Effect of social and behavioral change interventions on minimum dietary diversity among pregnant women and associated socio-economic inequality in Rajasthan, India | BMC Nutrition

Data

For this study, data were used from the cross-sectional survey conducted biannually to monitor the progress in outcome indicators of the program. Data was collected by third-party field enumerators, who had been thoroughly trained on survey tools, probing, and handling sensitive questions. Data from the latest survey round conducted between May and June 2023 was used for analysis for this study. Data quality was ensured through field monitoring, backcheck, and spot check during the survey.

A total of 6848 pregnant women (respondents) were surveyed using a structured survey tool on a mobile-based application. The survey tool was field-tested and modified after a pilot study to monitor the coverage and effectiveness of the program precisely. It consisted of various questions to identify respondent’s socio-economic status and demographic details, including closed-ended questions on age, level of education, ownership of household assets, and household built-up characteristics. The survey tool also included closed-ended questions on the consumption of food items, knowledge of prenatal care, cash incentive scheme, utilization of received cash incentives, and respondents’ exposure to various SBCC (Social and Behavior Change Communication) inputs given under the program in the last three months.

The survey was part of periodic program monitoring and data/report of the survey were not published anywhere in public domain.

The selection of food items for diet diversity was contextualised based on the findings of a qualitative study conducted to understand food consumption patterns and the availability of food groups among the target population. To capture the consumption of various food items in the last 24 h (open recall), enumerators asked a series of listed food items in the tool to help the respondent recall all foods and beverages consumed in the last 24 h and probed for ingredients in mixed dishes. Each food or beverage that the respondent mentioned was punched on the predefined list by the enumerator.

Study population & sampling

The target population for the cross-sectional survey was all pregnant women who had registered for nutrition counselling in the last six months from the reference date of the survey. The sample size was calculated using a population proportion determination formula based on the assumption of a 95% significance level, 4% margin of error, and 1.5 design effect, assuming prevalence for the unknown population 50% and 81% response rate. The calculated sample size was 6800 pregnant women in all five districts. 2000 Anganwadi Centers (AWCs) were selected as primary sampling units (PSUs) to achieve the sample size in each district. Simple random sampling was done while selecting the PSUs, assuming that at any point in one AWC, 5–6 pregnant women would be available. The number of AWCs per block was obtained by dividing the number of AWCs needed by the number of blocks in the districts.

Variable description

Outcome variables

Minimum dietary diversity (MDD)

The MDD was calculated as a proxy indicator to understand the nutritional adequacy of pregnant women’s diet during pregnancy. This study defines MDD as consumption of food from at least five food groups out of 10 in the last 24 h. Based on recommendations of FAO, collected data on food consumption grouped broadly under 10 food groups: (a) Grains, white roots and tubers, plantains, ( b) Pulses (beans, peas and lentils), (c)nuts and seeds, (d) Dairy Products, (e) Meat, poultry and fish, (f) egg (g) Dark green leafy vegetables, (h) vitamin A- rich fruits and vegetables, (i) other vegetables, (j) other fruits (FAO, 2016). Consumption of oils and sweet dishes are excluded from the construction of MDD. A diet diversity score was calculated using the information on all these 10-food groups ranging from 0 to 10-point scale (“0” signifies non-consumption of any food items from food groups, and “10” signifies consumption of food from maximum food groups). Respondent scoring less than 5 was categorized as “not having minimum dietary diversity”, and more than 5 was categorized as “having minimum dietary diversity”. Hence, MDD was identified if a woman has consumed food from five or more food groups in the last 24 h.

Explanatory variables

Household Wealth Status

The household wealth status was assessed using information on ownership of livestock (goat, camel, cow/buffaloes, sheep, chicken or ducks), ownership of material assets (access to electricity, mobile phone, computer, refrigerator, pressure cooker, bicycle, car etc.), source of drinking water, type of toilet facilities, type of cooking fuel, type of material used in house’s floor, walls and roof, number of sleeping rooms, number of household members and ownership of bank account. Scores were assigned to each household based on the mentioned characteristics. The household’s score was calculated using principal component analysis (PCA) (Gausman, 2018). The score was divided into five equal groups: poorest, poorer, middle, richer, and richest, with 20% of the population in each group.

SBCC exposure

Exposure to various SBCC inputs was captured retrospectively, taking three-months reference period. In this study, SBCC exposure measures the intensity of program inputs. Household Visits by Poshan champion, Anganwadi Worker (AWW), Axillary Nurse Midwives (ANM) and counselling during MCHN day under the program were considered for calculating respondent’s exposure to SBCC. Respondents were asked about their exposure to these four touch points as well as the frequency of these touch points in the last three months. Based on the frequency of exposure to four touch points, each respondent was assigned a score ranging from 0 to 10 points scale, where “0” signifies no exposure at all and “10” signifies maximum exposure. Further categorization was done using this score as “0–2 low exposure”, “3–5 medium exposure” and “>5 high exposure”. Table 1 denotes the SBCC exposure touchpoints.

Received cash incentives

In targeted districts, two conditional cash transfer schemes are operational, referring to women’s first and second parity. Central government-funded Pradhan Mantri Matru Vandna Yojna (PMMVY) provides a conditional cash transfer of INR 6000 in case of first parity birth, and state-funded Indira Gandhi Matritva Poshan Yojna (IGMPY) provides a conditional cash transfer of INR 8000 for second parity pregnant women. The conditionalities of both schemes rely on early registration for ANC, ANC compliance and institutional Delivery. During the survey, respondents were asked “if they have received cash incentives under any of these schemes”. During analysis, the affirmative response was categorized as “1 beneficiary received cash incentives under any of schemes” “0 beneficiaries who did not receive cash incentives under any schemes”.

Belief in food-related myths

Considering the locally prevailing belief, information was captured on myths and taboos in the survey. During the survey, women were asked about their belief in myths related to restrictions on consuming certain food items, bananas, milk and Jaggery during pregnancy. Respondents who believed in any of these myths were categorized as “believing in food-related myths”, and respondents who answered “No” to all three myths-related questions were categorized as “do not believe in any food-related myth”.

Social characteristics

Caste was coded as General/Other backward caste (OBC), Schedule Caste (SC), and Schedule Tribe (ST). The study population has a higher proportion of SC and ST population. SC and ST caste are socially and economically disadvantaged groups and have faced historical discrimination in India. Religion was grouped as Hindu, Muslim and Other. The education level of women was categorized into four groups: “illiterate”, “Primary Education”, “Up to High School”, “Higher Secondary and Above”.

Other explanatory variable

The age of beneficiaries was re-categoriesed as “0 = < 25 years” and “1>=25 years”. During the survey, women were asked about their birth parity. Under the program, only first and second-parity beneficiaries are eligible for intervention. Hence, the analytical sample only has women of first and second parity. During the survey, women were asked about their knowledge of ideal ANC visits for a pregnant woman. For analysis, women’s response has been re-coded as “0 < 4 ANC visits” and “1 = > 4 ANC visits”.

Statistical analysis

Descriptive statistics, including frequency distribution and cross-tabulations of each predictor and outcome variable, were used to describe variables for the study. Categorical variables were presented in percentages and frequency, whereas continuous variables were presented in mean and standard deviation. Pearson’s X2 test verified the association of the outcome variable with the predictors. Binary logistic regression was used to estimate odds ratios and 95% Confidence Intervals (CIs). Results were presented as crude odds ratio (cOR) and adjusted odds ratio (aOR) to assess the strength and presence of association, with a threshold of p < .05 used for determination of statistical significance.

Multivariate decomposition analysis was used to quantify the contribution of selected predictors in explaining the rich-poor gap in the prevalence of maternal dietary diversity. Multivariate decomposition technique uses the output from regression models to partition the components of a group difference in a statistic, such as a mean or proportion, into a component attributable to compositional differences between groups, i.e., differences in characteristics or endowments, and a component attributable to differences in the effects of characteristics, i.e., differences in the returns, coefficients or behavioural responses. The mean difference in Y between groups A and B can be decomposed as,

where Y denotes the N × 1 dependent variable vector, X is an N × K matrix of independent variables, and β is a K × 1 vector of the coefficient. The component labelled E refers to the part of the differential attributable to differences in endowments or characteristics, usually called the explained component or characteristics effects. C refers to the part of the differential attributable to differences in coefficients or effects usually known as the unexplained component or coefficient effects, where A is the pregnant woman from the richest household (comparison group), and B is the pregnant woman from the poorest household (reference group). Therefore, E reflects a counterfactual comparison of the difference in outcomes from the women from the richest household perspective, and C reflects a counterfactual comparison of outcomes from women from the poorest household perspective. STATA 18.0 software was used for data analysis.