The starting point for the Reference Case is to define the different purposes for which cost estimates are used. Ideally, any cost estimate could be used for multiple purposes (accepting that some adjustments may need to be made). In principle, the use of high-quality cost data can result in the improved allocation of resources to global health strategies, interventions, and services that maximize health gain and financial risk protection. Improved cost data can also result in cost savings and efficiency improvements that ultimately can be used to fund additional health improvement-related activities. Moreover, if cost data are used to inform the equity of financing (and costs) between different payers, then ultimately good cost data can be part of reducing any negative poverty impact associated with ill health.

In practice, budgets for cost estimation projects are often set with a specific purpose in mind, and the methodological choices will be driven by that purpose. Each of these purposes may require different approaches to definition and measurement (e.g., the scope, frequency, and unit of the cost reported), and there may be different emphases in areas such as sampling. We have, therefore, indicated throughout the Reference Case where principles may apply differently depending on purpose. The starting point is to define the purposes. For simplicity, four ‘buckets’ of purposes are defined:

This purpose is defined as the use of cost estimates in analytical approaches to assess allocative efficiency (see glossary) of investment and policy decisions. These include, for example, cost-effectiveness analysis, cost-benefit analysis, health technology assessment, essential package definition, benefits package definition, etc. It may also include investment cases, linking closely to the estimation of resource requirements as below.

Comparisons of the cost-effectiveness of alternative uses of resources are now recognized as a core piece of information in decisions around whether to invest in new technologies, or set priorities across different strategies and interventions. For example, cost data are often critical in shaping the design of health care benefit packages provided by governments or insurers and, as many low-income countries move towards national insurance schemes, are needed to estimate reimbursement levels.

This purpose describes the use of cost estimates to predict expenditures in the medium (3 to 5 years) and longer term. Examples include using costs to inform budget impact analyses, support medium-term expenditure frameworks, inform budgets for national strategic plans, develop financial plans for investment cases, and produce ‘global price tags’. These analyses both support national planning but can be used in both national and global fundraising efforts for increased investment in a specific global health area. For example, since 2009, the South African government has collected cost data to predict the medium- and long-term costs to the South African national public sector antiretroviral treatment (ART) program, which was then used to advocate for increasing funding for ARTs of funding.

Cost data may be of use to those planning both the incomes and expenditures of health providers (or funders). This purpose describes the use of costs to predict expenditures by specific budget holders and set prices for specific services. Budget settings would include annual program budgeting by managers for routine health services, or a specific provider, or could refer to an investment case for a specific project or a funding application. For some organizations, such as insurance companies or private providers, budgets involve planning incomes, and prices for specific goods and services for the coming year and costs are core element in this process.

This purpose describes the use of costs to explore differences and drivers of technical efficiency (see glossary) between providers and/or modes of delivery (integrated services, platforms, level of decentralization, etc.) for health interventions or services, usually conducted through the comparison or analysis of costs over multiple sites, or by comparing estimate costs to benchmarks. Cost data from studies that help to estimate technical efficiency provide critical information for improving the value for money on the supply side, such as identifying the minimum efficient scale of operation, or providing insights into areas of efficient or inefficient practices. For example, WHO, UNICEF, and GAVI use unit cost data to identify and design efficient supply chain logistic systems in immunization.

For each of these purposes, there may be different theoretical and practical reasons why a certain type of cost or methodological approach is preferred. For example, where countries are moving towards universal health coverage, the need to generate reimbursement rates and to understand the comparative value of new technologies (i.e., applying economic evaluation/and or priority setting) creates a demand for unit cost data that are comparable across diseases and health services, follow a standardized methodology, and reflect economic cost.

In contrast, cost data for technical efficiency studies may need larger sample sizes, have a different perspective, and need additional information about cost determinants collected to enable analysis. For the purposes of financial planning and resource requirement estimates, financial costs are generally needed rather than economic costs, and disaggregation of prices and quantities in unit cost reporting is helpful; in the South Africa example above disaggregated estimates were used to estimate the impact of changes such as introducing task-shifting to lower staff cadres and opening the South African antiretroviral drug market to international competition.

Where recommended methodological approaches differ by cost purpose, these differences are explained in the Reference Case and illustrated throughout using two examples of costing exercises. The first example is based on an economic evaluation of condom distribution using community health workers in India, the Avahan program. The second example is based on an exercise to help the South African Government to plan roll-out of the Xpert MTB/RIF diagnostic for tuberculosis. These are only two examples and should be interpreted as illustrations, rather than any prescription of methods.

The Reference Case is composed of a total of 17 methodological principles across four main topics: (1) study design; (2) resource use measurement; (3) pricing and valuation; and (4) analyzing and presenting results. For each principle, we provide an explanation as to why it is important, and information on the methods specification below.