Search tips
Search criteria 


Logo of archdischfnArchives of Disease in Childhood - Fetal & NeonatalVisit this articleSubmit a manuscriptReceive email alertsContact usBMJ
Arch Dis Child Fetal Neonatal Ed. 2007 September; 92(5): F361–F366.
Published online 2007 March 22. doi:  10.1136/adc.2006.104836
PMCID: PMC2675357

Inequities in neonatal survival interventions: evidence from national surveys



Nearly four million children die during the first four weeks of life every year, yet known and effective interventions exist. Neonatal mortality has to be addressed to reach the millennium development goal for child survival.


To determine the extent of within‐country inequities in neonatal mortality and effective intervention coverage.


Neonatal, infant and child (under 2 years) mortality rates were calculated from empirical data from Demographic and Health Surveys for eight countries using direct estimation techniques. Wealth groups were constructed using the World Bank wealth index; neonatal mortality inequities were evaluated by comparing low:high quintile ratios; concentration indices were calculated for intervention coverage rates.


The proportion of under‐2 deaths occurring in the neonatal period ranged from 24.3% (Malawi) to 49.4% (Bangladesh). In all countries (excluding Haiti) inequities in neonatal mortality and intervention coverage were evident across wealth groups with more deaths and less coverage in the poorest, compared with the richest, quintile; the largest mortality differential was 2.1 (Nicaragua) and the smallest was 1.2 (Eritrea). In Nicaragua 33% of the poorest women had a skilled delivery compared with 98% of the richest; in Cambodia for antenatal care this was 18% (poorest) and 71% (richest). Low coverage of interventions tended to show top inequity patterns whereas high coverage tended to show bottom inequity patterns.


Reducing inequity is a necessary step in reducing neonatal deaths and also total child deaths. Intervention efforts need to begin to integrate approaches relevant to equity in programme design, implementation, monitoring and evaluation.

Keywords: inequity, intervention coverage, neonatal mortality, DHS surveys, Millennium Development Goals (MDGs)

Ninety nine per cent of neonatal deaths occur in developing countries,1 and global estimates of health problems display the widening gap of health disparities within each country, where health problems are much more likely to be concentrated among the poorest people.2 Poorer people have a higher burden of disease and they are also less likely to use health facilities and services.3 For example, 90% of deliveries in the poorest 20% of households occur at home in less developed countries.4 Although these disparities in health status are probably avoidable, reflecting socioeconomic differences rather than different individual preferences,5 health initiatives to improve the health of the poor have been largely unsuccessful.6 Tackling the inequities that exist among populations should be made a fundamental element of any health initiative.

Nearly four million children die during the first four weeks of life every year, yet known and effective interventions exist.1,7,8 Reducing neonatal deaths is essential if the millennium development goal 4 (MDG‐4) is to be reached; it has been estimated that prevention of 70% of neonatal deaths would achieve a 25% reduction in under‐5s mortality.9 Many neonatal deaths are avoidable and could be prevented through several cost‐effective interventions delivered through primary healthcare systems and maternal health services.1,7,8 For example, deaths from neonatal tetanus account for about 14% of all neonatal deaths; the survival rate is low and more than 95% of those infected will die without treatment.10 However, poor socioeconomic status remains an important indirect risk factor for many neonatal health outcomes, such as tetanus infection, with many births occurring at home, without skilled attendance and with poor access to healthcare, often of poor quality. Antenatal and postnatal care (ANC/PNC) are also important factors in the health of the neonate; again the health‐seeking behaviour of the mother is indirectly influenced by socioeconomic status.11

A search of the peer‐reviewed literature using [neonat* AND socio‐economic* AND poverty AND trends AND developing countries] failed to identify a single study specifically looking at inequities in coverage of interventions for neonatal survival, although Lawn et al7 do mention this briefly as part of their study. This is in contrast to several recent findings concerning inequities in child and infant mortality.12,13,14,15,16

In this paper we present findings on within‐country inequities of neonatal mortality and coverage of key interventions using wealth quintiles and empirical data from Demographic and Health Surveys (DHS) in eight less developed countries.


Datasets/choice of countries

The DHS are designed with matching methodologies to allow nationally representative comparisons of various health and other indicators. A recent addition to the DHS database is the DHS/World Bank “wealth” index that seeks to measure the distribution of wealth across a country.17

Eight countries, Bangladesh, Benin, Cambodia, Eritrea, Haiti, Malawi, Nepal and Nicaragua, representing different global regions and meeting the selection criteria of having recent (2000 or later) surveys of sufficient sample size to allow for disaggregated analysis (table 11)) were chosen for this study. We compiled the empirical data for this analysis from the standard recode files (individual and child) available from Macro International (Calverton, Maryland, USA).

Table thumbnail
Table 1 Summary of sample sizes and mortality rates in the countries included in the study

Calculating mortality rates

Three indices of childhood mortality were calculated for this study: neonatal mortality, which includes all deaths from birth to 28 completed days of life, infant mortality, all deaths up to and including 1 year, and under‐2 mortality, all deaths up to 2 years. All indices are expressed as deaths/1000 live births. We used direct estimation techniques based on a life‐table approach. Mortality rates were measured by calculating the survival status of children born using a 10‐year observation period prior to the survey, up to two years before the interview date, and using the ‘stset' command in Stata 8.0 (StataCorp, College Station, Texas, USA). A 10‐year observation period was used to ensure sufficient statistical power while not being too long to minimise recall bias.

For each mother we calculated the number of children born and the number of those children surviving to age 2. To avoid censoring effects, births in the two years preceding the interview were not included. We used under‐2 rather than under‐5, because for measuring survival to age 5 a longer censoring period was included, resulting in older estimates of inequality. Also there is little difference between the two rates as most of under‐5 mortality occurs in the first two years of life.18

Using a wealth index to assess inequities

DHS survey questionnaires are not designed to measure household income or consumption expenditures. This has often been seen as a major limitation19 because these two indicators are usually used to rank households by economic status, allowing measurement of economic disparities across populations. However, an asset or “wealth” index, developed by the World Bank,17 can be derived from DHS questions, such as ownership of a number of different assets from a defined list, and has been used as a measure of the long‐run, rather than current, economic status by household in many countries.19,20,21,22

The wealth index is constructed from a set of asset indicators, using principal components analysis to derive the weights.17 An asset index is produced for each household, on the basis of which households are ranked and then divided into quintiles, from the poorest 20% to the richest 20%.23 The wealth index provides a relative rather than absolute indicator of wealth status across households in each country and so cannot be used to compare directly across countries. However, cross‐country comparisons can be made by looking at the poor–rich gap, expressed as ratios, for health outcomes.

Coverage rates for interventions

The interventions chosen for this particular study were those with proved efficiency in reducing neonatal deaths in less developed countries24,25 and with reported data available in the DHS datasets. These were tetanus toxoid vaccinations, ANC and skilled delivery. We also included PNC in the analysis as an indicator of the continuum of care during the neonatal period after pregnancy.

All intervention variables were modelled as binary variables reflecting the WHO recommendations for the minimum number of tetanus toxoid doses, two or more6; the minimum number of antenatal visits, four or more26; and the definition of a skilled attendant: doctor, nurse or midwife.27 PNC was taken as any postnatal care during the neonatal period that was given by a medically trained provider.

Statistical analysis

We first carried out descriptive analyses to look at the differences between neonatal, infant and child mortality rates across quintiles within each country to determine what percentages of child and infant deaths were due to neonatal mortality. Because it is not possible to compare directly quintiles for different countries, ratios of poorest/wealthiest were constructed to allow cross‐country comparisons of the poor–rich gap in mortality outcomes. We then calculated concentration indices,28 using the “glcurve” command in Stata, to determine inequities in intervention coverage rates across wealth quintile groups. The concentration index takes the value between −1 and +1 with zero representing equal coverage of the intervention, or health outcome, across wealth quintile groups. Negative values indicate unequal concentrations of the health intervention, to the benefit of the poor, whereas the opposite can be said for positive values where there is a disproportionate concentration of the intervention in favour of the rich.

All analyses were carried out using Stata 8.0 taking into account the frequency weights provided to reproduce the national population.


Throughout the analysis we have included the results for Haiti because it fitted the selection criteria, even though there was an obvious difference between this and the other seven countries. Without looking further into the data it might not be unreasonable to suggest some problem with the dataset.

Neonatal mortality rates were lowest in Nicaragua (16.2/1000 live births) and highest in Bangladesh (46.2/1000 live births) (table 11).). However, in both these countries, neonatal deaths accounted for over 40% of the total under‐2 deaths (40.5% and 49.4%, respectively). Malawi, Benin and Haiti had the lowest proportion of under‐2 deaths in the neonatal period (24.3%, 26.1% and 26.5%, respectively). In all eight countries, neonatal deaths accounted for over 40% of all infant deaths, the proportion ranging from 40.3% in Malawi to 63.8% in Bangladesh. Although all but one of the countries had a mean gross national income (GNI) of US$500 per head or less, annually, there was considerable variation in the neonatal mortality rate, not related to the GNI.

Table 22 presents neonatal mortality rates across wealth quintile groups within each country and the ratio of the neonatal mortality rate in the poorest group compared with the richest group. This ratio was greater than 1 in all countries except Haiti, where the neonatal mortality rate was higher in the richest quintile than the poorest quintile; the lowest rates were seen in the three middle quintiles. The greatest differential was seen in the country with the lowest neonatal mortality rate, Nicaragua, where the neonatal death rate among babies born to women living in the poorest households was more than double that of those born to women in the richest households.

Table thumbnail
Table 2 Distribution of neonatal mortality rates (NNMR) (deaths/1000 live births) by wealth quintile group

The smallest differential was seen in Eritrea, one of the two poorest countries, with a ratio of 1.2, which is still substantial and corresponds to a 20% increased risk of neonatal death among babies in the poorest quintile compared with the richest. For the six very poor countries (GNI below US$400 per head annually) the low/high ratio of the neonatal mortality rate averaged 1.4, with an overall mean neonatal mortality rate of 38.8 and mean infant mortality rate of 81.5. Bringing down the neonatal mortality rates of all quintiles to that experienced by the richest quintile will reduce the country neonatal mortality rate by the percentage given in the last column of table 22,, with an average value of 18.5%, and, if the peculiar Haitian data be excluded, average value of 26%.

These differences in death rates reflect, inter alia, the differences in access to and use of health services. services.FiguresFigures 1 and 22 show that for every country (except Haiti) an increase in the coverage of tetanus toxoid vaccinations, ANC, PNC and skilled deliveries coincided with a decrease in neonatal mortality rate. For example, in Nicaragua 33% of the poorest women had a delivery that was attended by a skilled attendant compared with 98% of the richest women; in Cambodia 18% of the poorest women attended four or more antenatal care sessions compared with 71% of the richest women, and in Bangladesh and Malawi the poor–rich ratios for coverage of tetanus, ANC, PNC and skilled delivery were 0.8, 0.3, 0.6, 0.1 and 0.9, 0.8, 0.6, 0.6, respectively compared with a ratio of 1.3 for neonatal mortality in both countries.

figure fn104836.f1
Figure 1 Coverage of interventions (%) by wealth quintile group.
figure fn104836.f2
Figure 2 Neonatal mortality rate (per 1000 live births) by wealth quintile group.

Figure 11 gives a crude picture of the patterns of inequities of intervention coverage (see also table 33),), and which we have defined here as either “top”, “linear” or “bottom” inequities, terminology used by Victora et al.29 Top inequity refers to distribution patterns in which it is just the rich that benefit, whereas bottom inequity refers to situations in which it is just the poorest that “lose out”. A linear distribution indicates that there is a steady change with wealth index. We defined intervention coverage as low (<50%), intermediate (50–74%) or high ([gt-or-equal, slanted]75%). Top inequity patterns were more commonly observed for interventions with low coverage levels, suggesting that for these interventions there is a disproportionate coverage among the richest quintiles. This is particularly noticeable for skilled delivery for which there was a top inequity pattern for all countries except Benin and Nicaragua, where coverage rates were much higher compared with other countries.

Table thumbnail
Table 3 Summary of coverage and inequality results

Cambodia and Bangladesh showed this top inequity pattern for three of the interventions (tetanus toxoid, ANC and skilled delivery; and ANC, PNC and skilled delivery, respectively) for which coverage was extremely low. Coverage of tetanus, and to some extent ANC, showed mainly linear patterns, even where coverage was low (tetanus: Eritrea, Nepal and Nicaragua; ANC: Nepal). The 2005 World Report describes this linear pattern as “queuing”, in which poor populations within countries “have to queue behind the better off, waiting to get access to health services and hoping that benefits will eventually trickle down”.30 Nicaragua was the only country to show bottom inequity patterns for two of the interventions (ANC and skilled delivery), and fig 11 shows the change in shape of the graph with marginalisation of the poorest groups for these two interventions in this country.


DHS data from the eight countries showed inequalities in neonatal survival within each country with children born to poorer women at higher risk of death compared with those born to richer women, except in Haiti. In general the increased relative risk of neonatal mortality associated with the poorest wealth quintile group compared with the richest varied from 20% in Eritrea to 110% in Nicaragua. Furthermore, access to interventions was directly related to socioeconomic level, with poorer mothers having reduced access to interventions compared with those in the wealthier groups.

The lack of relationship between GNI and neonatal mortality rate, except for Nicaragua possibly, was surprising but may well be explained by between‐country diversity because all the countries, except Nicaragua, had in common low GNI. Further investigation would be necessary to explain this, which is beyond the scope of this study. In general, coverage of interventions was higher in those households that were better off. This was particularly evident for skilled delivery and PNC, and to some extent ANC. All countries, except Benin and Nicaragua, showed the more typical patterns of inequity distribution of poor countries, described here as top inequities (the World Health Report calls this mass deprivation30), where mothers in the richest quintiles have greater access to, in this case, skilled delivery, ANC and PNC, compared with those in the poorer groups. As attempts at coverage were made in a more general way, there was a shift to a linear pattern, with a shape that indicated the degree to which poverty still hinders access: markedly in Nepal and much less so in Malawi, for tetanus toxoid.

What is already known on this topic

  • Neonatal mortality is a substantial part of total child mortality, and varies within countries.

As coverage increases still further, the right end of the curve will tend to flatten, giving a convex effect, because of approaching the asymptote plus the problems in achieving 100% coverage, even in the rich, and sometimes also with a massive failure of coverage in the very poor, as with skilled delivery in Nicaragua. Here the rates of coverage of both skilled delivery and ANC were high and it has been suggested that the reduction in neonatal mortality achieved in the 1990s was more likely due to an emphasis on skills training of nurses and doctors rather than an expanse of sophisticated technology,30 which explains the bottom inequity patterns of these two interventions in this country. Also, because the neonatal mortality rate was relatively low, the neonatal deaths were probably due to causes other than tetanus infection, such as preterm births and malformations (as opposed to asphyxia, tetanus and infections, in which mortality is higher).1,7,31 This explains why the coverage of tetanus immunisation was fairly similar across the wealth quintiles but the neonatal mortality rate was not; it was also lower than might be expected given the relatively lower neonatal mortality rate.

What this study adds

  • This seems to be the first such analysis specifically looking at inequities in the coverage of neonatal survival interventions.
  • The coverage of effective interventions, especially among the poor, needs to be increased if we are to see a decline in the number of neonatal deaths.

Neonatal mortality disproportionately affects poor people; this inequity is a composite of many factors, including reduced access to appropriate interventions, reduced quality of health services and fewer health benefits such as subsidised healthcare. Equity among populations, especially those consistent with the goals of poverty reduction, needs to be improved if there are to be adequate reductions in neonatal deaths to approach the MDGs. In the eight countries in this study, neonatal deaths accounted for between 26% and 49% of all under‐2 child mortality. The last column of table 22 shows the degree of reduction achievable by removal of inequity: in a few countries as much as 40% of the neonatal mortality rate, but overall 26% (excluding Haiti). This is substantial, even though hard to achieve, but on its own is not adequate to achieve the MDG target of 67% reduction. Differentials in the coverage of four key interventions that could prevent neonatal deaths across wealth groups are apparent, as is a decreasing risk of neonatal death with increasing wealth. Targeting appropriate and tailor‐made health interventions to poor people, as well as trying to establish universal coverage, is needed. In general most governments and international organisations have shaped health intervention programmes based on the national or regional averages of health outcomes; programmes have not been designed to explicitly influence within‐community inequities.

Overcoming the considerable inequities that exist will be essential (but not sufficient) if the child survival MDG is to be reached. Intervention efforts must begin to integrate approaches relevant to equity in programme design, implementation, monitoring and evaluation. Successful approaches can be adopted32 and new pro‐poor strategies are urged if improvements in neonatal health as well as child survival more generally are to be seen.


The authors would like to thank Andy Slogget (Centre for Population Studies, LSHTM) for his valuable input into calculating mortality rates and to Emma Slaymaker (Centre for Population Studies, LSHTM) who provided essential data management assistance.

Author contributions

B F participated in the design, carried out the analysis, participated fully in the interpretation, prepared the first draft of the paper and incorporated comments. B K had the original idea for this analysis, provided advice at all stages of its execution and in the interpretation of findings, and contributed to the preparation and finalisation of the paper. Z P contributed to the background research of the paper and initial analyses, and commented on the draft paper. D B participated in discussions on the design, analysis and interpretation of results, identification of key issues and how they should be tackled, and commented on the draft paper.


ANC - antenatal care

DHS - Demographic and Health Surveys


gross national income -


millennium development goal -

PNC - postnatal care


Competing interests: None.

David Bradley holds a Leverhulme Emeritus Fellowship.


1. Zupan J, Aahman E. Perinatal mortality for the year 2000: estimates developed by WHO. Geneva: World Health Organization, 2005
2. World Health Organization Macroeconomics and health: investing in health for economic development. Report of the Commission on Macroeconomics and Health. Geneva: WHO, 2001. 42–54.54
3. Victora C G, Wagstaff A, Schellenberg J A. et al Applying an equity lens to child health and mortality: more of the same is not enough. Lancet 2003. 362233–241.241 [PubMed]
4. Ball S. New research shows millions of baby deaths preventable at low cost. London: International Perinatal Care Unit, Centre for International Child Health, 2004, Search via (accessed 18 June 2005)
5. Wagstaff A. Poverty and health sector inequalities. Bull World Health Organ 2002. 8097–105.105 [PubMed]
6. World Health Organization The World Health Report 2000: health systems‐improving performance. Geneva: WHO, 2000
7. Lawn J E, Cousens S, Zupan J, for the Lancet Neonatal Survival Steering Team 4 million neonatal deaths: When? Where? Why? Lancet 2005. 365891–900.900 [PubMed]
8. Darmstadt G L, Bhutta Z A, Cousens S, Lancet Neonatal Survival Steering Team et al Evidence‐based, cost‐effective interventions: how many newborn babies can we save? Lancet 2005. 365977–988.988 [PubMed]
9. Vinod P. IMCI targets and millennium development goals. 2004. Search via (accessed 29 October 2005)
10. World Health Organization Immunization, vaccines and biologicals. 2003. (accessed 28 June 2005)
11. US Agency for International Development Focused antenatal care planning and providing care during pregnancy. Maternal and Child Health Division, Office of Health, Infectious Diseases and Nutrition, Bureau for Global Health, US Agency for International Development.‐Sp01e.pdf (accessed 28 September 2005)
12. Wang L. Determinants of child mortality in LDCs: empirical findings from demographic and health surveys. Health Policy 2003. 65277–299.299 [PubMed]
13. Rutstein S O. Factors associated with trends in infant and child mortality in developing countries during the 1990s. Bull World Health Organ 2000. 781256–1270.1270 [PubMed]
14. Sastry N. What explains rural‐urban differentials in child mortality in Brazil? Soc Sci Med 1997. 44989–1002.1002 [PubMed]
15. Kost K, Amin S. Reproductive and socioeconomic determinants of child survival: confounded, interactive, and age‐dependent effects. Soc Biol 1992. 39139–150.150 [PubMed]
16. Agha S. The determinants of infant mortality in Pakistan. Soc Sci Med 2000. 51199–208.208 [PubMed]
17. Gwatkin D R, Rutstein S, Johnson K. et alSocio‐economic differences in health, nutrition, and population. Washington, DC: World Bank, 2000
18. Macro International Inc An assessment of the quality of health data in DHS surveys. Calverton, MD: ORC, Macro, 1999
19. Sahn D E, Stifle D. Exploring alternative measures of welfare in the absence of expenditure data. Review of Income and Wealth 2003. 49463–489.489
20. Filmer D, Pritchett L. Estimating wealth effects without expenditure data‐or tears: an application to educational enrolments in states of India. Demography 2001. 38115–132.132 [PubMed]
21. Rutstein S. Wealth versus expenditure: comparison between the DHS wealth index and household expenditures in four departments of Guatemala. Calverton, MD: ORC, Macro, 1999
22. Hammer J. Health outcomes across wealth groups in Brazil and India. Washington, DC: Mimeo, DECRG, World Bank, 1998
23. DHS+ Dimensions: a semiannual newsletter of the Demographic and Health Surveys project. 2002. (accessed 28 June 2005)
24. Jones G, Steketee R W, Black R E. et al How many child deaths can we prevent this year? Lancet 2003. 36265–71.71 [PubMed]
25. Saving Newborn Lives The state of the world's newborns: a report from Saving Newborn Lives. Washington DC: Save the Children, 2001, 1–44. (accessed 28 June 2005)
26. World Health Organization Antenatal care. Report of a Technical Working Group. WHO/FRH/MSM/96. 8. Geneva: WHO, 1994
27. World Health Organization Making pregnancy safer: the critical role of the skilled attendant. A joint statement by WHO, ICM and FIGO. Department of Reproductive Health and Research. Geneva: WHO, 2004
28. Wagstaff A, Watanabe N. Socioeconomic inequalities in child malnutrition in the developing world, Volume 1. Policy Research Working Paper 2434. Washington: World Bank, 2000, Search via (accessed 4 August 2005)
29. Victora C, Fenn B, Bryce J. et al Co‐coverage of preventive Interventions and implications for child survival strategies: evidence from national surveys. Lancet 2005. 3661460–1466.1466 [PubMed]
30. World Health Organization The World Health Report. 2005: Make every mother and child count. Geneva: World Health Organization, 2005, (accessed 16 August 2005)
31. Zupan J. Perinatal mortality in developing countries. N Engl J Med 2005. 3522047–2048.2048 [PubMed]
32. Claeson M, Waldman R J. The evolution of child health programmes in developing countries: from targeting diseases to targeting people. Bull World Health Organ 2000. 781234–1245.1245 [PubMed]

Articles from Archives of Disease in Childhood. Fetal and Neonatal Edition are provided here courtesy of BMJ Publishing Group