NFHS-4: Districts in that could aspire to become malnutrition free early! (Part-2)


This is further to the preliminary analysis of the NFHS-4 data for rural part of Odisha.

This note identifies potential districts which can aspire to become malnutrition free early. This analysis shifts the emphasis from the high burden districts to the least burden districts. Such a shift is necessary given the goal of achieving a malnutrition free India with emphasis on district level planning, convergent action and monitoring. For this it is necessary to identify districts which have the best potential to become malnutrition free and become a role model other districts in the state. Ernakulum did provide an example of such a “role model” during the total literacy campaign.


Table – 1 Levels of malnutrition in districts of Odisha NFHS-4

The methodology can be used for other states as well. Every state has districts that have defied odds and have brought down levels of malnutrition well below the state average. These are, in certain sense, closest to the “goal – post” of becoming malnutrition free. These are “positive deviants” within a state which we need to identify, celebrate and push systematically towards becoming malnutrition free. We illustrate below one such method of identification of such districts based on the NFHS-4 data.

To recapitulate, the earlier analysis had presented the data on levels of malnutrition at district level, colour coded by high (red), medium (yellow) and low (green) levels of malnutrition (see table 1).

While the above table itself allows us to identify the top three to four, low burden districts, the pattern becomes clearer if we arrange the districts in ascending order of underweight and take a look at top 5-6 districts. This is done in Table – 2 below. For Odisha, Jagatsinghpur leads the pack with Cuttack, Puri and Khordha following closely. We select districts with least wasting and underweight rather than least stunting, for, removing stunting is a long haul while addressing wasting is relatively quicker and the number of children to be taken care of is less. We therefore, must tackle wasting first and to the extent possible underweight. We should also note the low levels of severe wasting in these districts as pointed in earlier analysis as well.


Table – 2 Levels of malnutrition in least burden districts in Odisha

Having identified these districts, we next look at various correlates of malnutrition and find out the district which have performed well above the state average in respect of maximum number of correlates. NFHS-4 provides data in respect of a number of such parameters. We have selected 43 such parameters and have arranged these in the life cycle sequence starting from the new born to the adolescent girl and the pregnant mother. The parameters have further been organized into major clusters i.e. child anthropometry, infant and young child feeding practices, immunization, management of diarrhoea and ARI, women’s marriage, health, ante natal care, post-natal care and educational background among others. These indicators have again been colour coded depending upon their levels with level to the state averages, and in some cases where the state average is itself rather low, the colour coding has been done in terms of absolute performance.

Table 3: Levels of malnutrition in least burden districts in Odisha (Source: NFHS-4)

Jagatsinghpur and Puri have the largest number of well performing indicators, not Cuttack and Khordha. Further, these two districts perform below par mainly in the IYCF segment, i.e. early initiation of breastfeeding, and adequacy of diet to young children. Other two important parameters are the sanitation coverage and mothers with 10 plus years of schooling. The above analysis is represented through radar diagram where each cluster has been assigned certain score and the achievement of the district marked against the full score. While doing an individual district analysis, it is also possible to do a full parameter radar diagram. Such a radar diagram allows the district to prioritise its areas of intervention. The issue of relative weightage of these parameters will however, remain. That will be taken up separately.


Figure 1: Radar diagram showing performance of best selected districts on different indicators that affect nutritional status of children

Burden of the problem: We now turn to an interesting aspect of the problem- the burden in absolute numbers. How many children are there likely to be moderately or severely malnourished? Every district will like to know this. We have used an easy but reasonably accurate estimate based on the census 2011 data for children in the 0-6 year age group. This number represents seven cohorts. As such 5/7th of this number will give rough estimate of the children below 60 months in the given district. Using NFHS-4 data we estimate number of malnourished children.


Table 4 above shows that Jagatsinghpur and Puri have the least burden of malnourished children in absolute number. In fact, the number of severely wasted children in both the districts is just below 3000. The corresponding figure for moderately wasted is about 9,000 and 12,000 respectively while children that are underweight are about 12,000 and 18,000 respectively. The average number of children within most Anganwadis will be well below a double digit figure. If we look at the ICDS project wise figure e.g. the 8 blocks of Jagatsinghpur, the numbers will be even more manageable even if we use the % underweight figures from an earlier survey CCM-II done for Odisha in 2011.


It can be nobody’s case that these districts cannot take up the task of reducing malnutrition in a campaign mode and achieve early results. The time to begin this is NOW !

[PS: Odisha is lucky to have block wise data on nutritional status and other parameters through an earlier, large sample, survey CCM-II. We will carry out similar exercise with block as a unit to identify the “closest to goal post” blocks in every district. That will be presented in the next part of our analysis.]

Satish B Agnihotri, Ayushi Jain

Nutrition Discussion Group, CTARA, IIT Bombay, Powai Mumbai 400076

Contact us: 9810307353 (Mobile), 022 576 6476

Email ID:,


NFHS-4: An analysis of district level malnutrition data for Odisha


NFHS – 4 provides, for the first time a district level data on nutritional status of children below the age of 5 years. This provides an excellent and timely opportunity to plan for eradication of child malnutrition at the district level. A quick preliminary analysis of the district level child malnutrition levels, reveals certain important aspects. This is presented below.

Table -1 below presents the data on wasting, underweight and stunting in a colour coded form. These three aspects of child malnutrition are interrelated. This relationship is brought out in Figure 1a. There is a clear linear relationship between underweight on one hand and stunting and wasting on the other. The very robust nature of both the linear regressions (R Sq of 0.88 for stunting and 0.76 for wasting), has a bearing on programme implemention. Collecting good quality data on underweight can give us a good indication of the levels of wasting and stunting as well. Hence we need not initiate routine measurement of height through Anganwadi workers or ASHAs. The task of estimating stunting can be left to periodical NFHS surveys which will now be taking place at 3 year intervals1. At Anganwadi level recording weight and use of MUAC tapes to identify wasting will be adequate at this stage.


Table – 1                                                                              Figure 1a and 1b

Measurement of height or length (child below 2 years) is not an easy task and is error prone if done by workers not adequately skilled and experienced. As such burdening the Anganwadi workers with this task is best avoided.

It is useful to rearrange Table – 1 in descending order of underweight. This brings out the low levels of severe wasting in coastal belt. Districts which have low figures of severe wasting have mostly got other parameters right, and, more importantly, where this has gone wrong, other parameters have gone wrong too.

Table -2, quite clearly, brings out the importance of reducing the incidence of severe wasting. It also shows the regional contiguity of the parameters and the need for separate planning for different districts.


The coastal districts of Jagatsinghpur, Cuttack, Kendrapada, Puri and Khordha have done well in all parameters. While Ganjam shows better result in underweight, it needs to improve on rest of the parameters. The case of Nayagarh is interesting, a modest reduction in underweight and stunting (not colour coded purposefully) would have put it in the league of the other coastal districts. Baleshwar and Jajpur are somewhat of a surprise and may need a closer scrutiny.

The Angul, Jharsuguda, Debgarh belt has intermediate position along with Gajpati, which shows a low incidence of wasting. Whether this is consistent or a one off case needs to be seen.

On the other end of the spectrum, districts of Southern and some of the districts in Western Odisha have not fared well. Mayurbhanj and Keonjhar two contiguous tribal districts have done relatively better in reducing wasting, but not other parameters.

The regional dimension of the situation can be readily appreciated if we look at the Odisha map as brought in Gigure 2a-2d below. The maps clearly show three separate clusters in green, yellow and red corresponding to the least, middle level and high malnutrition.


Figure 2a – Wastingwasting

Figure 2b – Stuntingtext

Figure 2c – Severe Wasting


Figure 2d – Underweight

There is a very clear case for taking Dhenkanal, Nayagarh, Gajapati and Ganjam belt from the yellow zone to the green zone. So is the case with the Jajpur – Bhadrak – Balasore zone.

One area of focus would be the Jagatsinghpur, Kendrapada and Cuttack belt which is‘closest to the goal-post’ of removing moderate and severe malnutrition. It may be a useful idea to monitor all the severely wasted and underweight children on an intensive basis and take remedial measures.

This preliminary analysis is useful in indicating where does the shoe pinch the most. A more detailed analysis needs to be done by looking at other parameters under NFHS-4 i.e. the correlates of child malnutrition. This is presented in the next stage of the analysis where we look at the districts with the best potential to achieve the status of being “malnutrition free”.

Satish B Agnihotri, Ayushi Jain Tel 9810307353 (Mobile)

‘Energy Access‘: Bringing the data together

While individual maps are helpful in understanding the spatial distribution of a given source of energy, these do not help us assimilate the overall picture at a glance. This can be done quietly elegantly by arranging the data in a table and colour code the cell to depict the high, medium and low values.

Table -1 shows such an arrangement using state-wise data for different sources of lighting. The table is quite revealing. It clearly shows that between themselves electricity and kerosene covers almost all the households barring small percentage of households covered by solar and ‘any other’.

Use of kerosene in urban areas for lighting needs to be viewed seriously and brought down speedily.

The position of Bihar is quite eye opener. Both rural and urban Bihar are doing quite badly both in electricity consumption and kerosene consumption. Rural UP also fares very badly though urban UP seems to have fared better than Bihar.

Table – 2 provides similar information in respect of sources of fuel. Here the variables are many more. Readers are expected to draw their own conclusion. The figure of kerosene consumption is urban Chandigarh is very intriguing. We also need to note similar situation in urban Tamil Nadu and Puducherry. Low LPG penetration in eastern states is also a matter of concern. Colour coding for the total (Rural + Urban) category has not been done. Reader is left to do this as a useful exercise.

We will next move to a state by state analysis with block / tehsil level data. Readers are requested to give a feedback on the colour coding methodology for a tabular data.