The resulted scores and ranks obtained by states and districts can be found in Additional File1.
3.1. Axioms Test
We first compared Linear aggregation and Geometric Mean method against MANUSH axioms, viz, Monotonicity, Anonymity, Normalization, Uniformity, Shortfall sensitivity and Hiatus sensitivity to level to bring out the differences in scoring patterns, if any, using districts as study unit. It is seen that all the three aggregation measures (Linear aggregation, Geometric Mean and MANUSH) satisfy the first three axioms while Geometric Mean satisfies additional condition of Uniformity (See Additional File 2). However, conditions of Shortfall Sensitivity and Hiatus Sensitivity to level is satisfied by MANUSH only, which makes it more robust compared to the other two aggregation measures, and have been presented below.
Shortfall Sensitivity. A measure of aggregation (LA, GM or MANUSH) should be such that for a given reduction in the aggregation value along its optimal path between two situations the reduction across dimensions should be in proportion to the shortfalls in the worse-off dimensions.
If weighted standardized values in situation (#1) are such that wasting (q1) is greater than stunting (r1) which is greater than overweight/obesity (s1) which is greater than anemia (t1), or, q1> r1> s1> t1, then, in situation (#2) the reduction in failures should be such that (q1- q2) ≥ (q1/ r1) (r1- r2), (r1- r2) ≥ (r1/ s1) (s1- s2), and (s1- s2) ≥(s1/ t1) (t1- t2). We see situation 1a (Upper Siang in Arunanchal Pradesh) and situation 1b (Kollam in Kerala) in Table 1 mimic the same order of indicators, where wasting represents the worse-off dimension and anaemia represents the best-off dimension. Kollam when compared to Upper Siang, shows a lower reduction, in the worse off dimension, i.e. in wasting, when compared to anaemia, it's best off dimension, and hence is penalized by MANUSH measure (α ≥2). We even notice that when we increase α measure, that is α=3, 5 and 10, Kollam is further penalized on account of disproportionate reduction in its best -off and worse-off dimension. Conversely, when we study the situation 2a and 2b in table below, of Dakshin Dinajpur of West Bengal and Kanniyakumari of Tami Nadu, which follow the same order of indicator values, where, stunting represents the worse off dimension, followed by anaemia, wasting and overweight/obesity, represents the best-off dimension, we see that Kanniyakumari when compared to Dakshin Dinajpur, shows more reduction in its worse-off dimension, i.e. stunting, in comparison to its better off dimensions, it is not penalized by MANUSH measure (α ≥2), on account of being sensitive to shortfall across its best-off and worse-off dimension.
Table 1: Cases explaining the condition of Shortfall Sensitivity
|
|
|
Weighted Standardized values (In descending order)
|
|
Situation
|
District
|
State
|
q = wyWA
|
r = wyST
|
s = wyOW
|
t = wyAN
|
q1-q2
|
(q1/r1)*(r1-r2)
|
r1-r2
|
(r1/s1)*(s1-s2)
|
(s1-s2)
|
(s1/t1)*(t1-t2)
|
1
|
a
|
Upper Siang
|
Arunanchal Pradesh
|
0.18
|
0.09
|
0.08
|
0.07
|
0.06
|
0.07
|
0.04
|
0.03
|
0.03
|
0.05
|
b
|
Kollam
|
Kerala
|
0.11
|
0.06
|
0.05
|
0.03
|
Situation
|
District
|
State
|
q = wyST
|
r = wyAN
|
s = wyWA
|
t = wyOW
|
q1-q2
|
(q1/r1)*(r1-r2)
|
r1-r2
|
(r1/s1)*(s1-s2)
|
(s1-s2)
|
(s1/t1)*(t1-t2)
|
2
|
a
|
Dakshin Dinajpur
|
West Bengal
|
0.13
|
0.11
|
0.10
|
0.06
|
0.06
|
0.06
|
0.05
|
0.05
|
0.05
|
0.04
|
b
|
Kanniyakumari
|
Tamil Nadu
|
0.07
|
0.06
|
0.05
|
0.04
|
Table 1 continued
|
|
|
Aggregation Scores
|
Situation
|
District
|
State
|
GM
|
LA
|
MANUSH (α = 2)
|
MANUSH (α = 3)
|
MANUSH (α = 5)
|
MANUSH (α = 10)
|
1
|
a
|
Upper Siang
|
Arunanchal Pradesh
|
0.439
|
0.426
|
0.447
|
0.470
|
0.510
|
0.559
|
b
|
Kollam
|
Kerala
|
0.255
|
0.251
|
0.273
|
0.294
|
0.325
|
0.357
|
Situation
|
District
|
State
|
State
|
LA
|
MANUSH (α = 2)
|
MANUSH (α = 3)
|
MANUSH (α = 5)
|
MANUSH (α = 10)
|
2
|
a
|
Dakshin Dinajpur
|
West Bengal
|
0.427
|
0.401
|
0.404
|
0.406
|
0.409
|
0.413
|
b
|
Kanniyakumari
|
Tamil Nadu
|
0.226
|
0.221
|
0.220
|
0.220
|
0.220
|
0.220
|
This indicates that neither geometric mean nor linear aggregation methods are shortfall sensitive. Few more cases that explain shortfall sensitivity, and are reported in Additional File 1.
Hiatus Sensitivity to Level. A measure of aggregation (LA, GM or MANUSH) should be such that the same gap (or hiatus) across dimensions should be considered worse-off as average failure decreases.
For instance, if standardized values of stunting, wasting, anaemia and overweight/obesity are 0.18, 0.17, 0.13 and 0.03, respectively, in case #1 and 0.17, 0.16, 0.12 and 0.02, respectively, in case #2, as observed in Surendranagar of Gujarat and Bhojpur of Bihar, which has the same gap across dimension, then value of aggregation in case #1, in comparison to case #2, should have a lower deviation from its average failure (µ). We see that except MANUSH aggregation method, neither linear aggregation nor geometric mean, has a lower deviation from the average failure, in case of Surendranagar. Hence, the gap in MANUSH score between the two districts, i.e. Surendranagar and Bhojpur, are found to be narrow, compared to the gap in linear and geometric mean scores between two districts. Hence MANUSH is also hiatus sensitive. What this condition means, in essence, is that the gap between different dimensions should continue to reduce as a district moves towards a greater reduction in failure.
Table 2: Cases explaining the condition of Hiatus Sensitivity to Level
|
|
|
Weighted Standardized values
|
Weighted Average
|
Aggregation Scores
|
ʋk
|
Case
|
Districts
|
State
|
wyST
|
wyWA
|
wyAN
|
wyOW
|
µ
|
GM
|
LA
|
MANUSH (α = 2)
|
GM - µ
|
LA - µ
|
MANUSH - µ
|
#1
|
Surendranagar
|
Gujarat
|
0.18
|
0.17
|
0.13
|
0.03
|
0.50
|
0.56
|
0.50
|
0.54
|
0.06
|
0.00
|
0.04
|
#2
|
Bhojpur
|
Bihar
|
0.17
|
0.16
|
0.12
|
0.02
|
0.46
|
0.51
|
0.46
|
0.51
|
0.05
|
0.00
|
0.05
|
We next highlight and discuss the implication of MANUSH indexing in assessing nutrition burden across states and districts in India and in developing national policies on eradicating malnutrition
3.2. Comparing the nutrition scenario in India at a different level of aggregation:
a) Nutrition scenario across State Boundaries in India:
Comparing the nutrition status in states of India over three different periods of surveys, i.e. from NFHS-3 (2005-06) to NFHS-4 (2015-16) to CNNS (2016-18), we see a significant upward shift at the state level, as seen in Figure 1a-c.
At the time of NFHS-3 (see Figure 1a), four states fell under extremely alarming category (MANUSH score ≥0.65), namely Madhya Pradesh (0.73), Meghalaya (0.72), Jharkhand (0.70) and Bihar (0.69) on account of extremely high prevalence of stunted and wasted children below the age of five. Seven states – Uttar Pradesh (0.61), Sikkim (0.61), Chhattisgarh (0.61), Gujarat (0.59), Haryana (0.56), Rajasthan (0.56) and Odisha (0.55) were found in the alarming category as per MANUSH scores. All the seven states but Sikkim were penalized for the high prevalence of stunting, whereas Sikkim was accounted for the high prevalence of both stunted and overweight/obese children. Only four states – Punjab (0.43), Goa (0.41), Manipur (0.39) and Kerala (0.36) were found in the moderate category with MANUSH score in the range of ≥0.35 and <0.45. Here we see that although Manipur fared better in all dimensions (weighted standardized score wasting 0.07, anaemia 0.08, overweight 0.06) except stunting (weighted standardized score 0.16), compared to Kerala (stunting 0.11, wasting 0.12, anaemia 0.08, overweight 0.03), yet Manipur was penalized on account of unbalanced development across its dimensions. None of the states was found to fall under low category (MANUSH score <0.35) and about 13 states were found to fall under the serious category (MANUSH score ≥0.45 and <0.55). We had to exclude Nagaland from NFHS-3 study as the anaemia prevalence was not known.
Figure 1: Map of India depicting state categorization on the severity scale based on MANUSH scores calculated based on a) NFHS-3 b) NFHS-4 c) CNNS
Next, we study the nutrition scenario across states in India after NFHS-4 round (See Figure 1b). On comparing the nutrition scenario in states surveyed in NFHS-4 with previous round, i.e. NFHS-3, it is found that none of the states falls in the extremely alarming category (MANUSH score ≥0.65), except Dadra and Nagar Haveli (MANUSH Score 0.67), which is a Union Territory (UT) and have been penalized on performing poorly on all four dimensions. Three states, Jharkhand (0.64), Madhya Pradesh (0.59) and Bihar (0.58), when compared to NFHS-3, have shifted from extremely alarming to alarming category on account of reduction in stunting and wasting, although overweight/obesity shows marginal rise in children below five years in these states with no or minimal reduction in anaemia. Sikkim (0.60) and Gujarat (0.57) show no change in the category on the severity scale and are still found in the alarming category in NFHS-4 round. Although, Gujarat shows a reduction in stunting, but is penalized on account of increased prevalence of wasting. Sikkim also shows a reduction in stunting but is penalized on account of rise in both wasting and overweight/obesity in children under the age of five. On the other hand, Karnataka (0.56) shows a downward fall from serious to alarming category between NFHS-3 and NFHS-4. It is penalized on account of rise in prevalence of both wasting and anaemia, with no reduction in overweight/obesity in children below five years.
Meghalaya (0.52) shows a significant upward transition from extremely alarming to serious category between two rounds of the survey and has been duly rewarded on account of a significant reduction in wasting, followed by a reduction in stunting and anaemia, although it shows a rise in overweight/obesity in children under the age of five. Uttar Pradesh (0.54), Rajasthan (0.54), Haryana (0.53), Chhattisgarh (0.52) and Odisha (0.47) also show one -scale upward shift from alarming to the serious category in NFHS-4 on account of robust decline in stunting, followed by anaemia. Although, except Odisha, we see an increase in wasting in all the four states. On the other hand, Goa (0.45) shows one-scale shift downwards, i.e. from moderate to serious category and has been penalized on account of an unprecedented rise in wasting and anaemia, despite a small reduction in stunting and children under five years. Maharashtra (0.53), Uttarakhand (0.50), Tamil Nadu (0.49), Arunanchal Pradesh (0.49), West Bengal (0.47) and Jammu and Kashmir (0.47), continue to be in serious category. Although, except Jammu and Kashmir, all the five states show a reduction in MANUSH scores, compared to NFHS-3. It is observed that Jammu and Kashmir, is penalized on account of a significant increase in the prevalence of overweight/obesity, despite the reduction in all the other three dimensions. Daman and Diu (0.50) and Puducherry (0.45), the union territories, are also found in serious category, with former penalized on account of a high prevalence of wasting and anaemia and later due to high prevalence of wasting with respect to other dimensions.
Assam (0.44), Delhi (0.44), Andhra Pradesh (0.44), Telangana (0.43), Tripura (0.40), Himachal Pradesh (0.38) and Mizoram (0.35) show a gradual shift from serious to moderate category. While Assam is rewarded for the decline in stunting and anaemia, Delhi shows a reduction in overweight/obese children followed by stunting. Tripura shows a reduction in all parameters but overweight, which is found to increase between the two rounds of the survey. Himachal Pradesh shows a modest decline in both stunting and wasting. Mizoram is the only state which shows a decline in all four indicators. Punjab (0.40) and Kerala (0.36) continue in the moderate category. Punjab is penalized on account of a dramatic rise in wasting, followed by overweight and a minimal reduction in anaemia. Whereas in case of Kerala, although there is a marginal decrease in stunting and anaemic children, with no reduction in wasting, however, there is a dramatic increase in the prevalence of overweight/obese children, and thus Kerala seems to have been penalized on this account. Andaman and Nicobar Island (0.42), Chandigarh (0.40), Lakshadweep (0.38) and Nagaland (0.37) were also found in the moderate category.
Manipur (0.33) jumps from moderate to low category and is the only state found in the low category on the severity scale. It shows a significant reduction in stunting, wasting and anaemia, although overweight children are found to increase in prevalence.
Except, Kerala, Jammu and Kashmir, Karnataka and Goa, all the states show a reduction in MANUSH scores, between NFHS-3 and NFHS-4 (See Figure 2). Meghalaya shows maximum improvement between the two rounds, followed by Tripura and Mizoram, accounting to 25% or more reduction in MANUSH scores. Himachal Pradesh also shows about 23% reduction in MANUSH scores. Madhya Pradesh and Bihar, the two states accounting to the maximum burden of undernourished children, also show significant reduction of 17% -18% in MANUSH scores, suggesting that these states are progressing in the right direction in terms of addressing malnutrition, although steadily. Same is the case with Delhi, Assam, Manipur, Chhattisgarh, Odisha, Andhra Pradesh, Uttar Pradesh and West Bengal that show 10%-15% reduction in MANUSH scores. States like Jharkhand, Uttarakhand, Punjab, Arunanchal Pradesh, Haryana and Tamil Nadu need to improve further as reduction between the two rounds is tardy, about 5-7%. Gujarat, Maharashtra, Rajasthan and Sikkim show the minimum decline (less than 5%), suggesting extremely marginal improvement between the two rounds. States like Kerala, Jammu and Kashmir and Karnataka, on the other hand, show a marginal increase (≤5%) in MANUSH scores between two rounds, suggesting unbalanced development across the dimensions. Goa shows about 11% increase in MANUSH scores and has been heavily penalized on account of the unprecedented increase in wasting in children under the age of five.
Figure 2: Percent reduction in MANUSH scores across states in India between NFHS-3 and NFHS-4 rounds of the survey
Recently released Comprehensive National Nutrition Survey (2016-18) conducted in 29 states of India and Delhi (Union Territory), paints a better picture of nutrition status in children across states in India, compared to previous two rounds of National Family Health Survey (See Figure 1c). None of the states is found in extremely alarming category (MANUSH score >0.65). Jharkhand continues to be an alarming category (0.56), although it shows a modest reduction in MANUSH scores between NFHS-4 and CNNS. Jharkhand shows a reduction in all parameters except wasting, which is found to be constant. Surprisingly, Nagaland (0.56) also joins the list of alarming category, showing the 2-scale downward shift from moderate category in NFHS-4. Nagaland is heavily penalized on account of a significant rise in overweight/obese children, followed by a marginal increase in wasting.
Madhya Pradesh (0.49), Gujarat (0.46) and Bihar (0.46) show an upward jump from severe to moderate category. While Madhya Pradesh and Bihar show a modest decline in all four dimensions, Gujarat, on the other hand, shows a significant decline in wasting and anaemia, although there is a marginal increment in stunting and no change is observed in overweight prevalence in children under the age of five. It is important to note that despite the decline in all parameters, Madhya Pradesh and Bihar have been penalized by MANUSH on account shortfall sensitivity since the reduction in the worse-off dimension (stunting) is not found to be proportionate to their best-off dimension (overweight). Assam (0.50) and Tripura (0.47) show a decline in performance and fall from moderate to severe category. Both Assam and Tripura have been penalized on account a significant increase in overweight/obese children. While wasting continue to increase in Assam, in Tripura, it is stunting, that is shown to increase between two surveys (NFHS-4 and CNNS). Maharashtra (0.47), Uttar Pradesh (0.47) and Meghalaya (0.46) continue to remain in a severe category despite a decrease in the MANUSH scores. While in Maharashtra, there has been a decrease in the prevalence of wasting and anaemia, no change is seen in the prevalence of stunting; moreover, overweight prevalence is on the rise. In Uttar Pradesh, the decline in stunting and anaemia is observed with increase in wasting and overweight in children at the same time. Meghalaya, on the other hand, shows a decline in all four dimensions, and yet is penalized by MANUSH on account of shortfall sensitivity, i.e. the decline in its worse-off dimension, stunting, is not in proportion to decline in its best-off dimension, anaemia.
A large group of states is found in the moderate category in the Comprehensive National Nutrition Survey (2016-18). Sikkim (0.37) and Karnataka (0.42) show 2-scale improvement and move from alarming to moderate category. While Sikkim shows a significant decline in all four dimensions, Karnataka shows a major decline in anaemia, followed by stunting; however, overweight and wasting are seen to increase dramatically. Eight states, namely, Chhattisgarh (0.45), Jammu & Kashmir (0.42), West Bengal (0.42), Rajasthan (0.41), Haryana (0.40), Tamil Nadu (0.40), Arunanchal Pradesh (0.39) and Odisha (0.36) show an upward shift from severe to moderate category. Rajasthan and Odisha show decline in all parameters, yet they are penalized by MANUSH on account of shortfall sensitivity. In West Bengal and Tamil Nadu, all parameters show a decline in prevalence but wasting, which remains unchanged in case of West Bengal and is found to increase in Tamil Nadu. Similarly, Haryana also shows a decline in all dimensions, except stunting, which shows a slight increase. Chhattisgarh manages to reduce the prevalence of overweight significantly, followed by stunting and wasting, with no reduction in anaemia. In Jammu and Kashmir, there is a significant decline in both stunting and anaemia, however wasting and overweight children tend to increase. On the contrary, Arunanchal Pradesh shows a significant decline in both wasting and anaemia, however stunting and overweight remain unaffected. Interestingly, Manipur (0.42), which ranked 1st in NFHS-4 survey drops to 17th rank in CNNS, thus falling from low to moderate category, on account of the unprecedented rise in overweight/obese children, no change in prevalence of stunted and wasted children and slow decline in anaemic children. Andhra Pradesh (0.41), Telangana (0.40), Mizoram (0.39) and Delhi (0.39) continue in moderate category. While the prevalence of overweight in children seems to be a rise in both Telangana and Andhra Pradesh, accompanied by a significant decline in anaemic children, a slight decrease in wasting in observed in Telangana, which remains unaffected in Andhra Pradesh along with the prevalence of stunting. On the other hand, Delhi which managed to reduce overweight significantly in children between NFHS-3 and NFHS-4 is penalized in CNNS round on account of a high prevalence of overweight children, while other indicators show a decline after NFHS-4. Mizoram also shows a slight decrease in stunting and wasting, with a slight increase in anaemic and overweight children. Uttarakhand (0.34) and Goa (0.34) move upwards from severe to low category. Uttarakhand shows a decline in all dimensions with a significant reduction in wasting and anaemia. Goa also shows a significant decline in wasting and anaemia, followed by overweight. However, no change in the status of stunting is observed.
Himachal Pradesh (0.35), Punjab (0.34) and Kerala (0.28) move upwards from moderate to low category. Both Himachal Pradesh and Punjab show a modest decline in wasting and anaemia, with a slight increase in stunting. Prevalence of overweight seems to increase significantly in Punjab between NHFS-4 and CNNS. Kerala, on the other hand, shows a rapid decline in anaemia and overweight, followed by wasting, while stunting remains unaffected.
b) Nutrition scenario across District Boundaries in India:
Next, we study the nutrition scenario across 640 districts of India, based on MANUSH score and ranking (See Figure 3). Since NFHS-4 is the only survey to provide district-level estimates for all districts in India, we use data of this survey for our study.
Pockets of districts falling under extremely alarming, alarming, serious and moderate range emerge from the central and spread across the posterior end of the country in the said order. Two districts (Pashchimi Singhbhum of Jharkhand and The Dangs of Gujarat) fall in the extremely alarming range (score >0.65), followed by 145 districts in the alarming range, a large proportion belonging to Madhya Pradesh, Uttar Pradesh, Jharkhand, Bihar, Rajasthan and Gujarat. 352 districts fall in the serious range, 140 districts in the moderate range and only one district (Mokokchung district of Nagaland) in low range.
Figure 3: Map of India depicting district categorization on the severity scale based on MANUSH scores calculated using CNNS data
Of the districts falling in alarming and extremely alarming category (≥0.50 MANUSH score), i.e. a total of 147 districts, 40% of theme have been penalized on account of high incidences of wasting, about 30% have been penalized due to high incidences of stunting and about 22% have higher prevalence of both stunting and wasting. Interestingly, Erode district in Tamil Nadu is severely penalized due to a very high incidence of overweight/obese children under the age of five. Few other districts like Chennai of Tamil Nadu, Ambala in Haryana, Banda of Uttar Pradesh, South District of Sikkim, and Ambala district of Haryana have been penalized on account of overweight/obese children in addition to undernourished children. All the districts in the bottom two category have medium to a high prevalence of anaemic children under the age of five.
Figure 4 below shows the performance of each state along with the MANUSH scores obtained by the best and worst-performing districts in each state with respect to the state averages arranged in descending order. For some states, the spread is quite compact, e.g. in Andhra Pradesh and quite large for some, e.g. Odisha and, surprisingly, even Tamil Nadu.
Figure 4: States arranged on the basis of moving average of MANUSH scores of districts. Note: On the y-axis, the value in brackets indicates the number of districts in that particular state.
Interestingly, Odisha's best performing district (Cuttack) with undernutrition score of 0.21, ranks 4th in the country while its worst-performing district, Nabarangpur, with a score of 0.60, ranks 623rd is amongst the worst-performing districts in the country. Similarly, Kanniyakumari of Tamil Nadu with a score of 0.22 ranks 7th in the country, while Chennai in Tamil Nadu ranks 611th and falls in the bottom list of worst-performing districts. Chennai shows higher prevalence in all the four indicators – overweight/obesity, stunting, wasting and anaemia in children under the age of five years. If one takes a closer look at the spread in the case of Odisha, the best and the worst-performing districts appear to form a cluster. A similar pattern is observed in West Bengal. In contrast to the case of Odisha and West Bengal, the states of Bihar and Madhya Pradesh each with 38 and 50 districts, have a low average MANSUH score for their state and a small deviation between the scores of their best performing and worst-performing districts, reflecting the uniformly poor performance across the state. Maharashtra, Karnataka and Rajasthan show a comparable spread. Such inequalities within and across states level reflect spatial heterogeneities that exist in India.
To study the spatial heterogeneity at the district level, we created Univariate LISA maps using MANUSH scores using GeoDa version 1.14.0. As shown in Figure 5a, we observe the formation of significant clusters (p<0.05) and the univariate Moran’s I value was found to be 0.619, depicting strong spatial autocorrelation. About 135 districts out of 640 districts, i.e. 20% were surrounded by districts with high MANUSH scores, signifying clusters of high malnutrition, while 108 districts, i.e. around 17% districts were surrounded by districts with low MANUSH scores, signifying better nutrition scenario. The high-high clusters were mostly found in the central belt of India, in the states of Madhya Pradesh, Jharkhand, Maharashtra, Bihar, Rajasthan, Uttar Pradesh, Gujarat and some parts of Chhattisgarh, Odisha, West Bengal and Karnataka. While low-low clusters were found in the peripheral states of India – primarily in the north-eastern states like Nagaland, Manipur, Mizoram, some parts of Assam, Arunanchal Pradesh and Tripura, also in the northern states like Jammu & Kashmir, Himachal Pradesh and Punjab, in the southern coastal belt of Karnataka, Kerala and Tamil Nadu and in eastern coastal belt of Odisha and West Bengal. Interestingly, low-low clusters are also found in few districts of Telangana, bordering Andhra Pradesh. Spatial analysis also brought two groups of outliers, i.e. districts with low MANUSH scores (eight in count) surrounded by districts with high MANUSH scores, that can be grouped under 'positive outliers', and second, districts with high MANUSH score (seven in count) surrounded by districts with low MANUSH scores, that can be grouped under ‘negative outliers’.
Figure 5: Univariate LISA cluster maps of India showing the geographic clustering based on MANUSH scores across districts of India, a) with state boundaries, 2015-16. b) with state and NSS boundaries (as per 68th round of NSS), 2015-16.
Also, when we study the spatial clustering at the National Sample Survey (NSS) region level, the differences within a state become much more evident. NSS regions are the ones used for survey by National Sample Survey Organisation (NSSO) in conducting surveys across India on various socio-economic aspects since 1950. These regions are formed by contiguous grouping districts within a state, in relatively homogenous regions based on geographical features, rural population densities and crop-pattern [29]. As observed in Figure 5b, if one takes a closer look at the spread in the case of Odisha, the best and the worst-performing districts appear to form significant clusters at coastal and southern NSS regions respectively. Similarly, in Karnataka and West Bengal, pockets of low performing and better performing districts appear at the anterior and posterior end, respectively, that are separated by NSS boundaries. Moreover, in states like Chhattisgarh and Bihar, significant clusters of poor-performing districts seem confined to a particular NSS region. A similar scenario is observed in Maharashtra, Gujarat and Rajasthan, where pockets of poor-performing districts seem to grouped together not only in a particular NSS region but also form clusters, across state boundaries, suggesting the spatial dependency of neighbours on malnutrition.
3.3. Implications of MANUSH indexing on nutrition policy:
Scoring and ranking of states and district have become a common phenomenon in recent times, in India, for ease of comparison and prioritization of issues and for development of policies. However, there are two limitations that have been observed in this context. First, if scoring is based on a combination of indicators, linear averaging is the only method of aggregation used, whose limitations have already been discussed in the sections above. Second, if the scoring is done on single parameters, for example, only stunting or wasting or anaemia, and policy is developed based upon it, there is an increased risk of overlooking other indicators.
To understand this, we looked at an actual policy priority measure in the recently launched National Nutrition Mission (NNM) in India. While the Mission objective is to reduce undernutrition in children, adolescent girls and women and achieve targets set for the year 2022 [8], it chose to group districts into three phases, for accelerated intervention, in order of priority. However, the priority was set solely based on the prevalence of stunting. The districts with the highest level of stunting were placed in the list of Phase I districts and so on. If the only objective were to reduce stunting, then this prioritization would have seemed rational. But to reduce undernutrition holistically – stunting, wasting and anaemia, dimensions which are independent and not a substitute for each other, need to be taken together. Not considering other two dimensions, might have led to misallocation of priorities, funds and resources. To comprehend this better, we ranked the same 640 districts according to MANUSH scores and grouped into 3 priority regions – priority 1, 2 and 3. Districts with MANUSH score ≥0.48 were accorded 1st priority, districts with MANUSH score between 0.38 and 0.48 were placed in the 2nd priority list, and districts with MANUSH score less than 0.38 was put in 3rd priority list (see Additional File 1). It is to be noted that MANUSH index discussed in this article, also takes into account wasted and overweight children under the age of five, along with stunted and anaemic children. It needs to highlighted because targets set under the National Nutrition Mission, 2018 do not talk about the reduction in wasting and overweight/obese children, who are on the rise in India.
After grouping the districts into priority 1, 2 and 3 regions, we compared these districts with the priority districts under National Nutrition Mission (NNM), rolled out in three phases, as shown in Figure 6 below. About 8 districts (Dehradun, Ambala, Jamnagar, Chitradurga, Tiruvannamalai, Erode and Dharmapuri) which should have been part of phase 1 of NNM round as per MANUSH ranking, are listed in the 3rd phase of NNM. All the eight districts but Erode have been penalised on account of a high prevalence of wasting. Erode district of Tamil Nadu, on the other hand, is penalised on account of a high prevalence of overweight children, followed by stunted, wasted and anaemic children. If we focussed solely on indicators of undernutrition, then this district would have been neglected, but taking overweight as a measure in MANUSH index, it pinpoints those districts as well who are becoming emerging capital of overnutrition, alongside undernutrition.
Similarly, 37 other districts, that should have been in phase 1 of NNM as per MANUSH ranking are part of the NNM 2nd phase. Majority of these districts too are penalised on the high prevalence of wasting, except South District of Sikkim and Mahe district of Pondicherry, who show a high prevalence of overweight/ obese children under the age of five.
Moving further, 33 districts which should have been part of phase 2 of NNM as per MANUSH ranking are found in the list of districts under NNM 3rd phase. Three districts (Faridabad, Rewari and Kolkata) out of 33 mentioned have been penalised on account of the high prevalence of anaemia while two districts (Tawang of Arunanchal Pradesh and Namakkal of Tamil Nadu) have been penalised on account high incidence of overweight/obese children, while the rest of the districts exhibit high incidences of wasting. Interestingly, Mahamayanagar of Uttar Pradesh, one of the districts among the 33 mentioned above, exhibits a very high burden of stunting, and lower incidence of wasting, anaemia and overweight/obese. It has been penalised by MANUSH on account of unbalanced development across the dimension.
Figure 6: Map of India depicting phase allocation of districts as listed in a) National Nutrition Mission and b) as per MANUSH scores