Change in nutritional status among women of childbearing age in India (1998–2016)

Summary Introduction In absolute numbers, India has more undernourished people than all the countries in sub‐Saharan Africa combined. In parallel with the high rates of hunger and undernutrition, the country has been undergoing rapid demographic and dietary transition marked by an increased prevalence of overweight/obesity, particularly among women. Objective To measure the changing prevalence of overnutrition during last two decades, as well as to identify the associated sociodemographic correlates among pregnant and non‐pregnant women in India. Methods This was a cross‐sectional study based on data from the latest round of National Family Health Survey (2015–2016) conducted among urban and rural women. Participants were 687,876 women (655,850 non‐pregnant and 32,026 pregnant) aged between 15 and 49 years. Nutritional status was assessed in terms of body mass index (BMI) using the cut‐off for Asian population. Results Since 1998–1999, the prevalence of underweight has decreased by 9.2%, while that of overweight (BMI = 23–27.4 kg/m2) and obesity (BMI ≥ 27.5 kg/m2) has increased by 6.7% and 3.4%, respectively. Results of multivariable regression analysis revealed significant association between nutritional status and age, parity residency, educational level, religious affiliation, household wealth quintile, and TV watching behaviour. Of those, age and wealth status appeared to be the strongest predictors among both pregnant and non‐pregnant women. Conclusion Since 1998, there has been a considerable drop in the prevalence of underweight and rise in the prevalence of overweight and obesity. Significant sociodemographic variations exist in nutritional status, notably age and financial situation, which should be highlighted in national nutrition policymaking and intervention programmes.


| BACKGROUND
Maintaining adequate nutrition is central to good health outcomes and quality of life at individual level and a healthy workforce and sustainable socioeconomic development at national level. 1,2 No other aspect of life has as pervasive an impact on physical, psychosocial and overall well-being as does nutrition. As promoting nutritional status is embodied as a key prerequisite for achieving the Millennium and Sustainable Development Goals of the United Nations, addressing global hunger and undernutrition remains a key priority of national governments and international donors. 3,4 Considerable progress has been made by many countries in terms of reducing undernutrition since the introduction of MDGs in 1990. At the same time, overnutrition has gradually emerged to be an equally important public health concern especially in the countries traditionally characterized by high rates of undernutrition, for example, countries in sub-Saharan Africa and South Asia. According to WHO estimates, globally about 1.9 billion adults are now affected by excess nutrition, while 462 million are underweight. 5 Growing body of evidence from the medical literature suggests that overnutrition is no longer a phenomenon unique to the affluent economies, as the burden of overweight and obesity is reaching epidemic proportions especially in the fast developing countries such as China, India and Bangladesh. [6][7][8][9] Commonly recognized as a global hotspot for maternal and child undernutrition, India now accounts for the highest number of overweight and fifth highest number of people with obesity in the world. 10 In India, the prevalence for adult population living with excess body weight has been increasing steadily during last two decades, with the prevalence being particularly higher among women compared with men. 11 Being a major risk factor of non-communicable chronic diseases (NCDs), the increasing prevalence of overweight and obesity has duly translated to a dramatic increase in the burden of diseases such as diabetes, hypertension, cardiovascular diseases, to name a few. 7,8 While the rate of undernutrition has been declining at the same time, the changing epidemiological landscape with higher burden of overnutrition and NCDs is posing significant challenges for population health and healthcare system in the country. 12,13 Obesity is a multifaceted problem whose risk factors vary across and within countries and having wide-ranging medical and socioeconomic consequences. [12][13][14] There has been a growing volume of epidemiological studies exploring the root causes overnutrition in India. [15][16][17] The findings indicate that the shift from undernutrition to overnutrition is mostly attributable to the demographic transition involving population ageing, rapid urbanization and socioeconomic transition that is triggering changes in lifestyle behaviour and dietary patterns. This epidemiological shift, also known as nutrition transition, is a phenomenon that involves the coexistence of undernutrition and overnutrition, which is commonly referred to as the double burden of malnutrition. 6,18,19 From the perspective of public health situation in India, addressing the challenges posed by double burden of malnutrition is particularly difficult due to the complex demographic distribution of the problem. For example, the occurrence of obesity in mothers with stunted child in the same family requires an integrated household based rather than individual-based approach to nutritional interventions for the double burden. 20 From this view, tackling obesity among women is a key public health priority for countries like India as maternal obesity that is associated with a host of pregnancy and child health issues (e.g., preterm birth, low birthweight) with potential impacts on child's nutritional status in the later stages of life. 21 Addressing maternal obesity is facilitated by effective health policymaking based on the evidences from sociodemographic analysis of the burden of distribution of the issue. Based on this understanding, this study was conducted to provide an update on the prevalence of overnutrition among married women in both rural and urban areas of

| Study variables
The outcome variable for this study was nutritional status measured in terms of BMI. The outcome variable was overweight/obesity status, which generally results from excessive intake of nutrients, generating an energy imbalance between food consumption and energy expenditure. 24 NFHS collects anthropometric information, for example, height and weight for a subsample of households only. Weight was measured using Seca 874 digital scale and height by Seca 213 stadiometer. BMI was calculated by using the standard formula (weight in kg/height in m 2 ) and was classified according to cut-off based on risk of type 2 diabetes and cardiovascular disease for Asian population: <18.5 kg/m 2 (underweight), 18.5-22.9 kg/m 2 (acceptable risk), 23-27.4 kg/m 2 (increased risk) and >27.5 kg/m 2 (high risk). 12 The terms 'overweight' and 'increased risk BMI' and 'obesity' and 'higher risk BMI' were used synonymously in the text. parity (nulliparous/ primiparous/multiparous); currently pregnant (yes/no); education (no education/primary/secondary/higher); residence type (urban/rural); wealth quintile (poorest/poorer/middle/richer/richest); religion (Hinduism/other); frequency of TV watching (never, less than once a week/at least once a week/almost every day).
For the calculation household wealth status, instead of direct income the volume of durable goods (e.g., TV, radio and bicycle) possessed by the household as well as and housing quality (e.g., type of floor, wall and roof) are taken into consideration. Each item is assigned a factor score generated through principal component analysis (PCA), which are then summed and standardized for the households. 25 These standardized scores place the households in a continuous scale based on relative wealth scores. The scores thus obtained from a continuous scale and subsequently categorized into quintiles to rank the household as poorest/poorer/middle/richer/richest to richest. 13

| Data analysis
Data were analysed using Stata version 14. As NFHS employs cluster sampling techniques, 'svy' command was used to account for the survey design. Using χ² tests, the sociodemographic characteristics of the sample population across the BMI categories were presented as percentages. The prevalence rates of overweight/obesity (23-27.4 and >27.5 kg/m 2 ) were shown as bar charts.
Second, multivariable logistic regression analyses were performed to identify the variables significantly associated with BMI status.
Regression models were further was stratified into subpopulation groups by pregnancy status, as pregnant women are less likely to have overweight/obesity compared with non-pregnant women. The level of significance was set at p < 0.05 for all analyses. After the multivariate analysis, variance inflation factor (VIF) test was performed to check for multicollinearity between independent variables has been checked. VIF values ranged from 1.01 to 2.01, denoting the absence of any multicollinearity.
Relative contribution of the explanatory variables to BMI status were also reported. This was performed as a regression postestimation procedure in Stata that calculates the individual contribution of a variable divided by the sum of the total contribution of all variables. This represents the proportional contribution of a variable in terms of total variance explained in the outcome variables. This was considered necessary as the odds ratios from regression analysis do not reflect the relative importance or weight of explanatory variables.
It is however, important to note that this procedure does not reflect any unexplained variance.

| Descriptive analysis
Sociodemographic profile of the sample population was shown in Table 1. A greater percentage of women who had underweight were in the lower age groups, whereas, those who had overweight and obesity were in the higher age groups. The percentage of overweight/obesity was also higher among women with more than two children, non-pregnant, had secondary level education, rural resident (for overweight), from the households with higher (richer/richest) wealth status, followers of Hinduism, and watched TV almost everyday.

| Factors associated with overweight and obesity
Results of multivariable regression analysis showed significant association between BMI with age, residency, educational level, religious affiliation and wealth index among both non-pregnant and pregnant women (

| DISCUSSION
The findings suggest a slow but steady progress in reducing the preva- tively, compared with about one-tenth and less than one-tenth in 1998-1999. The change in BMI across the survey years was statistically significant (p < 0.05). The prevalence of being in the underweight category was higher among women in the lower age groups, whereas, that of increased and higher risk was higher among women in the higher age groups, reflecting a clear age gradient in nutritional status.
Increasing age is known predictor of gaining excess body weight, which requires special age-specific intervention techniques. 26 Obesity is not only a risk factor for developing metabolic syndrome (cluster of conditions that occur together, e.g., cardiovascular diseases and type 2 diabetes) but also of maternal and neonatal morbidity mortality, a matter of concern for public health in India. With the growing burden of overweight/obesity, developing age appropriate obesity control and intervention programmes should be prioritized.
In line with previous findings, women with higher parity were found to be more likely to have overweight and obesity and less likely to have underweight, implying a beneficial effect of motherhood against underweight. [27][28][29] This finding also underscores the importance of regulating excess body weight among women with higher parity. As women tend to accumulate body fat at each pregnancy, higher parity can heighten the risk of developing moderate to extreme obesity and associated complications. Interestingly, the association between parity was reversed among pregnant women. However, pregnancy itself is a strong predictor of overweight/obesity, which can confound the association between parity and body weight.   34 Another review of NFHS reported that prevalence of obesity increased from 10.6% to 12.6% during the same period. 16 In comparison, the prevalence of overweight/obesity was slightly higher than in Nepal (27.5%) and lower than in Bangladesh (39.5%). 12 However, the prevalence rates are not directly comparable as they are likely to differ depending on the cut-offs values (Asian vs. international cut-off) and the inclusion/exclusion criteria applied.  35 and diabetes, 36 of which higher than normal BMI is a prominent risk factor. Addressing the rising prevalence of overweight/obesity should therefore be regarded as an urgent public health imperative, which needs to be facilitated by appropriate policy mix and actions by bringing together the relevant stakeholders to ensure an approach that is multidisciplinary and comprehensive in nature and effective in implementation. As a country characterized by stark geographic and sociocultural disparities, applying uniform health and nutrition policies are unlikely produce satisfactory outcomes in terms of addressing the challenges of malnutrition. 37 More population-based studies should be carried to explore the culture-specific factors that will help formulation of more nuanced and locally tailored policies to combat overweight/obesity in the population.
The key strength of this study was the large sample size and the use of nationally representative data which ensures greater robustness of the estimates (prevalence rates). As such, the findings are gen- were cross-sectional, and hence, the associations imply reciprocity rather than causality or directionality.

| CONCLUSIONS
This was a comprehensive analysis of NFHS data with the objectives of providing an updated scenario on overnutrition among adult nonpregnant female population in India. Consistent with the existing literature, the findings primarily indicate the co-occurrence of undernutrition and overnutrition, with undernutrition is predominant among the relatively lower wealth status and overnutrition among the relatively