Human Development

Despite improvements in child health, why do so many newborns still die?

  • Blog Post Date 02 January, 2018
  • Perspectives
  • Print Page
Author Image

Diane Coffey

University of Texas at Austin

The ‘Million Death Study’ shows that the death rate of under-five children in India dropped from about 90 per 1,000 to about 47 during 2000-2015. However, improvements in death rates in the first month of life have been slow, and deaths from low birthweight remained largely unchanged. In this article, Diane Coffey elaborates on these findings and recommends stronger efforts to understand and improve the health of pregnant women.

A recent article from the Million Death Study, published in The Lancet, brings good news for children: between 2000 and 2015, the death rate of children below five years of age in India dropped from about 90 per 1,000 to about 47 per 1,000. That means that the number of child deaths per year dropped from about 2.5 million1 to about 1.2 million and that many parents who might have experienced the loss of a child did not have to. However, the study also finds that improvements in death rates in the first month of life have been slow, and that deaths from low birthweight remained similar over this period. In this article, I elaborate on these findings and recommends stronger efforts to understand and improve the health of pregnant women.

Child death: Comparisons and causes

As encouraging as it is to see death rates among children improving, we might ask whether these gains are fast enough. Over this period in Bangladesh and Nepal, child death rates declined by 59% (from 87 per 1,000 to 36) and 56% (from 82 per 1,000 to 36), respectively, while India’s declined by 47%. The Million Death Study Collaborators explain that India’s child death rates would be falling much faster if it could achieve more improvement in neonatal death rates, or death rates in the first month of life. Indeed, declines have been too slow to achieve the UN’s Sustainable Development Goal of reaching 12 neonatal deaths per 1,000 by 2030. In order to meet the 2030 goal, reductions in neonatal deaths from 2015 onwards will have to increase from an annual decline of 3.3% to 5.3%.

Neonatal death accounts for a large fraction of child death in India: according to the Million Death Study, about 60% of deaths among under-five children happened in the first month. Neonatal deaths are also more common than economic indicators predict: in a descriptive, cross-country regression of neonatal death rates on GDP (gross domestic product) per capita, India has an excess death rate of about 7 per 1,000 beyond what it predicted by its GDP per capita (Coffey and Hathi 2016). This excess is larger than the total neonatal death rate of many countries.

In order to better understand why children are dying, the Million Death Study researchers collaborated with the Registrar General of India (RGI) to collect data on cause of death within the RGI’s existing Sample Registration System (SRS). The SRS is a representative survey of villages and urban places that has produced annual, state-level estimates of birth and death rates since the early 1970s. Since 2000, through the Million Death Study, the SRS has also measured cause of death using a tool called verbal autopsy, in which surveyors interview the family members of people who die, and then doctors read the surveyors’ notes and assign a probable cause of death. The addition of the verbal autopsies to the SRS is crucially important for improving health because these data can be used to direct attention and resources to those diseases and conditions which cause the most deaths.

The researchers identified separate causes of death in the neonatal period and in months 1-59. Neonatal deaths were categorised into one of nine major causes2, and deaths between months 1-59 were categorised into one of 11 major causes3. In 19 of the 20 cause of death categories, death rates fell between 2000 and 2015. Yet, the category in which death rates did not improve – prematurity and low birthweight4 – is a very important one. It is the leading cause of both neonatal death and of deaths of under-five year olds. Indeed, more than half of neonatal deaths, and nearly one third of all under-five deaths, were attributed to this cause alone. The researchers estimate that about 370,000 infants died of this cause in 2015.

The puzzling death rate from low birthweight

The researchers measured a death rate from low birthweight and prematurity of 12.3 per 1,000 in 2000 and of 14.3 per 1,000 in 2015. This is puzzling considering that death rates from all other causes fell and that there was substantial economic growth over this period.

Why did death rates from prematurity and low birthweight remain similar or even increase? The Million Death Study says that “most of the increase in prematurity or low birthweight deaths was in term births with low birthweight, not in preterm births”, so it is useful to begin by considering what causes low birthweight. Stress during pregnancy, poor healthcare during pregnancy, the mother’s exposure to infectious disease, and perhaps most importantly, poor nutrition during pregnancy, are causes of low birthweight.

We know that undernutrition during pregnancy is a widespread problem in India, far more so than in other, poorer, countries. Indeed, data from the National Family Health Survey (NFHS), 2005 show that over 40% of Indian women began pregnancy underweight and that the average pregnant woman gained only about 7 kgs. during pregnancy. In contrast, only about 15% of women in sub-Saharan Africa began pregnancy underweight (Coffey 2015). Unfortunately, and perhaps surprisingly for a democracy that is home to a fifth of the world’s births, there are no available data that would allow us to determine whether the nutrition of pregnant women in India improved or worsened between 2005 and 2015. Although the NHFS, 2015 measured the weights of pregnant women, the unit-level data, which are required for producing updated estimates of nutrition during pregnancy, have not yet been released.

Despite the fact that it is not currently possible to compute exactly how many women began pregnancy underweight in 2015, or how much weight they gained in pregnancy, it would be very surprising if the data did not show some improvement in these indicators of maternal nutrition. After all, the NFHS summary statistics that have been released show that the fraction of non-pregnant women who were underweight improved from 36% to 23% between 2005 and 2015. If there has likely been an improvement in maternal nutrition over this period, why then, did death rates from low birthweight not improve?

Possible explanations

The authors of the Million Death Study posit two possible explanations, which I elaborate on here. One possible explanation is that infants who, in 2000, would have died from another cause of neonatal death are now dying of low birthweight instead of one of the other causes. Let’s consider the example of birth asphyxia (breathing problems), the second leading cause of neonatal death. Low birthweight babies are more likely to have breathing problems than higher birthweight babies. If improvements in care at birth have prevented some asphyxia deaths, then some of the low birthweight babies who would have died shortly after birth from asphyxia now survive the first few hours, but eventually die from other complications of low birthweight. Their deaths would then be classified as low birthweight deaths, rather than as asphyxia deaths.

Another possible explanation is that the death rate from low birthweight remained similar because stillbirths – deaths that happen after 28 weeks gestation but while a foetus is still in the womb – went down. Research shows that undernourished women are more likely to have stillborn babies, and improvements in maternal nutrition lead to reductions in stillbirths (Imdad and Bhutta 2012). It is possible that modest improvements in maternal nutrition – in which maternal nutrition has improved by enough to cause more births to be live births rather than stillbirths, but not by enough for the babies to be born at healthy birthweights – could actually lead to stagnation or even an increase in neonatal death rates. Although an increase in death rates sounds undesirable, we have to remember that deaths in utero that would not previously have been categorised as neonatal deaths would now be categorised as deaths from low birthweight.

On the need for data and action

Are these explanations correct? High-quality, large-scale, up-to-date health survey data would be necessary to answer this question well. Data that measure the weights and heights of pregnant women, such as the NFHS, 2015, would allow us to quantify improvements in nutrition during pregnancy. The unit-level data from the NFHS, 2015 would also allow us to investigate whether or not there have been changes in the timing of neonatal deaths, such as those implied by the hypothesis that deaths are being shifted from other causes of death to low birthweight.

Assessing the hypothesis that the puzzling death rates from low birthweight result from a reduction in stillbirth would be more difficult. That is because, to my knowledge, high-quality data on stillbirth for India do not exist. The SRS’ estimates of stillbirth are implausibly low, and the NFHS asks about foetal loss in a question that combines stillbirth, early miscarriage, and abortion. The lack of data on stillbirth is unfortunate. As UCLA (University of California, Los Angeles) Sociologist Nicky Hart and others have pointed out, the experience of expectant parents whose baby is stillborn is similar to that of parents whose baby is born alive and dies in the first few days of life. Stillbirth rates are also an important indicator of maternal health and the extent to which women in a society are deprived (Hart 1998).

As important as it is to collect and release up-to-date data on maternal and infant health, it is nevertheless clear even from the existing evidence that maternal undernutrition plays an important role in India’s high rates of neonatal death from low birthweight, and that this problem urgently requires a policy response.

What sort of response might help improve maternal nutrition and reduce low birthweight and neonatal death? One promising policy, a cash transfer for pregnant women, is currently being rolled out by the government. The programme, called the Pradhan Mantri Matritva Vandana Yojana, was designed in response to the 2013 National Food Security Act, which legislates a cash transfer of Rs. 6,000 for each pregnant woman. Unfortunately, the programme currently covers only the first pregnancy, not subsequent ones. Considering the magnitude and the stubbornness of the problem of low birth weight, this programme should cover every birth; it should be given early in pregnancy; and the government should be experiment with ways to encourage families to use the cash to invest in nutrition during pregnancy.5


  • Data on the number of child deaths per year in each country is available at:
  • In decreasing order by the number of deaths caused, these are prematurity/low birthweight, neonatal infections, birth asphyxia or trauma, non-communicable, congenital, other conditions, injuries, tetanus, and “ill-defined”.
  • In decreasing order by the number of deaths caused, these are pneumonia, diarrhea, injuries, non-communicable, other infections, nutritional, malaria, congenital, meningitis or encephalitis, acute bacterial sepsis or severe infection, measles, “ill-defined”.
  • Prematurity and low birthweight describe different conditions − prematurity is when babies are born early, that is, before their mothers have completed 39 weeks of pregnancy. Often, babies who are born prematurely are small and need special care in the initial weeks of life. Low birthweight, on the other hand, tends to refer to babies whose mothers have completed full-term pregnancies, but who are nevertheless unhealthily small.
  • For more information on maternity entitlements, see Coffey and Hathi (2016) and

Further Reading

No comments yet
Join the conversation
Captcha Captcha Reload

Comments will be held for moderation. Your contact information will not be made public.

Related content

Sign up to our newsletter