Appendix: Data sources by category
Categorization of data sources
We have categorized our review of the data into two sets of categories: financial flows and resources flows (covering the physical goods and services required to deliver vaccinations). Financial flows are categorized into sources, channels and agents. Sources are defined as revenue streams of funds. Examples of sources include national treasuries and private individuals. Channels are institutions though which funding is passed or “channeled.” Organizations with the primary purpose of providing development assistance are generally standalone channels. Agents are entities which pool health resources collected from different channels, including actors that pay for health care from their own funds, such as households and firms. Some of the sources cover both sources and channels, or both channels and agents, and those are thus represented in separate categories.
Resource flows are categorized into: producers, distributors, commodities in country, vaccine and service delivery. Commodities include the vaccines themselves but also the syringes and other supplies required to administer vaccinations. In many lower-income and lower-middle income countries, production of vaccines and vaccination-related supplies would be done abroad, thus why producers and distributors are lumped into separate categories. Commodities would generally be imported and, in many cases, purchases are facilitated by multilateral development partners. Service delivery is distinguished from commodities because the relevant services are not generally traded. Service delivery is further segmented into vaccine delivery and service delivery. Service delivery includes: human resources, program management, training, social mobilization and disease surveillance. Vaccine delivery covers the expenditure on up-front investment and maintenance of cold chain equipment, cold chain overheads, vehicles and transport. These two categories could also be lumped into one classification: service delivery. We also include a special category for data on vaccination coverage, which could be connected to financial and resource flows to determine the ultimate impact of support provided to vaccinations. Finally, consideration of costs is not included in this section; we believe the general assessment found in the body of the text suffices.
1) Financial Flows: Sources
Overview:
The annual PhRMA Industry Profile provides data on total pharmaceutical sales to countries and regions around the world gathered through the PhRMA Annual Member Survey, but data on total vaccine sales is not available [1-2]. The total sales data are only available for a small number of countries in the framework, while other countries are grouped into regions. Compared with the other data sources listed in this section, the G-Finder data are more comprehensive as they report detailed expenditure data tabulated by disease, product, recipient, recipient country, and funder. The data are collected using surveys and by extracting data from the US National Institutes of Health (NIH) grant data. However, the G-Finder data suffer from issues of completeness as the sampling frame has expanded over time and subjects do not always respond to the survey.
Coverage:
No data source is publically available for the entire period of interest (2007 through 2012). The G-Finder database and tax forms are only available through 2010 at the time this document was written, while sales data from the PhRMA Industry Profile is only available for years 2009 to 2010. Pharmaceutical companies’ published data on charitable giving are available for more recent years. However, accessing historical charitable giving data on pharmaceutical companies’ websites is sometimes problematic. All data in the “Sources” category are updated on an annual basis.
Accessibility:
The G-Finder database is the most accessible data reviewed in this section as the database is easily queried and exported into Excel. The other data sources would be time-consuming to extract as the data would need to be copied and pasted into a spreadsheet to facilitate analysis.
Reliability:
Among those data sources that exclusively capture sources of funding, data quality is low. While data on the value of pharmaceutical vaccine donations exist, the data provided by pharmaceutical companies on their websites is neither disaggregated by recipient country nor consistently disaggregated by vaccine type. Tax forms from pharmaceutical companies’ foundations sometimes report cash grants for vaccine-related activities along with the recipient countries and institutions receiving the grants, but these data are not systematically reported.
Comparability:
The G-Finder Database and PhRMA Industry Profile data are collected using standardized methodologies to ensure comparability across years. In contrast, pharmaceutical companies’ published charitable giving data and grants data from pharmaceutical company foundation tax forms is not reported in a standardized manner across companies.
Data gaps:
Data on pharmaceutical companies’ sales of vaccines, vaccine donations, and charitable giving to vaccination programs in GVAP focus countries is not detailed or standardized enough to be useful for a comprehensive analysis of immunization financing. The G-Finder data are useful for tracking R&D funding for vaccines across disease groups, but these data also have limitations due to lack of completeness over time.
Source / Description / Geographic scope / Years availableG-Finder database [1] / Provides financial data on research and development expenditures on neglected diseases tabulated by disease, product, recipient, recipient country, and funder. The data are collected via survey and review of US NIH grant data. Completeness of the data is a problem due to survey non-response and the fact that the sampling frame has expanded over time. / Global—Only the country in which the recipient research institution is based is available, not the countries that the vaccines would benefit / 2007-2010
PhRMA Industry Profiles [2] / Using data collected through the PhRMA Annual Member Survey, the reports provide financial data on total pharmaceutical sales by region and total sales for selected middle- and high-income countries. Data for vaccine sales by country are not available. / Global—some sales data are reported for individual countries while others are grouped by region / Sales data for 2009, 2010 and 2011 available
Pharmaceutical companies’ published data on charitable giving / Leading pharmaceutical companies typically post data on their charitable giving on their websites and in annual corporate social responsibility reports. The value of vaccine donations is not always reported separately from other types of pharmaceutical donations. The reported value of product donations may be significantly higher than the value of the product in the recipient country. Furthermore, the quality of these product donations has not been independently assessed. / Giving to individual countries not reported separately, but only as lump sums / 2011, availability of historical data varies by company
Pharmaceutical companies’ foundation tax forms available through the Foundation Center [3] / On their tax forms, corporations’ foundations report their charitable donations. Data on pledges, project descriptions, recipient name, and target country and are sometimes available. These tax forms do not capture the value of product donations. As standardized reporting on individual grants is not required by the IRS, data are not available for all pharmaceutical companies’ foundations. For example, the Merck Company Foundation systematically reports pledges to support vaccination programs, but Pfizer and GlaxoSmithKline Foundations do not. / Global—donations to individual countries are not available for all donors / 2000 through 2011
1-2) Financial Flows: Sources/Channels
Overview:
The OECD Creditor Reporting System (CRS) contains commitment and disbursement data from Development Assistance Committee (DAC) member countries, two non-DAC members (Kuwait and the United Arab Emirates), BMGF, and multiple multilateral funders [4]. In addition to financial data, the OECD-CRS contains detailed project descriptions and information about the channel receiving the money as well as the recipient country. Data reported by Kuwait, the United Arab Emirates, and BMGF are only available for more recent years.
While AidData incorporates data from the OECD-CRS, its prime advantage is that it contains data on assistance from over 20 donors that are not members of the OECD-DAC. The researchers obtain data on non-DAC donors’ foreign assistance from a variety of sources such as project documents, donors’ websites, and directly from donors themselves. They convert the data to a standardized format and manually apply activity codes based on review of project titles and descriptions [5]. AidData researchers have applied activity codes to projects from non-DAC donors to facilitate rapid analysis of projects by focus area.
Foreign assistance data reported to the International Aid Transparency Initiative (IATI) are detailed, project-level data that are reported in a manner similar to that of the OECD-CRS [6]. The level of detail included in project descriptions varies across donors and organizations. Currently, the data reported by UK-based NGOs are largely incomplete as most only report data on DFID-funded projects.
The grants data in the BMGF Online Grants Database and the tax forms is highly disaggregated and detailed [7-8]. However, the Online Grants Database only contains commitments and lacks the disbursement information that is included in the tax forms. Both sets of data lack information on the countries that the grants are targeting, but the Online Grants Database includes information on the region served. BMGF reports detailed data to the OECD-CRS, but the data in the Online Grants Database is timelier and covers a larger number of years.
Coverage:
The data sources that contain data relevant for the time period of the GVAP analysis include the OECD-CRS, AidData, BMGF Online Grants Database, and BMGF tax forms. The OECD-CRS data and BMGF tax forms are available up to 2011 at the time of writing, while the BMGF online database is current through 2013. The time periods covered by AidData and IATI Registry data vary by donor and by organization.
The OECD-CRS database and BMGF tax forms are updated annually. BMGF Online Grants Database is continuously updated, while data in the IATI Registry are updated according to the time schedule determined by each donor or organization. AidData is updated continuously, but these data are not always complete or cited properly. AidData publishes an annual “research release” which corrects for completeness and citation concerns.
Accessibility:
Data from the OECD-CRS and AidData can be quickly exported into a variety of formats suitable for analysis. In IATI, each “activity file” containing grants must be downloaded individually, which is time-consuming. Alternatively, code can be written to download or “webscrape” data files automatically from the IATI Registry.
Data from the BMGF online database cannot be exported in spreadsheet form. Adding this functionality to the database would eliminate the time required to copy and paste the data into a spreadsheet for analytical purposes. Furthermore, the online database lacks disbursement data, but these data are located in BMGF’s tax forms. It is difficult and time-consuming to extract disbursement data from PDFs, so it would be best if BMGF could incorporate its disbursement data into the online database. Also, it would save data users time if BMGF added functionality to its Online Database to permit the data to be downloaded in spreadsheet format.
Reliability:
The data from the OECD-CRS, AidData, and the IATI registry are not systematically validated at the country level. While donors may state that they are devoting a certain amount of money to fund vaccination programs in a specific country, there is a lack of transparent tracking systems in place to monitor how much money actually reaches that country. Furthermore, it is not always possible to discern from project descriptions the types of vaccines donors are funding.
Over time, the reliability of the data found in the CRS has improved. To avoid double-counting, multilaterals do not report earmarked funds received from bilateral donors. From a vaccine resource tracking perspective, it may be best to track multilateral assistance through these organizations’ project databases or financial reports instead of through the CRS in the interest of capturing both core and earmarked funding closer to its ultimate point of disbursement. However, for the GAVI Alliance, we recommend using the data that they report to the OECD-CRS as it includes total expenditures from the organization whereas the financial data published on GAVI’s website only captures a subset of total expenditures. This complete dataset could be potentially obtained from GAVI itself.
Comparability:
The data within OECD-CRS data are comparable since donors report their aid data in a standardized manner using the CRS Directives [9]. The data in the IATI registry are also reported according to the “IATI Standard” [10]. Unlike the OECD-CRS, the data in the AidData database are not reported by donors in a standardized manner, but are compiled by AidData researchers. Therefore, the AidData database is not as comparable as the OECD-CRS data since the data are not reported by donors according to a standardized set of guidelines. The data from the BMGF Online Grants Database and BMGF’s tax forms differ from the BMGF data in the OECD-CRS database due to different reporting systems.
Data gaps:
There are is a plethora of data in the sources/channels category. Currently, the largest gap in this category is data on Chinese foreign assistance. Collaboration with the Peking University Institute for Global Health, who published “China health aid to Africa” in 2011, could be a possible avenue for obtaining data on Chinese funding of immunization programs in developing countries [11].
Unlike the many DAC bilateral organizations that report to the IATI Registry, complete, project-level data on US bilateral disbursements is not yet available past 2011. However, Secretary of State Hillary Clinton’s announcement that the US would join IATI indicates that these data are likely to be available in the near future [12].
Another step towards filling data gaps in this category would be to add disbursement data to the BMGF Online Grants Database. While the database is timely, the utility of the data would be increased through the addition of this important variable.
Source / Description / Geographic scope / Years availableOECD-CRS [4] / Contains project-level data from DAC members, Kuwait, United Arab Emirates (UAE), BMGF, and selected multilateral donors. Donors report information such as commitments, disbursements, project descriptions, sector, channel, and recipient country. Data quality varies by donor and has improved over time. Data from Kuwait, UAE, and BMGF are not available for all years. / Global / 1973-2011; not all donors/organizations provide data for entire time period
AidData [5] / Contains foreign assistance data from the OECD-CRS, the IATI Registry, project documents, donors’ websites, and project-level data obtained via correspondence. AidData contains data from many donors not captured in the OECD-CRS database. / Global / Varies by donor/organization
IATI Registry [6] / Data from 92 bilateral and multilateral donors and UK-based NGOs are represented in this database. The largest bilateral donor, the US, does not yet report its data. / Global / Varies by donor
BMGF Online Grants Database [7] / Contains project descriptions, commitments, recipient information, region served, program, and project length. / Global / 1995 onwards
BMGF tax forms [8] / Contains project descriptions, disbursements, and recipient information. / Global / 2001 through 2011
London School of Hygiene and Tropical Medicine’s Countdown to 2015 Initiative [9] / Contains estimates of the immunization funding provided by development assistance partners to 74 developing countries. / Global / 2003 through 2010
Institute for Health Metrics and Evaluation’s Development Assistance for Health database [10] / Estimates of development assistance for health by donor and recipient can potentially be pulled from dataset. Not currently publicly available. / Global / 1990 through 2012
2) Financial Flows: Channels
Overview: