Marginal Impact and Cost of Scaling Up Universal Neonatal Packages
Because newborn health depends on services in the continuum of care for mother, newborn, and child, a vertical program would be duplicative, expensive, and inappropriate (Tinker and others 2005). Hence costing and impact estimates will be based on marginal additions of neonatal-specific packages to existing maternal and child health (MCH) services (table 27.4). This scenario reflects the reality in many South Asian and Sub-Saharan African contexts, where MCH services exist but do not yet include newborn interventions. We will cost packages, because packages are more cost-effective than single interventions, and the emphasis is on the packages described for universal scale-up (table 27.3). The benefits take into account only neonatal deaths averted, whereas many of the interventions will also reduce maternal deaths, stillbirths, and childhood morbidity and disability—and therefore the benefits underestimate gains for both the fourth and the fifth MDGs.
[Table .]
Costing and impact simulations are provided using the "marginal budgeting for bottlenecks" tool, a prioritization tool developed by the United Nations Children's Fund (UNICEF), the World Bank, and WHO. The inputs for the analysis presented here are as follows:
-
Baseline epidemiology uses NMRs from the latest demographic and health surveys by country or state and recent local relevant demographic data, such as crude birth rates.
-
Cause-specific neonatal mortality estimates by country are from the Child Health Epidemiology Reference Group's neonatal estimates by country (Lawn, Cousens, and Wilczynska forthcoming).
-
Baseline coverage estimates for the neonatal packages presented in table 27.3 are taken from local data, if available (for example, exclusive breastfeeding prevalence), or drawn from coverage estimates in the Lancet newborn series (Darmstadt and others 2005; Knippenberg and others 2005; Lawn, Cousens, and Zupan 2005).
-
Impact estimates for neonatal mortality are from the literature, as presented in this chapter. The range uses the 95 percent confidence interval, rounded to the nearest 5 percent where available (table 27.3). If the data were from an efficacy trial or a before-and-after trial, the range in the literature was reduced to reflect the expected effectiveness, based on expert opinion. Cause-specific mortality was used to allow combinations of effects across packages, and the assumptions applied were aligned with those used in the Lancet neonatal series (Darmstadt and others 2005)—although the packages here differ, because this chapter is restricted to the neonatal period. The assumptions for cause-specific impact are detailed at http://www.fic.nih.gov/dcpp. The effect for outcomes other than neonatal ones was based on data in the marginal budgeting for bottlenecks tool, primarily from the Lancet Bellagio series (Jones and others 2003) and Cochrane reviews. Effects are combined in a residual manner; for example, deaths averted by preventive strategies are removed from the pool before curative approaches are applied, and hence the total effect is less than the sum of the effects. Years of life lost were calculated using local average life expectancy discounted at 3 percent per year. This measure equates to the fatal outcome component of disability-adjusted life years, as described in chapter 15.
-
Specific costs of adding the intervention packages are calculated on the basis of the cadre of worker, additional personnel time, in-service training, supervision, performance incentives, travel and subsistence costs for referral care, drugs, and equipment. Demand promotion and community mobilization are included. The costs of time, training, and incentives are based on national salary levels, using real country data or World Bank databases. The costs of commodities are based on the UNICEF supply system (http://www.supply.unicef.dk/Catalogue/). The cost of strengthening health systems, including improving management and logistics, constructing new facilities, and deploying and training new cadres of workers, is included in the comprehensive MCH package.
Table 27.4 presents the estimated NMR effects and per capita costs, in selected Indian states and Sub-Saharan African countries, of strengthening health systems to increase coverage with existing MCH packages (without neonatal care after birth). It then presents the additional specific costs of including neonatal packages at the family-community level and in clinical services and, finally, the combined costs for comprehensive MNCH. Results are shown for two coverage scenarios:
-
Scenario 1: increasing coverage of the interventions by 20 percent from the baseline
-
Scenario 2: increasing coverage to the level required to meet the fourth MDG, necessitating about a 45 to 60 percent reduction in NMR, depending on the baseline percentage of under-five mortality that is neonatal.
Table 27.4 shows that the addition of neonatal packages will reduce neonatal deaths at an average cost of about US$0.50 per capita per year for up to a 15 percent reduction in NMR at the family-community level and about US$0.20 per capita for a 22 percent NMR reduction at the clinical care level. Although the cost per capita is low for clinical care, the cost per case treated is higher, and the lag time to scale up is longer. The family-community neonatal package in India is estimated to cost US$100 to US$257 per year of life saved (table 27.4), which corresponds to about US$2,800 to US$7,800 per death averted. That is similar to the results of US$3,442 per neonatal death averted or US$111 per life year saved in a community participatory package in Nepal (US$4,397 and US$142 irrespectively, with health system strengthening) (Manandhar and others 2004).
The comprehensive MNCH package (the MCH package plus integrated neonatal packages) is more expensive than the neonatal packages alone: US$2.40 to US$2.80 per capita and per year for a 20 percent increase in coverage, and US$5.50 to US$5.70 to achieve the mortality reduction necessary to meet the fourth MDG (including the health system strengthening and demand-side approaches required). However, the effect of the MNCH packages on the NMR is more than double that of the neonatal packages alone—for example, a reduction of up to 58 percent in NMRs in Africa, compared with up to 22 percent using interventions in the neonatal period only. This finding emphasizes the advantages of a comprehensive approach across the continuum of care. Hence, the average cost per year of life saved is still low at US$380 (India) and US$432 (Sub-Saharan Africa) for a 20 percent increase in coverage, including costs of system strengthening. If the coverage of the MCH plus neonatal packages were to reach 90 percent, those packages would avert up to 71 percent of neonatal deaths in the African countries and up to 76 percent in the Indian states.
In settings where the current coverage of skilled care is low, opportunities exist to start with family care and extra care of LBW babies while building toward more challenging clinical packages. Some clinical care packages—such as simple extra care of the small baby or the provision of oral antibiotics for pneumonia later in the neonatal period—can be adapted for delivery through community health systems. Varying the cadres of worker involved or the level of health system at which the package is delivered may reduce the cost of the package, but it also necessitates extra supervision and attention to links with the formal health system. Box 27.2 describes the projected effect and cost of various packages in Ethiopia for a 12-year program to improve maternal and child survival targeted at achieving the fourth MDG by 2015. Outreach services such as prenatal care alone have an effect of about 10 percent on NMRs, but when they are combined with a family package using community health promoters, an additional 30 percent reduction in the NMR is projected in Ethiopia.
[Box 27.2]
Outreach and family care options are more feasible initially. Yet if commitment toward moving to strengthen the clinical care system is lacking, the potential reduction in NMRs over time from those options is limited, and the cost per death averted is higher. Although the estimated cost (averaged over 12 years, with gradually increasing amounts) is low, the input is higher than the current government and donor health expenditure of the countries examined. Thus, spending in India would have to be doubled, and in some African countries probably tripled. Considerable new funding is required at the national and international levels, as well as more efficient allocation and absorption of existing funds (Martines and others 2005).
