Determinants of dietary diversity and its relationship with the nutritional status of pregnant women

There is little information regarding factors that determine dietary diversity among pregnant women in Ghana. The present study, therefore, sought to assess the independent predictors of dietary diversity and its relationship with nutritional status of pregnant women in the Northern Region of Ghana. The present study was an analytical cross-sectional survey involving 423 pregnant women in different stages of gestation. The 24-h dietary recall method was used to assess minimum dietary diversity for women (MDD-W), and nutritional status was assessed using mid-upper arm circumference (MUAC) measurements. Binary logistic regression was performed to assess the association between maternal dietary diversity and maternal thinness and a P value of <0 .05 was considered statistically significant. Of the 423 women, 79 .9 % (95 % CI 76 .1, 83 .7) met the MDD-W and the prevalence of undernutrition among the pregnant women was 26 .0 %. The analysis showed that women of low household wealth index were 48 % less likely (AOR 0 .52, CI 0 .31, 0 .88) of meeting the MDD-W, whereas women from households of poor food insecurity were 88 % less likely (AOR 0 .12, CI 0 .05, 0 .27) of achieving the MDD-W. Women of low household size were three times more likely of meeting the MDD-W (AOR 3 .07, CI 1 .13, 8 .39). MDD-W was not associated with maternal underweight during pregnancy. In conclusion, the results of the present study showed that food insecurity and not low MDD-W, associated with mothers' thinness (underweight) during pregnancy in peri-urban setting of Northern Ghana. The present study was conducted in the Sagnarigu Municipality of the Northern Region of Ghana. The municipality, which is largely peri-urban, covers a total land area of 200⋅4 km2 and has a population of 163 513. The present study used a cross-sectional design to collect quantitative data. All women independent of their stage of pregnancy were asked to participate in the study when they attended antenatal care (ANC) in selected health facilities. The present study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects/patients were approved by the Institutional Review Board (IRB) of the Tamale Teaching Hospital, Ghana (Ref no. TTH/10/11/15/01). Written informed consent was obtained from all subjects/patients. The data were collected in all the three subdistricts (Taha-Kamina, Choggu and Sagnerigu) of the Sagnerigu District. The subdistrict health facilities (Choggu, Bagabaga, Garizeigu Clinic and Kalpohini Health Centres), private health facilities (Fulera maternity home and CSSH) and a CHAC facility (St. Louise) at Kpalsi were used as the data collection points. A systematic random sampling procedure was used to select the study participants. The attendance list of the women who sought ANC services served as the sampling frame in each facility. The sample size is determined using the formula for one-point sample estimation: where n is the required sample size, t is the statistical uncertainty chosen = 1⋅96 at a confidence level of 95 %, p is the estimated proportion of pregnant women using diversified diet (Unknown) = 50⋅0 % and m is the margin of error at 5 % (standard value of 0⋅05). A total required sample size large enough to detect a reliable smallest difference and the relationship between the variables tested in the study was, thus, estimated as 384. Allowing for a 10 % non-response rate (i.e. 39 respondents), the overall sample was adjusted to 423 respondents. The data were collected from the respondents using a structured questionnaire which was administered through face-to-face interviews at the household level. Anthropometric equipment includes mid-upper arm circumference (MUAC) tape and Seca electronic adult scale. Maternal height, haemoglobin concentration (Hb) and gestational age records were retrieved from Maternal Health Record Books (ANC cards). Maternal height was measured both as continuous and as a categorical variable with the following cut points: less than 145, 145–149⋅9, 150–154⋅9, 155–159⋅9 and at least 160⋅0 cm. Marital status was classified as married or as unmarried if a woman was divorced, separated, widowed or never married. Maternal occupation was classified according to whether the mother was not working or was working in a manual, non-manual or agricultural profession. The primary dependent variable was the nutritional status of pregnant women as measured by MUAC. The independent explanatory variable was dietary quality as measured by individual dietary diversity scores. The covariate variables included gestational age, maternal age, height, education, occupation, SES, number under-fives in household, parity, birth interval and ANC during the current pregnancy. The minimum dietary diversity for women (MDD-W) was used to assess the overall dietary quality of respondents since it has been shown to indicate adequate nutrient intake(6,19) and can be used as a proxy indicator for measuring nutrient adequacy among pregnant females(20). The MDD-W indicator is based on a 10-food group women dietary diversity score (WDDS-10). These food groups are starch staples (grains, white roots and tubers, and plantains); vitamin A-rich vegetables and fruits; dark green leafy vegetables; other vegetables; other fruits; flesh foods (meat, fish, poultry and liver/organ meats); eggs; pulses/legumes; nuts and seeds; and dairy products. WDDS, which is based on a 24-h dietary recall period(13), was applied to characterise the average usual dietary intake of pregnant women in the study area. The women were asked to recall all foods consumed from the above food groups on the previous day. Responses were recorded as ‘yes’ or ‘no’. A ‘yes’ response was scored as ‘1’, and a ‘no’ response was scored as ‘0’. The scores were summed up to create the women DD score. Available evidence suggests that WDDS is a good measure of household macronutrient adequacy and household nutrition insecurity. The dietary scores were classified into low and high diversity based on the MDD-W. Women having a diversity score of less than 5 were classified as having low dietary diversity and scores of 5–10 are classified in the high dietary diversity scores(21). Additionally, the FAO validated 11-item food groups frequency questionnaire (FFQ) was used to quantify maternal dietary intake based on 7-d dietary diversity score(13). This was derived based on the number of food groups consumed from a 7-d food frequency questionnaire and included 11 food groups. The food group frequency of consumption (past 7 d) was measured for each food group by assigning a score of 0 if not consumed during the previous week, 1 if consumed on 1–3 d and 2 if consumed for at least 4 d. This composite index of dietary diversity which took into account the weekly food frequency varied from a minimum of 0 to a maximum of 22. The eleven food groups were flesh meats (i.e. beef, pork, lamb, goat, poultry, etc.), fish, eggs, milk and milk products, organ meat (e.g. liver, kidney, etc.), legumes, cereals, roots and tubers, dark green leafy vegetables, vitamin A-rich fruits and fats and oils. Household food access was measured using the food consumption score (FCS), and it was calculated as per the World Food Programme (WFP)(22). The FCS as an index is expected to provide a more accurate measure of the quality of the household diet because it accounts for the nutritional value of food in addition to the number of different types of food consumed. The FCS is a proxy indicator of household caloric availability. MUAC is often used as a measure of fat-free mass, and in the present study, MUAC was used to assess the nutritional status of pregnant women. MUAC was used as a proxy for body weight, since it is not affected by gestational age(23). MUAC was also measured using a non-stretchable MUAC tape(24). MUAC was measured to the nearest 0⋅1 cm, and values below 25⋅0 cm were classified in the analyses as an indicator of low body weight. There is presently no internationally agreed MUAC cut-offs(25). A household wealth index based on household assets and housing quality was used as a proxy indicator for SES of households. Principal component analysis (PCA) was used to determine a household wealth index from information collected on housing quality (floor, walls and roof material), source of drinking water, type of toilet facility, the presence of electricity, type of cooking fuel and ownership of modern household durable goods and livestock (e.g. bicycle, television, radio, motorcycle, sewing machine, telephone, cars, refrigerator, mattress, bed, computer and mobile phone)(26–29). These facilities or durable goods are often regarded as modern goods that have been shown to reflect household wealth. A household of zero-index score for example means that household had not a single modern good. The scores were, thus, added up to give the proxy household wealth index. The main aim of creating the index was to categorise households into SES groupings in order that we could compare the difference in the prevalence of maternal thinness between the groups of lowest and highest SES. Data were analysed using SPSS version 21 (SSPS Inc. Chicago, IL, USA) statistical software. Both (bivariate and multivariable analysis) were performed to identify risk factors of maternal underweight during pregnancy. Only variables that showed significant association (P < 0⋅05) with each dependent variable in the bivariate analysis were selected and adjusted for in the multivariable binary logistic regression analysis. Multicollinearity among the predictor variables was checked before their inclusion in the final regression model. We assessed multicollinearity among independent variables by using the variance inflation factor (VIF), which assesses increment in regression coefficients if the independent variables are correlated. VIF > 5 is an indication that multicollinearity may be present, while VIF > 10 is certainly multicollinearity among the variables. We did not have any VIF exceeding 5, indicating no collinearity. Results were presented as adjusted odds ratio (AOR) with 95 % confidence intervals (CIs) to measure the strength of association.