Methodology
This chapter compares the cost-effectiveness of interventions that cover a broad spectrum of health conditions prevalent in developing countries. All results are presented in U.S. dollars discounted to the year 2001 using a 3 percent annual discount rate. Chapter 15 summarizes the general guidelines governing the analysis leading to the results reported for all LMICs.
Regional Variations
Where possible and appropriate, intervention cost-effectiveness ratios and other information have been disaggregated by World Bank region. In discussing the estimates, this chapter focuses on differences in the costs of interventions rather than on differences in their effectiveness in specific regional settings, although both contribute to differences in cost-effectiveness estimates across regions. Cost-effectiveness estimates also differ among regions because of variations in underlying mortality, age structure, disease prevalence, and efficiency with which interventions are implemented. The analyses take all but the last of these considerations into account.
Interregional cost differences are attributable to differences in the local costs of goods and services that are not easily tradable. For components that are tradable, such as patented drugs and specialized medical equipment typically imported from industrial nations, the analyses assume uniform international costs for all LMICs, usually adjusted for local transportation and distribution costs. By using a single composite set of resource costs for each region, the analyses mask intraregional differences in the costs of nontradable goods, such as physician time or hospitals, but this methodology is appropriate because results are presented only at the level of the region.4 Interventions may differ in cost-effectiveness because they are targeted more appropriately to some age groups rather than others, and important gender differences may also exist in cost-effectiveness for some conditions, but data to estimate such differences are lacking.
Caveats
The findings in this chapter are subject to a number of caveats. First, despite efforts to ensure the consistency of cost-effectiveness numbers across chapters, the approaches taken in arriving at these numbers vary significantly. Although some chapters rely on cost-effectiveness numbers drawn from the literature, other chapters have analyzed these numbers afresh using the standardized resource costs described elsewhere. Table 2.1 contains definitions of indicators used to assess the quality of evidence on which the cost-effectiveness estimates are based. The tables in annex 2.B and annex 2.C indicate the quality of evidence associated with each intervention.
[Table .]
Second, almost without exception, the cost-effectiveness numbers do not vary with the scale at which the intervention is undertaken, and this is probably not the case in reality (Birch and Donaldson 1987; Johns and Torres 2005). Some interventions, such as vaccination programs, have large setup costs but marginal costs of extending coverage that decline at least initially. Other interventions, such as educational campaigns for condom use, may be easy to target to urban populations, but the marginal costs of expanding such interventions to relatively inaccessible populations increase with coverage. Therefore, many of the cost-effectiveness ratios presented here are useful only for modest increments in coverage, and separate analyses may have to be conducted to determine their applicability to program start-ups and larger-scale intervention changes.
Third, the cost-effectiveness numbers presented apply to countries whose institutional and technical capacity in relation to health is close to the average for the region. This evaluation is restricted to what countries could do more (or less) of, and clearly a more ambitious analysis would also cover what countries could do better. This issue is discussed in detail in chapters 3 and 70.
Finally, the estimates are based on the best available data, which in many cases are somewhat weak. Statistically derived confidence bounds for the cost-effectiveness estimates are not provided, and in most cases, uncertainty analysis has not been carried out. Readers are encouraged to pay attention to the order of magnitude of each estimate rather than to the specific number presented.
