Menu
Menu
Menu
Menu
Dietary diversity before and during pregnancy is crucial to ensure optimal foetal health and development. We carried out a cohort study of women of reproductive age living in the Sô-Ava and Abomey-Calavi districts (Southern Benin) to investigate women’s changes in dietary diversity and identify their determinants both before and during pregnancy. Nonpregnant women were enrolled (n = 1214) and followed up monthly until they became pregnant (n = 316), then every 3 months during pregnancy. One 24-hr dietary recall was administered before conception and during each trimester of pregnancy. Women’s dietary diversity scores (WDDS) were computed, defined as the number of food groups out of a list of 10 consumed by the women during the past 24 hr. The analysis included 234 women who had complete data. Mixed-effects linear regression models were used to examine changes in the WDDS over the entire follow-up, while controlling for the season, subdistrict, socio-demographic, and economic factors. At preconception, the mean WDDS was low (4.3 ± 1.1 food groups), and the diet was mainly composed of cereals, oils, vegetables, and fish. The mean WDDS did not change during pregnancy and was equally low at all trimesters. Parity and household wealth index were positively associated with the WDDS before and during pregnancy in the multivariate analysis. Additional research is needed to better understand perceptions of food consumption among populations, and more importantly, efforts must be made to encourage women and communities in Benin to improve the diversity of their diets before and during pregnancy.
This study was part of the Retard de croissance intra‐utérin et paludisme (RECIPAL) cohort study, which has been fully described elsewhere (Accrombessi et al., 2018). Nonpregnant women of reproductive age were recruited at a community level from Sô‐Ava and Abomey‐Calavi, two semiurban districts of Benin, and followed up with monthly until they became pregnant; these women constituted the primary cohort (preconceptional follow‐up). The subsample of women who became pregnant was then tracked monthly at the maternity clinic from early pregnancy to delivery; they constituted the secondary cohort (gestational follow‐up). The present study collected dietary intakes of women from both the primary and secondary cohorts between November 2014 and December 2017. The RECIPAL study was approved in Benin by the ethical committees of the Institute of Applied Biomedical Sciences and the Ministry of Public Health, and in France by the French National Research Institute for Sustainable Development (IRD). The study was conducted according to the Helsinki Declaration for medical research. Before data collection, written informed consent was obtained from each participant after ensuring their understanding of the purpose, objectives, confidentiality rules, benefits, and risks of taking part in the study. The study took place in four subdistricts in Southern Benin as follows: So‐Ava, Houedo‐Aguekon, Vekky in the district of So‐Ava, and Akassato in the district of Abomey‐Calavi. Both districts are semiurban areas, but Sô‐Ava has the distinction of being a lake area mainly occupied by natives, whereas Abomey‐Calavi is more heterogeneous in terms of population. The climate is subequatorial and characterised by a long rainy season (April–July), a short dry season (August–September), a short rainy season (September–October), and a long dry season (November–March). Women were enrolled in the primary cohort when they met the following criteria: being 18–45 years old, married, nonpregnant, apparently healthy, not known to be sterile, using no current contraception, having no travel plans of more than 2 months during the 18 months after inclusion, willing to become pregnant, and planning to deliver in either the Sô‐Ava or Abomey‐Calavi districts. These women were visited at home every month and tested for pregnancy. Women with positive pregnancy tests were enrolled in the secondary cohort. Women who did not conceive after 1 year of follow‐up were invited to the district maternal care centre for a medical examination. In cases of genital infection, they received medical advice and were referred to a gynaecologist. Follow‐up stopped after 2 years when women did not become pregnant. Demographic and socio‐economic characteristics of both women and their households were collected once upon inclusion in the primary cohort via a structured questionnaire. Data included household size, assets, housing type, women’s ages, parity (number of children, alive or dead), type of union (polygamous/monogamous), ethnic group, education, and main activities. A multiple correspondence analysis (Sourial et al., 2010; Traissac & Martin‐Prevel, 2012) using socio‐economic data was performed to compute a wealth index and to classify households into low, middle, and high wealth levels according to tertiles. Dietary assessments of women were conducted before conception and at each trimester of pregnancy. The minimum number of dietary assessments per woman considered in this analysis was two, and the maximum was four. One quantitative 24‐hr dietary recall (Gibson, Charrondiere, & Bell, 2017; Gibson & Ferguson, 2008) was performed through face‐to‐face interviews. Women were asked to describe all foods, drinks, and snacks consumed over the last 24 hr, including a detailed description of the recipes. Food items consumed were classified into 10 food groups according to recommended classifications (Food and Agriculture Organization [FAO] & Family Health International 360, 2016): (a) grains, white roots and tubers, and plantains (also known as starchy staples); (b) pulses (beans, peas, and lentils); (c) nuts and seeds; (d) dairy; (e) meat, poultry, and fish; (f) eggs; (g) dark green leafy vegetables; (h) other Vitamin A‐rich fruits and vegetables; (i) other fruits; and (j) other vegetables. The number of food groups consumed was summed up and dichotomised using a cut‐off at five food groups to compute the minimum dietary diversity for women (MDD‐W) indicator, which has been recently developed and validated as a proxy of micronutrient adequacy (FAO & Family Health International 360, 2016; Martin‐Prevel et al., 2017). We also used the number of food groups consumed as a continuous variable, namely the women’s dietary diversity score (WDDS‐10), which ranged from 0 to 10 food groups. Four additional food groups—red palm oil, other oils and fats, sugar and sugary drinks, and alcoholic beverages—were used for the purpose of describing women’s dietary patterns. These groups were not used in the calculation of WDDS or MDD‐W. Women’s height was measured at inclusion. Women’s weight was measured twice during the preconceptional follow‐up, then once a month during the gestational follow‐up. Both weight and height were measured according to World Health Organization (WHO) standard procedures (Norgan, 1988). Height was measured to the nearest millimetre with a SECA 206 gauge (Hamburg, Germany). Weight was measured with calibrated electronic scales (Tefal, France) with a precision of 100 g. Body mass index (BMI) was calculated before pregnancy and women were classified as underweight (BMI < 18.5 kg/m2), normal (18.5 ≤ BMI ≤ 24.9 kg/m2), or overweight or obese (BMI ≥ 25 kg/m2) based on WHO classification (WHO, 2018). Data were collected by seven enumerators (five nurses and two nutritionists) holding at least bachelor's degrees, with experience in field data collection. They were trained over 6 days on the 24‐hr recall technique, the questionnaire and tools, and anthropometric measurements. The questionnaire was pretested by the enumerators during the training and was adjusted where needed. During data collection, the enumerators were supervised daily by an experienced nutritionist doubling as the principal investigator and supported by a team of experts in nutritional epidemiology. The supervisor checked the proper completion of the questionnaires daily as well as consistency of the answers. Data from dietary recalls were entered and cross‐checked by repeated entry using the Epidata entry 3.1 software (Lauritsen & Bruus, 2004), whereas anthropometric, socio‐economic, and demographic data were entered using ACCESS 2007. Statistical analyses were performed using Stata 13 (College Station, TX, USA). We first described the basic characteristics of the sample from the primary cohort and examined whether women who became pregnant during the project (n = 316) differed from women who did not (n = 581). We presented mean (SD) for continuous variables and frequencies (%) for categorical variables. The main analysis was restricted to women who had one assessment at preconception and at least one assessment during pregnancy (n = 234). The mean WDDS, the proportion of women who consumed different food groups, and the proportion of women reaching the MDD‐W were compared over the entire follow‐up using a linear mixed model (for continuous variables) or a logistic mixed model (for categorical variables) including a random intercept (the individual) to take into account repeated measurements for the same subject. In bivariate analyses, we examined factors that were associated with women's dietary diversity before pregnancy, using the WDDS as the continuous response variable in linear regression models and using. Factors tested included subdistricts (geographical factor); women's ages, household size, parity, type of union, ethnic groups, women's and their husband's education levels (socio‐demographic factors); women's and their husbands' professional activities and wealth index of the household (economic factors); women's body mass index (nutritional factor). Variables associated with the WDDS with a level of statistical significance of 0.20 were considered for the multivariate analysis. Blocks of factors were constituted based on conceptual reasons (factors belonging to a same dimension); these blocks were successively entered in the model using a manual ascending method. The final multivariate model was used to test if the WDDS changed between the visits of follow‐up (preconception, trimester 1, 2, or 3 of pregnancy). Interaction terms factor*visit were also tested in the final model to examine whether any change in the WDDS differed according to the modality of the factors. Univariate and multivariate analyses were systematically controlled for the season because of its known effect on food availability and hence on dietary diversity. Statistical level of significance was set at p < .05. This study has some limitations. There was only one 24‐hr recall administered at each time point; for this reason, we could not survey women's habitual dietary intake. We also focused on dietary diversity, a single dimension of diet quality, and did not take into account other dimensions or food quantities. Another limitation is the lack of a qualitative survey on the socio‐cultural component for objective measurement of attitudes, behaviours, and beliefs regarding diets before and during pregnancy. Such data would have helped us gain more insight into the trends of our results. However, this was beyond the primary focus of our study, which was to investigate whether there are changes in women's dietary diversity before and during pregnancy. Further research will focus on this purpose. Nevertheless, the cohort design of the study was a real strength and constituted a unique source of data in West Africa. As this study was part of a larger study for which biological samples were collected, we believe that the high rate of lost to follow‐up was precisely due to very strong endogenic belief and mistrust towards the research team.
N/A
