M.

Phil Synopsis (2018-19)

Assessment of Climatic Variability and Food Production on Child Health, India

Student: Mr.Subhojit Shaw Date: 12th March, 2019

Guide: Prof. Balram Paswan Venue: Presentation Hall

1.1. Introduction

Human wellbeing is linked with the environment through a complex web of vicious circle model describing the
interrelationships among population growth, poverty, health impairment, and environmental degradation such as climate
change (Jankowska, M.M. et al. 2012). There is a strong consensus among the scientists that with climate change,
meteorological factors like temperature, precipitation, humidity will increasingly influence human health (FAO, 2015) The
Intergovernmental Panel on Climate Change (IPCC) has declared with “very high confidence” that climate change already
contributes to the global burden of disease (Confalonieri et al., 2007). The change in the climate can result in drought and
famine. Indian agriculture is mostly dependent on rainfall and irregularities may result in crop failure, and food shortage
among the vulnerable population. Hence, rainfall is an important factor influencing the livelihoods of farmers (Haile M et
al. 2005). Therefore, the consequences have graved on crop production and child nutrition. Malnutrition is a condition
resulting from either an insufficient or excessive intake of various nutrients. It is a broad term that includes both
undernutrition and overnutrition. Globally, significant numbers of people are affected by malnutrition (undernutrition). The
Food and Agriculture Organization (FAO) estimated that the prevalence of undernutrition (PoU) is approximately 795
million people in 2014-16 across the world (Black RE et al., 2013). The prevalence of undernourishment and underweight
among children under age five were primarily used internationally to measure and compare the burden of undernutrition.

1.1.2. Global Burden of Undernutrition

Globally, for children under five years of age climate change is predicted to worsen all of the top five causes of death
namely, malnutrition, neonatal deaths, acute respiratory illness, diarrhoea, malaria and others (Kiang, K. et al., 2013). The
global trend in undernourishment has shown substantial improvement as the PoU had decreased globally from 18.6% in
1900-92 to 10.9% in 2014-16. However, there is a wide variation in PoU among regions. Two regions, southern Asia and
Sub-Saharan Africa, account for approximately 35.4 % and 27.7% of the shares of global undernourishment, respectively
(FAO, 2015). Next, to southern Asia, Sub-Saharan African countries, especially eastern African countries, have the highest
burden (share) of undernutrition than any region in the world. Globally, the prevalence of stunting (height-for-age Z-score
of less than -2SD) has declined from 47% in 1985 to 30% in 2011 (Stevens GA et al., 2012). Current estimates show that
the stunting prevalence is 23.8% globally, with approximately 159 million children affected (UNICEF, 2015). Within a
similar period, the prevalence of underweight (weight-for-age Z-score of less than -2SD) has decreased from 30.1% to

1|Page
19.4%, with more than 100 million children being underweight. The corresponding estimate for wasting (weight-for-height
Z-score of less than -2SD) is approximately 52 million children (Black RE et al., 2013).

1.1.3. Nutrition Situation in India

A large proportion of children and women are suffering from severe malnutrition, especially in developing countries like
India. Evidence suggests that there has been a significant decline in the level of stunting from 52% in 1992–93 to 38% by
2015–16. On the contrary, wasting had increased from 17% to 21% among the children (Khan j.et al. 2018). According to
NFHS-II (1998-99), there were 51, 20 and 43 per cent of children below three years suffer from stunting, wasting and under
weight respectively, which had decreased to 45, 23 and 40 (the corresponding figure for NFHS III, 2005-06) (Das S. et al.,
2011). Stunting, wasting and under weight is a composite measure of malnutrition (Gillespie and McNeil 1992; Arnold et
al. 2003). An iron deficiency that results in anaemia is a major health problem among young children in most parts of India
and Asia (Kotecha PV, 2011). According to NFHS-4 report, Fifty-eight per cent of children age 6-59 months have anemia
(hemoglobin levels below 11.0 g/dl), an improvement from the NFHS-3 estimate of 70 per cent. Iron deficiency affects
mainly women and children as a public health epidemic. The effects of anaemia on children can lead to ill- health, premature
death, lost hearing and others in early childhood.

1.2. Review of Literatures

It is anticipated that climate change will affect child health around the world with different consequences in different regions
and population groups (MacCracken et al. 2011; Bennett, C et al. 2014). Climate change is expected to amplify effect on
health among the poor and developing countries (Strazdins, L et al. 2011). Globally, the adverse effects of climate change
on agricultural productivity and food security are predicted to cause a 20% increase in the number of malnourished children
by 2050 (21).

1.2.1. Climate-Agricultural Production

Global average temperatures have risen by roughly 0.13°C per decade since 1950 with a pace of 0.2°C per decade
(S.Solomon et al. 2007), and the impact has had on agriculture is not well understood. (David B. Lobell et al.2011) Had
developed a database yield models to assess the impact of climate trends on major crop yields across the globe from (1980-
2008). A substantial study focuses on the areas of the African continent where poverty is the prohibition of development.
(Funk C, et al. 2008) Had quantified the potential impacts of precipitation and agricultural trends as a function of rainfall,
population, cultivated area, seed, and fertilizer use from (1951-80). Statistical analysis were employed to estimate the agro
climatic regional yield from previously simulated studies and have been summarised by (Iglesias et al., 2003). Association
of crop yield with temperature and precipitation anomalies over the pre-harvesting period were analyzed using correlation
coefficients. These studies served to identify significant variables that explain the yield variance. A research by (Perry M.
et al. 2005) has evaluated the implications of climate change for food production and risk of hunger in the African Continent.

2|Page
Another perspective of agricultural production is the consequence of the greenhouse gas-induced change in regional
productivity. (Parry, M.L. et al 2004) Has estimated the physiological CO2 effects on crop yields based on an extensive
review of previous simulations. Unstable food production may lead to food insecurity (Alderman, 2010; Haines &
McMichael, 1997).

1.2.2. Climate- Nutrition

Climate is one of the multiple factors that manifest to a wide range of health comes. The impacts resulting from
consequences of climate-induced environmental degradation has an indirect effects of health (Ebi, K.L., 2008). The
ecological degradation brought by climate change are likely to sway the availability of food and clean water causing
malnutrition, diarrhoea, poor growth and developmental delays in early childhood and increases in allergens causing
allergies and asthma (Bunyavanich, S. et al. 2003; Kistin, E.J. et al 2010). The development are being threatened with the
acceleration of climate variability in the recent past (Gellrich & Zimmermann, 2007) which are being reflected through food
security-related and health metrics like child nutritional status and infant or under-5 mortality. In a country like India, where
pluses are preferred over the meat as a source of protein, the climate change will affect the quality of food crops which in
turn will accelerate the epidemic of “hidden hunger” or micronutrient deficiency (Myers SS, et al. 2014). A handful of
empirical studies have investigated the correlation of weather and climate variability to crop yields and undernutrition.

1.2.3. Climate-Agricultural production- Nutrition

Most of developing world’s population resides in rural areas and are dependent on subsistence agriculture for survival, and
are subjected to climatic induced health vulnerabilities. A study conducted by (Kiersten Johnson et al. 2014) focused on
West African countries to understanding the impact of climate variability and associated environmental productivity, using
NDVI as a proxy, on child health and nutrition outcomes from DHS data. The study has identified the two driest countries,
Burkina Faso and Mali, which shows a strong response to changes in NDVI with mortality and severely wasted in Mali.
Another study for Mali (Jankowska, M.M.et al 2012) has examined and projected climate with potential evapotranspiration
(PPET) index and health trends in the African Sahel through the spatial coupling of climate data and DHS health data in
Mali. Results suggest that cluster measures of underweight and anemia appear influenced by livelihoods and mitigation
efforts should therefore focus on livelihood adaptation strategies. To link climate and under 5 nutritional status in Ethiopia
a study by (Hagos, S et al,2014) has concluded that rainfall and temperature are partly predicting the variation in stunting
and underweight in Ethiopia. Crop yield determines the quantity and the nutritional status as a significant reviewed
literatures by (Phalkey, R.K et al. 2015) implies the impact of climate/weather variability on crop yields on childhood
undernutrition, especially acute undernutrition.

3|Page
1.3. The rationale of the Study

What are the factors affecting child undernutrition? It is well- known that child undernutrition is affected by multiple factors
like socio-cultural, economic and demographic while Climate change is one of the challenges against the efforts undergoing
to combat child undernutrition through improved household food security. In this context, the present study tries to explore
the climatic determinants along with agricultural production on child undernutrition in India. This study can benefit us in
understanding the impacts on child nutrition at large by anticipating the effect of climate change on food productivity.

(Phalkey, R.K. et al.2015) Complex pathways from climate/weather variability to undernutrition in subsistence farming households.
The factors involved in and the probable impacts of weather variables on crop yields (blue arrows) and of food crop yields on
undernutrition (red arrows)

1.4. Research Questions:

 How has the climate/weather changed in the past (2000-15)?

 How the climatic variability ha influenced food productivity across India over time?
 Is there any association between climatic variability and food production on child health?

1.5. Objective of the Study

 Evaluate the spatial and temporal distribution of VHI, SDCI and Kendell’s API across India (2000-15)
 Characterize the pattern and interrelationship of Agricultural Productivity with VHI and SDCI through panel
analysis; and
 Analyze the variations of climate and agriculture productivity on under 5 child nutrition over time and across
India.
4|Page
1.6. Hypothesis
H0 = There is no association between API with VHI and SDCI.

H0 = There is no association between climate change and agricultural productivity with under 5 child nutrition.

2.1. Materials and Methods

2.1.1. Methodology flow chart

In this section, the underlying methodology of this study has been presented, consisting of data acquisition, pre-
processing and drought vulnerability based on remote sensing time series analysis gives a schematic overview of the major
methodological steps.

2.1.2. Study design and period

We employed a longitudinal panel study design to estimate the effect of drought/ vegetation health with agricultural
production on child undernutrition and anaemia for the period of 2000-2015. The datasets used for the study is mainly
comprised of two categories. Those are products derived from the satellite sensors and ancillary data from various sources.
Terra MODIS surface reflectance MOD13Q1 (250m) (2000-2015) was used to compute time series NDVI and VCI. MODIS
LST of MOD11A2 (1km) (2000-2015) was used to compute TCI, and TRMM 3B43 (0.25×0.25) precipitation estimate
(2000-2015) was used to compute PCI. The summary of datasets, is presented below:

5|Page
 Data Set Variable Temporal Coverage Temporal Resolution Spatial Resolution Source

MOD13Q1 NDVI, VCI 2000-15 Monthly 250 m Earth Explorer

MOD11A2 TCI 2000-15 Monthly 1000 m Earth Explorer
TRMM 3B43 SPI, PCI 2000-15 Monthly 27 km NASA Mirador
SRTM DEM Slope 30 m Earth Explorer
Source: Compiled by the author
2.2.1. Computation and analysis of meteorological indices:
Meteorological and vegetation Indices used in the study

Name of Indices Formula

Normalized Difference Vegetation Index (NDVI) ( NIR  red)

( NIR  red)
( NDVI  NDVI min )
Vegetation Condition Index (VCI)
( NDVI max  NDVI min )
( LSTmax  LST )
Temperature Condition Index (TCI)
( LSTmax  LSTmin )

Precipitation Condition Index (PCI) (TRMM  TRMM min )

(TRMM max  TRMM min )
Vegetation Health Index (VHI) VCI  (1 -  )TCI
Scaled Drought Condition Index (SDCI) 0.25𝑇𝐶𝐼 + 0.5𝑃𝐶𝐼 + 0.25𝑉𝐶𝐼

2.2.2. Vegetation Health Index (VHI)

The VHI is said to represent the overall vegetation health using a combined estimation of moisture and thermal conditions.
Where α is a coefficient determining the contribution of the two indices and is generally taken as 0.5.
It is computed using VCI and TCI as follows:

VHI  VCI  (1 -  )TCI

VHI Drought classification schema (Kogan 2002)

Drought Classes Vegetation Health Index (VHI)

Extreme drought < 10
Severe Drought ≥10 and < 20
Moderate Drought ≥ 20 and < 30
Mild Drought ≥ 30 and < 40
No drought ≥40

6|Page
2.2.3. Scaled Drought Condition Index (SDCI)
Scaled Drought Condition Index (SDCI) is a multi-sensor drought index proposed by (Rhee et al., 2010) considering
vegetation temperature and precipitation from NDVI, LST and TRMM datasets. SDCI index is composed of PCI, TCI, and
VCI.

SDCI  0.25TCI  0.5PCI  0.25VCI

Where the three elements have similar contributions its performance is better than the existing indices such as VHI in
different climate divisions. The values of TCI vary from 0 to 1. The low values of SDCI imply a serious condition of
drought. Under a drought process, the SDCI is close or equal to 0, and at wet conditions, the SDCI is close to 1.

2.3. Parameters from ancillary data:

In addition to the satellite-based data sets, some other important legacy data were used for the assessment of under-five
children nutritional status. The agricultural yield data was collected from the Crop Production Statistics Information system,
Agriculture Informatics Division, National Informatics Centre, Ministry of Communication and IT, Govt. Of India, New
Delhi.

2.3.1 Kendall’s Method of Agricultural Efficiency

Agricultural efficiency index was estimated by a statistician, M. G. Kendall on per acre production of different crops. The
scale effect was eliminated by converting the production per acre of each crop into ranks and the weight problem was
handled by assigning equal weights to each crop. The composite index of agricultural efficiency was constructed by
adding the ranks of different crops in each area. Thus If Rij is the ranks of the ith crop in the jth area, its composite index
of agricultural efficiency Ij is given by:
𝑛

𝐼𝑗 = ∑ 𝑅𝑖𝑗
𝑖=1

Where, n is the number of selected crops. For the study, we have considered 27 food crops based on the World Programme
for the Census of Agriculture 2010 classification across India.

2.3.2. Health and Population

NFHS data have been analyzed in hundreds of studies in the public health literature. However, the use of NFHS-4 data on
reproductive and child health to examine climate effects on humans has not been attempted to our knowledge. The survey
sample of 640 clusters was stratified, weighted, and representative at the national, state/UT, and district levels. The analysis
for this study revolves around the district level comprising of 259,627 children. Four commonly utilised measures of
nutrition were selected from the NHHS-4 for analysis: the child’s level of anaemia (children age 6-59 months have anaemia

7|Page
- haemoglobin levels below 11.0 g/dl), the child’s measure of stunting as the height-for-age is a measure of linear growth
retardation and cumulative growth deficits. Children whose height-for-age Z-score is below minus two standard deviations
(-2 SD) from the median of the reference population are considered short for their age (stunted), the child’s measure of
underweight status when the weight-for-age is a composite index of height-for-age and weight-for-height. It takes into
account both acute and chronic undernutrition. Children whose weight-for-age Z-score is below minus two standard
deviations (-2 SD) from the median of the reference population are classified as underweight, and the child’s wasting status
(Weight-for-height index measures body mass about body height or length and describes current nutritional status. Children
whose Z-score is below minus two standard deviations (-2 SD) from the median of the reference population are considered
as thin (wasted). Stunting, underweight and wasting are assessed from the guidelines of malnourishment (WHO & UNICEF,
2009).

3.1. Statistical Analysis

To fulfill the objectives, univariate Moran’s I and bivariate local Moran’s I was used to check the autocorrelation among
the variables. Multivariate OLS regression analysis was preliminarily used to check the association. Spatial lag model was
used to check the empirical association in a spatial setting. Fixed effect regression model was used to check the association
of VHI and SDCI with API for the period (2000-15) adjusting the elevation. We further examined the possible interaction
between under 5 nutritional status in relation to agricultural productivity and climatic variables over time across India.

3.1.1. Softwares used

The software used for data processing and analysis are as follows:
 ERDAS Imagine 2014
 Arc GIS 10.3
 MODIS Reprojection Tool (MRT)
 STATA 13
 Origin Pro 9.1

4.1. Results

Objective 1: Spatial and Temporal Pattern

Vegetation Health Index Analysis (VHI)

In this study, the spatially distributed VHI at various time scales has been computed. This part aims to identify regions
susceptible to low vegetation at different time scales (2000-15) fig 1. The spatial and temporal distribution of VHI indicates
that a good proportion of healthy vegetation occur in the northern plains, while the southern rain shadow zones of Deccan
plateau has lower frequencies. As the time scale increases significant changes can be observed over the sub-continent.

8|Page
Moderate vegetation account in the Indo-Gangetic plains. The VHI serves as the reflection of drought as the lower value
signifies drought condition while the reverse is true for insignificant drought condition.

Scaled Drought Conditioned Index (SDCI)

Spatial distribution is critically understood to measure drought events. Spatial and temporal distributions of drought
indicators SDCI for (2000-15) has been shown in Fig 2.The severity of SDCI is the most in the initial years and are restricted
to central and northern uplands of the sub-continent. The low values of SDCI imply a serious condition of drought. Under
a drought process, the SDCI is close or equal to 0, and at wet conditions, the SDCI is close to 1. With the time lag, the
events of drought have reduced, and the severity can be traced in the central and Deccan uplands.

Kendell’s Agricultural Productivity Index (API)

The spatial pattern of agricultural productivity for all season (2000-15) has been estimated in fig 3. The north and
northwestern part of the country has higher productivity throughout. Higher crop productivity is documented in some parts
of Southern India. The central upland has lower productivity throughout the study period. We have noted that zones with
higher drought condition have recorded lower agricultural productivity. However, with the time lag it has been found that
crop productivity has relatively increased from the initial seasons across the sub-continent.
A time –series of VHI, SDCI and Kendell’s API for twelve major stunting prevalent states of India among children (0-59
months) as per NFHS-4 report has been plotted in fig 4. The temporal evolution of VHI/ SDCI/ API on all the grids has
shown a significant change. As the frequency of VHI and SDCI has similar pace of fluctuations while the API has remained
mostly insignificant with major alterations during early 2010 across the major nutrition deficiency states of India. Thus it is
inferred from the graphs that there is an inverse relation among the higher API values and VHI/SDCI frequencies.

Objective 2: Characterize the interrelationship of Agricultural Productivity with VHI and SDCI

Table 1: Above results are have been calculated from the fixed effect regression model on panel data where we are interested
in analyzing the impact of variables that vary over time. Impacts of variables VHI and SDCI over time has been calculated
separately for studying the effect on API.

With VHI, we see that 36.7% of the variance is due to differences across panels while keeping the elevation at constant.
Under the reference category of the year 2000, API is predicted to decrease by 0.013 when VHI goes up by one in the year
2001. Similarly, under the reference category of the year 2000, API is predicted to decrease by 0.118, 0.121, 0.059, and
0.059 when VHI goes up by one in the year 2009, 2010, 2011, and 2015 respectively. Whereas, under the reference category
of year 2000, API is predicted to increase 0.067 , 0.076, 0.344, 0.086,0.071,0.046, 0.062, 0.541,0 .006, 0.027 when VHI

9|Page
goes up by one in year 2002, 2003 , 2004, 2005, 2006 ,2007, 2008, 2012, 2013 , 2014 respectively. Following this, we see
here that VHI has a significant influence on API from the year 2002 to 2012 and 2015.

In SDCI, we see that 36.7% of the variance is due to differences across panels. Under the reference category of the year
2000, API is predicted to decrease by 0.01 when SDCI goes up by one in the year 2001. Similarly, under the reference
category of the year 2000, API is predicted to decrease by 0.118, 0.11, 0.047, 0.049, when SDCI goes up by one in the year
2009, 2010, 2011, 2015 respectively. Whereas, under the reference category of year 2000, API is predicted to increase
0.067 , 0.084, 0.352, 0.095,0.073,0.057, 0.074,0 .548, 0.021, 0.034 when SDCI goes up by one in year 2002, 2003 , 2004,
2005, 2006 ,2007, 2008, 2012, 2013 , 2014 respectively. Following this, we see here that SDCI has a significant influence
on Nutritional Index from the year 2002 to 2012 and 2015.

Objective 3: Variations of climate and agriculture productivity on under 5 child nutrition over time and across
India.

Table 2 lists out the all districts across the states of India identifying spatial autocorrelation of Under 5 Children Health-
Variables (Stunting, Under Weight, Wasting, Anemia), Geographical–Climate Variables (API, VHI, SDCI, Elevation),
Socio-demographic Variables (Proportion of Male, Rural, Hindus, SC/ST, Illiterate Mothers, Poor, No Media Exposure).
Estimated Univariate Moran’s I statistic values indicate a highly significant spatial dependence in Stunting whereas wasting
shows lowest spatial dependence across the districts of India. For Geographical-Climate Variables, Univariate Moran’s I
recognizing a highly spatial dependence in API and Elevation shows the lowest dependency across India. Similarly, for
Socio-Demographic Variable, Univariate Moran’s I pointing out of high dependency in Proportion of Poor and Proportion
of Hindus while Proportion of Rural and Proportion of SC/ST locating lowest spatial dependence in India. For almost all
the variable Univariate Moran’s I value giving a Clustered Pattern Spatial Autocorrelation. Through Randomization for
almost all the variable it is also suggesting Univariate Moran’s I is Significant.

Table 3: Bivariate LISA estimates of spatial autocorrelation, compelling evidence of spatial autocorrelation between the
different components of Child-Nutrition with most of the exposure variables considered in this study. Of the different
exposure variables Proportion of Hindus, Illiterate Mothers, & No Media Exposures showed a positive & statistically
significant spatial autocorrelation with each of Under 5 Child Nutrition indicators and Anemia. In case of Wasting, for VHI,
a Geographical-Climatic variable showed a negative & insignificant spatial autocorrelation, For Socio-demographic
Variable Proportion of Male, Proportion of Rural & Proportion of SC/ST showed a very low (insignificant) with positive,
positive and negative spatial autocorrelation respectively. Variables like Proportion of Male in Socio-demographic Variable
showed low spatial autocorrelation. In case of Stunting VHI, SDCI, Elevation & Proportion of SC/ST showed negative but
significant autocorrelation. Likewise, the pattern of spatial autocorrelation (and the statistical significance) between all the
Under 5 Nutrition Index and exposure variables are shown in the table 3.
10 | P a g e
Table 4: The API is significant with anemia with the standardized β coefficient at -3.801 for α= 0.01 indicating a positive
impact on anemia. The proportion of male child is also statistically significant for anemia with β coefficient 0.378 at α=
0.01. Proportion of rural population is found to be significant with under-weight and wasting, β coefficient being -0.074 and
-0.049 respectively at α= 0.01. However more variability was explained by Under-Weight as indicated by improved R2
value. The proportion of hindus is significant among all four under 5 Health indicators. Proportion of SC/ST is significant
for only Wasting. Proportion of Illiterate Mothers is statistically significant for Under-Weight, Wasting and Anemia at α=
0.01. The proportion of poor has significance with Stunting and Under-Weight with β coefficients 0.094 and 0.129
respectively. The no media exposure population category has high significance with Stunting and Anemia.

Table 5 :The VHI is slightly significant with wasting with the standardized β coefficient at 4.246 for α= 0.05. The proportion
of male children is also statistically significant for Anemia with β coefficient 0.397 at α= 0.01 indicating a positive impact
with anemia. Proportion of rural population is found to be significant with Under Weight and Wasting, β coefficients being
-0.078 and -0.056 respectively at α= 0.01. However more variability was explained by Under-Weight as indicated by
improved R2 value. The proportion of hindus is significant among all four under 5 nutrition health indicators. Proportion of
SC/ST is significant for only Wasting. Proportion of illiterate mothers is statistically significant for stunting, under-weight
and anemia at α= 0.01. The Proportion of poor population has significance with the under-weight indicator with β coefficient
0.141. While, no media exposure category has high significance with stunting and anemia.

Table 6: The SDCI is significant with stunting with the standardized β coefficient at -3.788 for α= 0.05. The proportion of
male children is also statistically significant for anemia with β coefficient 0.398 at α= 0.01 indicating a positive impact with
anemia. Proportion of rural population is found to be significant with under-weight and wasting, where β coefficient being
-0.073 and -0.052 respectively at α= 0.01. The variability is explained more by under-weight due to more improved R2
value. The proportion of hindus is significant among all four under 5 nutrition health indicators. Proportion of SC/ST is
significant for only wasting. Proportion of illiterate mothers is statistically significant for stunting, under-weight and anemia
at α= 0.01. The proportion of poor has significance with the under-weight indicator with β coefficient 0.127, indicating a
positive impact with under-weight. While, no media exposure category has high significance with stunting and anemia.

Table 7: Elevation is significant with stunting and under-weight with the standardized β coefficients at -0.001 and -0.002
for α= 0.05. The variability is explained more by under-weight due to more improved R2 value. The proportion of male
children is also statistically significant for anemia with β coefficient 0.396 at α= 0.01 indicating a positive impact with
anemia. The Proportion of rural population is found to be significant with Wasting, β coefficient being -0.047 at α= 0.01.
The proportion of hindus is significant among all four under 5 nutrition health indicators. Proportion of SC/ST is significant
for only Wasting. Proportion of Illiterate Mothers is statistically significant for Stunting, Under-Weight and Anemia at α=
0.01. The Proportion of Poor has significance with stunting and under-weight indicators with β coefficients 0.082 and 0.127
respectively .While, no media exposure category has a slightly lower significance with stunting and anemia at α= 0.05.
11 | P a g e
5.1. Discussion:

The variations in the observed associations, between VHI/SDCI/API and our outcomes of interest differs in socio
demographic situations across the study and over the time. These variations may be due to better frequencies of agricultural
production, vegetation health status and diminishing drought condition in the prolonged time (2000-15) to support child
nutrition. Historical records accounts that the failure of rain results in crop failure, impact food productions and usually
results in food shortages in vulnerable parts of the population. The association was empirically estimated as with higher
VHI / SDCI the food production is higher. Further analyses have incorporate remote sensing data on district level at the
time of survey in order to clarify the observed relationships. The results has shown an unexpected inverse association
between stunting/anemia and higher VHI/SDCI/API values. The reasons for this association could be many. It could be that
the VHI/SDCI values are higher because of irrigation and modernization of agriculture (and thus more wet condition).This
possibility implies that there may have been a good production of certain crops such as food crops and vegetables that are
important for child nutrition. While the geographical space (elevation) shows an inverse relationship among the stunted and
under-weight children. Another unexpected positive association was found between under-weight and wasting with VHI in
2015 across India. While our measure captures hemoglobin levels in children, it does not discern among etiologies.

Thus, our regression results suggest that we cannot isolate climate change and food productivity as the cause of a statistically
significant number of these children suffering from nutritional deficiencies.

6.1. Conclusion

In this study, the focus was on understanding the impact of climate variability and agricultural productivity, using VHI,
SDCI and API as a proxy, on child health and nutrition outcomes. The VHI and SDCI parameter is often used by researchers
determine if crop growing conditions are below average. This study provides evidence of the broader, long-term
implications of drought and climate variability on health and nutrition outcomes. With this empirical study in a country like
India, we have demonstrated the relationship between variations in geographical and climatic variables on child nutrition.
The results of our analyses suggested that the extent to which human populations are unable to mount an adaptive response
to climatic variability is the extent to which we can observe associations between environmental growing conditions and
food security-related outcomes. The most vulnerable people in our society might be facing a barrage of compounding
challenges: increased climate variability and change with the consequences being found among children under 5 years, an
age when undernutrition can cause irreparable damage to a child's intellectual and physical development. To understanding
the empirical relationship between environmental conditions and child nutrition it is critical for informing evidence-based
policies and programs designed to mitigate the worst impacts of the current global economic and environmental crisis. This
paper has taken initial steps in this direction; however, there is much more to do. We recommend further work but at a micro
level using similar analysis methods to assess the effect of climate on stunting, wasting, underweight and anemia and using
new approaches and modelling to better understand the inherent complexities of these relationships which, will further
improve the knowledge of human vulnerability and adaptability to an increasingly variable climatic context.

12 | P a g e
References:
Alderman H. Safety nets can help address the risks to nutrition from increasing climate variability. The Journal of
Nutrition 2010. 140(1), 148S.

Arnold F, Nangia P, Kapila U. Indicators of Nutrition for Women and Children in India: Current Status and
Programme Recommendations. Paper Presented at a Workshop on National Family Health Survey 2003,
Hyderabad: Centre for Economic and Social Studies.

Bennett C, Friel S. Impacts of Climate Change on Inequities in Child Health. Children 2014, 1, 461–473.

Bhuiyan, C, Singh RP, and Kogan FN. Monitoring drought dynamics in the Aravalli region (India) using different
indices based on ground and remote sensing data. International Journal of Applied Earth Observation and
Geoinformation 8, no. 4 (2006): 289-302.

Black RE, Victora CG, Walker SP, Bhutta ZA, Christian P, de Onis M et al.: Maternal and child undernutrition and
overweight in low-income and middle-income countries. Lancet 2013, 382.

Bunyavanich S, Landrigan CP, McMichael AJ, Epstein PR. The Impact of Climate Change on Child Health. Ambul.
Pediatr. 2003, 3, 44–52.

Confalonieri U, Menne B, Akhtar R, Ebi K. L, Hauengue M, Kovats RS, et al.: Human health. Climate change 2007:
Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change, In Parry ML, Canziani OF, Palutikof JP, dLinden PJv, & Hanson CE
(Eds.) Cambridge, UK: Cambridge University Press 2007, 391-431.

Crop Production Statistics Information System, Agriculture Informatics Division, National Informatics Centre,
Ministry Of Communication & IT,Govt. Of India, New Delhi.

Das S, Sahoo H. An Investigation into Factors Affecting Child Undernutrition in Madhya Pradesh. The
Anthropologist 2011 Jul 1; 13(3):227-33.

Ebi KL. Healthy people 2100: modeling population health impacts of climate change. Climatic Change 2008, 88(1),
pp.5-19.
F.A.O. The State of Food Insecurity in the World, 2015. Rome, Italy.

Funk C, Dettinger MD, Michaelsen JC, Verdin JP, Brown ME, Barlow M and Hoell A. Warming of the Indian Ocean
threatens eastern and southern African food security but could be mitigated by agricultural development. Proceedings
of the National Academy of Sciences 2008, 105(32), pp.11081-11086.

Gellrich M & Zimmermann NE. Investigating the regional-scale pattern of agricultural land abandonment in the Swiss
Mountains: a spatial statistical modeling approach. Landscape and Urban Planning. Landscape and Urban Planning
2007. 79(1), 65e76

Gillespie S, McNeil G. Food Health and Survival in India and Developing Countries. Delhi: Oxford University
Press 1992

Hagos S, Lunde T, Mariam DH, Woldehanna T and Lindtjørn B. Climate change, crop production and child under
nutrition in Etiohpia; a longitudinal panel study. BMC public health 2014, 14(1), p.884.
13 | P a g e
Haile M. Weather patterns, food security and humanitarian response in sub-Saharan Africa. Philosophical
Transactions of the Royal Society B Biological Sciences 2005, 360:2169–2182.
Haines A & McMichael AJ. Climate change and health: implications for research, monitoring, and policy. British
Medical Journal 1997, 315(7112), 870e874.

IPCC, in Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth
Assessment Report of the Intergovernmental Panel on Climate Change, S. Solomon et al., Eds. Cambridge Univ.
Press, Cambridge, 2007

Jankowska MM, Lopez-Carr D, Funk C, Husak GJ and Chafe ZA. Climate change and human health: Spatial
modeling of water availability, malnutrition, and livelihoods in Mali, Africa. Applied Geography 2012, 33, pp.4-15.

Johnson K and Brown ME. Environmental risk factors and child nutritional status and survival in a context of climate
variability and change. Applied Geography 2014, 54, pp.209-221.

Kendall MG. A Course in Multivariate Analysis. Charles Griffin Co. London 1968, p. 35.

Kendall MG. The Geographical Distribution of Crop Productivity in England. Journal of Royal Statisticial society
102, 21 (1939)

Khan J and Mohanty SK. Spatial heterogeneity and correlates of child malnutrition in districts of India. BMC
Public Health 2018, 18(1), p.1027.

Kiang K., Graham S, Farrant BM. Climate change, child health and the role of the paediatric profession in under-
resourced settings. Trop. Med. Int. Health 2013, 18, 1053–1056.

Kistin EJ, Fogarty J, Pokrasso RS, McCally M, McCornick PG. Climate change, water resources and child health.
Arch. Dis. Child. 2010, 95, 545–549.

Kotecha PV. Nutritional Anemia in Young Children with Focus on Asia and India. Indian journal of community
medicine: official publication of Indian Association of Preventive & Social Medicine 2011 Jan-Mar;36(1):8.
Lobell DB, Schlenker W and Costa-Roberts J. Climate trends and global crop production since 1980. Science
2011, 333(6042), pp.616-620.

MacCracken MC. Climate Change Impacts Will be Pervasive. Int. Public Health J. 2011, 2, 371–376.
Myers SS et al. Increasing CO2 threatens human nutrition. Nature 510 (7503): 139–142.

Parry M, Rosenzweig C and Livermore M. Climate change, global food supply and risk of hunger. Philosophical
Transactions of the Royal Society B: Biological Sciences 2005, 360(1463), pp.2125-2138.

Parry ML, Rosenzweig C, Iglesias A, Livermore M and Fischer G. Effects of climate change on global food
production under SRES emissions and socio-economic scenarios. Global environmental change 2004, 14(1), pp.53-
67.

Phalkey RK, Aranda-Jan,C, Marx S, Höfle B and Sauerborn R. Systematic review of current efforts to quantify the
impacts of climate change on undernutrition. Proceedings of the National Academy of Sciences 2015, 112(33),
pp.E4522-E4529.

Phalkey, R.K., Aranda-Jan, C., Marx, S., Höfle, B. and Sauerborn, R., 2015. Systematic review of current efforts to
quantify the impacts of climate change on undernutrition. Proceedings of the National Academy of Sciences, 112(33),
pp.E4522-E4529.

14 | P a g e
Rhee J, Im J, Carbone GJ. Monitoring agricultural drought for arid and humid regions using multi-sensor remote
sensing data. Remote Sensing of Environment. 2010 Dec 15;114(12):2875-87

Rosenzweig, C, Iglesias A. Potential impacts of climate change on world food supply: data sets from a major crop
modeling study, 2003

Stevens GA, Finucane MM, Paciorek CJ, Flaxman SR, White RA, Donner AJ et al.: Trends in mild, moderate, and
severe stunting and underweight, and progress towards MDG 1 in 141 developing countries: A systematic analysis of
population representative data. Lancet 2012, 380: 824-834

Strazdins L, Friel S, McMichael A. Butler SW, Hanna E. Climate change and child health in Australia: Likely futures,
new inequities? Int. Public Health J. 2011, 2, 493–500.

The International Food Policy Research Institute (IFPRI). Climate Change: Impact on Agriculture and Costs of
Adaptation; IFPRI: Washington, DC, USA, 2009.

UNICEF W, World Bank. Levels and trends in child malnutrition. 5-36, 2015. New York.

Unicef. Improving child nutrition. The achievable imperative for global progress. New York: United Nations
Children’s Fund Google Scholar;2013. p. 2016.

15 | P a g e
Fig1: Spatial and temporal distribution of Vegetation Health Index (VHI): (2000-15)

16 | P a g e
Fig 2: Spatial and temporal distribution of Scaled Drought Condition Index (SDCI) : (2000-15

17 | P a g e
Fig 3: Spatial and temporal distribution of Kendell’s Agricultural Productivity Index (API) : (2000-15)

18 | P a g e
Fig 4: Time-series of VHI, SDCI and API for twelve major stunting prevalent states as per NFHS-4 report

19 | P a g e
Table 1: Estimated result from the fixed effect regression model API with VHI and SDCI

Beta Std. Error Probability Beta Std. Error Probability

VHI -0.002 0.028 0.94 SDCI -0.041 0.029 0.162
Period Period
2000 ® 2000 ®
2001 -0.013 0.019 0.498 2001 -0.01 0.018 0.573
2002 0.067*** 0.018 0.000 2002 0.067*** 0.017 0
2003 0.076*** 0.019 0.000 2003 0.084*** 0.018 0
2004 0.344*** 0.040 0.000 2004 0.352*** 0.035 0
2005 0.086*** 0.020 0.000 2005 0.095*** 0.019 0
2006 0.071*** 0.021 0.001 2006 0.073*** 0.020 0
2007 0.046** 0.022 0.038 2007 0.057*** 0.020 0.005
2008 0.062** 0.026 0.019 2008 0.074*** 0.025 0.003
2009 -0.118*** 0.019 0.000 2009 -0.118*** 0.019 0
2010 -0.121*** 0.020 0.000 2010 -0.11*** 0.018 0
2011 -0.059** 0.024 0.015 2011 -0.047* 0.024 0.052
2012 0.541*** 0.075 0.000 2012 0.548*** 0.076 0
2013 0.006 0.023 0.780 2013 0.021 0.022 0.329
2014 0.027 0.039 0.486 2014 0.034 0.041 0.402
2015 -0.059** 0.026 0.027 2015 -0.049** 0.025 0.049
Sigma_u 0.463 Sigma_u 0.463
Sigma_e 0.608 Sigma_e 0.608
roh 0.367 (fraction of variance due to u_i) roh 0.367 (fraction of variance due to u_i)
***p<0.01, **p<0.05, *p<0.10, ® Reference Category

Table 2: Univariate Moran’s I results for all districts (n=640) among under 5 children (n=259627)

Moran’s I P-value
Under 5 Children Health-Variables
Stunting 0.635 0.001
Under Weight 0.742 0.001
Wasting 0.482 0.001
Anemia 0.604 0.001
Geographical-Climatic Variables
API 0.878 0.001
VHI 0.822 0.001
SDCI 0.817 0.001
Elevation 0.579 0.001
Socio-demographic Variable
Proportion of Male 0.644 0.009
Proportion of Rural 0.400 0.001
Proportion of Hindus 0.736 0.001
Proportion of SC/ST 0.572 0.001
Proportion of Illiterate Mothers 0.702 0.001
Proportion of Poor 0.754 0.001
Proportion of No Media Exposure 0.736 0.001

20 | P a g e
Table 3: Bivariate Locals Moran’s I results for all districts (n=640) among under 5 children (n=259627)
Stunting Under Weight Wasting Anemia
Moran’s I p Moran’s I p Moran’s I p Moran’s I p
Geographical-Climatic Variables
API -0.011 0.261 0.049 0.003 0.110 0.001 -0.209 0.001
VHI -0.300 0.001 -0.195 0.001 -0.002 0.435 -0.005 0.381
SDCI -0.355 0.001 -0.306 0.001 -0.109 0.001 -0.059 0.002
Elevation -0.295 0.001 -0.368 0.001 -0.192 0.001 -0.091 0.001
Socio-demographic Variable
Proportion of Male 0.081 0.001 0.091 0.001 0.038 0.013 0.091 0.001
Proportion of Rural 0.222 0.001 0.150 0.001 0.012 0.237 0.002 0.456
Proportion of Hindus 0.242 0.001 0.456 0.001 0.391 0.001 0.323 0.001
Proportion of SC/ST -0.062 0.001 -0.123 0.001 -0.049 0.005 -0.202 0.001
Proportion of Illiterate Mothers 0.518 0.001 0.430 0.001 0.119 0.001 0.324 0.001
Proportion of Poor 0.496 0.001 0.435 0.001 0.153 0.001 0.052 0.003
Proportion of No Media Exposure 0.537 0.001 0.437 0.001 0.115 0.001 0.107 0.001

Table 4: Estimated result from the multivariate OLS regression for all districts (n=640) among under 5 children (n=259627)
Stunting Under Weight Wasting Anemia
Adjusted R2 0.545 0.560 0.222 0.330
β s.e P β s.e P β s.e P β s.e P
API -0.505 0.521 0.332 0.108 0.594 0.856 0.587 0.529 0.268 -3.801*** 0.953 0.000
Proportion of male child 0.162** 0.071 0.022 0.059 0.081 0.465 0.065 0.072 0.365 0.378*** 0.130 0.004
Proportion of rural -0.036** 0.017 0.029 -0.074*** 0.019 0.000 -0.049*** 0.017 0.004 -0.023 0.030 0.456
Proportion of hindus 0.057*** 0.011 0.000 0.181*** 0.013 0.000 0.123*** 0.012 0.000 0.148*** 0.021 0.000
Proportion of sc/st -0.026* 0.015 0.074 -0.001 0.017 0.931 0.048*** 0.015 0.001 0.031 0.027 0.238
Proportion of illiterate mothers 0.178*** 0.024 0.000 0.171*** 0.028 0.000 0.051** 0.024 0.036 0.562*** 0.044 0.000
Proportion of poor 0.094*** 0.033 0.005 0.129*** 0.038 0.001 0.053 0.033 0.117 -0.008 0.060 0.900
proportion of no media exposed 0.108*** 0.041 0.009 0.074 0.047 0.115 -0.013 0.042 0.753 -0.252*** 0.075 0.001
***p<0.01, **p<0.05, *p<0.10

Table 5: Estimated result from the multivariate OLS regression for all districts (n=640) among under 5 children (n=259627)
Stunting Under Weight Wasting Anemia
Adjusted R2 0.545 0.561 0.226 0.313
β s.e P β s.e P β s.e P β s.e P
VHI -1.233 1.857 0.507 3.048 2.116 0.150 4.246** 1.881 0.024 -0.243 3.441 0.944
Proportion of male child 0.167** 0.071 0.019 0.054 0.081 0.502 0.056 0.072 0.433 0.397*** 0.131 0.003
Proportion of rural -0.033** 0.017 0.045 -0.078*** 0.019 0.000 -0.056*** 0.017 0.001 -0.011 0.031 0.713
Proportion of hindus 0.057*** 0.011 0.000 0.179*** 0.013 0.000 0.122*** 0.012 0.000 0.143*** 0.021 0.000
Proportion of sc/st -0.027* 0.014 0.060 -0.004 0.016 0.817 0.047*** 0.015 0.001 0.014 0.027 0.595
Proportion of illiterate mothers 0.183*** 0.024 0.000 0.166*** 0.027 0.000 0.042* 0.024 0.087 0.586*** 0.045 0.000
Proportion of poor 0.08** 0.032 0.012 0.141*** 0.036 0.000 0.077** 0.032 0.018 -0.079 0.059 0.185
Proportion of no media exposed 0.114*** 0.041 0.005 0.076 0.046 0.101 -0.017 0.041 0.672 -0.196*** 0.075 0.009
***p<0.01, **p<0.05, *p<0.10

21 | P a g e
Table 6: Estimated result from the multivariate OLS regression for all districts (n=640) among under 5 children (n=259627
Stunting Under Weight Wasting Anemia
Adjusted R2 0.548 0.560 0.221 0.314
β s.e P β s.e P β s.e P β s.e P
SDCI -3.788** 1.895 0.046 -1.753 2.168 0.419 1.16 1.932 0.548 -1.665 3.520 0.636
Proportion of male child 0.168** 0.071 0.018 0.06 0.081 0.459 0.061 0.072 0.394 0.398*** 0.131 0.002
Proportion of rural -0.031* 0.017 0.064 -0.073*** 0.019 0.000 -0.052*** 0.017 0.002 -0.01 0.031 0.749
Proportion of hindus 0.055*** 0.011 0.000 0.181*** 0.013 0.000 0.124*** 0.012 0.000 0.142*** 0.021 0.000
Proportion of sc/st -0.024 0.015 0.103 0.001 0.017 0.941 0.05*** 0.015 0.001 0.016 0.027 0.550
Proportion of illiterate mothers 0.19*** 0.024 0.000 0.174*** 0.028 0.000 0.045* 0.025 0.069 0.589*** 0.045 0.000
Proportion of poor 0.076** 0.032 0.017 0.127*** 0.036 0.000 0.066** 0.032 0.041 -0.082 0.059 0.165
Proportion of no media exposed 0.106*** 0.041 0.010 0.068 0.047 0.143 -0.019 0.041 0.649 -0.2*** 0.076 0.008
***p<0.01, **p<0.05, *p<0.10

Table 7: Estimated result from the multivariate OLS regression for all districts (n=640) among under 5 children (n=259627)
Stunting Under Weight Wasting Anemia
Adjusted R2 0.553 0.581 0.221 0.316
β s.e P β s.e P β s.e P β s.e P
Elevation -0.001*** 0.000 0.001 -0.002*** 0.000 0.000 0.000 0.000 0.340 0.001 0.001 0.113
Proportion of male child 0.178** 0.070 0.012 0.082 0.079 0.298 0.066 0.072 0.360 0.386*** 0.131 0.003
Proportion of rural -0.02 0.017 0.243 -0.047** 0.019 0.014 -0.047*** 0.017 0.007 -0.025 0.032 0.438
Proportion of hindus 0.047*** 0.012 0.000 0.163*** 0.013 0.000 0.121*** 0.012 0.000 0.151*** 0.022 0.000
Proportion of sc/st -0.024* 0.014 0.089 0.006 0.016 0.689 0.052*** 0.015 0.000 0.011 0.027 0.691
Proportion of illiterate mothers 0.189*** 0.024 0.000 0.183*** 0.027 0.000 0.05** 0.024 0.041 0.579*** 0.044 0.000
Proportion of poor 0.082*** 0.031 0.009 0.127*** 0.035 0.000 0.063* 0.032 0.050 -0.076 0.058 0.193
Proportion of no media exposed 0.097** 0.040 0.017 0.039 0.045 0.388 -0.027 0.041 0.515 -0.18** 0.075 0.017
***p<0.01, **p<0.05, *p<0.10

22 | P a g e