NFHS Policy Tracker

The National Family and Health Survey (NFHS) has provided critical data on India's health, nutrition, and population indicators. The Government of India (GOI) utilizes many of the NFHS indicators to formulate policies and track its progress. Since 2015-16, the fourth round of the NFHS (NFHS-4) became the only data representative at the district level for India. The fact sheets from the fifth round of the NFHS (NFHS-5) conducted in 2019-20 are now available for 342 districts from 17 states and five union territories (UTs).

We present an interactive dashboard that systematically allows multiple stakeholders (e.g., policymakers, donors, journalists, and researchers) to compare 55 indicators for 342 districts in India. We also compare these indicators for 285 Parliamentary Constituencies in India with the motivation to engage the Members of Parliament (MP) who are responsible for envisioning policy priorities for their constituents.

Geographic Visualization

NOTE: Changes in indicators are represented as “reversed” (red color) and “improved” (blue color). Red color means that the situation of the district has reversed (e.g., percentage of stunting in Pune, Maharashtra increased from 22.4 to 30.7). Blue color means that situation in the district has improved (e.g., percentage of stunting in Palghar, Maharashtra decreased from 38.5 to 33.0). We further distinguished reversed as light/dark red and improved as light/dark blue. Districts with absolute changes higher than the median values among all reversed/improved districts were coded in dark red/blue, and those with changes lower or equal to the median values were coded in light red/blue. While rare, we considered light red when the indicator experienced zero change between NFHS-4 and NFHS-5, since no improvement can be regarded as not desirable.

Explore Change-On-Change Correlations

The scatter plot above shows the "change on change" of any two indicators. Users can filter to specific states and select multiple districts or parliamentary constituencies within or across states to learn how the correlations between two change indicators differ across geographies.

Explore Data

The spreadsheet below provides users to customize data views using filters. To enable filtering click on the filter icon as shown in the image below:

Filter Icon

Citation: Akhil Kumar, Weiyu Wang, Julie Kim, Weixing Zhang, Rockli Kim, Rakesh Sarwal, S V Subramanian. NFHS Policy Tracker for Districts and Parliamentary Constituencies of India. Apr 2021. Geographic Insights Lab at the Harvard Center for Population and Development Studies; Center for Geographic Analysis at Harvard University, Cambridge, MA.

Download Data


Data Sources

We gathered information from the recently released district factsheets (i.e., the phase 1 results of the NFHS-5 conducted throughout the 2019-2020 period), which presented indicator values for NFHS-5 and NFHS-4. The first phase results covered 22 states and UTs covering 342 districts, which corresponds to nearly half of districts in India. For Ladakh UT, we used the state factsheet in lieu of the district factsheet since there has only one district in the Ladakh UT.

As the creation of new districts is an ongoing process in India, there are 79 districts, either newly created or being carved out, that do not have valid NFHS-4 indicator values. To make the NFHS-4 indicators comparable to the NFHS-5, we imputed the missing indicators using the NFHS-4 district factsheets in 2016 based on the following methodology:

  • If   a   district   was   carved   out from   one   district, we   assumed   the   same   2016 indicators for both new and carved districts. For example, Palghar was created from Thane, Maharashtra. We made the reasonable assumption that the NFHS-4 indicators for the newly created Palghar were the same as the indicators of Thane in NFHS-4.
  • If a district was created from multiple districts, we considered the mean of the indicators from all the districts from which the new district was created. For example, Botad was created partially from Ahmadabad and partially from Bhavnagar, Gujarat in 2013. We assume the 2016 indicators of Botad is the mean of the indicators of Bhavnagar and Ahmadabad.
  • If a pre-existing district was available in the NFHS-4 survey but experienced a change due to the creation of a new district, we retained the NFHS-4 indicators for 2016.

The NFHS-5 results were released in phases: as of December 2020, the NFHS published phase 1 of district factsheets conducted throughout the 2019-2020 period and covered 342 districts in 22 states/union territories (UTs) for 104 indicators. Among 66 indicators that commonly appear in both the NFHS-4 and NFHS-5 district factsheets, this dashboard presents district- and PC-level changes for 55 indicators that commonly appear and are most relevant. The following 11 indicators with multiple missing values are not presented for visualization concern.


Number of missing districts

Number of missing districts




Pregnant women age 15-49 years who are anaemic  



Children with fever or symptoms of ARI in the 2 weeks preceding the survey taken to a health facility or health provider 



Births in a private health facility that were delivered by caesarean section 



Children with diarrhoea in the 2 weeks preceding the survey taken to a health facility or health provider 



Children with diarrhoea in the 2 weeks preceding the survey who received zinc 



Current users ever talked about side effects of current method 



Children born at home who were taken to a health facility for a check-up within 24 hours of birth



Non-breastfeeding children age 6-23 months receiving an adequate diet 



Children under age 6 months exclusively breastfed 



Children age 6-8 months receiving solid or semi-solid food and breastmilk 



Children with diarrhoea in the 2 weeks preceding the survey who received oral rehydration salts (ORS) 



Crosswalk Methodology (Districts to Parliamentary Constituencies)

We generated a crosswalk between districts and parliamentary constituencies. Assuming the indicators were distributed uniformly across the population in one district, crosswalk uses GIS techniques to estimate population distribution in granularity and reaggregate to PC level for estimation. See the workflow below and read the working paper for more details.

  1. Calculate district area (dist_area): perform 'Calculate Geometry' function on the district shapefile, using the Kalianpur 1975/India Zone lla (EPSG: 24379) coordinate system.
  2. Intersect PC and district boundaries, link each segment with district ID and PC ID: perform 'intersect' command to district and PC shapefile, create a new shapefile called Dist_PC_intersect.
  3. Calculate segment area (seg_area): perform 'Calculate Geometry' function on the district shapefile, using the Kalianpur 1975/India Zone lla (EPSG: 24379) coordinate system.
  4. Calculate segment population (seg_ment): perform 'zonal statistics' command using segment shapefile as the 'zone' and the WorldPop2020 as the data raster; derive pc level population by aggregating segment population by PC ID.
  5. Calculate the percentage of district area (Pct_area) by dividing segment area by district area. Remove segments that were less than a hundredth of a percent of the original district area. Those extremely small areas are created because of the slight boundary inaccuracies between district and PC shapefiles. In total, there are 3468 segments.
  6. Calculate segment headcount = district rate * segment population, assuming district prevalence rate unchanged across the whole district
  7. Aggregate segment headcounts by PC ID to get PC level headcount.
  8. Calculate PC level prevalence rate by PC headcount divided by PC population.