Changes in Child Undernutrition and Overweight in India From 2006 to 2021: An Ecological Analysis of 36 States

India has historically displayed high levels of child stunting and low levels of child overweight. Despite improvements in human development indicators between 2006 and 2021, population-level reductions in child stunting have slowed and child overweight is rising faster than predicted by human development indicators.


INTRODUCTION
S ince the first cross-national studies of child growth, India has consistently ranked among the countries with the highest levels of growth faltering. 1-4 India's pernicious child undernutrition problem has been attributed to a complex interaction among structural factors such as inadequate food systems; poor water, sanitation, and hygiene infrastructure; and household poverty, which ultimately determine individual dietary intake and morbidities that impact child nutrition. 5,6 Since 2005, several ambitious national-level development programs have targeted many of these structural factors through income, education, sanitation, and direct nutrition campaigns. Between 2006 and 2016, nationwide child stunting decreased by 10 percentage points while child overweight increased. 7, 8 These shifts in undernutrition and overweight occurred across the socioeconomic spectrum. 7, 8 Between 2015 and 2019, several new nationallevel campaigns were launched to improve living conditions such as sanitation (Swachh Bharat Mission), clean cooking fuel (Pradhan Mantri Ujjwala Yojana), electrification, and health (Ayushman Bharat-PMJAY). 9 Simultaneously, the Indian economy experienced shocks such as currency demonetization (November 2016), which worsened the slower economic growth of the 2010s relative to the previous decade and may have adversely affected some districts more than others. 10 Whether and how this dynamic development policy landscape between 2015 to 2019 impacted trends in child undernutrition and child overweight is not clear.
We conducted state-and district-level analyses of nationally representative surveys of children to evaluate changes in child growth indicators across survey rounds conducted in 36

METHODS
We analyzed data from 3 cross-sectional surveys conducted over 15 years in India. We tracked changes in child undernutrition, overweight, and household development indicators. Some of the indicators listed are used in monitoring India's progress toward achieving the Sustainable Development Goals. 11

Data Sources
We used individual-level data from the National Family Health Surveys (NFHS) conducted from 2005 to 2006 (Round 3), 2015 to 2016 (Round 4), and 2019 to 2021 (Round 5). We refer to the NFHS-3, NFHS-4, and NFHS-5 surveys as 2006, 2016, and 2021, respectively, corresponding to the last year of data collection. NFHS-4 and NFHS-5 employed a multistage sample survey design to generate samples that are representative at the district, state, and national levels for children aged younger than 5 years and adults (women aged 15-49 years, men aged 15-54 years). NFHS-3 was a multistage sample survey that represented state and national levels. Rural villages or urban census enumeration blocks served as sampling clusters, from which households were randomly sampled. NFHS-3 (November 2005-August 2006) surveyed 109,041 households in 28 states and Delhi. 12 NFHS-4 (January 2015-December 2016) surveyed 601,509 households in 29 states, 7 union territories, and 640 districts. 13 Phase 1 (June 2019-March 2020) of NFHS-5 surveyed 307,554 households in 17 states, 5 union territories, and 341 districts before the first wave of coronavirus disease (COVID-19). 14 Phase 2 (November 2020-April 2021) of NFHS-5 surveyed 329,145 households in 11 states, 3 union territories, and 366 districts. Additional information on changes in geographic boundaries between rounds is provided in Note 1 and Figure 1 in Supplement 1. We used shapefiles provided by the Demographic and Health Surveys program (www.dhsprogram.com) to classify NFHS-5 primary sampling units into NFHS-4 districts.

Child Growth Outcomes
State-and district-level prevalence of child growth indicators were treated as the outcomes. All outcomes are expressed as the percentage of children aged younger than 5 years exhibiting the indicator. Stunting was defined as height-for-age z-score < À2 standard deviations (SD) from the median based on World Health Organization growth standards. Underweight was defined as weight-forage z-score < À2 SD. 15 Wasting was defined as weight-for-height z-score < À2 SD. 15 Overweight or obesity in children was defined as weight-forheight z-score > 2 SD. 15 We excluded those children with extreme or unavailable values for any of height-for-age z-score (<À6 or >6), weightfor-age z-score (<À6 or >5), and weight-forheight z-score (<À5 or >5) as per World Health Organization standards. Additional information on measurement protocol and quality control is provided in Note 2 of Supplement 1.

Human Development Indicators
We focus on 2 domains of human development: human capital and standard of living. Human capital was measured by 4 indicators (percentage of women who are literate, percentage of women with 10 or more years of schooling, percentage of female children at birth, and percentage of women aged 20-24 years who married or first cohabited at age 18 or later), and standard of living was measured by 5 indicators (households with improved sanitation, improved drinking water, electricity, clean cooking fuel, or covered by any health insurance scheme).

Statistical Analysis
We conducted an ecological analysis of surveyweighted estimates derived from individual-level We conducted state-and districtlevel analyses of nationally representative surveys of children to evaluate changes in child growth indicators across 36 states and union territories.
data of child malnutrition and development indicators for 36 states and 640 districts (2015 boundaries) at the state and district levels, respectively. We estimated annualized state-level change between surveys as average absolute change in percentage points per year. We tested for difference in statelevel annualized change between 2006 and 2016 and 2016 and 2021 using a paired Wilcoxon signedrank test. We also estimated district-level changes in growth indicators between 2016 and 2021.
We benchmarked changes in growth indicators according to targets set by POSHAN Abhiyaan (National Nutrition Mission), India's nationally coordinated flagship nutrition campaign launched in March 2018. 16,17 POSHAN Abhiyaan announced targets to reduce stunting and underweight by 2 percentage points per year between 2019 and 2022. 18,19 In the absence of a national target, we assumed a target of 2 percentage points per year for wasting.
In addition, we examined expected changes in child growth indicators from 2016 to 2021 based on changes in the set of human development indicators by estimating a 2-way Blinder-Oaxaca decomposition (unweighted, bivariate). In the bivariate decomposition, the expected change reflects the change in the child growth indicator that is predicted ("explained") by the human development indicator. The difference between the expected and observed changes reflects the change in the nutrition outcome that was not predicted ("unexplained") by the human development indicator. For example, a smaller observed reduction in an undernutrition indicator relative to the expected reduction suggests unrealized improvements that were anticipated based on secular trends predicted by changes in human development. Our primary measure of human development was a weighted composite score of 9 indicators of district-level human development that was derived using principal component analysis of these indicators. In addition, we report the decomposition findings for each of the human development indicators individually. All analysis was conducted using R version 3.6.1 and Stata version 13.

Ethical Approval
The study was determined to be human subjects exempt research by the Emory University Institutional Review Board.

Child Growth Outcomes
From 2006 to 2021, the population-weighted prevalence of stunting declined from 47.8% to 35.5% (12.3 percentage point reduction) and that of underweight declined from 42.4% to 32.1% (10.3 percentage point reduction) (Table 1). Over the same 15-year period, the prevalence of wasting was unchanged (about 20%) and that of overweight increased from 1.5% to 3.4% (1. . Annualized rates of change for stunting and overweight were significantly different between the 2 periods (P<.01) at the state level. As of 2021, the prevalence of stunting, wasting, and underweight for the nation exceeded the 2022 POSHAN Abhiyaan targets for stunting, wasting, and underweight by 11 percentage points, 10 percentage points, and 8 percentage points, respectively (Table 1).
At the state level in rural settings, we observed slowing reductions (mean difference in annualized rates) in stunting (0.    Child stunting was defined as height-for-age < À2SD, child wasting was defined as weight-for-height < À2SD, child underweight was defined as weight-for-age < À2SD, and child overweight was defined as weight-for-height > 2SD. The modeled differences between the observed and expected district-level changes in child growth indicators associated with changes in human development between 2016 and 2021 are shown in Table 3. Changes in the human development score predicted statistically significant reductions in all 3 measures of undernutrition. The observed reductions fell short of the predictions for stunting and underweight but were larger than predicted for wasting. Taking stunting as an example, given the change in the development score from 2016 to 2021, we would predict a 5.1 percentage point (95% confidence interval [CI]=À5.9, À4.3) reduction in stunting. Instead, the prevalence of stunting changed by just À2.5 percentage points, so that observed levels of stunting were 2.6 percentage points (95% CI=1.8, 3.4) higher than what we would expect based on changes in human development. In contrast to predicted undernutrition outcomes, the human development score predicted a modest increase in overweight (0.6 percentage points; 95% CI=0.4, 0.7). Observed changes in overweight exceeded this prediction by 1.1 percentage points (95% CI=0.8, 1.4).
With respect to individual development indicators, 8 of 9 predicted statistically significant reductions in 1 or more child undernutrition indicators and 7 of 9 predicted significant increases in overweight. For many development indicators, reductions in undernutrition were smaller than expected while the rise in overweight was larger than expected. Our findings also indicate that India is unlikely to achieve the national targets set by POSHAN Abhiyaan for stunting and underweight of 26% and 24%, respectively, by the end of 2022. 17,20 While there is no national target for the prevalence of child overweight in 2022, the projected prevalence in 2022 of nearly 4% is closer to the average among low-and middle-income countries in 2017 (6.0%). [21][22][23] Given that the rise in the proportion of overweight children is expected to offset any reductions in undernutrition, the total prevalence of unhealthy weight among children in India will likely remain unchanged. 24,25 Mirroring the national scenario, in both rural and urban areas, we found that progress in undernutrition slowed, while overweight increased. The slowing progress in undernutrition occurred against the backdrop of improvements in most development indicators from 2016 to 2021. Our findings highlight a potentially worrisome disconnect between the apparent expansion of human development and improvements in child nutrition outcomes. 26 Specifically, the bivariate decomposition analysis predicted moderate to large declines in populationlevel stunting and wasting based on the rise in favorable development indicators such as women's literacy, improved sanitation, and households with electricity. Our analyses showed that these  Child stunting was defined as height-for-age < À2SD. Child wasting was defined as weight-for-height < À2SD. Child underweight was defined as weight-forage < ÀSD. Child overweight was defined as weight-for-height > 2SD. All values are coefficient (bootstrapped 95% confidence interval). b Difference between observed and expected change between 2016 (NFHS-4) and 2021 (NFHS-5). Contributions are modelled through 2-way Blinder-Oaxaca decomposition with each indicator entered separately. Equal weights (w=0.5) were applied for both survey rounds. The decomposition model gives each of the 640 districts analyzed equal weight, and district grand mean will not sum to the state population-weighted mean.
predicted reductions in undernutrition did not occur. Moreover, overweight is rising at a rate that is greater than what would be expected by improvements in human development observed in the past 5 years. At a national level, we found that the prevalence of stunting decreased by 0.6% between 2016 and 2021, with nearly 15% of districts experiencing a 5 percentage points or larger rise in the prevalence of stunting despite improvements in development indicators. In other settings, the impact of nutrition-sensitive and nutrition-specific interventions on child growth outcomes has been mixed, with some studies reporting reductions 27,28 and others reporting no change 29 in child stunting. The observed slowing declines in child stunting also paralleled a reversal in the decline in annual infant mortality rates, 30 suggesting alarming trends in nutrition-sensitive child health outcomes. Targeted monitoring of nutrition outcomes and careful investigation of district-level heterogeneity in trends may be informative to develop locally tailored strategies to more effectively address the lingering burden of child stunting, especially in areas that are worsening.
From 2016 to 2021, there was virtually no national-level change in child wasting, a reflection of current deficits in tissue and fat mass. The stagnation in wasting may signal entrenched insufficient access to diets required for optimal growth among children at the population level. It must be noted, however, that the prevalence of wasting is sensitive to seasonal variations in food availability and infectious disease. 31 Given the differential timing of data collection across NFHS survey rounds, it is possible that the observed stagnation in prevalent wasting is an artifact of seasonality. The rapid rise in child overweight presents a challenge to India. Cohort studies from low-and middle-income countries suggest that childhood overweight persists into adulthood, 32,33 with implications for risk of diabetes and cardiovascular disease. Moreover, South Asians as a group are at higher risk for cardiometabolic disorders at earlier ages and lower body mass index. 34 Even in children aged 5 to 19 years, metabolic abnormalities such as high triglycerides are present at all levels of weight status, with progressively higher levels of metabolic morbidities with increasing body mass index-for-age. 26 Rising overweight in childhood combined with the elevated risk of metabolic diseases in ethnic South Asians creates a perfect storm for higher morbidity, mortality, and medical costs due to cardiometabolic disease over the life course.

Strengths and Limitations
Overall, our analysis has several strengths. We used the latest available data to evaluate and benchmark ongoing efforts to reduce child malnutrition in India and set priorities for global reductions in child undernutrition. The application of decomposition analysis permitted us to explore the hypothetical reductions that should have been achieved based on improvements in human development indicators in the population. However, our analyses must be interpreted in light of data limitations. NFHS-5 data was collected partly after the first wave of COVID-19 lockdowns in 14 states. The present ecological analysis did not separate the trends in child health due to economic changes before the pandemic and trends due to the disruption to the economy and society during the lockdowns from the first wave of COVID-19. Restricting our analysis to the 22 states and union territories that included only pre-pandemic data, we found similar patterns for stunting, wasting, and overweight, and slowing of reductions of underweight (data not shown). In addition, precision of outcome estimates for the ecological analysis varied by survey round, as seen for NFHS-3 and NFHS-4 (Supplement 2). We did not explore the association of changes in micronutrient supplementation and of provision of supplementary nutrition with changes in malnutrition outcomes since this was beyond the scope of our analysis. 35 Finally, our decomposition analysis described associations between human development indicators and child nutrition measured over the same period. In doing so, we were not able to evaluate a lag between human development and improvements in child undernutrition.

CONCLUSIONS
In light of the unprecedented economic and psychosocial strain resulting from the COVID-19 pandemic, attention to child nutrition will be even more important in the coming years. 36,37 India's economy shrunk for the first time in between the periods of NFHS-5 data collection due to the nationwide lockdowns. As a result, many nutritionsensitive programs were affected, which may have resulted in a disruption of the trend that we were reporting on and may have contributed to higher undernutrition. Projections show that in the absence of additional social safety net programs to tackle mounting food insecurity and financial distress incurred through the COVID-19-related economic shutdowns, there may be a doubling of the number of wasted and stunted children in South Asia alone by 2022. 38  burden of child undernutrition, such as the urban and rural poor, also tended to be the most vulnerable to pandemic-related disruptions. 39 In these circumstances, careful investigation of the impacts of the pandemic on nutritional indicators is an urgent priority, with a particular focus on the most marginalized and vulnerable. 40,41 In summary, the observed slowing progress in measures of child undernutrition warrants further exploration when considered against apparent indicators of progress. Further investigation is prudent to determine whether inequitable human development across segments of the population or broader social and economic shocks-such as demonetization or the COVID-19 pandemic-related disruptions-contribute to slowing rates of reductions in undernutrition. Furthermore, investigation of outcomes beyond anthropometric growth, such as cognition, would provide a more comprehensive picture of the impact of changes in socioeconomic conditions on child social and developmental outcomes.