Health Policy and Planning Advance Access published online on February 11, 2008
Health Policy and Planning, doi:10.1093/heapol/czn001
Saving newborn lives in Asia and Africa: cost and impact of phased scale-up of interventions within the continuum of care
1 Associate Professor, Department of International Health; Director, International Center for Advancing Neonatal Health, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD, USA.
2 Senior Scientist, Department of International Health, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD, USA.
3 Senior Research and Policy Advisor, Saving Newborn Lives, Save the Children-US; and Health Systems Research Unit, Medical Research Council, Cape Town, South Africa.
4 Husein Lalji Dewraj Professor of Paediatrics & Child Health, The Aga Khan University, Karachi, Pakistan.
5 Department of International Health, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD, USA.
6 Professor of Epidemiology and Medical Statistics, London School of Hygiene & Tropical Medicine, London, UK.
* Corresponding author. Department of International Health, E8153, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD 21205, USA. Tel: +1 443–287–3003. Fax: +1 410–614–1419. Email: gdarmsta{at}jhsph.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussion and conclusionsReferences |
|---|
Background Policy makers and programme managers require more detailed information on the cost and impact of packages of evidenced-based interventions to save newborn lives, particularly in South Asia and sub-Saharan Africa, where most of the world's 4 million newborn deaths occur.
Methods We estimated the newborn deaths that could be averted by scaling up 16 interventions in 60 countries. We bundled the interventions in a variety of existing maternal and child health packages according to time period of delivery and service delivery mode, and calculated the additional running costs of implementing these interventions at scale (90% coverage) in sub-Saharan Africa and South Asia. The phased introduction and expansion of interventions was modelled to represent incremental strategies for scaling up neonatal care in developing country health systems.
Results Increasing coverage of 16 interventions to 90% could save 0.59–1.08 million lives in South Asia annually at an additional cost of US$0.90–1.76 billion. In sub-Saharan Africa, 0.45–0.80 million lives saved would cost US$0.68–1.32 billion. Additional costs for increased antenatal interventions are low, but given relatively high baseline coverage and lower impact, fewer additional newborn lives can be saved through this package (5–10%). Intrapartum care has higher impact (19–34% of deaths averted) but is costly (US$1.66–3.25 billion). Postnatal family-community care, with potential for high impact at low cost (10–27%, US$0.38–0.75 billion), has been neglected. A first phase of scaling up care in 36 high (NMR 30–45) and 15 very high (NMR >45) mortality countries would cost approximately US$0.56–1.10 and US$0.09–0.17 billion annually, respectively, and would avert 15–32% and 13–29% of neonatal deaths, respectively, in these countries. Full coverage with all interventions in the 51 high and very high mortality countries would cost US$2.23–4.37 billion, and avert 38–68% of neonatal deaths (1.13–2.05 million), at an extra cost per death averted of US$1100–3900.
Conclusions Low-cost, effective newborn health interventions can save millions of lives, primarily in South Asia and sub-Saharan Africa. Modelling costs and impact of intervention packages scaled up incrementally as health systems capacity increases can assist programme planning and help policy makers and donors identify stepwise targets for investments in newborn health.
Key Words: Neonatal survival, neonatal mortality, scaling up, MDG-4, evidence-based interventions, developing countries, health systems, service delivery
KEY MESSAGES
|
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussion and conclusionsReferences |
|---|
The Lancet Neonatal Survival Series, published in March 2005, provided an analysis of the global burden of neonatal mortality (Lawn et al. 2005a
); the potential impact of simple, cost-effective interventions to save newborn lives (Darmstadt et al. 2005c); the inputs needed to scale up country programmes to address neonatal health (Knippenberg et al. 2005); and the policy, programme and research action needed to fill key gaps in the delivery of interventions where they are most needed (Horton 2005; Martines et al. 2005; Tinker et al. 2005). The series has helped to stimulate policy and programme change at global and national levels (Lawn et al. 2006). However, more detailed information is needed to enable countries to predict the impact and cost of combinations of interventions introduced in locally adapted packages and in a phased manner over time. Such information would facilitate decisions for wise investment in existing programmes and development of cost-effective new programmes. In this paper, we provide estimates of the impact and additional running costs of providing increased coverage with evidence-based interventions in 60 countries identified as high priority countries by UNICEF. The lifesaving potential of these interventions and the additional costs needed to provide high (90%) coverage are described for a variety of intervention combinations that might be implemented at country level. We also examine the impact and cost of introducing interventions incrementally in phases in country settings with very high (>45/1000 live births) and high (30–45/1000) neonatal mortality rates (NMRs). The costs presented here represent the additional annual running costs for programmes at expanded coverage but do not include all the costs that might be incurred in expanding facilities and infrastructure to achieve high coverage.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussion and conclusionsReferences |
|---|
The methods used to estimate impact and cost of evidence-based interventions have been described previously (Darmstadt et al. 2005c
), with minor modifications. We included the 60 UNICEF priority countries (where 93% of all under-five deaths occur) (Table 1) and utilized the recent data available by country population figures (UNDP 2005), intervention coverage (UNICEF 2005a), and NMRs, the latter coming from recent Vital Registration data in countries with full coverage Demographic Health Surveys for the majority of countries and estimates from the World Health Organization (WHO) for the remainder (WHO 2006). The country population figures, intervention coverage estimates and NMRs used in this analysis are summarized in Web Table 1. Interventions were packaged by time period of implementation (antenatal, childbirth/intrapartum, or postnatal); service delivery mode (outreach, family-community care, or clinical care); and whether the intervention is appropriate for all settings (‘universal’), for certain settings where a condition targeted by the intervention poses a public health problem (‘situational’), or for settings with more advanced health systems capacity (‘additional’) (Tables 2 and 3). These categorizations were chosen based on their importance in developing a rational, phased approach to increasing coverage of interventions over time.
|
|
|
Impact estimation
For each country and each intervention, we applied the increase in coverage from current levels to target coverage, together with estimates of intervention effectiveness and neonatal deaths by cause (Bhutta et al. 2005; Darmstadt et al. 2005c) to estimate the number of avertable deaths, as described previously (Darmstadt et al. 2005c). Target coverage generally was set at 90% of those in need (unless otherwise specified); the population in need was calculated by identifying the general target population (e.g. all newborns, pregnant women) and multiplying by the incidence or proportion of the target population with the condition that the intervention was targeted against (see Web Table 2). Web Table 3 provides the evidence for impact of single interventions, and Web Table 2 shows the cause-specific impact for intervention packages used in the analysis.
We identified interventions for which there was good evidence of efficacy or effectiveness in reducing neonatal mortality. Since a given intervention may impact more than one cause of death, and deaths due to a specific cause may be averted by more than one intervention, we required estimates of the effect of interventions on specific causes of death in order to derive impact estimates. For many interventions, however, cause-specific effects were not reported. Therefore, for each intervention we estimated its likely effectiveness in reducing cause-specific mortality using the available literature (Bhutta et al. 2005; Darmstadt et al. 2005c), the work of the Neonatal Group of the Child Health and Epidemiology Reference Group (Bryce et al. 2003) and, when required, expert opinion (see ref 2, Table 3; Web Table 2). This process inevitably introduced substantial uncertainty. Consequently, we used a range of effectiveness estimates from low (pessimistic) to high (optimistic). These cause-specific effects were then applied to estimates of the numbers of neonatal deaths due to each cause in each of the 60 countries (WHO 2006), assuming that the population impact increases linearly with coverage. This strong but simple assumption about the relationship between coverage and impact was made in the absence of data to support a more complex form of relationship. Furthermore, we assumed a multiplicative model which estimates the relative reduction, rather than absolute reduction, in neonatal deaths due to a given intervention or intervention package. Thus, the number of deaths that would be expected to be prevented for a given cause of death by a given intervention was calculated as:
|
|
The equation was derived as follows. If the number of deaths at 0% coverage is N0, then the number of deaths at 100% coverage is N0 x (1 – I). For any given coverage P, the number of deaths can be written as: Number of deaths = N0(1 – P x I). Then, at current coverage P0, the current number of deaths is N = N0(1 – P0 x I), and so N0 = N/(1 – P0 x I). Increasing coverage to P1 will reduce the number of deaths to N0(1 – P1 x I). Therefore, the deaths prevented by increasing coverage from P0 to P1 will be N0(1 – P0 x I) – N0(1 – P1 x I) = N0 x I x (P1 – P0), and substituting N (current deaths) for N0 produces the formula above for deaths prevented. The number of deaths due to a given cause and prevented by the intervention was then subtracted from the current number of deaths, before calculating the impact of the next intervention. We calculated the deaths that could be prevented for a number of different scenarios with different combinations of interventions. Some scenarios modelled different phases in scaling up, and by comparing these scenarios the impact of phased introduction of interventions can be examined. However, within any given scenario, the order in which interventions are entered into the model has no importance because of the multiplicative nature of the model. For example, if we start with 100 deaths, reduce these by 10% (leaving 90) and then reduce the deaths which are left by 20%, we end up with 72 deaths. If we reduce 100 deaths first by 20% (leaving 80) and then those remaining by 10%, we again end up with 72 deaths.
We purposefully have not shown data on impact by individual intervention by country. This modelling exercise, while generating such figures for use in deriving summary global and regional estimates for intervention impact, is meant to be illustrative and serve to facilitate the formulation of policy and programme guidelines, while avoiding either comparisons among countries or a ‘one size fits all’ approach.
Cost estimation
Running costs, or the value of the resources used when the services are up and running at an expanded level of coverage of interventions over and above costs for current levels of coverage, were estimated using a spreadsheet approach as described previously (Bryce et al. 2005; Darmstadt et al. 2005c). Details regarding costing assumptions and data used in calculations are summarized in Web Table 1. Costs included commodities (medicines, equipment, supplies), in-service training, salaries and supervision of the health care providers, and depreciation and maintenance costs of health care facilities, amortized over 30–50 years. The costs for an intervention were calculated by: adding a unit cost for the service provision (either facility- or community-based) plus the costs for the drugs or equipment required for the intervention times the estimated number of interventions in the country. In addition, for some interventions, hospital and/or operating costs were added. We estimate only the costs to providers of delivering these interventions, and have not included other economic costs borne by the family. We did not include the costs for expanding infrastructure related to these interventions, given the lack of country-specific information required to make these estimates. There was no discounting as costs are an estimate of the current costs to provide 90% coverage of the critical interventions in 2006, if the infrastructure were available.
To obtain costs for a given intervention or package of interventions, we derived and summed costs for each component of each intervention in each country pertinent to the analysis. Interventions that were packaged together and delivered at the same time period or in a single visit to a clinic or by a health worker shared the cost of the clinic or health worker visit. We calculated the increase from current to target coverage, which enabled calculation of expanded or running costs, or new investments needed to support the incremental increase in coverage. We then determined the health service delivery inputs needed to increase the provision of the interventions from current to target levels of coverage, either in domiciliary settings by a family member or community health worker, at a clinic by a health care provider, or in a hospital (including days of hospitalization required). Drug costs were calculated based on the amount of first-line drug that would be used for treatment, and the estimated current price on the UNICEF Supply website (UNICEF 2005b). Similarly, commodity costs were obtained from the UNICEF Supply website and applied to all countries, since the commodities are available in all countries at these costs; we did not take into account local variations in supply or costs. Unit costs for inpatient care and outpatient services were obtained from the WHO CHOICE database (Adam et al. 2003). Unit costs for service provision increased stepwise with level of coverage, from 0–50%, 50–80% and 80–90%, to reflect the increasing unit costs for providing services to those harder-to-reach; the scaling-up factor was derived from WHO CHOICE (Mulligan et al. 2003). Unit costs for community-based care were based on 75% of the hourly costs for a midwife (including 30% additional costs for training and supervision), with data on midwife costs drawn from previous work (World Bank 2004). Costs were calculated using the year 2006 US$ applied to the base populations from the year 2006. Therefore, all costs have to be increased to reflect population growth in order to estimate the actual costs for target levels of service provision based on the year the targets would be reached.
While we have utilized the basic framework and assumptions for interventions as stated in The Lancet Neonatal Survival Series (Darmstadt et al. 2005c) (see Web Tables 1 and 2), the costs of some delivery strategies may differ in certain situations; for example, use of peer counselling versus community support groups for family and community packages. To reflect the uncertainty and variation inherent in the cost estimates, we performed a sensitivity analysis to obtain a range on the estimates for additional costs by varying the three primary assumptions driving cost estimates: (1) the costs of services provided, (2) the costs of drugs and equipment, and (3) the current level of coverage of these services. We set the costs of service provision and costs of drugs and equipment at 25% less than currently calculated to generate the lower bound in costs. In addition, we assumed that current coverage would be 25% higher (yielding a smaller population in need). Similarly, for the upper bound, we assumed 25% higher costs for services, drugs and equipment, and 25% lower (or 0, whichever is higher) current coverage levels. These ranges do not represent a statistical confidence interval around the estimated additional costs, but represent a sensitivity analysis of the effect of the key assumptions driving the cost estimates.
The range of cost was utilized to calculate cost per capita. Cost per death averted was calculated using ranges for both cost and impact, a sensitivity analysis which resulted in the widest possible uncertainty. Thus, the lower end for the range was calculated as the higher bound for the cost estimate divided by the low estimate for impact. Similarly, the upper end of the range was calculated as the lower bound for cost divided by the high estimate for impact.
Phased introduction and expansion of interventions
To derive estimates of incremental impact and health systems costs during programme expansion, we created scenarios wherein interventions were packaged by time period and service delivery mode of implementation to optimize feasibility and model programme approaches to implementation (Figure 1), with interventions introduced within existing maternal and child health packages and coverage increased in phases to simulate developing health systems. Prior cost-effectiveness analyses guided this approach, including an Expansion Path (Evans et al. 2005) showing the order in which interventions would be purchased at given levels of resource availability, if cost-effectiveness was the only consideration (Adam et al. 2005; Darmstadt et al. 2005c). The scenarios are purely illustrative and not prescriptive for any particular country or region; a phased programme would require tailoring to the local epidemiological and health system context. Due to anticipated differences in initial programme emphases and time required to achieve full coverage, we modelled different phasing scenarios for 15 very high NMR (>45) and 36 high NMR (30–45) countries (Figure 1). Costs were then summed for these 51 countries for each phase in scale-up. The total number of annual deaths in these 51 countries is 2.88 million, approximately three-quarters of the global burden of neonatal mortality (Lawn et al. 2005a).
|
In creating the phased approach to scaling up intervention coverage, we used several guiding principles: (1) initial emphasis, particularly in very high mortality countries, was placed on expanding outreach coverage [particularly tetanus toxoid (TT) immunization against high tetanus incidence in settings with poor outreach services] and family-community care; (2) clinical care was simultaneously strengthened in an integrated, graduated fashion; (3) integrated programmes for mothers and newborns were developed across service delivery modes; and (4) additional interventions were phased in as health systems capacity developed, achieving full coverage earlier in high compared with very high mortality countries. Thus, for example, in very high mortality countries, expansion in coverage of skilled maternal and immediate newborn care was assumed to begin with phase 2, following an initial phase emphasizing expansion in coverage of family-community and outreach services, and some strengthening of facility-based emergency obstetric and neonatal care. In high mortality countries, however, expansion of skilled maternal and immediate newborn care was assumed to start immediately by building on existing health systems capacity.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussion and conclusionsReferences |
|---|
The additional cost of delivering all proven interventions at 90% coverage in the 60 countries in our analysis was US$3.89 billion. This investment would save an estimated 36–66%, or 1.33–2.45 million of the current 3.7 million, deaths in these countries (Table 4). A sensitivity analysis produced lower and upper bounds on cost of US$2.68 and US$5.24 billion. This yielded a relative range around the point estimate of –31% on the lower bound and +35% on the upper bound. While uncertainty estimates are not shown for all cost figures presented in this analysis, the same relative range applies to the other cost estimates. The additional cost per death averted for the 60 countries was estimated at US$1100–4000, and additional cost per capita was US$0.62–1.22.
|
To facilitate country planning and in hopes of stimulating investments in neonatal health, we disaggregated the costs and impact for delivering interventions at high coverage. This was done for regions and for smaller packages of care by time period and by service delivery mode. This information is needed when planning to add certain interventions for newborns to existing maternal and child health programmes, or when starting new programmes with limited resources. We then synthesized and applied the lessons from this analysis in a set of scenarios illustrative of a phased, incremental approach to scaling-up programmes in a health systems context in countries.
Regional impact and cost
We first calculated the cost and the potential of interventions to save newborn lives on a regional basis. The greatest number of births and of newborn deaths is in Asia and hence most lives would be saved there (0.75–1.47 million annually). Approximately three-quarters of the deaths averted (0.59–1.08 million) would be in four countries of the South Asia sub-region (Table 4). An estimated 0.48–0.86 million newborn lives could be saved annually in Africa, 93% of these (0.45–0.80 million) in the sub-Saharan sub-region. Eighty per cent of the lives saved in sub-Saharan Africa would be realized in 10 countries (Nigeria, Ethiopia, Democratic Republic of Congo, Tanzania, Uganda, Côte D’Ivoire, Mozambique, Angola, Mali and Kenya). The proportion of lives saved is similar for the South Asia (38–70%) and sub-Saharan Africa (38–67%) sub-regions. The total additional cost is higher in South Asia (US$0.90–1.76 versus US$0.68–1.32 billion) but the per capita cost in South Asia may be lower (US$0.59–1.15 versus US$0.95–1.86 billion) as the cost is spread over a larger population (Table 1). Beyond these two regions, the large populations and high numbers of neonatal deaths in China and Indonesia in the East Asia and Pacific region warrant concerted programme action, as do several other countries outside these sub-regions.
Intervention packages
Next, we modelled the impact and cost of interventions individually (Table 2) and packaged (Table 3) according to time period of implementation and linking with existing maternal and child health packages (Table 5), an important factor in programme feasibility.
|
Antenatal
Interventions delivered to mothers during the antenatal period have limited additional potential to save newborn lives (Table 5). Intermittent presumptive treatment (IPT) for malaria would make little difference in South Asia, due to the low burden of malaria-related neonatal deaths there. In sub-Saharan Africa, only an estimated 2–5% of neonatal deaths would be averted, at an additional cost of US$0.03–0.05 billion. IPT has benefits, however, which extend well beyond the neonatal period (Jones et al. 2003) and also extend to the mother (Shulman et al. 1999; Mbaye et al. 2006). An estimated 4–7% of neonatal deaths could be averted in 60 countries if routine antenatal care (ANC)—including routine antenatal visits and examinations (as recommended by WHO, Villar et al. 2001), TT immunization, screening and treatment of syphilis in endemic areas, and IPT—was implemented at high (90%) coverage. Even if testing and treatment for bacteriuria was added to the ANC package, which requires a relatively well-developed health system, only 5–10% of neonatal deaths could be averted. All antenatal interventions would have greater impact in sub-Saharan Africa than in South Asia (7–14% versus 4–9% of deaths averted), despite the higher current levels of ANC attendance in Africa compared with Asia (AbouZahr and Wardlaw 2004), because conditions addressed by antenatal interventions, such as malaria, neonatal tetanus and syphilis, are more prevalent in Africa.
Overall, the additional cost for all antenatal interventions (US$0.23–0.45 billion) is 8% of the total needed to deliver all 16 interventions (US$2.68–5.24 billion).
Childbirth/intrapartum
Achieving high coverage of clean childbirth care practices during home and facility-based births would avert an estimated 5–8% of deaths in 60 countries (Table 5) at a cost of US$0.06–0.11 billion, assuming that clean childbirth kits are used during home births. Comprehensive skilled care at birth, including facilitated referrals for emergency obstetric and neonatal care, could avert 13–24% of deaths. Focusing solely on emergency obstetric care would have relatively low impact (3–11%) on newborn deaths, although maternal deaths would be reduced. However, the additional cost is high (US$0.63–1.24 billion). Adding corticosteroids for preterm labour to the emergency obstetric care package increases the impact (8–21%), but is usually feasible only in referral hospitals, and even in industrialized countries is often poorly implemented. High coverage of all intrapartum interventions, including clean home childbirth, skilled maternal and immediate neonatal care, emergency obstetric care, and additional interventions could avert up to one-third of neonatal deaths (19–34%), but at relatively high cost (US$1.66–3.25 billion), representing 62% of the additional funds needed for all interventions in 60 countries. Skilled intrapartum care becomes less costly, however, when bundled with emergency postnatal care for newborns, since the same health facilities can care for both mothers and newborns.
Postnatal
In settings with high NMRs, limited resources and poor health systems, initial emphasis on family and community-based interventions can avert a substantial proportion of neonatal deaths at relatively low cost (Darmstadt et al. 2005c; Knippenberg et al. 2005). Promotion of immediate and exclusive breastfeeding, keeping the baby warm, and clean cord care for all newborns (i.e. family package I, see Table 3), with special attention to feeding and hygiene for low birth weight (LBW) infants (i.e. community-based care of LBW infants), can avert 6–19% of deaths in 60 countries at a cost of US$0.36–0.71 billion (Table 5). Adding pneumonia case management by community-based health workers boosts the impact to 10–27% at little additional cost, since the same health workers who promote healthy home practices can be trained and equipped to detect and manage pneumonia (Bang et al. 2005). Implementation of all postnatal interventions, including facility-based emergency neonatal care, could avert 17–39% of deaths. This is slightly higher than the impact of intrapartum interventions, but similarly short of the goals of Millennium Development Goal (MDG) 4.
Delivering all postnatal care interventions globally would cost US$0.67–1.31 billion, 40% of the cost of all intrapartum interventions, and 25% of the total price tag for all antenatal, intrapartum and postnatal interventions.
Packaging of interventions by service delivery mode to achieve a continuum of care
Integrating care across a continuum of time periods and service delivery modes maximizes benefits and minimizes the costs for mothers, newborns and children (Table 5). Combining family and community care across time periods (i.e. family packages 1 and 2, community-based care of LBW infants, community-based pneumonia case management) could avert 14–32% of neonatal deaths in 60 countries, at a cost of US$0.72–1.41 billion. Combining all outreach (antenatal interventions) and family and community care interventions could avert 17–37% of deaths for US$0.82–1.61 billion. Emergency care for newborns in-hospital could avert 9–24% of deaths at a relatively low expanded cost of US$0.29–0.57 billion. By bundling emergency obstetric care with emergency neonatal care, 12–34% of deaths could be averted for US$0.93–1.82 billion.
Reductions in neonatal deaths of 50% and upwards are required if MDG-4 for child survival is to be realized (de Zoysa et al. 1998; Bryce et al. 2003; Darmstadt et al. 2005c; Freedman et al. 2005; Lawn et al. 2005a; Paul 2006). An integrated and equitable approach that simultaneously strengthens outreach and family-community care, while creating demand for and providing clinical services, is required (Knippenberg et al. 2005). Such an approach combining all but additional interventions across all time periods and service delivery modes could avert 31–60% of neonatal deaths. Incorporating the additional interventions as health systems capacity improves could increase impact to 36–66%.
Phased introduction and expansion of interventions
Packaging interventions across time periods and service delivery modes increases cost effectiveness (Knippenberg et al. 2005). However, countries require information on the cost and impact of programmes that introduce various intervention packages at different points in programme maturity and are scaled up incrementally. For this analysis, we focused on the highest mortality countries, largely in Asia and sub-Saharan Africa, resulting in higher estimates of impact at lower cost than might be found in lower mortality settings.
We estimated that an initial phase (i.e. Phase 1 for both high and very high mortality countries, Figure 1) of implementing interventions packaged across various time periods (i.e. antenatal, intrapartum, postnatal) and service delivery modes at incrementally increased coverage would avert an estimated 14–31% of deaths at an estimated cost of US$0.65–1.27 billion per year in the 51 very high (NMR >45) and high (NMR 30–45) mortality countries combined (Figure 2). Considering very high and high mortality countries separately, Phase 1 would cost US$0.09–0.17 billion per year and avert 13–29% of deaths in 15 very high mortality countries, and would cost US$0.56–1.10 billion per year and avert 15–32% of deaths in 36 high mortality countries. In very high mortality countries alone in Phase 1, outreach and family-community care combined would cost US$0.04–0.07 billion per year (Figure 3). By the end of Phase 2, 23–47% of deaths in both high and very high mortality countries could be averted at a total cost of US$1.28–2.51 billion.
|
|
Figure 3 shows the relatively greater cost in the early phases of programme development for outreach and family-community care, and the relatively higher investment in clinical care services in later phases, as health system capacity expands. The lower incremental impact during Phase 2 (an additional 9–16% of deaths averted) compared with Phase 1 reflects the fact that high coverage of outreach interventions is reached in Phase 1 in a number of countries. By the end of Phase 4, for an additional cost of US$2.23–4.37 billion, all interventions would be delivered at 90% coverage and could avert a total of 38–68% of neonatal deaths, or a total of 1.13–2.05 million lives each year. The extra cost per life saved when interventions are at full coverage in high and very high mortality countries is US$1100–3900, at US$0.86–1.68 per capita. In very high mortality countries, saving 0.40–0.70 million lives would cost an additional US$0.49–0.96 billion, or US$700–2400 per death averted, at US$0.98–1.92 per capita. In high mortality countries, saving 0.73–1.35 million lives would cost an additional US$1.74–3.41 billion, or U$1300–4700 per death averted, at US$0.83–1.62 per capita.
|
Discussion and conclusions |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussion and conclusionsReferences |
|---|
Several principles for scaling up programmes to avert neonatal deaths emerge from this analysis (Box 1). Consideration of the geographic distribution of neonatal deaths (Lawn et al. 2005a
) and regional analysis of the number of preventable deaths confirms that South Asia and sub-Saharan Africa have the greatest need. Both regions face unique challenges that demand locally adapted solutions. While the absolute burden of deaths averted tends to be slightly higher in Asia than in Africa, neonatal mortality rates, which more closely reflect inequity than total burden, tend to be higher in Africa. | Box 1 Prioritizing, phasing and pricing intervention packages Priority regions
Packages along the continuum of care
Packages according to service delivery mode
The price
Phasing of the cost
Priority gaps
|
Our data suggest that expanded coverage of antenatal interventions may have limited additional potential to save newborn lives. Two factors explain this finding: (1) current ANC coverage levels are high relative to many other interventions, and thus the incremental increase in reaching 90% coverage is small, and (2) antenatal interventions are less effective at preventing neonatal deaths than many intrapartum and postnatal interventions (Darmstadt et al. 2005b). Nevertheless, ANC remains a foundation for health education and birth and child-rearing preparedness, and is a critical point of contact between health services and the community, encouraging care-seeking and fostering a continuum of care (Tinker et al. 2005). If women come for ANC and receive high quality services, they may be encouraged to deliver in a facility, bring a sick neonate to a facility for care, or seek immunizations (Kwast 1998). Our previous cost-effectiveness analyses for sub-Saharan African regions (Adam et al. 2005; Darmstadt et al. 2005c) suggested that ANC interventions are highly cost-effective (Adam et al. 2005; Darmstadt et al. 2005c; Hogan et al. 2005).
Clean childbirth practices to prevent tetanus and other serious neonatal infections, such as sepsis arising from umbilical cord infections, could have substantial impact. Hygienic cord care along with high TT coverage as a routine part of ANC is thus a high priority, particularly in very high mortality settings where neonatal tetanus remains a problem (Lawn et al. 2005a). For clean childbirth kits (Tsu 2000; Crook 2002; Beun and Wood 2003; Winani et al. 2005) to be effective they must be introduced as part of a behaviour change programme that increases users’ awareness, understanding and use of clean childbirth techniques in the context of other improved essential newborn care practices, such as hygiene practices (e.g. hand-washing) (Beun and Wood 2003; Winani et al. 2005). Thus, our figure for the cost of clean childbirth is an underestimate, as we did not consider the cost of social marketing and behaviour change.
Universal access to skilled care at birth is fundamental to reducing maternal mortality, and is also important and cost-effective in reducing neonatal deaths (Adam et al. 2005; Darmstadt et al. 2005c). Major challenges exist, however, in training, placing and supervising skilled birth attendants, particularly in the underserved, very high mortality areas where they are needed most (Piper 1997). For skilled birth attendants to be optimally effective, they must be linked to a health facility that provides emergency obstetric and neonatal care to manage complications. Thus, major investments in health systems are required and these costs are not captured by this current work. Postnatal interventions, including essential newborn care (e.g. breastfeeding, thermal care, hygiene) promoted through behaviour change communications, community-based care of LBW infants and case management for pneumonia can all be provided by the same worker through a family-community service delivery mode, contributing to the relatively low cost (US$0.38–0.75 billion) of integrated postnatal family and community care. Despite the fact that postnatal interventions are overall more cost-effective than antenatal or intrapartum interventions (Adam et al. 2005; Darmstadt et al. 2005c), postnatal care remains one of the most neglected aspects of health care in developing countries (Seims 2004). Further integration of family and community care with facility-based emergency newborn care as well as emergency maternal care is essential to meet the community demand to provide for the care of sick newborns and mothers created through effective family-community-based interventions (Darmstadt et al. 2005a).
Continuum of care
The strategy advanced by The Lancet Neonatal Survival Series (Darmstadt et al. 2005c; Knippenberg et al. 2005; Martines et al. 2005; Paul 2006), and voiced by others too (Costello et al. 2004; Freedman et al. 2005; WHO 2005), for averting neonatal deaths, calls for an integrated and equitable system of outreach, family-community and facility-based clinical care. Emergency obstetric services must be available to substantially reduce maternal mortality (Koblinsky et al. 1999; Paxton et al. 2005; Rosenfield and Schwartz 2005), whereas effective community-based care is recognized as a key to child survival (Bryce et al. 2003; Victora et al. 2003; Costello et al. 2004). Thus, neonatal health serves as a bridge between maternal and child care, and development of health systems approaches to neonatal health that link community and facility-based care will also strengthen health services for mothers and older children. Two-thirds of the per capita cost to implement the 16 interventions considered here will also benefit mothers and older children (Martines et al. 2005). Further integration of maternal, neonatal and child health care, as advanced by the newly formed Partnership for Maternal, Neonatal and Child Health (PMNCH 2005), as well as reproductive health, including optimal birth spacing (Rutstein 2002), will further improve programme cost-effectiveness due to utilization of common health workers and health systems whenever possible to deliver the interventions (Bryce et al. 2005; Darmstadt et al. 2005c; Lawn et al. 2005a; Martines et al. 2005; Morris et al. 2005). Work is underway to determine the combined cost of averting deaths across the maternal–neonatal–under-five continuum of care, which will further demonstrate the gains possible through integrating care.
Scaling-up
We modelled programme scale-up in phases to include more interventions and higher coverage as health systems capacity grows. This analysis suggests that the proportion of deaths that can be averted is similar in very high and high mortality countries. In the phased approach that we propose, initial investments are focused on developing outreach and family-community care, with greater emphasis on expanding clinical care coming later, although investments in health systems development are needed from the outset.
The programme expansion phases that we used in modelling impact and cost in health systems in very high and high mortality countries may be useful as a general guide to programme expansion, but are not meant to be prescriptive. Selection of interventions, their order of introduction and expansion of coverage will need to be decided locally.
The reality of scaling up care to achieve the level of mortality reduction that our modelling suggests is possible is fraught with many challenges and constraints that have not been fully factored into our analyses (Knippenberg et al. 2005). In our exhaustive review of the evidence for impact of interventions on neonatal mortality, we found only 10 studies that took place in a health systems context, and all of these studies tested just a single intervention, not an integrated package, under conditions of efficacy, not effectiveness (Bhutta et al. 2005). Furthermore, in our review of the impact of packages of interventions on neonatal health, we could not identify any data that addressed the issue of effectiveness in a health systems context (Haws et al. 2007). A similar lack of effectiveness data for under-five children has been reported (Bryce et al. 2003; Claeson et al. 2003). Thus, many important questions remain about how to deliver interventions effectively in a health systems context.
We have included estimates of uncertainty around our impact and cost figures based on sensitivity analysis. However, at present there is a near complete lack of information on the process and actual costs and impact (effectiveness) of scaling up interventions in a health systems context. Only two studies have reported cost-effectiveness of a programme of neonatal health interventions implemented at high coverage, albeit in small populations [e.g. 40 000 in rural India (Bang et al. 1999), 180 000 in rural Nepal (Manandhar et al. 2004; Borghi et al. 2005)]. Cost per life saved was derived using widely different methods and reported to be US$95 (Bang et al. 1999) and US$3442 (Manandhar et al. 2004) (US$4397 including costs of health-service strengthening activities), a range which encompasses the estimates derived through our analysis. None of these studies, however, have included the costs for construction of new health facilities or the roads and other infrastructural investments that may be needed to support an expanded health system. A recent analysis based on five sub-Saharan African countries and five Indian states produced a range of US$2800–7800 per newborn life saved for the first 20% increase in coverage from baseline, including extensive infrastructural investments (Lawn et al. 2005b). This is in line with the Nepal findings (Manandhar et al. 2004).
Emphasis must be placed on obtaining data from developing country health systems on the actual costs and impact of programmes, an objective of Countdown to 2015 (Lawn 2005). While this work is underway, in the meantime, we have tried to present in this analysis figures to guide donors, policy makers and programme managers regarding programme costs and impact that could be anticipated under good programme conditions. We hope these figures will provide impetus for the investments that are needed to realize the MDGs for maternal and child survival. Furthermore, these analyses provide a framework for investing in health systems in a rational, incremental manner that evidence suggests can save millions of lives.
Finally, these analyses are not meant to be used to evaluate the cost-effectiveness of different interventions, either generically or for specific countries. Rather, one purpose was to develop estimates of the costs associated with running these interventions at scale so that they could be compared with other costing exercises at the global level, such as those done for HIV and AIDS and child survival using similar methods (Schwartländer et al. 2001; Stover et al. 2002; Bryce et al. 2005; Stover et al. 2006). This type of broad comparison suggests that efforts to reduce neonatal mortality can be as effective as those proposed for HIV and AIDS, and at lower estimated costs. Moreover, as argued in The Lancet Neonatal Survival Series (Darmstadt et al. 2005c; Martines et al. 2005), investment leading to broad expansion of neonatal health interventions (US$0.59–1.15 in additional costs per capita per year to achieve 90% coverage with proven interventions in South Asia, US$0.95–1.86 in sub-Saharan Africa) is achievable, and an opportunity that countries cannot afford to miss.
Country by country, feasibility and impact will vary based on existing health system resources, the cost of labour and supplies, and unique barriers to intervention implementation (Knippenberg et al. 2005). Calculating the cost of scaling up a package of 16 neonatal and child health interventions, Stenberg et al. (2007) found that the per capita scale-up cost was greatest (US$3.40) in those countries least ready to scale up, compared with those most ready (US$1.00). Readiness to scale up is strongly inversely correlated with mortality; the highest mortality countries are the least ready to scale up. The countries most ready to scale up, despite lower per capita costs, accounted for approximately half of the global price tag, primarily because this category had many countries with large populations. On average, governments will have to increase their budgets by at least 26% to implement the interventions Stenberg et al. (2007) assessed, but some of the poorest countries will require increases of 74% or more.
In The Lancet Neonatal Survival Series, Knippenberg et al. (2005) considered not only intervention implementation costs per capita but also considerable inputs for health system strengthening activities, and estimated that Ethiopia (a high mortality country) will require additional inputs of US$9.50 per capita per year to implement a package of interventions of benefit to neonatal health because of its weak health system, as opposed to Madagascar (a high mortality country with a more developed health system) which will require only US$5 per capita per year. Still, investments for both countries are significant, reflecting a 3-fold increase in health expenditures for Ethiopia and a 2-fold increase for Madagascar. Very high mortality countries, which will need to increase their budgets by a higher percentage, are unlikely to be able to reallocate within their existing budgets. Countries need guidance on how to proceed most efficiently and effectively toward this goal.
External donor inputs, ideally solicited by countries themselves (Martines et al. 2005), will be required in most countries, despite efforts of some to reallocate existing government budgets to improve health services for maternal, neonatal and child health programmes and services (e.g. Tanzania, Armstrong Schellenberg et al. 2004). Stenberg et al. (2007) note the difficulties among the countries with weakest health systems to mobilize the comparatively greater sums needed for scale-up. For countries that need external assistance, international funding must increase. Powell-Jackson et al. (2006) found that influxes of international aid in 2003–2004 for maternal, newborn and child health accounted for just 2% of gross aid disbursements to developing countries, and the 60 priority countries accounting for the majority of child and newborn deaths received just US$3.10 per capita, insufficient to reach MDGs 4 and 5 by our and others’ estimates if health systems strengthening activities are undertaken to support the running costs, as shown here, of recommended intervention packages. Clearly, an increased international commitment to funding maternal, newborn and child health programmes is needed alongside national efforts to prioritize maternal and newborn health.
As countries prepare to scale up interventions that can save newborn lives, several potential tools exist that donors, policy makers and programme managers can use to inform programme planning, intervention selection and resource allocation, including development of estimates of impact and cost of country-specific interventions. These include the modelling approach used here and in the Child Survival and Neonatal Survival series in The Lancet (Bryce et al. 2005; Darmstadt et al. 2005c; Knippenberg et al. 2005), Marginal Budgeting for Bottlenecks (MBB) by the World Bank and UNICEF (Soucat et al. 2002; Knippenberg et al. 2005), CHOICE (Edejer et al. 2003; Evans et al. 2005) and other costing tools by WHO (Stenberg et al. 2007). There is now an urgent need for a consensus planning toolkit to guide local selection and phasing of interventions and cost-effective allocation of resources for maternal, neonatal and child health programmes.
|
Acknowledgements |
|---|
Development of this paper was supported by the Bill & Melinda Gates Foundation. Support was also provided by The Office of Health, Infectious Diseases and Nutrition, Global Health Bureau, United States Agency for International Development, Washington, DC, Award No. GHS-A-00–03–00019–00, to the Department of International Health at The Johns Hopkins Bloomberg School of Public Health. Joy E Lawn was supported by the Saving Newborn Lives initiative of Save the Children-US through a grant from the Bill & Melinda Gates Foundation. Simon Cousens received some salary support from the Bill & Melinda Gates Foundation through the WHO for his contribution to this work, Neff Walker was supported by UNICEF, and Zulfiqar Bhutta was supported by Aga Khan University. The funding sources had no role in determining the content of the paper. The opinions expressed herein are those of the authors and do not necessarily reflect the views of any of the agencies.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussion and conclusionsReferences |
|---|
AbouZahr C, Wardlaw T. Antenatal care in developing countries – promises, achievements, and missed opportunities: an analysis of trends, levels, and differentials, 1990–2001 (2004) Geneva: UNICEF and WHO joint publication.
Adam T, Evans D, Murray C. Econometric estimation of country-specific hospital costs. Cost Effectiveness and Resource Allocation (2003) 1:3.[CrossRef]
Adam T, Lim SS, Mehta S, et al. Cost effectiveness analysis of strategies for maternal and neonatal health in developing countries. British Medical Journal (2005) 331:1107.
Armstrong Schellenberg JRM, Adam T, Mshinda H, et al. Effectiveness and cost of facility-based Integrated Management of Childhood Illness (IMCI) in Tanzania. The Lancet (2004) 364:1583–94.
Bang A, Bang R, Baitule S, et al. Effect of home-based neonatal care and management of sepsis on neonatal mortality: field trial in rural India. The Lancet (1999) 354:1955–61.
Bang AT, Reddy HM, Deshmukh MD, et al. Neonatal and infant mortality in the ten years (1993 to 2003) of the Gadchiroli field trial: effect of home-based neonatal care. Journal of Perinatology (2005) 25(Suppl. 1):S92–107.[CrossRef]
Beun M, Wood S. Acceptability and use of clean home delivery kits in Nepal: a qualitative study. Journal of Health, Population and Nutrition (2003) 21:367–73.[Web of Science][Medline]
Bhutta Z, Darmstadt G, Hasan B, Haws R. Community-based interventions for improving perinatal and neonatal outcomes in developing countries: a review of the evidence. Pediatrics (2005) 115:519–617.
Borghi J, Thapa B, Osrin D, et al. Economic assessment of a women's group intervention to improve birth outcomes in rural Nepal. The Lancet (2005) 366:1882–4.
Bryce J, el Arifeen S, Pariyo G, et al. Reducing child mortality: can public health deliver? The Lancet (2003) 362:159–64.
Bryce J, Black RE, Walker N, et al. Can the world afford to save the lives of 6 million children each year? The Lancet (2005) 365:2193–200.
Claeson M, Gillespie D, Mshinda H, et al. Knowledge into action for child survival. The Lancet (2003) 362:323–7.
Costello A, Osrin D, Manandhar D. Reducing maternal and neonatal mortality in the poorest communities. British Medical Journal (2004) 329:1166–8.
Crook B. The feasibility of clean delivery projects within the Saving Newborn Lives Initiative Global Program (2002) Washington, DC: Save the Children-US.
Darmstadt G, Kumar V, Singh P, et al. Community mobilization and behavior change communications promote evidence-based essential newborn care practices and reduce neonatal mortality in Uttar Pradesh, India [poster]. (2005a) Countdown to 2015: Tracking Progress in Child Survival. London, UK.
Darmstadt G, Saha S, Ahmed A, et al. Effect of topical treatment with skin barrier-enhancing emollients on nosocomial infections in preterm infants in Bangladesh: a randomized controlled trial. The Lancet (2005b) 365:1039–45.
Darmstadt GL, Bhutta ZA, Cousens S, et al. Evidence-based, cost-effective interventions: how many newborn babies can we save? The Lancet (2005c) 365:977–88.
de Zoysa I, Bhandari N, Akhtari N, Bhan MK. Careseeking for illness in young infants in an urban slum in India. Social Science and Medicine (1998) 47:2101–11.[CrossRef]
Edejer T, Baltussen R, Adam T, et al, eds. Making choices in health: WHO guide to cost-effectiveness analysis. (2003) Geneva: World Health Organization.
Evans DB, Edejer TT, Adam T, Lim SS. Methods to assess the costs and health effects of interventions for improving health in developing countries. British Medical Journal (2005) 331:1137–40.
Freedman L, Waldman R, de Pinho H, et al. Transforming health systems to improve the lives of women and children. The Lancet (2005) 365:997–1000.
Haws R, Thomas A, Bhutta Z, Darmstadt G. Impact of packaged interventions on neonatal health: a review of the evidence. Health Policy and Planning (2007) 22:193–215.
Hogan D, Baltussen R, Hayashi C, et al. Cost effectiveness analysis of strategies to combat HIV/AIDS in developing countries. British Medical Journal (2005) 331:1431–7.
Horton R. Newborn survival: putting children at the centre. The Lancet (2005) 365:821–2.
Jones G, Steketee R, Black R, et al. How many child deaths can we prevent this year? The Lancet (2003) 362:65–71.
Knippenberg R, Lawn J, Darmstadt G, et al. Coverage low, progress slow, inequity high; scaling up newborn care in countries. Neonatal Survival Series Paper 3. The Lancet (2005) 365:1087–98.
Koblinsky M, Campbell O, Heichelheim J. Organizing delivery care: what works for safe motherhood? Bulletin of the World Health Organization (1999) 77:399–406.[Web of Science][Medline]
Kwast BE. Quality of care in reproductive health programmes: monitoring and evaluation of quality improvement. Midwifery (1998) 14:199–206.[CrossRef][Web of Science][Medline]
Lawn J. The price tag for newborn and child survival [conference presentation]. (2005) Countdown to 2015: Tracking Progress in Child Survival, London, UK.
Lawn J, Cousens S, Zupan J. Four million neonatal deaths: Where? When? Why? Neonatal Survival Series Paper 1. The Lancet (2005a) 365:891–900.
Lawn J, Zupan J, Begkoyian G, Knippenberg R. Newborn survival. In: Disease control priorities in 5 sub-Saharan African countries and 5 Indian states—Jamison D, Measham A, eds. (2005b) 2nd. Washington, DC: The World Bank and the National Institutes of Health.
Lawn J, Wilckzynska-Ketende K, Cousens S. Estimating the causes of 4 million neonatal deaths in the year 2000. International Journal of Epidemiology (2006) 35:706–18.
Manandhar D, Osrin D, Shrestha B, et al. Effect of a participatory intervention with women's groups on birth outcomes in Nepal: cluster-randomised controlled trial. The Lancet (2004) 364:914–5.
Martines J, Paul VK, Bhutta ZA, et al. Neonatal survival: a call for action. The Lancet (2005) 365:1189–97.
Mbaye A, Richardson K, Balajo B, et al. A randomized, placebo-controlled trial of intermittent preventive treatment with sulphadoxine-pyrimethamine in Gambian multigravidae. Tropical Medicine and International Health (2006) 11:992–1002.[CrossRef]
Morris S, Black R, Shibuya K, et al. How many child deaths can we prevent? 2003 update [poster]. (2005) Countdown to 2015: Tracking Progress in Child Survival. London, UK.
Mulligan J, Fox-Rushby J, Adam T, et al. Unit costs of health care inputs in low and middle income regions. In: DCPP Working Paper No. 9. (2003) Bethesda, MD: Disease Control Priorities Project.
Paul VK. The current state of newborn health in low income countries and the way forward. Seminars in Fetal and Neonatal Medicine (2006) 11:7–14.[CrossRef]
Paxton A, Maine D, Freedman L, et al. The evidence for emergency obstetric care. International Journal of Gynaecology and Obstetrics (2005) 88:181–93.[CrossRef]
Piper C. Is there a place for traditional midwives in the provision of community-health services? Annals of Tropical Medicine and Parasitology (1997) 91:237–45.[CrossRef][Web of Science][Medline]
PMNCH. The Delhi Declaration. Lives in the Balance: The Partnership Meeting on Maternal, Newborn and Child Health. (2005) New Delhi: Partnership for Maternal, Neonatal and Child Health.
Powell-Jackson T, Borghi J, Mueller D, et al. Countdown to 2015: tracking donor assistance to maternal, newborn, and child health. The Lancet (2006) 368:1077–87.
Rosenfield A, Schwartz K. Improving the health of women in developing countries: the time is now. Journal of Midwifery and Women's Health (2005) 50:272–4.[CrossRef]
Rutstein S. Birth spacing and newborn health. In: American Public Health Association, 130th Annual Meeting and Exposition, Program and Abstracts (2002) Washington, DC: APHA.
Schwartländer B, Stover J, Walker N, et al. Resource needs for HIV/AIDS. Science (2001) 292:2434–6.
Seims L. Newborn care and survival. In: American Public Health Association, 132nd Annual Meeting and Exposition, Washington, DC (2004).
Shulman C, Dorman E, Cutts F, et al. Intermittent sulphadoxine-pyrimethimine to prevent severe anaemia secondary to malaria in pregnancy: a randomised, placebo-controlled trial. The Lancet (1999) 353:632–6.
Soucat A, Van Lerberghe W, Diop F, et al. Marginal Budgeting for Bottlenecks: a new costing and resource allocation practice to buy health results. Using health sector's budget expansion to progress towards the Millennium Development Goals in Sub-Saharan Africa. Africa Region Human Development (2002) Washington, DC: The World Bank.
Stenberg K, Johns B, Sherpbier RW, Tan-Torres Edejer T. A financial road map to scaling up essential child health interventions in 75 countries. Bulletin of the World Health Organization (2007) 85:305–14.[CrossRef][Web of Science][Medline]
Stover J, Walker N, Garnett GP, et al. Can we reverse the HIV/AIDS pandemic with an expanded response? The Lancet (2002) 360:73–7.
Stover J, Berozzi S, Guitierrez J-P, et al. The global impact and net costs of scaling-up prevention programmes for HIV/AIDS in low- and middle-income countries, 2005–2015. Science (2006) 311:1474–6.
Tinker A, ten Hoope-Bender P, Azfar S, et al. A continuum of care to save newborn lives. The Lancet (2005) 365:822–5.
Tsu V. Nepal clean home delivery kit: evaluation of health impact (2000) Washington, DC: Program for Appropriate Technology in Health.
UNDP. World Population Prospects: The 2004 Revision (2005) New York: United Nations Population Division, Department of Economics and Social Affairs.
UNICEF. State of the world's children 2006 (2005a) New York: UNICEF.
UNICEF. UNICEF Supply Division Catalogue (2005b) New York: UNICEF.
Victora CG, Wagstaff A, Schellenberg JA, et al. Applying an equity lens to child health and mortality: more of the same is not enough. The Lancet (2003) 362:233–41.
Villar J, Ba'aqeel H, Piaggio G, et al. WHO antenatal care randomised trial for the evaluation of a new model of routine antenatal care. The Lancet (2001) 357:1551–64.
WHO. The World Health Report 2005: making every mother and child count (2005) Geneva: World Health Organization.
WHO. Neonatal and perinatal mortality – Country, regional and global estimates (2006) Geneva: World Health Organization.
Winani S, Coffey P, Wood S. Evaluation of a clean delivery kit intervention in preventing cord infection and puerperal sepsis in Mwanza, Tanzania: Final research report. (2005) Seattle, WA: Program for Appropriate Technology in Health.
World Bank. The World Development Report 2004: Making services work for the poor (2004) Washington, DC: The World Bank.
Accepted for publication 17 December 2007.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  G. L Darmstadt, S. E. Arifeen, Y. Choi, S. Bari, S. M Rahman, I. Mannan, P. J Winch, A. N. U. Ahmed, H. R. Seraji, N. Begum, et al. Household surveillance of severe neonatal illness by community health workers in Mirzapur, Bangladesh: coverage and compliance with referral Health Policy Plan., November 16, 2009; (2009) czp048v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. J. Stoltzfus Research Needed to Strengthen Science and Programs for the Control of Iron Deficiency and Its Consequences in Young Children J. Nutr., December 1, 2008; 138(12): 2542 - 2546. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||