|
The Costs Of Scaling Up Vaccination In The World’s Poorest Countries
We examine the relationship between country-level average costs and coverage levels for diptheria-pertussis-tetanus (DTP) vaccines. Coverage data are from the World Health Organization, and cost data are from financial sustainability plans filed with the Global Alliance for Vaccines and Immunization (GAVI) by forty countries from 2000 to 2003. In this data set, average costs are lower for countries that vaccinate more children. At the highest numbers of covered children, there was no trend toward higher average costs. Vaccine programs in this set of poor countries have not yet scaled up to the point at which diminishing marginal returns are observed.
VACCINES ARE AMONG THE MOST EFFECTIVE and cost-effective interventions to improve child survival in low- and middle-income countries. Nevertheless, millions of children fail to receive the routine traditional six vaccines against diphtheria, pertussis, and tetanus (DPT); polio; bacillus Calmette-Guérin (BCG); and measles. The World Health Organization (WHO) estimates that in 2003 there were 528,400 cases of measles, 106,135 cases of pertussis, 13,831 cases of tetanus, and 6,654 cases of diphtheria.1 Scaling up immunization programs entails improving performance and assuring adequate long-term funding. Planning for long-term financial sustainability requires estimates of short-and long-term financial requirements. In this paper we extend the literature on financial sustainability by studying the statistical correlation between country-level immunization coverage and national immunization program spending from 2000 to 2003. Estimating costs. A common procedure in estimating the costs of scaling up vaccination programs is to identify the cost per fully immunized child and to implicitly assume constant returns to scale as the number of immunized children increases.2 One important question to consider is whether it will become progressively more expensive per child to increase vaccination coverage levels. If the unreached populations are more remote or more resistant to vaccinations, then one might predict that the cost per child will likely increase. On the other hand, if national vaccine programs achieve economies of scale because of up-front investments in physical or human capital, then the cost per child will likely decrease. Reporting requirements. Although national vaccine programs have routinely reported rates of vaccine coverage, until recently there were very few systematic collections of the costs incurred in delivering vaccines to children. Since 2002 the Global Alliance for Vaccines and Immunizations (GAVI) has required all countries receiving Vaccine Fund grants to prepare detailed financial sustainability plans (FSPs).3 These are documents that help national governments analyze and list the costs associated with immunizing children in their countries, assess the key financial challenges facing the national immunization program, and then present their strategies for mobilizing and effectively using financial resources to support the medium- and long-term immunization program objectives. National program staffs have to prepare these reports approximately 2.5 years after they have begun receiving GAVI support. The purpose is to have the national governments plan adequately for sustaining the expansions in their immunization programs once GAVI support has been withdrawn. By late 2005, forty countries had reported the costs of their vaccination programs, and, with one exception, each had prepared reports for two different years: the base year in which GAVI support was initiated and a subsequent year. The small number of countries reporting these costs limits our ability to make broad generalizations. Nevertheless, an initial examination of the association between average costs and coverage rates might shed light on some possible regularities. Foremost, we can test whether average costs appear to be larger or smaller in countries that vaccinate more children. Expected cost patterns. A national vaccine program can be likened to a factory engaged in the production of immunized children. The goal is to combine non-immunized children, vaccines, infrastructure, and health workers to produce "immunized children." National vaccine programs might be expected to have average short- and long-run cost patterns similar to general patterns experienced by factories in other industries.
Both long- and short-run average costs in the long run typically have a U shape when plotted against output (Exhibit 1
In looking at national vaccine programs in multiple countries, one would expect that a single country would not have as much scope to vary inputs during a three-to-four-year period as would a group of many countries whose vaccine programs had grown over decades. The experience of a single country over three to four years would inform us about that country’s short-run average costs. The experiences of a full set of countries together could indicate the long-run relationship between average costs and the number of children vaccinated. However, the long-run average cost curve is actually not attainable by any single country’s vaccine program as long as a fixed or shrinking number of children are born in a given country. The long-run average cost curve is of interest because it indicates what is possible under the maximal efficiency attainable with all constraints removed. An important constraint in the long term is that a small nation might have a limited number of non-immunized children and so might never be able to achieve efficiencies that are available on the scale of a large country or region. Inputs versus outputs. A key technical question is whether long-run average costs increase or decrease as output is increased. This can be answered either visually or through a statistical test to determine output elasticity. If output elasticity is greater than 1 in the long run, then a 10 percent increase in spending on program inputs would result in more than a 10 percent increase in the number of children vaccinated. The converse would be a regime of diminishing marginal returns in which one must use increasingly higher amounts of an input or resource to keep output increasing.
Data. We obtained data on the costs of national vaccine programs, vaccine coverage rates, and median gross domestic product (GDP) per capita from countries’ FSPs.4 In our baseline analysis, we followed the WHO’s recommendation to focus on just the costs of routine immunization (not national immunization program costs), eliminating the costs of vaccination campaigns, newly introduced vaccines, and shared costs.5 Spending on immunization campaigns and adoption of costly new vaccines can be quite variable. Shared costs (such as in-kind contributions of personnel and equipment by primary care providers) were optional for reporting countries and not consistently reported in the FSP; thus, we subtracted shared costs from our estimates when they were reported.6
Methods.
The estimates of average cost in the baseline analysis are a fraction of the full costs but provide an estimate purged of cost items that might have been inconsistently reported. Analyses that include shared costs and campaign costs were also conducted, and we summarize those results also. Exhibit 2
Costs of routine immunization. We prepared two versions of the costs of routine immunization: (1) including the costs to purchase four traditional vaccines (DPT, polio, measles, and BCG); and (2) excluding all vaccine purchase costs. Excluding vaccine costs corrects for the noncomparability of vaccine schedules across countries.8 Some countries give more than the recommended numbers of doses of DTP and measles, for example, through mass campaigns or additional booster shots. Since we focused on routine vaccination costs, the costs of immunization campaigns were not included in the baseline analysis. Costs per fully vaccinated child. We first measured coverage based on the number of children who have received three doses of DTP vaccine (DTP3), reasoning that DTP3 is a close measure of routine immunization program performance.9 We calculated the number of children covered by multiplying reported DTP3 coverage rates by the number of infants in each country surviving to age one. This approach overestimates per capita costs because some children who were vaccinated are now deceased. All estimates of the number of surviving infants and DTP3 vaccine coverage were from a single WHO source.10 Measures of the average cost per child covered with DTP3 were constructed as costs divided by the number of children receiving DTP3.
We first created scatter plots of costs per DTP3-immunized child versus DTP3 coverage to assess the general nature of the relationship (Exhibit 3
Predicting resource needs. In studying the elasticity, we examined costs versus number of children vaccinated. We considered (and subsequently rejected) using models of costs versus proportion vaccinated, controlling for the population of infants. If one thinks of a vaccination program as a service industry, the best measure of output that will predict resource needs is not the proportion of a community served but rather the total number served. To control for confounding that is attributable to institutional and infrastructural factors related to the general level of development, we included a dummy variable marking countries that had GDP per capita greater than the median for all countries in the sample.
Average costs. Exhibit 3  shows the relationship between the average cost per immunized child and the number of children immunized among the countries studied in the analytical sample of countries. It shows a monotonic decline in average costs, such that the countries that immunize more children tend to have lower average costs than countries that immunize fewer. This same monotonic decline was found in all versions of average cost studied. A quadratic curve fitted to the data in Exhibit 3 would have minimum costs at about 1.5 million vaccinated children.12 Alternative analyses. Limiting the analysis to only those countries whose shared costs could be explicitly extracted from the estimates led to a major reduction in sample size from ninety-three country-years of data to fifty-two. The number of countries where two waves of cost data were sufficiently detailed to permit an analysis of changes in cost was also limited. In thirteen of the fourteen cases where good longitudinal cost data were available for a single country, the average costs were rising over time. In nine cases, DPT3 coverage fell while costs rose. However, it should be noted that external funding was being made available to each country based in part on the reports of program spending in the baseline year. The phenomenon of rising average costs within a country despite declining average costs overall was seen no matter which version of average costs was used.
Costs of increasing vaccination.
To study the overall connection between costs and coverage, we estimated that for the whole sample of countries, a 10 percent increase in the number of children vaccinated with DTP3 would be associated with an 8.4 percent rise in costs.13 This elasticity estimate suggests increasing marginal returns (as in the dotted line in Exhibit 1
In this set of the world’s poorest countries, the average cost of vaccinating children is lower in countries where more children are immunized. Limitations. Our study could be strengthened in several respects. Ideally, one would like to have data from countries that have already attained very high vaccination rates. This would eliminate the need to extrapolate to the highest-coverage segment of the function as we did here. Second, more detailed cost data would allow us to entertain other important covariates such as logistics and health-sector productivity. In the future, these more detailed analyses will be feasible as data accumulate from more countries. Policy implications. The benefits of immunization have been noted widely and include preventing disability and death, reducing strains on health systems, creating opportunities for spending on other health programs, and adding to a country’s economic and social development.14 Also, scaling up immunization is seen as vital to achieving the 2015 Millennium Development Goal for child health.15 Improving our collective understanding of the costs of scaling up vaccination has important policy implications in that it will enable better long-term financial planning for national immunization programs. The GAVI Financing Task Force has stated that such planning can enable a country to "mobilize and efficiently use domestic and supplementary external resources on a reliable basis to achieve current and future target levels of immunization performance in terms of access, utilization, quality, safety and equity."16 Averaging fixed costs. The new data collected by GAVI for a selected set of low-income countries and presented here show a negative relationship between the average cost per child immunized and the total number of children immunized. Part of this relationship reflects the fixed costs required to establish the infrastructure of offices and trained administrators required to launch a national vaccine program in a country regardless of its population size. Larger countries’ ability to average these fixed costs over more children accounts for some of the lower average costs. However, if the vaccine program of a large country were to immunize very few children, its cost per child immunized would be quite large because of the smaller denominator. Thus, at least modest success in immunizing children is required for large countries to demonstrate a low average cost per child immunized. One might theorize that diminishing marginal returns would make the average cost begin to rise with increases in children immunized in the full sample of countries. According to this view, larger and larger scale inefficiencies in operation would begin to raise the average cost. Our findings show that this is not the case in this set of the world’s poorest countries. If there were no political and social obstacles, moving the scale of an immunization program from several hundred thousand children immunized per year to millions per year would improve technical efficiency. For the sample of countries studied, increasing the number of covered children by 10 percent increases costs by only 8.4 percent. The increased efficiency from scaling up applies to the sample as a whole and might not apply to any given country.
Advantages of regional programs.
Given our diverse, multicountry data set, it is unlikely that we will observe a rise in average costs at the far right of the average costs curve indicated by Exhibit 1 Importance of sharing experiences. The collection and dissemination of the FSP data illustrate concretely how the global immunization community can create opportunities for countries to share their experiences. Regardless of the specific strategies employed, the provision of evidence-based technical support by global partners such as GAVI, the WHO, and the World Bank to assist countries in assessing their current performance and identifying opportunities to improve efficiency continues to be of paramount importance.
David Bishai (dbishai{at}jhsph.edu) is an associate professor and Michael McQuestion is an assistant professor at the Johns Hopkins Bloomberg School of Public Health in Baltimore, Maryland. Rochika Chaudhry is an independent consultant based in Washington, D.C. Alyssa Wigton is a doctoral student at the Bloomberg School. The authors declare that they have no competing interests. Financial support from the Vaccine Fund is gratefully acknowledged. Helpful comments were received from Scott Barrett, Logan Brenzel, Bob Davis, Steve Landry, Miloud Kaddar, Patrick Lydon, and staff in Immunization, Vaccines, and Biologicals at the World Health Organization. All errors are the authors’ own.
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Study Data And Methods
Results
