Household food insecurity and children's dietary diversity and nutrition in India. Evidence from the comprehensive nutrition survey in Maharashtra

Transcription

1 Received: 20 October 2016 Revised: 30 January 2017 Accepted: 8 February 2017 DOI: /mcn SUPPLEMENT ARTICLE Household food insecurity and children's dietary diversity and nutrition in India. Evidence from the comprehensive nutrition survey in Maharashtra S. Chandrasekhar 1 Víctor M. Aguayo 2 Vandana Krishna 3 Rajlakshmi Nair 4 1 Indira Gandhi Institute of Development Research, General A K Vaidya Marg, Goregaon (E), Mumbai , India 2 United Nations Children's Fund (UNICEF), New York, NY, USA 3 Indian Administrative Service, Government of Maharashtra, Mumbai, India 4 UNICEF Maharashtra, Mumbai, India Correspondence S Chandrasekhar, Indira Gandhi Institute of Development Research, General A K Vaidya Marg, Goregaon (E), Mumbai India chandra@igidr.ac.in Abstract We analyse data from the 2012 Comprehensive Nutrition Survey in the State of Maharashtra, India, which surveyed 2,630 households. This is a unique dataset because in addition to nutritional status of mothers and children, it has information on diets of women and children and household food security. This rich dataset allows us to address three issues: whether household food security predicts higher diet diversity in children aged 6 23 months; whether household food security predicts lower risk of undernutrition; and whether the lower risk of undernutrition in children who live in food secure households is mediated by improved diet diversity for children. We find that children from moderately food insecure and severely food insecure households are more likely to have lower diet diversity scores. We find that the odds of a child being severely stunted, severely underweight, or severely wasted are higher in severely food insecure households. After controlling for children's diet diversity, and other child, maternal and household characteristics, we find that household food security is no longer statistically associated with stunting, wasting, or underweight. However, diet diversity of children is statistically significantly associated with whether a child is stunted or underweight. Our results although not causal provide evidence for understanding the extent to which household food insecurity affects children's diet diversity and how both these factors affect nutrition outcomes in children. Our analysis informs Government of Maharashtra's and India's National Nutrition Mission in their efforts for formulating appropriate policies and programmes to address child undernutrition. KEYWORDS children dietary diversity, household food security, India, Maharashtra, stunting, underweight, wasting 1 INTRODUCTION The conceptual framework for maternal and child undernutrition by UNICEF identifies three levels of causes: immediate, underlying, and basic (United Nations Children's Fund, 2015). Most of the empirical literature on maternal and child undernutrition has focussed on understanding the strength of the association between the underlying and immediate causes of undernutrition. The three underlying causes are household food insecurity, inadequate care for women and children including feeding practices, and inadequate access to health services and a healthy environment. Inadequate dietary intake and/or disease are identified as the immediate causes of maternal and child undernutrition. The bulk of the empirical literature on the degree of association between children's diet diversity and children's nutritional status uses data from the Demographic and Health Survey (DHS). One of the early contributions by Arimond and Ruel (2004) using DHS data found an association between dietary diversity of children and children's height for age Z score after controlling for other socioeconomic factors. However, such analysis of DHS data is constrained by the fact that information on household food security is not available in DHS datasets. Recent surveys in a handful of countries have information on indicators of household food security and children's diet diversity and nutrition. The findings from these studies on the association between indicators of child undernutrition on one hand and indicators of child dietary intake and household food security on the other are far from Matern Child Nutr. 2017;13(S2):e wileyonlinelibrary.com/journal/mcn 2017 John Wiley & Sons Ltd 1of8

2 2of8 robust. A more recent contribution to the literature by Ali et al. (2013), who analyse data collected as part of the Alive & Thrive project in Bangladesh, Ethiopia, and Vietnam, relies on the Household Food Insecurity Access Scale to group households into the four categories: food secure and mild, moderately, and severely food insecure. For each country, they estimate three separate logit models where the child outcome variables of interest are stunted, wasted, and underweight. When they model these outcomes only as a function of household food insecurity, they find that household food insecurity is indeed an important predictor of child undernutrition. However, when they control for all other explanatory variables, household food insecurity is not statistically significant in predicting wasting. In the case of stunting and underweight, the results are mixed as household food insecurity is not unequivocally an important predictor of either outcome. Ali et al. also report that child diet diversity did not mediate the relation between household food insecurity and undernutrition in any of the countries (p. 2015). A synthesis paper by Jones et al. found that WHO infant and young child feeding indicators showed mixed associations with child anthropometric indicators across countries. Indicators of diet diversity and overall diet quality were positively associated with HAZ in Bangladesh, Ethiopia, India and Zambia (p.1 Jones et al., 2014). They concluded that additional measures of dietary quality and quantity may be necessary to understand how specific IYCF behaviours relate to child growth faltering (p.1). The objective of this paper is to contribute to the literature investigating the relationship between household food security and children's diet diversity and anthropometry by exploring evidence from India, home to a large number of undernourished children. We quantify the impact of household food insecurity (underlying cause) on diet diversity (immediate cause) and nutritional status (outcome) using a comprehensive household survey dataset from Maharashtra, India's second most populous state. We set out to answer three interrelated questions: First, does household food security predict higher diet diversity in children aged 6 23 months?; second, does household food security predict lower risk of undernutrition measured as stunting, underweight, or wasting in children aged 6 23 months old; and third, is the lower risk of undernutrition in children who live in food secure households mediated by improved diet diversity for children? By answering these three questions, we aim to strengthen the evidence base for policy development, programme design, and research prioritization on maternal and child nutrition in Maharashtra and India. 2 DATA AND METHODS We analyse unit level data from the Comprehensive Nutrition Survey in Maharashtra (CNSM), which is representative for the rural and urban population of Maharashtra (IIPS and UNICEF 2013). With a population of 112 million, Maharashtra accounted for 9.3 per cent of India's population in 2011 (last census data available). Although Maharashtra is one of the richer Indian states, it was a concern that the reductions in the prevalence of child undernutrition over the period and lagged that of economic growth. In particular, stunting Key messages Children from moderately food insecure and severely food-insecure households are more likely to have lower diet diversity scores. The odds of a child being severely stunted, severely underweight, or severely wasted are higher in severely food insecure households. Our results, although not causal, provide evidence for understanding the extent to which household food insecurity affects children's diet diversity and how both these factors affect nutrition outcomes in children. Our analysis informs Government of Maharashtra's and India's National Nutrition Mission in their efforts for formulating appropriate policies and programmes to address child undernutrition. declined meagrely from 49.6% in (IIPS and ORC Macro 2000) to 44% in (IIPS and ORC Macro. 2008). Consequently, the Government of Maharashtra decided to launch the Rajmata Jijau Mother and Child Health and Nutrition Mission (i.e., the Nutrition Mission). The objective of the Nutrition Mission is to reduce child undernutrition in the state by focusing on the 1,000 days from conception to age 2 years. The CNSM was conducted from February to May 2012 following the completion of the first phase of the Nutrition Mission. The CNSM provided information on 2,630 households, 2,565 mothers, and 2,650 children aged 0 23 months. The questionnaire used for CNSM includes three modules: household module, mother's module, and child's module. The CNSM is a unique dataset in that it allows for a detailed exploration of the association between indicators of undernutrition in children aged 0 23 months on one hand and children's diets and household food security on the other hand. Although there are detailed surveys of consumption expenditure in India, the CNSM is the first household survey in India with information on household food security. Chandrasekhar and Pandey (2015) have pointed out that National Sample Survey Organisation's survey of consumption expenditure is inadequate for estimating the number of food insecure households in India. They have argued that by collecting information on the four pillars of food security: availability, access, nutritional adequacy or utilisation, and stability, CNSM data appears to capture the different elements of food security. Thus, the CNSM is a unique dataset for assessing household food security in India. Details on the sampling methodology and data collection are available in the final CNSM report (IIPS and UNICEF 2013). Before we proceed further, we discuss the key variables of interest: prevalence of stunting, wasting, and underweight in children, measures of child dietary diversity, and measures of household food security. Three standard indicators were used to assess the nutritional status of children: height for age, weight for height, and weightfor age.

3 3of8 Children's dietary diversity was measured by collating information on index child's food consumption from a set of 21 food items during the 24 hours preceding the survey. These 21 items were grouped into seven food groups: grains, roots, and tubers; legumes and nuts; dairy products; flesh foods; eggs; vitamin A rich fruits and vegetables; and other fruits, dry fruits, and vegetables. The value corresponding to each of these food groups takes the value one if the child was fed any one of the food items corresponding to the food group in consideration, else the value was coded as zero. The seven food group score, a measure of dietary diversity, is arrived at adding up the value of the variables reflecting the seven food groups. For each child, the outcome variable of interest in diet diversity can potentially take any value between zero and seven. A child would get the value zero if the child does not consume any one of the 21 items and hence was not fed any one of the seven food groups. At the other end, a child would get the value seven if the child was fed items from all seven food groups. CNSM measured household's food security using a set of nine questions developed by USAID's Food and Nutrition Technical Assistance project. The response to the nine questions captured a) feelings of uncertainty or anxiety over food, b) perceptions that the food is of insufficient quality, c) perceptions that food is of insufficient quantity, and d) reported reductions of food intake (p. 14 IIPS and UNICEF 2013). On the basis of this information, households were grouped into four categories: food secure, mildly food insecure, moderately food insecure, or severely food insecure. The proportion of households classified as food secure in the rural and urban areas of the state was 49.4% and 66.2%, respectively. Furthermore, the proportion of households classified as severely food insecure in rural and urban areas was 18.0% and 8.3%, respectively. A more detailed discussion on the measurement of household food insecurity is given in the CNSM report (IIPS and UNICEF 2013). 3 RESULTS 3.1 Sample characteristics The sample we analyse includes all children aged 6 23 months. We exclude children less than 6 months of age since they are expected to be exclusively breastfed and not fed any complementary foods. The average age of children in the sample is 14.1 months. The characteristics of the children included in the analytical sample and those of their mothers and the households to which they belong are presented in Table 1: 55% of the children included in the sample are boys, 42% have a mother with 10 or more years of formal education, and 59% live in households that are food secure. The prevalence of undernutrition among children aged 6 23 months in the sample is as follows: 18.3% of children are stunted (9.6% severely stunted), 18% are underweight (6.9% severely underweight) and 10.7% are wasted (3.7% severely wasted). 3.2 Regression results Household food security and children's diet diversity Figure 1 gives the distribution of our dependent variable, the seven food group score, in children aged 6 23 months. The seven food group TABLE 1 Child, household, and maternal characteristics Child anthropometry (%) Child stunting 27.9 Child underweight 24.9 Child wasting 14.4 Household food security Food secure 58.6 Mildly food insecure 17.1 Moderately food insecure 11.7 Severely food insecure 12.6 Children's characteristics Girl 45 Household characteristics Social group Scheduled caste 14.7 Schedule tribe 11.9 Other backward class 26.8 Vimukta Jati and nomadic tribes 11 Others 35.6 Religion Hindu 84.4 Muslim 9 Others 6.6 Wealth Index (dummy variables for the five quintiles Ration Card No card 25.7 White 4.7 Orange 45.2 Yellow/antyodaya 24.4 Mother's Education No education 10.4 Less than 5 years years years years 18.9 Over 12 years 22.8 Location of Residence Rural 50 Nonslum urban 34.3 Slum 15.7 Place of residence (region) Amravati Aurangabad Konkan Nagpur 7.80 Nashik Pune score, reflecting diet diversity of the child, is a count measure. In our sample, the seven group food score ranges from zero to six, with 75% of children being fed two food groups or fewer. If the mean and standard deviation of this variable were the same then it would be appropriate to estimate a Poisson model. However, this is not true in our case since the mean is 1.86 and the standard

4 4of8 FIGURE 1 Distribution of children aged 6 23 months by diet diversity (seven food group score) deviation is Also, we have nearly 11 per cent of children whose seven food group score is zero. In this scenario, it is preferable to estimate the negative binomial model. It should be noted that in the literature authors have converted the seven food group score into a binary variable that takes the value one if the score is greater than or equal to four and value zero if the score is less than four. We do not opt to define the dependent variable as a binary variable. The reason for this is that less than 11 per cent of our sample will take the value one, and we also loose the variation in the score that is present in the seven food group score. In Model 1, we include the dummy variable reflecting the index of household food security as a control. In Model 2, we include characteristics of the household, mother and the child. The socioeconomic characteristics of the households that we control for include wealth index (as a measure of asset possession by the household) and the type of ration card possessed by the household for accessing the Public Distribution System (PDS), which provides subsidised food grains. Whether a household is eligible to access the subsidized food grains and entitlement from the PDS is reflected by the type of ration card it holds. There are four types of households: those without a card and those with a white, orange, or a yellow (antyodaya) card. In addition to reflecting the access to the PDS, these cards implicitly reflect the income of the household. The yellow card is for households with an income of less than Rs 15,000 per annum; the orange card for households with an annual income of Rs 15,000 Rs 100,000; the white card is for households with an annual income of Rs 100,000 and above. We use the type of ration card as a control for the socioeconomic status of the household rather than whether the household drew food grains from the PDS. The estimates based on the two models are reported in Table 2. We find that children from households that are mildly food insecure, moderately food insecure, and severely food insecure are more likely to have a lower diet diversity score that children from households that are food secure. When we include additional controls, we find that there is no significant difference in children's dietary diversity between households that are food secure and mildly food insecure. However, children from households that are moderately or severely food insecure are more likely to have lower diet diversity scores that children from households that are food secure. Our finding that household food security is significantly associated children's diet diversity is a robust result Household food security and child anthropometry In order to understand the extent of association between household food security and child nutrition outcomes, we explore three separate logit models where the outcome variables relating to each child are stunted (equals one if stunted, else zero), underweight (equals one if underweight, else zero), and wasted (equals one if wasted, else zero). We find that children from severely food insecure households have a 53% higher odds of being stunted (AOR 1.528; 95% CI ) and a 64% higher odds of being wasted (AOR 1.640; 95% CI ) than children from food secure households. Similarly, children from moderately food insecure households have a 67% higher odds of being underweight (AOR 1.674; 95% CI ) than children from food secure households while children from severely food insecure households have a 91% higher odds of being underweight (AOR 1.912; 95% CI ) than children from food secure households (Table 3). Although estimating the logit we did not make a distinction between children who were stunted and severely stunted, wasted and severely wasted, and underweight, severely underweight. TABLE 2 Children's diet diversity (seven food group score) and household food insecurity estimates from negative binomial regression model Coefficient Model 1 Model 2 Robust standard error p value Coefficient Robust standard error p value Household food security: reference food secure Mildly food insecure ( 0.136, 0.007) ( 0.104, 0.026) Moderately food insecure ( 0.262, 0.068) ( 0.173, 0.006) Severely food insecure ( 0.316, 0.133) ( 0.198, 0.010) N = 1997, 1979 in Model 1 and 2 respectively Sample restricted to children aged 6 23 months Figures in parenthesis are the 95% confidence intervals Model 1: Only dummies reflecting household's food insecurity are included as explanatory variables. Model 2: In addition to the variables in Model 1 we include the following variables: Place of Residence (rural, slum, nonslum urban area), Access to PDS (without a card and those with a white, orange, or a yellow (antyodaya) card), gender of child (boy or girl), age of child, square of age of child, social group (Scheduled Tribe, Scheduled Caste, OBC, VJ/NT, and others), Religion (Hindu, Muslim, Others), wealth index (dummy variables for the five quintiles), mother's education (no education, <5 years, 5 7 years, 8 9 years, years, > 12 years), mother's age at time of birth of child, region (place) of residence (Amravati, Aurangabad, Konkan, Nagpur, Nashik, Pune)

5 5of8 TABLE 3 Association between household food insecurity and stunting, underweight, and wasting estimates from a logit model Odds ratio Standard error p value Stunting (HAZ) Household food security: reference food secure Mildly food insecure (0.888, 1.520) Moderately food insecure (0.928, 1.718) Severely food insecure (1.142, 2.045) Underweight (WAZ) Mildly food insecure (0.980, 1.713) Moderately food insecure (1.228, 2.283) Severely food insecure (1.423, 2.571) Wasting (WHZ) Mildly food insecure (0.865, 1.711) Moderately food insecure (0.864, 1.887) Severely food insecure (1.149, 2.341) N = 1997 Recognizing that household food security could affect the severity of the problem of undernutrition, we also estimated a multinomial logit model where the dependent variable takes three plausible values for each of the outcomes: stunted (equals two if severely stunted, one if stunted, else zero), underweight (equals two if severely underweight, one if underweight, else zero), and wasted (equals two if severely wasted, one if wasted, else zero). Our key findings can be summarized as follows. First, in households that are moderately food insecure or severely food insecure the odds of children being severely stunted are higher and statistically significant. Second, a child from a food insecure household, irrespective of whether the household is mildly, moderately or severely food insecure, is more likely to be underweight. Third, children from severely food insecure households are more likely to be wasted. The detailed results are available on request Children's diet diversity and child anthropometry We next explore the association of diet diversity (values zero seven) with stunting, underweight, and wasting. The outcome variables are coded as categorical variable. We find children's diet diversity score to be an important correlate of children's anthropometric indicators, namely, weight for age and weight for height (Table 4). This suggests that the higher the diet diversity of the child the lower is the odds of the child being underweight or wasted are lower. We do not find any significant association between child diet diversity and stunting (Table 4). In order to understand whether this result is robust, instead of logit model, we estimate a multinomial logit model where we group children into three categories: not stunted, stunted, and severely stunted; not wasted, wasted and severely wasted; and not underweight, underweight and severely underweight (Results available on request). We find that children with a more diverse diet are less likely to be underweight or severely underweight or wasted Household food security, children's diet diversity and child anthropometry When we include household food security, children's diet diversity, and other standard control variables in the empirical model, we find that household food security is no longer a statistically significant predictor of stunting, underweight, or wasting. However, unlike in the parsimonious specification where we did not find an association between children's diet and stunting, we do find that children with a more diverse diet are less likely to be stunted or underweight (Table 5). 4 DISCUSSION UNICEF's conceptual framework for nutrition recognizes that household food insecurity and inadequate dietary intake have important implication for child undernutrition. In our analysis, we find that even after controlling for household, maternal and child characteristics, TABLE 4 Impact of children's diet diversity (seven food group score) on child anthropometry: estimates from logit model Coefficient (odds ratio) on diet diversity 95% confidence interval Standard error p value Stunting (HAZ) (0.911, 1.071) Underweight (WAZ) (0.785, 0.935) Wasting (WHZ) (0.802, 0.993) N = 2007 Sample restricted to children aged 6 23 months

6 6of8 TABLE 5 model Association between children's diet diversity household food insecurity and stunting, underweight, and wasting estimates from a logit Odds ratio Standard error p value Stunting (HAZ) Seven food group score (Children's diet diversity) (0.750, 0.931) Mildly food insecure (0.623, 1.159) Moderately food insecure (0.692, 1.370) Severely food insecure (0.640, 1.280) Underweight (WAZ) Seven food group score (Children's diet diversity) (0.734, 0.917) Mildly food insecure (0.750, 1.370) Moderately food insecure (0.899, 1.765) Severely food insecure (0.855, 1.665) Wasting (WHZ) Seven food group score (Children's diet diversity) (0.822, 1.065) Mildly food insecure (0.690, 1.408) Moderately food insecure (0.674, 1.534) Severely food insecure (0.801, 1.744) N = 1953 In addition to the child's diet diversity score and household's food insecurity, we include the following variables: Place of residence (rural, slum, nonslum urban area), access to PDS (without a card, and those with a white, orange or a yellow [antyodaya] card), gender of child (boy or girl), age of child, square of age of child, social group (Scheduled Tribe, Scheduled Caste, OBC, VJ/NT, and others), religion (Hindu, Muslim, and others), wealth index (dummy variables for the five quintiles), mother's education (No Education, <5 years, 5 7 years, 8 9 years, years, > 12 years), Mother's age at time of birth of child, region (place) of residence (Amravati, Aurangabad, Konkan, Nagpur, Nashik, Pune) children from moderately food insecure and severely food insecure households are more likely to have lower diet diversity scores. Our results are similar to Saha et al. (2008) who found that in Bangladesh household food security status was associated with better infant feeding practices among children of 6 12 months of age. Ali et al. (2013) found a strong association between household food security and minimum diet diversity of children in Bangladesh, Ethiopia, and Vietnam. In general, studies have found that as the level of household food insecurity increases, children are less likely to achieve the minimum diet diversity. Our findings in India are consistent with this pattern that has been documented in the literature. Next, we explored the impact of household food insecurity and children's diet separately on nutrition outcomes. We find that children from severely food insecure households are more likely to be stunted, underweight, or wasted. We also find that the odds of a child being severely stunted, severely underweight, or severely wasted are higher in severely food insecure households. Ali et al. (2013) in their sample from Bangladesh, Ethiopia, and Vietnam, found a robust association between food insecurity and stunting but not wasting. When examining the individual impacts of children's diet diversity independently of household food security status, we find that children with higher diet diversity scores are less likely to be underweight or wasted. When we control for children's diet diversity, and all other child, maternal and household characteristics, we find that household food security is no longer statistically associated with stunting, wasting, or underweight. This result of ours is unlike that of Ali et al. (2013) who do find household food security to be associated with stunting and underweight. However, our important finding is that children's diet matters for stunting and underweight. Although Saha et al. (2009) find a similar result to ours in the context of Bangladesh, they did not control for household food security. They find that diet of children matters for height for age. In the context of Bangladesh too, Zongrone, Winskell, and Menon (2012) find that higher diet diversity index reduces the probability of stunting and underweight among children. Analysing India's National Family Health Survey , Menon, Bamezai, and Subandoro (2015) find a strong statistical association between children's diet diversity and all three measures of child anthropometry. However, they are unable to control for household food insecurity in the absence of this variable in the National Family Health Survey. The issue of whether dietary diversity captures food security or dietary quality has been discussed in the literature (see Hoddinott & Yohannes, 2002; Ruel, 2002). But this issue has hardly been discussed in the Indian context. Such discussion on diets and household food security has been hampered by the absence of data from official sources. This serious data gap needs to be bridged and the CNSM provides evidence of feasibility and utility of such an exercise. Our results although not causal provide the evidence base for understanding the extent to which household food insecurity affects children's diet diversity and how both these factors affect nutrition outcomes in children. Our aims to inform Government of Maharashtra's efforts to formulate appropriate policies as part of the

7 7of8 next phase of the State Nutrition Mission. It also informs the efforts by the National Nutrition Mission in addressing India's nutrition challenge.. In Maharashtra, 42 per cent of children live in households that cannot be classified as food secure. This pattern is likely to be true in other Indian states too. In light of the fact that four out of every 10 households are food insecure in Maharashtra, the PDS, which provides subsidised food, could have an important role to play in tackling household food insecurity, particularly in households below the poverty line. Tackling food insecurity is the first step towards improving diets and, in this regard, it is heartening that India's National Food Security Act (NFSA) 2013 seeks to cover up to 75 per cent of rural and 50 per cent of the rural and urban population respectively under the PDS. In a recent contribution, Rahman (2016) establishes a robust link between consumption of subsidized food grains from PDS and nutrient intake. He argues that that the savings resulting from purchases at PDS shops where prices are lower as compared to open market prices could be spent by households on nutritious foods. He provides empirical support for this conjecture by analysing consumption expenditure patterns of households from hunger prone districts in Odisha in eastern India. He finds that making PDS accessible to all households in these districts led to a significant reduction over time in the proportion of households consuming below the recommended dietary allowance of calorie, fats and protein (p. 73). Further, there are specific provisions in NFSA aimed at improving diets of households. Schedule II of NFSA states that the nutritional standards for children in the age group of 6 months to 3 years, age group of 3 to 6 years and pregnant women and lactating mothers will be met by providing take home rations or nutritious hot cooked meals through the Integrated Child Development Services programme while children in lower and upper primary classes will benefit from the Mid Day Meal programme in the schools (Government of India, 2013). The nutritional quality of the foods delivered by these programmes (i.e., are they contributing to meet the age specific nutrition needs of these population groups?) and the effective coverage of these social protection programmes (are they delivering timely and quality services to the population groups that are most vulnerable from a nutritional point of view?) will be of the essence. There are limitations in examining the association between household food insecurity and children's diets and anthropometry using cross sectional data. One limitation is that we are unable to identify temporal variations in household food security and the resultant implications for children's diet and nutrition outcomes. Furthermore, we hardly have any understanding of the dynamics in households that experience food insecurity for longer time and households experiencing food insecurity at the time of the survey. This is in our view an important area for future work. ACKNOWLEDGEMENT This paper benefitted from comments received at the, First Foods: A Global Meeting to Accelerate Progress on Complementary Feeding for Young Children, November 2015, Mumbai India. We are grateful for comments received from two anonymous referees of this journal. The views expressed in this article are personal and do not represent those of the organisation to which the authors belong. SOURCE OF FUNDING Chandrasekhar acknowledges support from Rajmata Jijau Mother Child Health and Nutrition Mission, Government of Maharashtra. CONFLICTS OF INTEREST The authors declare that they have no conflicts of interest. CONTRIBUTIONS S Chandrasekhar, Víctor M. Aguayo, and Rajalakshmi Nair discussed the scope of the paper and the empirical framework for the analysis. Chandrasekhar conducted the statistical analysis. Chandrasekhar and Víctor M. Aguayo wrote the manuscript. Rajilakshmi Nair and Vandana Krishna reviewed the manuscript. REFERENCES Ali, D., Saha, K. K., Nguyen, P. H., Diressie, M. T., Ruel, M. T., Menon, P., & Rawat, R. (2013). Household food insecurity is associated with higher child undernutrition in Bangladesh, Ethiopia, and Vietnam, but the effect is not mediated by child dietary diversity. The Journal of Nutrition, 143(12), Arimond, M., & Ruel, M. T. (2004). Dietary diversity is associated with child nutritional status: Evidence from 11 demographic and health surveys. The Journal of Nutrition, 134(10), Chandrasekhar, S., & Pandey, V. L. (2015). Food sufficiency in India: Addressing the data gaps. Economic and Political Weekly, Vol 50 No.(9), February 28, Government of India (2013). The National Food Security Act, 2013, Ministry of Law and Justice, Government of India Hoddinott, J., & Yohannes, Y. (2002). Dietary diversity as a food security indicator, Food Consumption and Nutrition Division Discussion Paper No. 136, International Food Policy Research Institute International Institute for Population Sciences (IIPS) and ORC Macro (2000). National family health survey (NFHS 2), : India. Mumbai: IIPS. International Institute for Population Sciences and Macro International (2008). National Family Health Survey (NFHS 3), India, : Maharashtra. Mumbai: IIPS. International Institute for Population Sciences and United Nations Children's Fund (2013). Comprehensive nutrition survey in Maharashtra (CNSM) Mumbai: IIPS. Jones, A. D., Ickes, S. B., Smith, L. E., Mbuya, M. N. N., Chasekwa, B., Heidkamp, R. A., Stoltzfus, R. J. (2014). World Health Organization infant and young child feeding indicators and their associations with child anthropometry: A synthesis of recent findings. Maternal and Child Nutrition, 10, Menon, P., Bamezai, A., & Subandoro, A. (2015). Mohamed Ag Ayoya and Victor Aguayo (2015) Age appropriate infant and young child feeding practices are associated with child nutrition in India: Insights from nationally representative data. Maternal and Child Nutrition, 11, Rahman, A. (2016). Universal food security program and nutritional intake: Evidence from the hunger prone KBK districts in Odisha. Food Policy, 63, August Ruel Marie T. (2002). Is dietary diversity an indicator of food security or dietary quality? A review of measurement issues and research needs, Food Consumption and Nutrition Division, Discussion Paper No. 140, International Food Policy Research Institute. Saha, K. K., Frongillo, E. A., Alam, D. S., Arifeen, S. E., Persson, L. A., & Rasmussen, K. M. (2008). Household food security is associated with infant feeding practices in rural Bangladesh. The Journal of Nutrition, 138(7), Saha, K. K., Frongillo, E. A., Alam, D. S., Arifeen, S. E., Persson, L. A., & Rasmussen, K. M. (2009). Household food security is associated with

8 8of8 growth of infants and young children in rural Bangladesh. Public Health Nutrition., 12(9), United Nations Children's Fund (2015). UNICEF approach to scaling up nutrition for mothers and their children UNICEF New York: Programme Division, UNICEF. Zongrone, A., Winskell, K., & Menon, P. (2012). Infant and young child feeding practices and child undernutrition in Bangladesh: Insights from nationally representative data. Public Health Nutrition., 15(9), How to cite this article: Chandrasekhar S, Aguayo VM, Krishna V, Nair R. Household food insecurity and children's dietary diversity and nutrition in India. Evidence from the comprehensive nutrition survey in Maharashtra. Matern Child Nutr. 2017;13(S2):e