District-level trends of child mortality and child growth failure in India

Published on 12 May 2020

Two scientific papers released in The Lancet and EClinicialMedicine published by The India State-Level Disease Burden Initiative. The initiative was launched in 2015 as a collaborative effort between the Indian Council of Medical Research, Public Health Foundation of India, Institute for Health Metrics and Evaluation, and a number of other key stakeholders in India, including academic experts and institutions, government agencies and other organizations, under the aegis of the Ministry of Health & Family Welfare. Over 300 leading scientists and experts representing about 100 institutions across India are engaged with this collaborative work.

The paper in The Lancet reports the first comprehensive estimates of district-level trends of child mortality in India from 2000, and the paper in EClinicalMedicine reports detailed district-level trends of child growth failure. The findings show that although the child mortality and child growth failure indicators have improved substantially across India from 2000 to 2017, the inequality between districts has increased within many states, and that there are wide variations between the districts of India. The child mortality and child growth failure trends reported in these papers utilized all accessible georeferenced survey data from a variety of sources in India, which enabled more robust estimates than the estimates based on single sources that may have more biases. The district-specific findings described in these scientific papers highlight the extent of the effort needed in each district to achieve the national and global targets for the child mortality and child growth failure indicators.

The findings reported in the papers published today are part of the Global Burden of Disease Study 2017. The analytical methods of this study have been refined over two decades of scientific work, which has been reported in over 16,000 peer-reviewed publications, making it the most widely used approach globally for disease burden estimation. These methods enable standardized comparisons of health loss caused by different diseases and risk factors, between different geographies, sexes, and age groups, and over time in a unified framework.

India State-Level Disease Burden Initiative Child Mortality Collaborators. Subnational mapping of under-5 and neonatal mortality trends in India: the Global Burden of Disease Study 2000–2017. Lancet. 12 May 2020.
http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(20)30471-2/fulltext

India State-Level Disease Burden Initiative CGF Collaborators. Mapping of variations in child stunting, wasting and underweight within the states of India: the Global Burden of Disease Study 2000-2017. EClinicalMedicine. 12 May 2020.
https://www.thelancet.com/pb-assets/Lancet/pdfs/ECM-GBD-India.pdf

Key findings from the child mortality paper published in The Lancet:

State-level child mortality variations

U5MR in India reduced by 49% from 83 in 2000 to 42 per 1000 livebirths in 2017, and NMR reduced by 38% from 38 to 23 per 1000 livebirths during this period.
There were 1.04 million under-5 deaths in India in 2017, of which 0.57 million were neonatal deaths, down from 2.24 million under-5 deaths including 1.02 million neonatal deaths in 2000.
The highest number of under-5 deaths in 2017 were in the state of Uttar Pradesh (312,800 which included 165,800 neonatal deaths) and Bihar (141,500 which included 75,300 neonatal deaths).
U5MR and NMR was lower with the increasing level of development of the states. In 2017, there was 5.7 fold variation in U5MR ranging from 10 per 1000 live births in the more developed (high SDI) state of Kerala to 60 in the less developed (low SDI) state of Uttar Pradesh, and 4.5 fold variation for NMR ranging from 7 per 1000 live births in Kerala to 32 in Uttar Pradesh.
The annual rate of reduction from 2010 to 2017 for U5MR ranged among the states from 2.7% in a small north-eastern state of Nagaland to 6.5% in the middle SDI state of Telangana, and for NMR from 1.8% in Nagaland to 5.5% in the high SDI state of Tamil Nadu.
The annual rate of reduction of NMR was lower than that of U5MR in all states during 2010-2017, but this varied considerably between the states.

District-level child mortality variations

U5MR varied 10.5 times between the 723 districts of India in 2017, ranging from 8 to 88 per 1000 livebirths, and NMR varied 8.0 times, ranging from 6 to 46 per 1000 livebirths. The highest district-level U5MR and NMR in 2017 were comparable to the highest rates globally among some Sub-Saharan Africa countries.
U5MR was 40 or more per 1000 livebirths in 88% of the districts in the less developed (low SDI) states, but only in 2% of the districts in the more developed (high SDI) states.
Similarly, NMR was 20 or more per 1000 livebirths in 93% of the districts in the low SDI states, but only in 13% of the districts in the high SDI states.
The annual rate of change 2010-2017 varied widely among the districts from 9.0% reduction to no significant change for U5MR, and from 8.0% reduction to no significant change for NMR.
Inequality between the districts within the states, measured as coefficient of variation, varied extensively in 2017, ranging 11-fold for U5MR and 13-fold for NMR among the states.
Despite the decrease in U5MR and NMR in most of the districts from 2000 to 2017, the inequality in these rates increased in 74% of the states for U5MR and in 77% states for NMR.
· The highest increases in inequality between districts within states were in Assam and Odisha among the low SDI states, in the small north-eastern states of Meghalaya and Arunachal Pradesh, and in Haryana among the middle SDI states.

Identification of priority districts

Priority districts for child mortality reduction were identified within states as those that fell in the category of high U5MR and NMR in 2017 and low annual rate of reduction from 2010 to 2017 for the distribution of rates within the states. Using this approach, priority districts for the nationwide distribution of U5MR and NMR and the rate of reduction were also identified to enable a complimentary understanding of the standing of each district with respect to all districts in India.
In Uttar Pradesh, which had the highest child mortality rate in 2017 among the states, the districts in the highest priority category of high NMR and U5MR and low annual rate of reduction included a cluster of eight districts in the north-central part (Bahraich, Balrampur, Barabanki, Gonda, Hardoi, Kheri, Shravasti, and Sitapur), three districts in the south (Allahabad, Banda and Chitrakoot), and Lalitpur district in the south-west corner of the state.
In Assam, which had the second highest child mortality rate in 2017, the highest priority category of high U5MR and NMR and low annual rate of reduction was concentrated in the southern handle of the state (Cachar, Dima Hasao, Hallakandi, Karbi Anglong, Karimganj, and West Karbi Anglong)
In Bihar, the highest priority was scattered in the north-east (Kishanganj and Purnia) and the south-west of the state (Aurangabad and Kaimur).
Based on the nationwide district-level distribution of child mortality rate, two-thirds of the districts in the less developed low SDI states fell in the high category of U5MR and NMR.
In Uttar Pradesh, 48% of the districts fell in the highest priority category of high NMR and low rate of reduction for the nationwide distribution of the district-level rates.

Comparison of child mortality trends with targets

If the trends up to 2017 were to continue, India would not achieve the National Health Policy (NHP) 2025 U5MR target of 23 per 1000 live births or the NHP 2025 NMR target of 16 per 1000 live births. With these trends, India would achieve the SDG 2030 U5MR target of 25 per 1000 live births but not the SDG 2030 NMR target of 12 per 1000 live births.
In order to achieve these NHP 2025 and SDG 2030 targets individually, most of the less developed low SDI states would need a higher rate of improvement in U5MR and NMR than they had up to 2017.
Of the 723 districts in India, 34% would need a rate of improvement higher than they had up to 2017 to individually meet the SDG 2030 target for U5MR.
59% districts would need a rate of improvement higher than these had up to 2017 to individually meet the SDG 2030 target for NMR; this proportion was 91% in the less developed low SDI states and 21% in the more developed high SDI states.

Causes of child mortality

Lower respiratory infections (17·9%), preterm birth (15·6%), diarrhoeal diseases (9·9%), and birth asphyxia and trauma (8.1%) were the leading causes of under-5 death in India in 2017.
Preterm birth (27·7%), birth asphyxia and trauma (14·5%), lower respiratory infections (11.0%), and congenital birth defects (8.6%) were the leading causes of neonatal deaths in India in 2017. 80% of the neonatal deaths were in the early neonatal period of 0–6 days.
There were wide variations in the percentage of under-5 deaths due to various causes across the states even within the same SDI group. For example, within the low SDI states, the percentage for lower respiratory infections ranged from 15% in Odisha to 27% in Rajasthan, for diarrheal diseases from 6% in Chhattisgarh to 16% in Bihar, and for preterm birth from 11% in Bihar to 19% in Chhattisgarh.
The rates for most causes of under-5 death in India were lower in the more developed states than in the less developed states.
The death rate for all major causes of under-5 death reduced in India from 2000 to 2017, with the highest decline in measles (82%), followed by diarrhoeal diseases (69%), and lower respiratory infections (57%) and least for congenital birth defects (15%). There were wide variations in the magnitude of decline between the states, even within the same SDI group.

Risk factors for child deaths

The dominant risk factor for under-5 death in India in 2017 was child and maternal malnutrition, to which 68% of the deaths could be attributed. The largest contributors to this were low birth weight and short gestation (46%) followed by child growth failure (21%).
11% of the under-5 deaths in India in 2017 could be attributed to unsafe water and sanitation and 9% to air pollution.
For neonatal death, child and maternal malnutrition was the predominant risk factor to which 83% of deaths could be attributed, almost all of which was due to low birth weight and short gestation.
The proportion of under-5 deaths attributable to child and maternal malnutrition varied between the states from 51% to 73%, unsafe water and sanitation from 1% to 14%, and air pollution from 2% to 14%.
The proportion of neonatal deaths attributable to child and maternal malnutrition varied between the states from 63% to 87%, unsafe water and sanitation from 1% to 6%, and air pollution from 2% to 9%.
The contribution of these risk factors to under-5 and neonatal deaths was relatively higher in the less developed low SDI states.

Implications of these findings

This study provides the most comprehensive understanding of child mortality trends across the states and districts of India over the past two decades, highlighting the enduring disparities in child survival between the states and districts.
The comparison of child mortality trends with the India and the SDG targets in this study identifies the states and districts that have gaps where more attention is needed.
An approach combining the different levels of mortality rates and the rate of decline to identify priority districts in each state as used in this study could be used by policy makers to target districts that have persistently high child mortality rates and low rates of mortality reduction.
The trends of the causes of under-5 and neonatal deaths reported in this study are a useful guide for the relative effort needed to deal with particular causes of child mortality in each state.
The risk factor analysis reported in this study for each state highlights that child mortality can be reduced substantially with more effective improvements in the leading risk factors.
The estimation of granular child mortality trends across all districts of the country combined with estimation of causes of death and risk factors at the state level, using all accessible data sources from India in a single framework as reported in this study provides crucial inputs for further planning of child mortality reduction in India.

Targets set by the National Health Policy 2025 and the SDG 2030

National Health Policy 2025 targets	SDG 2030 targets
U5MR: 23 per 1000 live births by 2025	U5MR: 25 per 1000 live birth by 2030
NMR: 16 per 1000 live births by 2025	NMR: 12 per 1000 live births by 2030

Key findings from the child growth failure paper published in EClinicalMedicine:

District-level variations

The prevalence of stunting among children under five years of age in India in 2017 was 39%, which varied 3.8 times between the 723 districts, ranging from 16% to 63%. Stunting prevalence was 40% or more in the 67% districts in the less developed (low SDI) states, but only in 1% of the districts in the more developed (high SDI) states.
The prevalence of wasting among children under five years of age in India in 2017 was 16%, which varied 5.4 times between the districts ranging from 6% to 30%.
The prevalence of underweight among children under five years of age in India in 2017 was 33%, varying 4.6 times between the districts from 11% to 51%.
The annual rate of change from 2010 to 2017 varied widely among the districts. Stunting prevalence declined in 99% of the districts with a maximum decline of 41% and underweight in 95% of the districts with a maximum decline of 54%, but wasting declined only in 61% of the districts with a maximum decline of 44%.
A higher proportion of the districts in the low SDI states had less than 20% reduction in stunting prevalence compared with the high SDI states (67% versus 46%), while for wasting a relatively higher proportion of districts in the high SDI states had less than 20% reduction compared with the low SDI states (44% versus 35%).

Inequality within states

Inequality between the districts within the states, measured as coefficient of variation (CV), varied extensively in 2017, ranging 7-fold for stunting, 12-fold for wasting, and 9-fold for underweight among the states.
The magnitude of inequality in stunting, wasting and underweight varied widely even between states at similar levels of socio-demographic development. For example, the CV for stunting in 2017 varied from 4% in Bihar to 21% in Odisha among the less developed (low SDI) states, and from 3% in Delhi to 19% in Kerala among the more developed (high SDI) states.
Despite the decrease in stunting, wasting and underweight in most of the districts from 2000 to 2017, the inequality in these indicators increased in 90% of the states for stunting, in 52% states for wasting, and in 65% states for underweight.
For stunting, the highest increase in inequality between districts within states was in Odisha among the low SDI states, in Telangana and Haryana in the middle SDI states, and in Nagaland and Delhi in the high SDI states.
For underweight and wasting, the inequality between the districts within the states increased from 2000 to 2017 in some of the states, while it decreased for the others, spread across the low, middle and high SDI states.

Identification of priority districts in states

Priority districts for child growth failure reduction were identified within states as those that fell in the category of high prevalence of stunting, wasting or underweight in 2017 and low annual rate of reduction from 2010 to 2017 for their distribution within the states. Using this approach, priority districts for the nationwide distribution of the prevalence of stunting, wasting and underweight and the rate of reduction were also identified to enable a complimentary understanding of the standing of each district with respect to all districts in the country.
In Odisha, which had the highest inequality between districts for all the three CGF indicators in 2017, the districts in the highest priority category of high prevalence and low annual rate of reduction for stunting, wasting and underweight included a cluster of three districts in the south-west handle of the state (Kalahandi, Koraput, and Rayagada), and additionally for underweight and wasting in the neighbouring three districts (Nuapada, Nabarangapur and Malkangiri), and for stunting and underweight in Balangir district.
In Uttar Pradesh, which had the highest stunting prevalence and medium level of inequality in 2017, the districts in the highest priority category of high prevalence and low rate of reduction for stunting included a cluster of 13 districts in the northern part (Pilibhit, Shahjanpur, Lakhimpur Kheri, Sitapur, Bahraich, Sharavasti, Balrampur, Siddharth Nagar, Gonda, Barabanki, Faizabad, Basti, and Maharajganj).
Based on the nationwide district-level distribution of the prevalence of CGF indicators, all 38 districts in Bihar were in the high tertile of stunting and none were in the high tertile for their rate of reduction, while in Uttar Pradesh, 97% of the districts fell in the high tertile for stunting and only 12% were in their high tertile for the rate of reduction.
Interestingly, for wasting, 60% of the districts in the Odisha were in the high nation-wide tertile in 2017, while 67% in Uttar Pradesh were in the low tertile, indicating the contrast even within the less developed (low SDI) states.

Comparison of CGF indicators trends with targets

Of the 723 districts in India, 83% of the districts would need a rate of improvement higher than they had up to 2017 to individually meet the NNM stunting target of 25% prevalence in 2022; this proportion was 98% in the less developed (low SDI) states and 56% in the more developed (high SDI) states.
In order to achieve the WHO/UNICEF 2030 target of 50% reduction in stunting prevalence from 2012 to 2030, 80% of the districts would individually need a rate of improvement higher than they had up to 2017; this proportion was 89% in the low SDI states and 63% in the high SDI states.
99% of the districts would need a rate of improvement higher than they had up to 2017 to individually reach the NNM 2022 underweight target of 2 percentage point reduction annually, and all the districts to reach the WHO/UNICEF wasting target of less than 3% prevalence in 2030.

Correlation between major national surveys

This report assessed the correlation for the district-level estimates of CGF indicators between the National Family Health Survey- 4 (NFHS-4, 2015–2016) and the two complementary household surveys (District-Level Household Survey [DLHS-4, 2012–2014] and Annual Health Survey [AHS, 2014]) for the 27 states covered by these surveys.
The correlation between the NFHS-4 and AHS, which covered the same nine states, for district-level estimates of the CGF indicators was significant only in three states for stunting, three states for wasting, and two states for underweight, with Pearson correlation coefficient of more than 0.7 only in Odisha for stunting and underweight.
In Bihar and Uttar Pradesh, that had the highest prevalence of stunting in 2017, there was no correlation between NFHS-4 and AHS, and no or very poor correlation for wasting and underweight.
The correlation between NFHS-4 and DLHS-4, which covered the same 18 states, for district-level estimates of the CGF indicators was significant only in four states for stunting, in three states for wasting, and in two states for underweight, but with Pearson correlation coefficient of more than 0.7 only in four small north-eastern states and in none of the other larger state.

Implications of these findings

This study provides a comprehensive understanding of the trends of CGF indicators for every district of India since 2000, highlighting the persistently high inequality in child nutrition between the districts.
The comparison of the trends of CGF indicators with the India and the global targets in this study identifies the district that have gaps and need more attention.
An approach combining the different levels of CGF indicators and their rate of decline to identity priority districts in each state used in this study can help policy makers in better targeting of the districts that have persistently high prevalence and low rates of reduction.
The poor correlation between the national surveys for the district-level estimates of CGF indicators highlights the need to standardize the survey design and collection of anthropometric data in India.
This find-grid geospatial mapping, using all accessible data sources from India in a single framework as reported in this study can facilitate better strategic targeting of resources at sub-state levels to improve child nutrition as suggested under NNM.

Targets set by the National Nutrition Mission 2022 and the WHO/UNICEF 2030

National Nutrition Mission 2022 targets	WHO/UNICEF 2030 targets
Child stunting: prevalence of 25% in 2022	Child stunting*: 50% reduction in number of children under-five who are stunted from 2012 to 2030
Child underweight: 2% points reduction in prevalence annually from 2017 to 2022	Child wasting: prevalence of less than 3% by 2030

*A relative reduction in the prevalence of stunting was estimated instead of the absolute numbers for consistency with other indicators, as all other targets are based on prevalence.

These persons can be contacted for discussion on the child mortality findings and their implications:

Prof Balram Bhargava, Indian Council of Medical Research, New Delhi
Prof Rakhi Dandona, Public Health Foundation of India, Gurugram
Prof Siddarth Ramji, Maulana Azad Medical College, New Delhi
Prof Subodh S Gupta, Mahatma Gandhi Institute of Medical Sciences, Wardha
Prof Rashmi Kumar, King George’s Medical University, Lucknow, India
Prof Rakesh Lodha, All India Institute of Medical Sciences, New Delhi
Prof Anita Kar, School of Health Sciences, Savitribai Phule Pune University, Pune
Prof Anura V Kurpad, St John’s Medical College, Bengaluru
Dr Hendrik J Bekedam, World Health Organisation, New Delhi
Dr R S Sharma, Indian Council of Medical Research, New Delhi
Prof Lalit Dandona, Indian Council of Medical Research, New Delhi; Public Health Foundation of India, Gurugram

Spokespersons for the child growth failure findings and their implications:

Prof Balram Bhargava, Indian Council of Medical Research, New Delhi
Dr R Hemalatha, National Institute of Nutrition, Indian Council of Medical Research, Hyderabad
Prof Siddarth Ramji, Maulana Azad Medical College, New Delhi
Prof Rakesh Lodha, All India Institute of Medical Sciences, New Delhi
Prof Subodh S Gupta, Mahatma Gandhi Institute of Medical Sciences, Wardha
Prof Anita Kar, School of Health Sciences, Savitribai Phule Pune University, Pune
Prof Anura V Kurpad, St John’s Medical College, Bengaluru
Dr Hendrik J Bekedam, World Health Organisation, New Delhi
Prof Rakhi Dandona, Public Health Foundation of India, Gurugram
Prof G S Toteja, Indian Council of Medical Research, New Delhi
Prof Lalit Dandona, Indian Council of Medical Research, New Delhi; Public Health Foundation of India, Gurugram

Public Health Foundation of India