.015
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
YCLNU2805_proof ■ 29 April 2016 ■ 6/7
6
T. Papazian et al. / Clinical Nutrition xxx (2016) 1e7
obtained after both administrations implies the strong reproducibility of our tool, even though it was conducted to a restricted number of participants. The values of ICC were superior to those observed by Erkkola et al. (2001) for a similar study in Finnish pregnant women [22] and to those of Li et al. (2014) in China [25], but join the values published by Loy et al. (2011) in Malaysian pregnant women [23]. As in other validation studies, our FFQ overestimated food consumption and intakes of nutrients as compared with the 24 h recall [20,21]. Pinto et al. proposed that this overestimation is partly due to difficulties in accurately estimating the portion size consumed [17]. On the other hand, Iqbal et al. suggested that a FFQ containing a large number of food items may be the cause of this overestimation [26]. Moreover, according to Beaton, an overestimating FFQ is less of a concern than an underestimating one, since the latter can precipitate more wrong assessment and misclassification of individuals [27]. Spearman correlation coefficients obtained while comparing the two tools were 0.294 for iron, 0.762 for energy intake, and 0.361 for cholesterol. Similar results were achieved in other validation studies, assessing nutrient intake during pregnancy [5,18,23]. However, statistical analyses were highly significant for almost all studied nutrients compared to those published by Shatenstein et al. in 2011 [20], Mouratidou et al. in 2006 [13] and Loy et al., in 2011 [23]. Willet suggested that acceptable correlations for FFQ validations are between 0.5 and 0.7 [5]. However, these coefficients differed according to Hankin et al. who suggested considering <0.3 as a poor correlation, between 0.3 and 0.49 fair and >0.5 a good level of correlation [28]. Cross-classification permitted to classify the participants in the same or different categories by both tested tools [4]: more than three quarter of women fell into the same quartiles by both methods, which proves that the degree of misclassification was small (<10%), similar to other studies [17,23]. In the present study, the BlandeAltman scatter plots were used to provide a graphical interpretation of agreement between the FFQ and the 24 h dietary recall and to obtain further information that the correlation coefficient itself cannot provide. The plots indicated that, for some nutrients the difference between the FFQ and the 24 h dietary recall increased with increasing intake pointing to overestimates by the FFQ. This shows that participants who consumed higher amounts of food and energy reported more errors during diet assessments. These findings are consistent with the study conducted among pregnant women by Loy et al. [23] but differ than the findings noted by Mouratidou et al. [13] and Shatenstein et al. [20]. In addition, the LOA e defined as the bias (2SD) of the difference e were wide for some nutrients while others were narrow. The reason for this is unclear. New methods of dietary assessments are proposed by Boeing, such as the use of smart phones to translate food labels or pictures consumed by the patient to a server, for a direct nutrient analysis [29]. Another proposition is to rely on biochemical markers, which are more specific but often expensive, invasive, nutrient-selective and influenced by physiological changes [4]. In addition, those indicators are more related to recent nutrient status, unlike the FFQ which predicts more long-term exposure [30]. Therefore, valid FFQs and DRs remain valid tools that predict food intake and nutritional status in all population subgroups. The strengths of this study were the development of a FFQ containing foods culturally adapted to the Mediterranean area and the Arab world, the use of visual aids to achieve higher precision on portion size, the inclusion of the same participants in the reproducibility and the validity phases and the face to face interview administered by a trained dietitian. No one retrieved during the reproducibility phase and no missing data was reported which represent a high sense of motivation and commitment from both
parties. Finally, the use of different statistical analysis permitted to better highlight the reproducibility and the validity of our tool. However, some limitations were encountered in our study: bias related to memory, misreporting of portion sizes and cooking techniques, seasonal variations in some fruits and vegetables and the choice of a single recall as the ideal reference method. Multiple recalls spread along gestation insure a better estimate of nutritional status, however the research team had a limited time frame and could not reach a larger percentage of the respondents'. In addition, the computerized program englobing nutrient databases was not fully compatible with our food items, since policies related with folate enrichment of flour are not implemented by the Lebanese authorities. Therefore, the folate content of the diet of Lebanese pregnant women may be overestimated, since it's mostly derived from USDA food composition tables, where the enrichment of the flour by folic acid is a must. 6. Conclusion Monitoring nutritional status during pregnancy is crucial to implement prenatal nutritional strategies. However, dietary assessment techniques vary depending on the tools and the ways of data collection and analysis. The aim of this research was to develop a semi-quantitative FFQ, culturally convenient to Mediterranean and Middle Eastern countries and to study its reproducibility and validity among a sample of pregnant women. Thus, it has been proven that our newly developed tool is a valid and a reliable method, that provides a real assessment of nutritional intake during pregnancy and its future application on a larger number of participants will permit to assess the effect of dietary intake of pregnant Lebanese women on the neonatal outcomes in our prospective study. However, later work should aim in omitting some food items present in the FFQ to shorten its administration time and validating it in other population groups such as children, adolescents, adults and athletes. Authorship All authors reviewed and approved the submitted manuscript. T.P. conceived the study, supervised the recruitment of the participants, did the face to face interviews, conducted the nutrient analysis and prepared the manuscript. H.H contributed to the design of the study, did the face to face interviews, conducted the data analysis, the statistical interpretation and the drafting of the manuscript. D.S. contributed to the design of the study, did the face to face interviews and participated in the data analysis. N.H. helped in the design of the study material. N.O. carried out the statistical analysis and helped with the data interpretation. L.K. provided complete supervision, critical revision, data interpretation and correction of the manuscript. H.Y. participated in the study conception and design. All authors constituted the research team. Contributors All authors have approved the final article. Financial support This research received no specific grant from any funding agency, commercial or not-for-profit sectors. Conflict of interest The authors declare that they have no conflict of interest.
Please cite this article in press as: Papazian T, et al., Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.04.015
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
YCLNU2805_proof ■ 29 April 2016 ■ 7/7
T. Papazian et al. / Clinical Nutrition xxx (2016) 1e7
Acknowledgments The authors gratefully acknowledge the participants and the gynecologists.
References [1] Procter SB, Campbell CG. Position of the Academy of Nutrition and Dietetics: nutrition and lifestyle for a healthy pregnancy outcome. J Acad Nutr Diet 2014 Jul;114(7):1099e103. [2] Olson CM. Achieving a healthy weight gain during pregnancy. Annu Rev Nutr 2008;28:411e23. [3] Scientific Advisory Committee on Nutrition. The influence of maternal, fetal and child nutrition on the development of chronic disease in later life. 2011. Available at: http://www.sacn.gov.uk/pdfs/sacn_early_nutrition_final_report_ 20_6_11.pdf [accessed on June 2015]. [4] Cade J, Thompson R, Burley V, Warm D. Development, validation and utilisation of food-frequency questionnaires e a review. Public Health Nutr 2002 Aug;5(4):567e87. [5] Willett W. Nutritional epidemiology. 2nd ed. New York: Oxford University Press; 1998. [6] Barbieri P, Crivellenti LC, Nishimura RY, Sartorelli DS. Validation of a food frequency questionnaire to assess food group intake by pregnant women. J Hum Nutr Diet Off J Br Diet Assoc 2015 Jan;28(Suppl. 1):38e44. [7] Tayyem RF, Abu-Mweis SS, Bawadi HA, Agraib L, Bani-Hani K. Validation of a food frequency questionnaire to assess macronutrient and micronutrient intake among Jordanians. J Acad Nutr Diet 2014 Jul;114(7):1046e52. [8] Naja F, Nasreddine L, Itani L, Chamieh MC, Adra N, Sibai AM, et al. Dietary patterns and their association with obesity and sociodemographic factors in a national sample of Lebanese adults. Public Health Nutr 2011 Sep;14(9): 1570e8. [9] Bonaccio M, Iacoviello L, de Gaetano G, Moli-Sani Investigators. The Mediterranean diet: the reasons for a success. Thromb Res 2012 Mar;129(3): 401e4. [10] Issa C, Darmon N, Salameh P, Maillot M, Batal M, Lairon D. A Mediterranean diet pattern with low consumption of liquid sweets and refined cereals is negatively associated with adiposity in adults from rural Lebanon. Int J Obes 2005 2011 Feb;35(2):251e8. [11] Galobardes B, Shaw M, Lawlor DA, Lynch JW, Davey Smith G. Indicators of socioeconomic position (part 1). J Epidemiol Community Health 2006 Jan;60(1):7e12. [12] Report of a WHO Expert Committee Physical status: the use and interpretation of anthropometryWorld health organ tech rep ser, vol. 854; 1995. p. 1e452. [13] Mouratidou T, Ford F, Fraser RB. Validation of a food-frequency questionnaire for use in pregnancy. Public Health Nutr 2006 Jun;9(4):515e22.
7
[14] US Department of Agriculture: agricultural research Service, USDA National Nutrient Database for standard reference, release 27. Available from: http:// ndb.nal.usda.gov/ndb/search [accessed April 2015]. [15] Pellet P, Shadarevian S. Food composition table for use in the Middle East. 3rd ed. Lebanon: American University of Beirut; 1975. [16] Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet Lond Engl 1986 Feb 8;1(8476): 307e10. [17] Pinto E, Severo M, Correia S, dos Santos Silva I, Lopes C, Barros H. Validity and reproducibility of a semi-quantitative food frequency questionnaire for use among Portuguese pregnant women. Matern Child Nutr 2010 Apr;6(2): 105e19. [18] Baer HJ, Blum RE, Rockett HRH, Leppert J, Gardner JD, Suitor CW, et al. Use of a food frequency questionnaire in American Indian and Caucasian pregnant women: a validation study. BMC Public Health 2005;5:135. [19] Potosky AL, Block G, Hartman AM. The apparent validity of diet questionnaires is influenced by number of diet-record days used for comparison. J Am Diet Assoc 1990 Jun;90(6):810e3. [20] Shatenstein B, Xu H, Luo Z-C, Fraser W. Relative validity of a food frequency questionnaire for pregnant women. Can J Diet Pract Res 2011;72(2):60e9. Q4 [21] Mouratidou T, Ford F, Prountzou F, Fraser R. Dietary assessment of a population of pregnant women in Sheffield, UK. Br J Nutr 2006 Nov;96(5):929e35. €s a €nen L, Knip M, Virtanen SM. Val[22] Erkkola M, Karppinen M, Javanainen J, Ra idity and reproducibility of a food frequency questionnaire for pregnant Finnish women. Am J Epidemiol 2001 Sep 1;154(5):466e76. [23] Loy SL, Marhazlina M, Nor AY, Hamid JJM. Development, validity and reproducibility of a food frequency questionnaire in pregnancy for the Universiti Sains Malaysia birth cohort study. Malays J Nutr 2011 Apr;17(1):1e18. [24] Cheng Y, Yan H, Dibley MJ, Shen Y, Li Q, Zeng L. Validity and reproducibility of a semi-quantitative food frequency questionnaire for use among pregnant women in rural China. Asia Pac J Clin Nutr 2008;17(1):166e77. [25] Li M, Halldorsson TI, Bjerregaard AA, Che Y, Mao Y, Hu W, et al. Relative validity and reproducibility of a food frequency questionnaire used in pregnant women from a rural area of China. Acta Obstet Gynecol Scand 2014 Nov;93(11):1141e9. [26] Iqbal R, Ajayan K, Bharathi AV, Zhang X, Islam S, Soman CR, et al. Refinement and validation of an FFQ developed to estimate macro- and micronutrient intakes in a south Indian population. Public Health Nutr 2009 Jan;12(1):12e8. [27] Beaton G. Interpretation of results from diet history studies. L. Kohlmeier. London: Smith-Gordon; 1991. [28] Hankin JH, Wilkens LR, Kolonel LN, Yoshizawa CN. Validation of a quantitative diet history method in Hawaii. Am J Epidemiol 1991 Mar 15;133(6):616e28. [29] Boeing H. Nutritional epidemiology: new perspectives for understanding the diet-disease relationship? Eur J Clin Nutr 2013 May;67(5):424e9. ~ oz E-M, Gimenez-Monzo D, García-de-la-Hera M, [30] Vioque J, Navarrete-Mun Granado F, Young IS, et al. Reproducibility and validity of a food frequency questionnaire among pregnant women in a Mediterranean area. Nutr J 2013;12:26.
Please cite this article in press as: Papazian T, et al., Development, reproducibility and validity of a food frequency questionnaire among pregnant women adherent to the Mediterranean dietary pattern, Clinical Nutrition (2016), http://dx.doi.org/10.1016/j.clnu.2016.04.015
41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80