Child Survival in India
http://www.100md.com
《美国医学杂志》
1 Child Development and Nutrition, United Nations Children's Fund, India Country Office, New Delhi, India
2 Health, United Nations Children's Fund, India Country Office, New Delhi, India
Abstract
About 2.1 million Indian children under 5 years of age die each year. In spite of reductions in child mortality rate over the past two decades, the rate remains high at 87 per 1000 live births. The main causes are diarrhoea, pneumonia, and for deaths among the neonates asphyxia, pre-term delivery, sepsis and tetanus. The major underlying cause of death is undernutrition. Child survival interventions of proven impact, feasible for use at high coverage in India were identified, and their effect on child mortality was calculated if high coverage were to be achieved. Exclusive breastfeeding, oral rehydration therapy, and adequate complementary feeding were among the most effective interventions. If these interventions would be applied universally 57% of mortality among pre-schoolers could be prevented. No cause specific mortality data were available from individual Indian states. Nevertheless, the range of child mortality, as well as the proportion of neo-natal deaths, occurring across the states, suggests that at state level 50-70% of deaths can be prevented. The results show that the targets set in the millennium development goals as well as in the Tenth Five Year plan can be reached.
Keywords: Child survival, Undernutrition, Breastfeeding, Oral rehydration therapy, India
Around the world more than 10 million children under the age of five die every year and the world's poorest countries continue to bear the burden of these deaths [1]. Levels of under-5 mortality vary widely across countries - from 4 to over 280 deaths per 1000 live births. About 41 percent of under-5 deaths occur in sub-Saharan Africa and another 34 percent in South Asia.[1]
Child health has been high on the international agenda for over two decades now. The World Summit for Children in 1990 called for a worldwide reduction in child mortality below 70 deaths per 1000 live births by the year 2000 [2]. Unfortunately this goal was not met. The Millennium Development Goal on under-five mortality is to reduce the rate by two-thirds between 1990 and 2015 [3]. For developing regions this would mean a reduction from 105 to 35 deaths per 1000 live births, and for South Asia a reduction from 126 to 42. According to UNICEF estimates, child mortality rates have been reduced by around 12 percent over the period 1990 - 2002. Given this average annual rate of reduction of 1 percent, the pace of progress will need to increase significantly during the remaining years in order for the millennium development goal (goal IV) to be met.
In India, about 2.1 million child deaths occur every year, which is the highest number within a single country worldwide.[4] The national under-5 mortality rate is around 87 per 1000 live births, but there is a wide variation between states. For example, in 1998-99 the under-5 mortality rate varied from 19 per 1000 births in Kerala to 138 in Madhya Pradesh.[5] For India, attaining the millennium development goal would imply a reduction of the under-5 mortality rate to 41 by 2015. There has been a substantial decrease in child mortality in the past two decades in India. The reduction was more marked in the 1980's than in the 1990's.[6] Despite these impressive gains, India compares poorly in the pace of child mortality reduction to several other countries in south and south-east Asia, including Bangladesh. More disturbing are the data which indicate that the decline in the child mortality rate is slowing in India [6].
The main causes of deaths in low-income countries are diarrhoea, pneumonia, measles, malaria, HIV/AIDS, and for deaths among the neonates asphyxia, pre-term delivery, sepsis and tetanus [1]. The major underlying cause of death is undernutrition, including micronutrient deficiencies, among children younger than 5 years, which is a determinant in more than 50% of deaths. The question can be raised whether the goals on mortality reduction can be achieved with the currently available interventions.
It is in this context that a series of articles on child survival appeared in The Lancet in mid-2003. The aim was to provide an evidence-based high impact approach to the task of reducing under-5 deaths. The Lancet Child Survival series describe how many of the under-5 deaths could be prevented with currently available interventions, whether delivery systems can achieve high coverage with these interventions and how to ensure that the poorest children receive the interventions of which they are most in need.[7] In particular, the second paper of the series showed that about two-thirds of child deaths could be prevented by interventions that are available today and are feasible for implementation in low-income countries if high levels of population coverage were to be achieved (referred to as Lancet approach in this paper).[7]
The aim of the present paper is to report on possible reductions in child deaths in India, using the Lancet approach, including how the approach might be applied at the state level. This is of particular relevance since India is presently going through a major revision of its health policy under the National Rural Health Mission, the Reproductive and Child Health Program phase 2 (RCH-2) as well as expanding the Integrated Child Development Services. The findings from this paper could help in identifying and developing the most effective and feasible interventions for reducing child mortality and malnutrition in India.
The paper briefly describes the Lancet approach at the global level and for India alone. An application of the Lancet approach at the State level, though limited by lack of cause of death data by state, then identifies the substantial range in the proportion of deaths that could be prevented across the states if universal coverage of key interventions are achieved.
Methodology
The second paper in The Lancet Child Survival series reviewed the state of evidence for interventions that reduce mortality for each of the major direct and underlying causes of death in children under-5 [7]. The focus was on interventions addressing proximal determinants of child mortality and those that can be delivered mainly through the health sector. For each of the main causes of under-5 deaths the effectiveness of available interventions was assessed.
Estimates of the effectiveness of interventions were taken either from published articles that summarized earlier research results or from systematic reviews by the authors and participants in the Bellagio Child Survival Study Group, together with input from other experts. Interventions were identified for each cause, and these were categorized by whether they are preventive or therapeutic. These interventions had to be feasible for delivery at high coverage levels in countries where most under-5 deaths occur; however, only the impact of an intervention on death was considered with no determination as to how the intervention will be delivered.
Each potential intervention was assigned to one of three levels of evidence based on the strength of evidence for its effect on under-5 mortality. Level 1 identified sufficient evidence - that a causal relationship between the intervention and cause-specific reductions of under-5 mortality had been established. Level 2 identified limited evidence - that an effect is possible, but data are not sufficient to establish a causal relationship. Level 3 included those interventions with inadequate evidence of effect. They include those that hold promise of substantial effects on under-5 mortality, but have not yet been fully assessed. A diagram of exposure, onset of disease or condition, and the subsequent resolution in terms of survival or death gives a visual summary of the interventions by cause with level 1 in green and level 2 in yellow Figure1Figure1a. Interventions are listed with prevention on the left and treatment on the right. Rotavirus, listed at the bottom of the diarrhoea diagram is classified as a level 3 intervention that holds considerable promise for the future.
Once the interventions were identified for each of the countries in the first paper of the Lancet series[1], the number of under five deaths that could be prevented was calculated with the coverage levels around the year 2000 increased to universal coverage. Universal coverage was taken as 99% except for exclusive breastfeeding, where 90% was used. The calculations divided into three types:
Exclusive and continuing breastfeeding, as this involved three levels: exclusive breastfeeding, partial breastfeeding and no breastfeeding
Complementary feeding, which utilized the underweight distribution of under-5's within a country
All other interventions, where the components of the calculation were intervention coverage (current and universal), efficacy, affected fraction of the population and evidence level
In applying the methodology described above and in the Lancet paper to the 42 countries, including India, a conservative approach was taken to the estimation of preventable under-5 deaths. There are four reasons why this is so.
Only interventions for which cause-specific evidence of effect was available were included
The methodology was restricted to interventions that are feasible at high coverage in low-income countries. As a result, some interventions were excluded for which there is evidence of effect, but that are only feasible for implementation in countries with higher levels of human, health-system and financial resources.
The methodology excluded promising interventions that are currently being assessed, for example rotavirus and pneumococcal vaccines.
It is limited to interventions that address the major causes of child death and selected underlying causes. Childhood anemia is one such example of conditions that contribute to child mortality, but for which the underlying causes and risk factors are not yet understood.
For India, and each of the other 42 countries, the number of under-5 deaths that could be prevented was calculated. Intervention coverage data for both India and the 42 countries as a group are reported in table1. For interventions where coverage ranges are given, the coverage data was obtained primarily from national household surveys. Where no coverage ranges are given, the coverage has been estimated by the authors of paper 2, after discussion with other experts, and is the same across all countries.
For each specific cause, the under-5 deaths preventable were calculated on the basis of all relevant interventions for that cause. In order to avoid double counting, after the first intervention each subsequent intervention was applied to the deaths not prevented by the previous intervention. The resulting number of deaths preventable is independent of the order of the application of the interventions.
Application of the Lancet approach at state level requires data both on the distribution of under-5 deaths by cause and the current coverage levels of the key interventions. While the Sample Registration System (SRS), carried out regularly by the Office of the Registrar General of India (RGI), provides the best estimates of neonatal, infant and under-five mortality at national and at state levels for the major states, cause of deaths data is more problematic.
The RGI used to conduct a survey that provided estimates of deaths by cause based on the International Classification of Diseases. However, the survey was limited to rural areas, produced estimates of uncertain quality and was discontinued in 1998. It is being replaced by what should be a much more representative system, based on the SRS, which will classify detailed cause of death information using a dual system with independent adjudication. Unfortunately, the system had not produced any results yet at the time of the preparation of this manuscript.
However, state level data are available on the proportion of under-5 deaths that occur in the neonatal period, from two sources -the two National Family Health Surveys (NFHS) (5, 8) and SRS.[9],[10] Furthermore, information on the coverage of interventions at state level was available from NFHS and from the Multiple Indicator Cluster Survey.[11] These types and sources of information were therefore used to assess potential impact of proven interventions at state level.
Results
Nationwide
Information on the coverage of interventions around the year 2000 is shown in table1. Compared to the 42 countries India has markedly lower coverage rates for vitamin A supplementation and for use of clean water, sanitation, hygiene and has higher malnutrition rates. table2 shows that the main specific causes of death are diarrhoea, pneumonia, and asphyxia, pre-term delivery, sepsis and tetanus for deaths among the neonates.
The under-5 deaths preventable through universal application of level 1 and level 2 interventions for India and for the 42 countries as a group were calculated for individual intervention, specific cause and group of interventions by location. Figure2Figure3 show the under-5 deaths preventable through individual intervention for India and the 42 countries as a group for prevention and treatment respectively. The ranking of effective interventions and the magnitude of their impact are very similar for India and the 42 countries. The most striking difference between India and the group of 42 countries is in the prevention and treatment of malaria, because reported mortality due to malaria in India is low and intermittent preventive treatment with anti-malarial drugs is not part of India's strategy for control of malaria. For India, individual preventive interventions that can avert the largest percentage of deaths are early and exclusive breastfeeding (more than 15%), clean delivery (more than 6%), and complementary feeding (more than 4%), with breastfeeding being the most effective intervention by far. Oral rehydration therapy is the most effective treatment intervention and could avert about 14% of deaths.
The overall proportion of deaths preventable through universal coverage of the listed interventions is a little lower for India table2 than for the group of 42 countries table3 - 57% vs 63%.
Interventions are also grouped on the basis of locations where they are applied Figure4. The percentages shown cannot be added since each group of interventions is applied independently of the others. While results show that the under-5 deaths preventable for interventions that are health facility centric are slightly higher for India than for the group of 42 countries, what is more important is that home care interventions prevent the largest proportion of deaths of all the location groups.
Individual States
A comparison of the under-5 mortality rate and proportion of deaths that occur in the neonatal period from the two NFHS surveys (5, 8) and SRS 1996 and 2000 (9, 10) is given in table4. As can be seen from the table, neonatal as a percentage of under-5 mortality is fairly consistent both intra- and inter- source. Where they differ considerably, as in the case of Goa and Kerala, this is likely because of sampling procedures in the NFHS, since the relative sampling error for Kerala, for example, is over 30% for the neonatal mortality estimate. If the smaller sample size states are excluded, the proportion of under-5 deaths that are neonatal varies across states from 40% to 55% based on the NFHS 1998-99 and, for a smaller range of states, from 43% to 57% based on the SRS.
State level data on coverage for interventions was obtained from NFHS (1998-99) and the Multiple Indicator Survey (MICS 2000) [11] table5. Because data on breast-feeding at 6 to 11 months was not readily available, percent coverage for breastfeeding at the time of the survey data collection at less than two years of age was used as a related proxy. Most of these coverage figures vary widely across states and are comparable in most cases to the range of coverage found across the 42 countries used in Lancet approach. The range of coverage of each intervention across the 42 countries and across Indian states is shown in table6.
For India in paper 2 the proportion of under-5 deaths that are neonatal differs from that reported by NFHS and SRS. The Lancet uses 36% as the neonatal proportion, whereas the NFHS 1998-99 reports 46% and SRS 2000 reports 50%. What might happen if 46% from the NFHS is used instead of the 36% used in the Lancet approach The result is very little change in the overall number or proportion of deaths prevented - a change from 57% deaths prevented to 56%. The robustness of the package of interventions is clearly evident when the overall difference in deaths prevented is only one percentage point.
Although there are coverage data at state level on the key interventions to reduce child mortality, and these have been noted above as mostly comparable to their range across the 42 countries, the comparison with cause of death data is more limited since only data on neonatal causes in total across the states are available. The proportion neonatal across the Indian states varies from 35% to 55% and that across the 42 countries from 18% to 54%. The under-5 mortality rate (U5MR) also differs between the two groups, ranging from 19 to 138 per 1000 live births for the states and from 30 to over 300 for the 42 countries.
Nevertheless, useful inferences can be drawn using this more limited mortality data, as described in the remainder of this section. Figure5 links the proportion neonatal to U5MR for the group of 42 countries and shows the well-known relation that as U5MR declines, the proportion of neonatal deaths increases. As can be seen from the figure, the proportion of neonatal deaths varies from less than 20% when U5MR is above 200, to more than 50% at low levels of U5MR.
Figure6 shows that the proportion of deaths preventable is related to the level of U5MR, with more deaths preventable by the interventions at high levels of U5MR than at low levels. Figure7 shows the relationship between proportions of deaths preventable by proportion neonatal, with a higher proportion of deaths preventable at lower levels of neonatal mortality. A conclusion from these last two figures is that in order to make comparisons between the Indian states and the 42 countries, both the proportion neonatal and the level of U5MR have to be comparable.
Figure8 shows the relation between deaths prevented and neonatal deaths, when the proportions of neonatal deaths and the level of U5MR are limited to a range comparable to the Indian states, that is, 35% and above for proportion of neonatal deaths and a U5MR below 130. The key inference from this figure is that for countries with comparable levels of neonatal and U5MR, the proportion of deaths preventable range from about 50% to 70%. The lower level of preventable deaths is likely more comparable with Kerala, and the higher levels of preventable deaths more comparable to Uttar Pradesh, Madhya Pradesh and Orissa.
Clearly the results obtained from the comparison cannot be equated to the results of the Lancet approach using a full set of cause of death data by state. Nevertheless, the above analysis utilizes comparable countries from the 42 country set to provide a useful and indicative assessment of the deaths that could be prevented at state level if the Lancet set of interventions were applied in the Indian states.
Conclusion
The Lancet approach indicates that 57% of the annual deaths among under-fives can be prevented through achievement of high coverage of basic public health and nutrition interventions. An effective implementation of these interventions would lead to achievement of the Millennium Development Goal on child mortality. Exclusive breastfeeding for the first 6 months of life is a highly effective way to reduce child mortality. Breastfeeding is widely practiced in India, but early and exclusive breastfeeding is not common. Too many mothers provide pre-lacteal food, throw away colostrum, and start giving fluids and foods too early before the child reaches 6 months of age.[5] Adequate complementary feeding is also still too rare in India. Mothers start either too early or too late, do not feed frequently enough, and do not provide enough protein, and vitamin and mineral rich foods. Unfortunately, the widespread availability of ORS and application of oral rehydration therapy, a highly effective therapeutic intervention, is also lacking.
Given the paucity of representative data for cause of death, it was difficult to directly apply the Lancet approach to State level data. The Office of the Registrar General of India is collecting this information for few states and representative data on cause of death on a regular basis is expected to be available soon. In the context of the lack of a full range of cause of death data by state, the paper has utilized the available data on neonatal causes of death and intervention coverage by state to show what would be the likely proportion of deaths preventable at state level if these interventions reached full coverage. The results show that as the levels of under-5 mortality falls, the proportion of deaths that are neonatal increases due to a increase in pace of decline in post neonatal mortality rate. This has also been reported elsewhere [6]. The data suggest that the proportion of deaths which are preventable ranges from about 50% to 70% when analysis is limited to countries with proportion of neonatal deaths and of under five mortality rates within the range found in Indian states. Clearly the results obtained from the comparison cannot be equated to the results of the Lancet approach using a full set of cause of death data by state, but it does provide an indication of the deaths that could be prevented at state level.
As indicated in paper 2, with the exception of interventions for neonatal sepsis, there is limited evidence of effect for addressing causes of death in the neonatal period, showing an urgent need for research in this area. A study to evaluate the quality of research published in major Indian journals showed that only 12% of the studies published in these pertained to neonates; of these less than 20% were studies of national interest and only 5% were community based [12]. A recent editorial in the Indian Peadiatrics highlights the neglect of newborn care in the country from primary to tertiary levels and calls for a rethinking on the issues related to delivery of newborn care; both strengthening of existing intervention and to identify and introduce new ones.[13] The specific issues related to the reduction of newborn deaths were also addressed in a recent publication in The Lancet.[14] In that publication a set of antenatal, intrapartum, and postnatal interventions were identified which could lead to a 41-72% reduction in neonatal mortality. In the context of the present paper primarily intra-partum and post-natal interventions were analyzed. Additional interventions listed in the neo-natal mortality paper[14] which were not used for calculation in the present paper, and which could possibly add to the mortality reduction potential during the post-natal period, are additional special care of low birth weight infants.
Planning and implementation of child health and nutrition and child development programmes should prioritize the full implementation of the mentioned high-impact interventions if the national goals on child mortality reduction are to be achieved [15]. A rapid implementation of the Indian Integrated Management of Neo-natal and Childhood Illnesses programme should help to achieve this. Expansion of the coverage of the integrated child development scheme, especially focusing on children under-three would also be crucial. In view of the fact that at higher levels of neonatal mortality the proportion of under-five deaths preventable are lower and the wide inter-state differentials in child mortality exist, it is essential that recommendations of differential strategies for interventions for states with high and low levels of under 5 mortality be an integral part of planning future programmes.
Acknowledgement
The assistance and advice of Dr R Sankar in the writing of this manuscript was highly appreciated.
References
1. Black RE, SS Morris, J Bryce. Where and why are 10 million children dying each year The Lancet 2003; 361 : 2226-2234.
2. UNICEF. Progress since the World Summit for children: a statistical review. New York: UNICEF, 2001. http://www.unicef.org/pubsgen/wethechildren-stats/sgreport_adapted_stats_eng.pdf (accessed November16, 2004)
3. UN. General assembly, 56th session. Road Map Towards the Implementation of the United Nations Millennium Declaration: report of the Secretary General (UN document no. A/56/326). New York: United Nations, 2001.
4. UNICEF. State of the World's Children 2005. New York: UNICEF, 2004.
5. National Family Health Survey (NFHS-2) 1998-99. International Institute for Population Sciences. Mumbai, India and ORC Macro, Maryland USA, October 2000.
6. Claeson M, Bos ER, Mawaji T, Pathmanathan I. Reducing child mortality in India in the new millennium. Bulletin of the World Health Organization 2000; 78 : 1192-1199.
7. Jones G, Steketee RW, Black RE, Bhutta ZA, Morris SS and the Bellagio Child Survival Study. How many child deaths can we prevent this year. The Lancet 2003; 362 : 65-71.
8. National Family Health Survey 1992-93. International Institute for Population Sciences, Bombay, India, August 1995.
9. SRS Bulletin 1996. Registrar General, India. Sample Registration System Bulletin, 1998;32(2):1-3.
10. SRS Bulletin 2000. Registrar General, India. Sample Registration System Bulletin, 2000; 34 (2): 1-4
11. Multiple Indicator Survey (MICS-2000) - India Summary Report , Department of Women and Child Development, Government of India and UNCIEF.Dec.2001
12. Narang A, Murki S. Research in Neonatology: Need for Introspection. Indian Pediatrics 2004; 41 : 170-174
13. Santosham K. Bhargava. The Challenge of Neonatal Mortality in India. Indian Pediatrics 2004; 41 : 657-662.
14. Darmstadt G, Bhutta Z, Cousens S, Adam T, Walker N, de Bernis L, for the Lancet Neonatal Survival Steering Team. Evidence-based, cost-effective interventions: how many newborn babies can we save The Lancet 2005; 365: 977-988.
15. Tenth Five Year Plan (2002-2007). Volume II Sectoral Policies and Programmes. Chapter 2.10 Family Welfare. Planning Commission Government of India, New Delhi.(Jones Gareth, Schultink W)
2 Health, United Nations Children's Fund, India Country Office, New Delhi, India
Abstract
About 2.1 million Indian children under 5 years of age die each year. In spite of reductions in child mortality rate over the past two decades, the rate remains high at 87 per 1000 live births. The main causes are diarrhoea, pneumonia, and for deaths among the neonates asphyxia, pre-term delivery, sepsis and tetanus. The major underlying cause of death is undernutrition. Child survival interventions of proven impact, feasible for use at high coverage in India were identified, and their effect on child mortality was calculated if high coverage were to be achieved. Exclusive breastfeeding, oral rehydration therapy, and adequate complementary feeding were among the most effective interventions. If these interventions would be applied universally 57% of mortality among pre-schoolers could be prevented. No cause specific mortality data were available from individual Indian states. Nevertheless, the range of child mortality, as well as the proportion of neo-natal deaths, occurring across the states, suggests that at state level 50-70% of deaths can be prevented. The results show that the targets set in the millennium development goals as well as in the Tenth Five Year plan can be reached.
Keywords: Child survival, Undernutrition, Breastfeeding, Oral rehydration therapy, India
Around the world more than 10 million children under the age of five die every year and the world's poorest countries continue to bear the burden of these deaths [1]. Levels of under-5 mortality vary widely across countries - from 4 to over 280 deaths per 1000 live births. About 41 percent of under-5 deaths occur in sub-Saharan Africa and another 34 percent in South Asia.[1]
Child health has been high on the international agenda for over two decades now. The World Summit for Children in 1990 called for a worldwide reduction in child mortality below 70 deaths per 1000 live births by the year 2000 [2]. Unfortunately this goal was not met. The Millennium Development Goal on under-five mortality is to reduce the rate by two-thirds between 1990 and 2015 [3]. For developing regions this would mean a reduction from 105 to 35 deaths per 1000 live births, and for South Asia a reduction from 126 to 42. According to UNICEF estimates, child mortality rates have been reduced by around 12 percent over the period 1990 - 2002. Given this average annual rate of reduction of 1 percent, the pace of progress will need to increase significantly during the remaining years in order for the millennium development goal (goal IV) to be met.
In India, about 2.1 million child deaths occur every year, which is the highest number within a single country worldwide.[4] The national under-5 mortality rate is around 87 per 1000 live births, but there is a wide variation between states. For example, in 1998-99 the under-5 mortality rate varied from 19 per 1000 births in Kerala to 138 in Madhya Pradesh.[5] For India, attaining the millennium development goal would imply a reduction of the under-5 mortality rate to 41 by 2015. There has been a substantial decrease in child mortality in the past two decades in India. The reduction was more marked in the 1980's than in the 1990's.[6] Despite these impressive gains, India compares poorly in the pace of child mortality reduction to several other countries in south and south-east Asia, including Bangladesh. More disturbing are the data which indicate that the decline in the child mortality rate is slowing in India [6].
The main causes of deaths in low-income countries are diarrhoea, pneumonia, measles, malaria, HIV/AIDS, and for deaths among the neonates asphyxia, pre-term delivery, sepsis and tetanus [1]. The major underlying cause of death is undernutrition, including micronutrient deficiencies, among children younger than 5 years, which is a determinant in more than 50% of deaths. The question can be raised whether the goals on mortality reduction can be achieved with the currently available interventions.
It is in this context that a series of articles on child survival appeared in The Lancet in mid-2003. The aim was to provide an evidence-based high impact approach to the task of reducing under-5 deaths. The Lancet Child Survival series describe how many of the under-5 deaths could be prevented with currently available interventions, whether delivery systems can achieve high coverage with these interventions and how to ensure that the poorest children receive the interventions of which they are most in need.[7] In particular, the second paper of the series showed that about two-thirds of child deaths could be prevented by interventions that are available today and are feasible for implementation in low-income countries if high levels of population coverage were to be achieved (referred to as Lancet approach in this paper).[7]
The aim of the present paper is to report on possible reductions in child deaths in India, using the Lancet approach, including how the approach might be applied at the state level. This is of particular relevance since India is presently going through a major revision of its health policy under the National Rural Health Mission, the Reproductive and Child Health Program phase 2 (RCH-2) as well as expanding the Integrated Child Development Services. The findings from this paper could help in identifying and developing the most effective and feasible interventions for reducing child mortality and malnutrition in India.
The paper briefly describes the Lancet approach at the global level and for India alone. An application of the Lancet approach at the State level, though limited by lack of cause of death data by state, then identifies the substantial range in the proportion of deaths that could be prevented across the states if universal coverage of key interventions are achieved.
Methodology
The second paper in The Lancet Child Survival series reviewed the state of evidence for interventions that reduce mortality for each of the major direct and underlying causes of death in children under-5 [7]. The focus was on interventions addressing proximal determinants of child mortality and those that can be delivered mainly through the health sector. For each of the main causes of under-5 deaths the effectiveness of available interventions was assessed.
Estimates of the effectiveness of interventions were taken either from published articles that summarized earlier research results or from systematic reviews by the authors and participants in the Bellagio Child Survival Study Group, together with input from other experts. Interventions were identified for each cause, and these were categorized by whether they are preventive or therapeutic. These interventions had to be feasible for delivery at high coverage levels in countries where most under-5 deaths occur; however, only the impact of an intervention on death was considered with no determination as to how the intervention will be delivered.
Each potential intervention was assigned to one of three levels of evidence based on the strength of evidence for its effect on under-5 mortality. Level 1 identified sufficient evidence - that a causal relationship between the intervention and cause-specific reductions of under-5 mortality had been established. Level 2 identified limited evidence - that an effect is possible, but data are not sufficient to establish a causal relationship. Level 3 included those interventions with inadequate evidence of effect. They include those that hold promise of substantial effects on under-5 mortality, but have not yet been fully assessed. A diagram of exposure, onset of disease or condition, and the subsequent resolution in terms of survival or death gives a visual summary of the interventions by cause with level 1 in green and level 2 in yellow Figure1Figure1a. Interventions are listed with prevention on the left and treatment on the right. Rotavirus, listed at the bottom of the diarrhoea diagram is classified as a level 3 intervention that holds considerable promise for the future.
Once the interventions were identified for each of the countries in the first paper of the Lancet series[1], the number of under five deaths that could be prevented was calculated with the coverage levels around the year 2000 increased to universal coverage. Universal coverage was taken as 99% except for exclusive breastfeeding, where 90% was used. The calculations divided into three types:
Exclusive and continuing breastfeeding, as this involved three levels: exclusive breastfeeding, partial breastfeeding and no breastfeeding
Complementary feeding, which utilized the underweight distribution of under-5's within a country
All other interventions, where the components of the calculation were intervention coverage (current and universal), efficacy, affected fraction of the population and evidence level
In applying the methodology described above and in the Lancet paper to the 42 countries, including India, a conservative approach was taken to the estimation of preventable under-5 deaths. There are four reasons why this is so.
Only interventions for which cause-specific evidence of effect was available were included
The methodology was restricted to interventions that are feasible at high coverage in low-income countries. As a result, some interventions were excluded for which there is evidence of effect, but that are only feasible for implementation in countries with higher levels of human, health-system and financial resources.
The methodology excluded promising interventions that are currently being assessed, for example rotavirus and pneumococcal vaccines.
It is limited to interventions that address the major causes of child death and selected underlying causes. Childhood anemia is one such example of conditions that contribute to child mortality, but for which the underlying causes and risk factors are not yet understood.
For India, and each of the other 42 countries, the number of under-5 deaths that could be prevented was calculated. Intervention coverage data for both India and the 42 countries as a group are reported in table1. For interventions where coverage ranges are given, the coverage data was obtained primarily from national household surveys. Where no coverage ranges are given, the coverage has been estimated by the authors of paper 2, after discussion with other experts, and is the same across all countries.
For each specific cause, the under-5 deaths preventable were calculated on the basis of all relevant interventions for that cause. In order to avoid double counting, after the first intervention each subsequent intervention was applied to the deaths not prevented by the previous intervention. The resulting number of deaths preventable is independent of the order of the application of the interventions.
Application of the Lancet approach at state level requires data both on the distribution of under-5 deaths by cause and the current coverage levels of the key interventions. While the Sample Registration System (SRS), carried out regularly by the Office of the Registrar General of India (RGI), provides the best estimates of neonatal, infant and under-five mortality at national and at state levels for the major states, cause of deaths data is more problematic.
The RGI used to conduct a survey that provided estimates of deaths by cause based on the International Classification of Diseases. However, the survey was limited to rural areas, produced estimates of uncertain quality and was discontinued in 1998. It is being replaced by what should be a much more representative system, based on the SRS, which will classify detailed cause of death information using a dual system with independent adjudication. Unfortunately, the system had not produced any results yet at the time of the preparation of this manuscript.
However, state level data are available on the proportion of under-5 deaths that occur in the neonatal period, from two sources -the two National Family Health Surveys (NFHS) (5, 8) and SRS.[9],[10] Furthermore, information on the coverage of interventions at state level was available from NFHS and from the Multiple Indicator Cluster Survey.[11] These types and sources of information were therefore used to assess potential impact of proven interventions at state level.
Results
Nationwide
Information on the coverage of interventions around the year 2000 is shown in table1. Compared to the 42 countries India has markedly lower coverage rates for vitamin A supplementation and for use of clean water, sanitation, hygiene and has higher malnutrition rates. table2 shows that the main specific causes of death are diarrhoea, pneumonia, and asphyxia, pre-term delivery, sepsis and tetanus for deaths among the neonates.
The under-5 deaths preventable through universal application of level 1 and level 2 interventions for India and for the 42 countries as a group were calculated for individual intervention, specific cause and group of interventions by location. Figure2Figure3 show the under-5 deaths preventable through individual intervention for India and the 42 countries as a group for prevention and treatment respectively. The ranking of effective interventions and the magnitude of their impact are very similar for India and the 42 countries. The most striking difference between India and the group of 42 countries is in the prevention and treatment of malaria, because reported mortality due to malaria in India is low and intermittent preventive treatment with anti-malarial drugs is not part of India's strategy for control of malaria. For India, individual preventive interventions that can avert the largest percentage of deaths are early and exclusive breastfeeding (more than 15%), clean delivery (more than 6%), and complementary feeding (more than 4%), with breastfeeding being the most effective intervention by far. Oral rehydration therapy is the most effective treatment intervention and could avert about 14% of deaths.
The overall proportion of deaths preventable through universal coverage of the listed interventions is a little lower for India table2 than for the group of 42 countries table3 - 57% vs 63%.
Interventions are also grouped on the basis of locations where they are applied Figure4. The percentages shown cannot be added since each group of interventions is applied independently of the others. While results show that the under-5 deaths preventable for interventions that are health facility centric are slightly higher for India than for the group of 42 countries, what is more important is that home care interventions prevent the largest proportion of deaths of all the location groups.
Individual States
A comparison of the under-5 mortality rate and proportion of deaths that occur in the neonatal period from the two NFHS surveys (5, 8) and SRS 1996 and 2000 (9, 10) is given in table4. As can be seen from the table, neonatal as a percentage of under-5 mortality is fairly consistent both intra- and inter- source. Where they differ considerably, as in the case of Goa and Kerala, this is likely because of sampling procedures in the NFHS, since the relative sampling error for Kerala, for example, is over 30% for the neonatal mortality estimate. If the smaller sample size states are excluded, the proportion of under-5 deaths that are neonatal varies across states from 40% to 55% based on the NFHS 1998-99 and, for a smaller range of states, from 43% to 57% based on the SRS.
State level data on coverage for interventions was obtained from NFHS (1998-99) and the Multiple Indicator Survey (MICS 2000) [11] table5. Because data on breast-feeding at 6 to 11 months was not readily available, percent coverage for breastfeeding at the time of the survey data collection at less than two years of age was used as a related proxy. Most of these coverage figures vary widely across states and are comparable in most cases to the range of coverage found across the 42 countries used in Lancet approach. The range of coverage of each intervention across the 42 countries and across Indian states is shown in table6.
For India in paper 2 the proportion of under-5 deaths that are neonatal differs from that reported by NFHS and SRS. The Lancet uses 36% as the neonatal proportion, whereas the NFHS 1998-99 reports 46% and SRS 2000 reports 50%. What might happen if 46% from the NFHS is used instead of the 36% used in the Lancet approach The result is very little change in the overall number or proportion of deaths prevented - a change from 57% deaths prevented to 56%. The robustness of the package of interventions is clearly evident when the overall difference in deaths prevented is only one percentage point.
Although there are coverage data at state level on the key interventions to reduce child mortality, and these have been noted above as mostly comparable to their range across the 42 countries, the comparison with cause of death data is more limited since only data on neonatal causes in total across the states are available. The proportion neonatal across the Indian states varies from 35% to 55% and that across the 42 countries from 18% to 54%. The under-5 mortality rate (U5MR) also differs between the two groups, ranging from 19 to 138 per 1000 live births for the states and from 30 to over 300 for the 42 countries.
Nevertheless, useful inferences can be drawn using this more limited mortality data, as described in the remainder of this section. Figure5 links the proportion neonatal to U5MR for the group of 42 countries and shows the well-known relation that as U5MR declines, the proportion of neonatal deaths increases. As can be seen from the figure, the proportion of neonatal deaths varies from less than 20% when U5MR is above 200, to more than 50% at low levels of U5MR.
Figure6 shows that the proportion of deaths preventable is related to the level of U5MR, with more deaths preventable by the interventions at high levels of U5MR than at low levels. Figure7 shows the relationship between proportions of deaths preventable by proportion neonatal, with a higher proportion of deaths preventable at lower levels of neonatal mortality. A conclusion from these last two figures is that in order to make comparisons between the Indian states and the 42 countries, both the proportion neonatal and the level of U5MR have to be comparable.
Figure8 shows the relation between deaths prevented and neonatal deaths, when the proportions of neonatal deaths and the level of U5MR are limited to a range comparable to the Indian states, that is, 35% and above for proportion of neonatal deaths and a U5MR below 130. The key inference from this figure is that for countries with comparable levels of neonatal and U5MR, the proportion of deaths preventable range from about 50% to 70%. The lower level of preventable deaths is likely more comparable with Kerala, and the higher levels of preventable deaths more comparable to Uttar Pradesh, Madhya Pradesh and Orissa.
Clearly the results obtained from the comparison cannot be equated to the results of the Lancet approach using a full set of cause of death data by state. Nevertheless, the above analysis utilizes comparable countries from the 42 country set to provide a useful and indicative assessment of the deaths that could be prevented at state level if the Lancet set of interventions were applied in the Indian states.
Conclusion
The Lancet approach indicates that 57% of the annual deaths among under-fives can be prevented through achievement of high coverage of basic public health and nutrition interventions. An effective implementation of these interventions would lead to achievement of the Millennium Development Goal on child mortality. Exclusive breastfeeding for the first 6 months of life is a highly effective way to reduce child mortality. Breastfeeding is widely practiced in India, but early and exclusive breastfeeding is not common. Too many mothers provide pre-lacteal food, throw away colostrum, and start giving fluids and foods too early before the child reaches 6 months of age.[5] Adequate complementary feeding is also still too rare in India. Mothers start either too early or too late, do not feed frequently enough, and do not provide enough protein, and vitamin and mineral rich foods. Unfortunately, the widespread availability of ORS and application of oral rehydration therapy, a highly effective therapeutic intervention, is also lacking.
Given the paucity of representative data for cause of death, it was difficult to directly apply the Lancet approach to State level data. The Office of the Registrar General of India is collecting this information for few states and representative data on cause of death on a regular basis is expected to be available soon. In the context of the lack of a full range of cause of death data by state, the paper has utilized the available data on neonatal causes of death and intervention coverage by state to show what would be the likely proportion of deaths preventable at state level if these interventions reached full coverage. The results show that as the levels of under-5 mortality falls, the proportion of deaths that are neonatal increases due to a increase in pace of decline in post neonatal mortality rate. This has also been reported elsewhere [6]. The data suggest that the proportion of deaths which are preventable ranges from about 50% to 70% when analysis is limited to countries with proportion of neonatal deaths and of under five mortality rates within the range found in Indian states. Clearly the results obtained from the comparison cannot be equated to the results of the Lancet approach using a full set of cause of death data by state, but it does provide an indication of the deaths that could be prevented at state level.
As indicated in paper 2, with the exception of interventions for neonatal sepsis, there is limited evidence of effect for addressing causes of death in the neonatal period, showing an urgent need for research in this area. A study to evaluate the quality of research published in major Indian journals showed that only 12% of the studies published in these pertained to neonates; of these less than 20% were studies of national interest and only 5% were community based [12]. A recent editorial in the Indian Peadiatrics highlights the neglect of newborn care in the country from primary to tertiary levels and calls for a rethinking on the issues related to delivery of newborn care; both strengthening of existing intervention and to identify and introduce new ones.[13] The specific issues related to the reduction of newborn deaths were also addressed in a recent publication in The Lancet.[14] In that publication a set of antenatal, intrapartum, and postnatal interventions were identified which could lead to a 41-72% reduction in neonatal mortality. In the context of the present paper primarily intra-partum and post-natal interventions were analyzed. Additional interventions listed in the neo-natal mortality paper[14] which were not used for calculation in the present paper, and which could possibly add to the mortality reduction potential during the post-natal period, are additional special care of low birth weight infants.
Planning and implementation of child health and nutrition and child development programmes should prioritize the full implementation of the mentioned high-impact interventions if the national goals on child mortality reduction are to be achieved [15]. A rapid implementation of the Indian Integrated Management of Neo-natal and Childhood Illnesses programme should help to achieve this. Expansion of the coverage of the integrated child development scheme, especially focusing on children under-three would also be crucial. In view of the fact that at higher levels of neonatal mortality the proportion of under-five deaths preventable are lower and the wide inter-state differentials in child mortality exist, it is essential that recommendations of differential strategies for interventions for states with high and low levels of under 5 mortality be an integral part of planning future programmes.
Acknowledgement
The assistance and advice of Dr R Sankar in the writing of this manuscript was highly appreciated.
References
1. Black RE, SS Morris, J Bryce. Where and why are 10 million children dying each year The Lancet 2003; 361 : 2226-2234.
2. UNICEF. Progress since the World Summit for children: a statistical review. New York: UNICEF, 2001. http://www.unicef.org/pubsgen/wethechildren-stats/sgreport_adapted_stats_eng.pdf (accessed November16, 2004)
3. UN. General assembly, 56th session. Road Map Towards the Implementation of the United Nations Millennium Declaration: report of the Secretary General (UN document no. A/56/326). New York: United Nations, 2001.
4. UNICEF. State of the World's Children 2005. New York: UNICEF, 2004.
5. National Family Health Survey (NFHS-2) 1998-99. International Institute for Population Sciences. Mumbai, India and ORC Macro, Maryland USA, October 2000.
6. Claeson M, Bos ER, Mawaji T, Pathmanathan I. Reducing child mortality in India in the new millennium. Bulletin of the World Health Organization 2000; 78 : 1192-1199.
7. Jones G, Steketee RW, Black RE, Bhutta ZA, Morris SS and the Bellagio Child Survival Study. How many child deaths can we prevent this year. The Lancet 2003; 362 : 65-71.
8. National Family Health Survey 1992-93. International Institute for Population Sciences, Bombay, India, August 1995.
9. SRS Bulletin 1996. Registrar General, India. Sample Registration System Bulletin, 1998;32(2):1-3.
10. SRS Bulletin 2000. Registrar General, India. Sample Registration System Bulletin, 2000; 34 (2): 1-4
11. Multiple Indicator Survey (MICS-2000) - India Summary Report , Department of Women and Child Development, Government of India and UNCIEF.Dec.2001
12. Narang A, Murki S. Research in Neonatology: Need for Introspection. Indian Pediatrics 2004; 41 : 170-174
13. Santosham K. Bhargava. The Challenge of Neonatal Mortality in India. Indian Pediatrics 2004; 41 : 657-662.
14. Darmstadt G, Bhutta Z, Cousens S, Adam T, Walker N, de Bernis L, for the Lancet Neonatal Survival Steering Team. Evidence-based, cost-effective interventions: how many newborn babies can we save The Lancet 2005; 365: 977-988.
15. Tenth Five Year Plan (2002-2007). Volume II Sectoral Policies and Programmes. Chapter 2.10 Family Welfare. Planning Commission Government of India, New Delhi.(Jones Gareth, Schultink W)