The Coexistence of Dual Malnutrition At Household Level in India: Evidence From The Comprehensive National Nutrition Study 2016–18

Objective To examine the prevalence of DBM at national, state, regional level and its proximate determinants in households with a mother and child under 5 years of age. Methods The present study includes data on 38,060 children under 5 years of age and their biological mothers, drawn from the nationally representative Comprehensive National Nutrition Survey (CNNS) of children and adolescents aged 0–19 years in India. The outcome variable for this study was the double burden of malnutrition at the household level. A child was classied as stunted if the height was at least -2SD below the mean for their age. A mother with a body mass index (BMI) of more than or equal to 25kg/m2 (BMI ≥ 25 kg/m2) was considered as overweight. A binary variable with coexistence of both conditions was created and categorized as 1 (existence of stunted child and overweight mother), and 0 otherwise. Multivariate regression analysis was used to examine the association between DBM and covariates at maternal, child and household level. between the DBM and covariates. Adjusted odds ratios with 95% condence interval (CI) were estimated to assess the strength of the association of DBM with predictors and covariates. In view of the complexity of the survey design, sampling weights were used and accounted for stratication and clustering in the sample design to estimate the proportion and 95% CIs for estimates. All the analyses were performed using the Stata software version 16.0 (STATA Corp LP, College Station, USA).

Previous studies reported the co-existence of these two forms of malnutrition within the same country [16][17][18], among same household members [5,19,20], among mother-child pairs [3,4,21,22] or within same individual [23,24]. However, the co-existence of these two forms of malnutrition within the same household is not usually considered in the design and implementation of nutrition interventions.
Interventions designed to reduce undernutrition contradicts the obesity prevention programme and vice versa. Supplying food to reduce undernutrition and recommending reduced fat diet to prevent obesity in household will have adverse effects on overweight and underweight members of the same household [19]. To address both forms of malnutrition, nutritious food and a healthy lifestyle is essential, as both over-and undernutrition have equally devastating effects on the future health of individuals as well as society [9,10].
In India, malnutrition is found to be one of the major contributors to the disease burden among children under 5 years of age. Of the 1.04 million deaths of children under 5 years of age in India in 2017, 68.2% could be attributed to malnutrition [25]. The proportion of deaths attributable to malnutrition in India has been persistently high and changed only slightly from 70.8% to 68.2 % during 1990-2017 [25]. While child malnutrition continues to be high in India, adult overweight and obesity have also shown an increasing trend over the past few decades [26], possibly because of a rise in sedentary lifestyles and more energy dense, fat and animal based dietary habits [19]. Recent studies document a 21% overall prevalence of overweight and obesity among women [27]. With persistent child undernutrition and growing adult overweight and obesity, India is now facing a complex situation-that of the double burden of malnutrition. At the household level, the double burden of malnutrition (DBM) can be de ned as the coexistence of a stunted child along with an overweight mother within the same household [3].
Previously, a study conducted across 131 developing countries reported that DBM at the household level ranged from 1.8 % to 15.9% [28]. Recent studies in bordering countries like Nepal and Bangladesh reported prevalence of DBM at the household level as 2-4% [29,30]. In India, several studies have documented DBM at the household level but are limited to certain geographies [21,31]. A study, using the National Family Health Survey (NFHS) 2015-16 data, highlights prevalence of DBM at the household level, but no state or region-speci c variations were documented [32]. The present study aimed to ll this gap in literature by documenting prevalence of DBM at the household level among mother-child pairs for at the country, state and regional level and further examined the factors associated with it.
With the aim to meet the UN Sustainable Development Goal (SDG) of eliminating malnutrition by 2030 [33], in 2017, the Indian government released the National Nutrition Strategy with measures to address malnutrition across the lifecycle [34]. In 2018, the Indian government launched the Poshan Abhiyaan programme with an overarching goal of reducing child and maternal malnutrition [35].The ndings of this study can play an important role in understanding the factors associated with DBM and the variations in the prevalence of DBM across the country. These ndings can further help device a comprehensive nutrition policy focusing on both forms of malnutrition.

Dataset and sample covered
This study used nationally representative data from the CNNS that was conducted under the aegis of the Ministry of Health and Family Welfare (MoHFW) in collaboration with UNICEF and the Population Council, India with approval from the National Statistical Commission. CNNS was designed to provide nationally representative and comprehensive nutritional pro ling of preschoolers (0-4 years), school-age children (5-9 years) and adolescents (10-19 years

Study sampling and Participants
CNNS used a multi-stage strati ed sampling design to obtain a nationally representative sample of households and individuals aged 0-19 years across 29 states and the capital Delhi. In each state, the sample was selected in two stages. In the rst stage, PSUs were selected using probability proportional to size (PPS) sampling. The second stage involved a systematic random selection of households within each PSU. In large PSUs, the sampling design involved three stages, with the addition of a segmentation procedure to reduce enumeration areas to manageable sizes. Children who had a chronic illness, physical deformity, mental illness, cognitive disability, or an ongoing current illness (high fever, infection) were not included in the study. The overall study design and methods have been described elsewhere [36]. For children 0-4years, caregivers (mostly the mother) were interviewed, and anthropometric measurements of eligible children and biological mothers were recorded. Weight was recorded in kilograms and height in centimetres using a digital SECA scale and a three-piece wooden height/length board, respectively. Before each measurement, the instrument was set up on a portable wooden square and spirit-level was used to ensure even measurement surface. Recumbent length was measured in children who were either less than 2 years of age or 85 cm in height and for the rest, standing height was measured [26]. The present study included data from 38,060 children aged 0-4 years and their 35,452-biological mothers. The procedure of arriving at nal the analytical sample size is described in Table 1. Outcome measures for nutritional status of children and their mothers According to the World health organization (WHO), the double burden of malnutrition is the coexistence of undernutrition along with overweight, obesity or diet related non -communicable diseases, within individuals, households and populations and across the life-course [37]. The outcome variable of this study-double burden of malnutrition at household level-is de ned as mother-child pair, with mother as overweight or obese and child as stunted. The WHO Anthro-Plus software was used to calculate Z-scores (standard deviation scores) for height-for-age of the child. A child was classi ed as stunted if the height was at least two standard deviation (< 2SD) below the mean for their age. A mother was considered as overweight if her BMI was more than or equal to 25kg/m2 (BMI ≥ 25 kg/m2). A binary variable with coexistence of both the conditions was created and categorized as 1 (existence of stunted child and overweight mother), and 0 otherwise.

Covariates
Based on UNICEF's conceptual framework of factors associated with malnutrition[38] and other relevant literature on overweight mothers and stunted children [3,19,21,28,29], a variety of socio-demographic characteristics at child, mother and household level were included in the analysis. Child covariates included age categorized as less than 24 months and 24 months to 59 months. Sex of child as dichotomous variable was categorized as male or female. Birth order of child was classi ed as 1st to 3rd order and 4th or more. Breastfeeding status denoted that the child was ever breastfed and was categorized as yes, or no. Morbidity was classi ed as yes if the child suffered from diarrhea, acute respiratory infection, or fever in the 2 weeks prior to the survey.
Maternal covariates included age categorized as less than 20 years, 20-24 years, 25-29 years, and 30 or more years. Maternal education was categorized as no education, completed primary schooling, Table 1 presents the analytical sample for this study. A total of 38,060 households with children aged 0-4 years were interviewed. Out of these households, in 38,454 households' biological mothers were interviewed. Considering the exclusion of implausible values and missing cases, the nal analytical sample for this study was 31,334. Figure 1 shows the percentage of households with stunted children, overweight mothers and DBM at the household level. Overall, 33.9% of children and 15.3% mothers were found stunted and overweight, respectively. The prevalence of stunting was found to be higher in rural areas (36.3% vs. 26.5%); whereas prevalence for both overweight mothers and DBM was higher in urban areas. Of the total number of households, 3.4% were found having DBM. Table 2 shows the maternal, child and socio-demographic characteristics of the study population among households with one mother-child pair. More than 30% of the mothers were less than 25 years of age and the same proportion had no formal education. About one-fourth of the mothers were working and 60% reported exposure to mass media. Of the total, 61% of the children were in the age group of 24-59 months and 4% were never breastfed. Half of the children were reported having morbidity in the two weeks prior to the survey. More than three-fourths of the households were in rural areas and more than one third of the households belonged to Scheduled Caste or Scheduled Tribe.

Results
completed secondary schooling, above secondary schooling. Working status was classi ed as working or not working. Type of delivery of the index child was categorized as normal or cesarean.
Finally, household level characteristics included wealth index, place of residence (urban or rural), caste, and exposure to mass media. Wealth index was computed using data on household's ownership of selected assets. The detailed description of methodology for wealth index calculation has been documented elsewhere [36]. Caste was classi ed as Scheduled Caste or Scheduled Tribe and Others including Other Backward Castes.

Statistical analysis
Descriptive statistics were used to describe the study variables. The association of DBM at the household level with socioeconomic and demographic characteristics was veri ed by the Pearson's χ2 test.
Multivariate logistic regression models were used to assess the association between the DBM and covariates. Adjusted odds ratios with 95% con dence interval (CI) were estimated to assess the strength of the association of DBM with predictors and covariates. In view of the complexity of the survey design, sampling weights were used and accounted for strati cation and clustering in the sample design to estimate the proportion and 95% CIs for estimates. All the analyses were performed using the Stata software version 16.0 (STATA Corp LP, College Station, USA).

Discussion
This study attempted to document the prevalence and factors associated with DBM at the national, regional, and state levels using a nationally representative dataset. DBM at the household level was reported as 3.4%, which is similar to previous studies conducted in India and other south Asian countries.
A study conducted using NFHS 2015-16 data reported 4% households with overweight mothers and stunted child [32]. Further, a study conducted in 42 countries based on the demographic health survey (DHS) reported the prevalence of DBM as 1%, 1.2% and 1.5% in Bangladesh, India and Nepal utilizing data collected in 2000, 1998 and 2001 respectively [21]. Recently, a study reported the prevalence of DBM as 4.2%, 1.5%, 3.9% and 5.5% in Bangladesh, Nepal, Pakistan and Myanmar, respectively [29]. Another study reported that 11% and 4% of households in rural Indonesia (2000-03) and Bangladesh (2003-06), respectively, were double burdened [39]. These results indicate that the prevalence of DBM is increasing over time, highlighted the urgency for targeted interventions and attention of policy makers. This present study further showcased the prevalence of DBM at the state and regional level. The prevalence of DBM varied from 0.5% in Jharkhand to 10.4% in Kerala. A previous study conducted in Kerala reported a similar prevalence of DBM at the household level as 10.7% [31].
This study found that the prevalence of stunting among children less than 5 years of age was 33.9%.
NFHS has reported 38% of stunted children across India ranging from 20% in Kerala to 48% in Bihar [40]. A study conducted in high burden pockets (districts) of India reported stunting as 37.9% in Madhya Pradesh to 70.7% in Jharkhand [41].This shows a huge variation in the prevalence of stunting in India at the state and district level. The prevalence of stunting was higher in rural areas (36.3%) than in urban areas (26.3%) and these ndings corroborated with the ndings from NFHS 2015-2016 [40]. This study also revealed that 15.3% of mothers were overweight or obese, whereas, NFHS 2015-2016 reported 21% of women as overweight or obese. This may be due the fact that the mothers of young children practice breastfeeding and therefore lose pregnancy weight [42].
This study further examined the maternal, child and household factors that are associated with DBM.
Maternal characteristics, including mothers' age, education, working status, caesarean delivery and mass media exposure were found to be associated with DBM. It was found that households with older mothers were at an increased risk of DBM as compared to those with younger mothers. This nding is in line with other studies [29,39] and a possible explanation can be that with increasing age, chances of obesity increase due to a sedentary lifestyle and reduced metabolic rates. Another study also reported that maternal weight has been negatively associated with breastfeeding [42,43]. Mothers' education level was negatively associated with the prevalence of DBM at the household level. The nding corroborated with previous studies conducted in Pakistan [29] and Indonesia [39]. However, few studies documented that the risk of obesity among educated mothers was higher and can be attributed to a sedentary lifestyle [44]. Study households with mothers who had c-section deliveries were found to be at a higher risk of DBM.
Studies suggest that obese mothers have a higher cesarean delivery rate [45]. In addition, children born via c-section delivery are less likely to have timely initiation of breastfeeding [46,47].
Child characteristics, including age, sex, birth order and morbidity were examined and were found to be associated with DBM. Household with older children in age group of 24-59 months were at higher risk of DBM. These nding were similar to the studies conducted in Pakistan, Myanmar, Indonesia, and Bangladesh [29,39]. One of the probable reasons can be that decreased breastfeeding and improper complementary feeding results in stunting among children age 24-59 [48]. Further, households with higher birth order of child were at increased risk of DBM. As the birth order increased so did the chances of a child being stunted, due to a preference for an older male child [49]. Further, with an increase in the number of live births, chances of maternal obesity increased by 7% per live birth [50].
The results of this study suggested that DBM was positively associated with higher wealth index of household and urban residence. Other studies in the South Asian region also reported the association of higher wealth index with DBM [19,29,39]and urban residence [29]. This is contradictory to the studies conducted in Latin American countries that report DBM association with lower wealth quintile and rural areas [51]. This contrast can be because of dietary habits. For instance, households in the higher wealth index had a higher consumption of energy dense food and aerated drinks. As the prevalence of obesity among women was high in urban areas [52], it contributed to the prevalence of DBM among urban households.
This study has several limitations. As this study used cross-sectional data, it limits the ability to draw causal inferences. Further, few important variables due to non-availability were not included in this study, such as physical activity and time use pattern of mothers to understand their lifestyles. In addition, genetic factors, which might have in uenced both mother and child nutritional statuses, have not been assessed in this study.
DBM at the household level indicated the dual nutrition challenge facing India [37]. Addressing this complex issue is a crucial step towards achieving SDG-2. With unprecedented investment and focus on nutrition through the Poshan Abhiyaan programme, customized intervention, programme and policy can help overcome DBM and eventually smoothen the way to achieve SDG-2. The nutrition policies should focus not only on the diet of children but also on that of the mothers. Most of the child health programs focus on weight gain; therefore, stunting continues to remain a problem. Focus needs to be put on recommending a nutritious diet to overcome stunting along with other forms of malnutrition which coexist in children. Mothers should also be recommended to have a nutritious diet and reduce energy-dense foods which lack nutrients. This will help reduce overweight and obesity and will also help in reducing other forms of malnutrition (micronutrient de ciency).

Conclusion
The study computed the prevalence of DBM at the household level and the determinants associated with it. The ndings of this study pose a challenge for policy makers and programmers in a rapidly developing and urbanizing country like India. While policy formulation for these two problems as standalone is relatively easier, however, in this case it cannot be assumed that households with malnourished children need more food. Issues of dietary quality, changes in individual dietary patterns and feeding behaviors need to be addressed. Customized policies and programmes are needed to combat the con icting situation of dietary excess and deprivation at the household level.

Declarations
Funding This study was funded by UNICEF

Con icts of interest/Competing interests
On behalf of all authors, the corresponding author states that there is no con ict of interest.

Ethics approval
The CNNS received ethical clearance from the Ethical Review Board of the Post Graduate Institute for Medical Education and Research (PGIMER) and the Institutional Review Board of the Population Council in New York.

Consent to participate
All aspects of the survey were informed to participants, following which written consent was obtained from caregivers of children aged 0-4 years.

Consent for publication
Not Applicable Availability of data and material Reasonable request for data used in this article may be made to the corresponding author.
Code availability (software application or custom code) Reasonable request for code used in this article may be made to the corresponding author.  Prevalence of stunted child, overweight mother and DBM at household level, India, CNNS 2016-18