The 10 largest public and philanthropic funders of health research in the world: what they fund and how they distribute their funds

Background Little is known about who the main public and philanthropic funders of health research are globally, what they fund and how they decide what gets funded. This study aims to identify the 10 largest public and philanthropic health research funding organizations in the world, to report on what they fund, and on how they distribute their funds. Methods The world’s key health research funding organizations were identified through a search strategy aimed at identifying different types of funding organizations. Organizations were ranked by their reported total annual health research expenditures. For the 10 largest funding organizations, data were collected on (1) funding amounts allocated towards 20 health areas, and (2) schemes employed for distributing funding (intramural/extramural, project/‘people’/organizational and targeted/untargeted funding). Data collection consisted of a review of reports and websites and interviews with representatives of funding organizations. Data collection was challenging; data were often not reported or reported using different classification systems. Results Overall, 55 key health research funding organizations were identified. The 10 largest funding organizations together funded research for $37.1 billion, constituting 40% of all public and philanthropic health research spending globally. The largest funder was the United States National Institutes of Health ($26.1 billion), followed by the European Commission ($3.7 billion), and the United Kingdom Medical Research Council ($1.3 billion). The largest philanthropic funder was the Wellcome Trust ($909.1 million), the largest funder of health research through official development assistance was USAID ($186.4 million), and the largest multilateral funder was the World Health Organization ($135.0 million). Funding distribution mechanisms and funding patterns varied substantially between the 10 largest funders. Conclusions There is a need for increased transparency about who the main funders of health research are globally, what they fund and how they decide on what gets funded, and for improving the evidence base for various funding models. Data on organizations’ funding patterns and funding distribution mechanisms are often not available, and when they are, they are reported using different classification systems. To start increasing transparency in health research funding, we have established www.healthresearchfunders.org that lists health research funding organizations worldwide and their health research expenditures. Electronic supplementary material The online version of this article (doi:10.1186/s12961-015-0074-z) contains supplementary material, which is available to authorized users.


Background
Approximately 40% of all health research in highincome countries is funded by public and philanthropic funding organizations [1]. These organizations play a central role in the development of new knowledge and products, particularly in areas that are not sufficiently profitable [2]. For example, the involvement of public and philanthropic funding organizations has been key in the development of new medical products to combat neglected diseases [1,2] and, since recently, these organizations are increasingly taking action to address the lack of development of new antibiotics [3][4][5].
Transparency on who the main funding organizations of health research are, on what they fund (their funding patterns) and on how they decide on what gets funded (their priority setting mechanisms and funding distribution mechanisms) can help funding organizations to synchronize their efforts, potentially preventing the duplication of research and improving collaboration on research priorities, and has various other strategic and practical benefits for funders [2,[6][7][8][9][10][11][12]. Such transparency also allows for external evaluation of funding organizations' portfolios and decision-making processes [7,13]. This is particularly important for public funding organizations, since they distribute public funds. For philanthropic funders, such accountabilities are less clear, but given the substantial impact of some of these funders on the global landscape for health research, it might be reasonable to make similar demands from this group of funders [14,15].
Although substantial insight has been created in recent years into countries' expenditures on health research [1,[16][17][18][19][20], there has been relatively little scrutiny of the funding patterns and mechanisms of individual funding organizations. Mappings of individual funding organizations' expenditures on health research are often limited to one or several countries [7,10,[21][22][23][24][25][26] or to a select group of diseases [25,[27][28][29]. To increase the available information on major public and philanthropic funders of health research, we present a mapping in this article that had a simple target: to identify the 10 largest public and philanthropic funders of health research in the world, to report on what they fund, and on their mechanisms for distributing these funds (funding organizations' priority setting mechanisms were beyond the scope of this studysee Limitations section for more detail).

Methods
Here, we outline the methods used to identify the 10 largest funding organizations of health research in the world, and to assess the funding patterns and funding distribution mechanisms of these organizations. A more detailed description of these methods is provided in Additional file 1. All data were collected from November 4, 2013, to August 12, 2014.

Identifying the 10 largest funders of health research Search strategy
This study distinguished between four types of public and philanthropic health research funders: (1) public national or regional funders (excluding funders of official development assistance (ODA) and multilateral funders), (2) philanthropic funders, (3) ODA funders, and (4) multilateral funders. The mandate of the funding body did not need to be limited to funding health research. Funding organizations were identified through a search strategy that had several components: key funding organizations in the 20 countries with the highest spending on health research [1] were identified, membership lists of collaborative groups of funders (i.e. groups where major funders of health research collaborate on a global or regional level) were reviewed, publicly available lists of funding organizations that included annual spending on health research were searched, and data on Development Assistance for Health were used to identify key ODA funders. For every funder type, a specific search strategy was used to identify the largest funders of health research (Additional file 1). Private for-profit funding organizations were not included in our analysis; we only aimed to map public and philanthropic funders (private for-profit health research funders are mapped elsewhere [30]). Product development partnerships (PDPs) and other public private partnerships (PPPs) were also excluded because they are intermediate funding organizations, who are funded in turn by governments, philanthropies and the for-profit sector. Furthermore, we excluded single disease funders; although the majority of philanthropic funders of health research focuses on one disease [21], the largest philanthropic funders of health research tend to fund across multiple disease areas (with some exceptions [31,32]). We note that the annual health research expenditures of the largest PDP, PPP and single-disease funders that we are aware of are lower than the annual expenditures of the 10 largest public and philanthropic funders reported in this study (see Additional file 1). Finally, in two cases (the United States Department of Defense (US DoD) and the European Commission (EC)) we included both the overarching organization at its largest sub-organizations or sub-programmes, because of the substantial differences between the funding distribution mechanisms of these sub-organizations and sub-programmes.
To aid future analyses of this kind, we provide an overview of various sources that helped us identify the main public and philanthropic funders of health research globally in Additional file 2.

Assessing health research expenditures
For all the funding organizations that followed from our search, publicly available data were collected on the organizations' annual health research expenditures (from annual reports and websites). Data were collected for the most recent year available. When we were not able to find data on organizations' annual expenditures in the public domain, we contacted funders to ask if they could provide us with their annual expenditures on health research.
Funding organizations differ on at least three aspects in terms of how they report their annual health research expenditures. First, expenditures can be reported as actual expenditures, commitments or budgets. Second, there can be differences in terms of what the expenditures cover. They can cover the organization's total expenditures on health research excluding operational costs (for managing the funding organization), its total expenditures including operational costs, or its total overall turnover over a single fiscal year (this was only collected if the funding organization exclusively funded health research). Third, there can be differences in terms of the research areas that the reported expenditures pertain to: only health research, or broader categories such as health and biological research or life sciences research. For each funder we extracted data on annual health research expenditures in a step-wise manner, always reporting the actual expenditures excluding operational costs in the area of health research when possible. When these numbers were not available, we reported the next best available number, following the order in the categories provided above. We note that the data from the funding organizations in the top 10 all relate only to health research, all concern actual expenditures or commitments, and for all, except one, operational costs were excluded.
Training support and research education were not included in the overall amount for health research expenditures. In addition, for government ministries, we excluded two types of funding flows. First, when funding was provided by ministries to funding agencies for distribution, we included the funding for the funding agencies, but not for the ministries. Second, for government ministries, such as ministries of education or health, we excluded block funding to universities or hospitals (similar to other initiatives that have reported on health research funding flows [24]). For funding agencies, we did include institutional funding.
Finally, organizations' expenditures were made comparable using methods by Young et al. [17,20]. To do so, we first deflated organizations' expenditures in the national currency to the year 2013 using Gross Domestic Product deflators from the International Monetary Fund World Economic Outlook Database of April 2014 [33]. Second, we converted the inflation-corrected expenditures to US dollars using the World Bank Official exchange rates for the year 2013. As a secondary outcome, we calculated funding organizations' health research expenditures as 2013 purchasing power parity-adjusted US dollars (these are not reported in this article, but are available on www.healthresearchfunders.org) [17,20].
Assessing the funding patterns and funding distribution mechanisms of the 10 largest funders of health research After the 10 largest funding organizations of health research were identified, data were collected on their funding patterns and funding distribution mechanisms. For each organization, data were collected on: 1. The funding mechanisms used to distribute funding: intramural funding or extramural funding. For extramural funding, we distinguished between project grants, 'people grants' , programme grants, funding distributed to organizations and other extramural research funding. For project grants, data were collected to assess if the distribution was untargeted, targeted or highly targeted (for definitions see Additional file 1). 2. The amount of funding allocated to a list of 20 key health areas from the Global Burden of Disease classification [34].
Funding for operational expenditures was excluded. Finally, we denoted whether funding organizations used a classification system to classify funding to various health areas and whether they reported statistics on funding for various research types (e.g. biomedical research, clinical research, epidemiological research or health systems research [35]) and recipient countries or regions.
All data were collected from online reporting databases, annual reports, official websites, or other information sources. After this, each funder was invited to participate in an interview. Before the interview, a document with collected data was made available to a representative of the funder. Before and during the interviews, representatives were asked to add, amend or confirm the data.

Results
Identifying the 10 largest funding organizations of health research Public and philanthropic funding organizations Our search identified 55 public and philanthropic funders that were candidates for being one of the 10 largest funders of health research in the world (Table 1), excluding ODA funders and multilaterals (we searched separately for these and report on them later). For 41 organizations, data on the organizations' annual health research expenditures were available. For five of these organizations, this information was received through personal communications (not publicly reported). Fourteen funders did not provide figures about their annual health research expenditures. Often, these organizations   The annual expenditures for the Wellcome Trust are a slight overestimation for health research spending. Under Medical Humanities and engagement, various non-research grants are provided as well as other activities (e.g. running the Wellcome library), but these were not reported on separately, and are therefore included under 'Research' e Information was collected from official websites and annual reports. For these five organizations, information was not publicly available, and was acquired through personal communication with a representative of the organization were general funders of research and did provide overall expenditure data but not for health research specifically.
For the 10 largest funders, health research funding totalled to $ 37.1 billion, approximately 40% of all spending on health research globally by public and philanthropic sources [1].

ODA-agencies and multilaterals
The expenditures of ODA-agencies and multilaterals on health research were substantially smaller than the expenditures of the largest public and philanthropic funding organizations (Tables 2 and 3). The largest funder of health research through ODA was USAID ($ 186.4 million) and the largest multilateral funder was WHO ($ 135.0 million).
Assessing the funding patterns and funding distribution mechanisms of the 10 largest funding organizations of health research Funding mechanisms used to distribute funding There was considerable diversity in organizations' funding distribution mechanisms (Table 4). Five funders funded research fully extramurally, five allocated at least a proportion of their funding to intramural research institutes, and one funder, the Institut national de la santé et de la recherche médicale (Inserm), funded research (almost) exclusively intramurally (total is 11 because for the EC and the US DoD we analysed the suborganizations or sub-programmes: the US Congressionally Directed Medical Research Program (CDMRP), the Health theme of the EC FP7 Cooperation programme and the European Research Council (ERC)).
Of the 10 funding organizations that provided extramural funding, for six, the main mechanism for extramural funding distribution was the allocation of funding through untargeted competitive project or investigator grants (often, there were also some smaller programmes that used a more targeted distribution). Two funders, the Health theme of the European Commission FP7 Cooperation programme and the US CDMRP, used a more targeted approach and issued calls under prioritized areas. Funders also made use, in varying degrees, of highly targeted funding schemes, such as research contracts, tenders or prizes, but this was never the dominant form of funding distribution. The last two funders, the United Kingdom Medical Research Council (MRC) and the Deutsche Forschungsgemeinschaft (DFG), used a mixed approach to allocate funding, with substantial contributions made through different funding distribution mechanisms. Lastly, the funding model of the NIH and the untargeted part of the MRC deserve separate mentioning because, although they adhered largely to an untargeted model and research funding was available for all areas of health research, the amounts of funding available for various broad research areas were earmarked (in the case of the NIH, for example, through budgets for the NIH institutes). This differs from targeted approaches, where not all areas have to be funded and the prioritization is often more specific, but it is also not completely untargeted.
Finally, most funders mainly dispensed funding via project grants, with smaller programmes that provide grants to excellent individual researchers. However, others put more focus on individual excellence. The HHMI has traditionally been a proponent of such people-focused funding. Since recently, other funders, such as the Wellcome Trust and the NIH, are increasingly making use of 'people grants' as well [36].

Funding patterns towards diseases
The funding organizations' research expenditures towards 20 specific health areas are shown in Table 5. We could report data for at least one health area for seven funders. However, as the table makes clear, these data were often not available.
Funding patterns varied, with some funders showing preferences for investing in non-communicable over communicable diseases and others showing the opposite. For example, the NIH spent less on infectious disease research in total than on cancer research alone, while the Wellcome Trust spent 14 times more on infectious disease research than on cancer research. Similar variations arose when comparing more specific disease areas within the non-communicable or communicable diseases. For example, the NIH spent almost three times more on cancer research than on cardiovascular research while the EC under the FP7 programme spent roughly equal amounts on both, and while HIV/AIDS funding comprised more than half of the infectious disease research funding at the US NIH, it comprised less than 10% of that funding at the Australian National Health and Medical Research Council (NHMRC).
Six funders used classification systems to classify their funding to health areas (Table 6); five different classification systems were used by these funders (the two funders from the United Kingdom used the same system). Besides using different categories for health problems, these systems also varied on other aspects, such as who enters the data (e.g. the researcher or a specialist employed by the funder) and whether grants can be indexed as belonging to one or multiple health problems. Seven funders reported amounts of funding allocated to various research types and the same seven reported how much funding was allocated to various recipient countries or regions.

Discussion
In this article, we have identified the 10 largest funding organizations of health research globally and shed more light on their funding distribution mechanisms and     All data are from publicly available documents, except for CDMRP and DFG, for which data were received through personal communication, and for CIHR for which the data were largely publicly available but were supplemented through personal communication Zero's ("0") indicates an amount of zero million in funding; dashes ("-") indicate that no information on the amount of funding was available; NA indicates that we did not extract information for overarching funding organizations (only for the more specific organization(s) next to it). Some of the funders we identified (the EC and the US DoD) consisted of several sub-organizations that distributed the research funding in different ways, in which case we compiled total funding for the organization as a whole, but analysed the distribution mechanisms of the largest sub-organization ( Inserm funds research almost exclusively intramurally. However, Inserm does provide for some extramural funding. We were not able to find any information about the proportion of funding by Inserm that is distributed extramurally, but it is likely very small, so we classified all of Inserm's funding as intramural. Historical research into Inserm's expenditures put the organization's intramural/extramural proportions of distributed funds at 10/90 [23] c Technically, HHMI funds research 100% intramurally because it employs its own researchers. However, its researchers are mostly located at external research institutes or universities (with the exception of HHMI's Janelia Research Campus), and therefore, for this table, we classified HHMI's funding distribution mechanism as largely extramural. The 100 million dollar for intramural funding is based on an annual budget estimate from 2010 for the Janelia Research Campus by HHMI itself and is not a precise number Zero's ("0") indicates an amount of zero million in funding; dashes ("-") indicate that no information on the amount of funding was available. Health areas were chosen to be a representative sample of health areas in the Global Burden of Disease (GBD) classification [34]. Adding spending on the various health areas per funder will not total to the funder's total expenditures, because the categories are a selection of health areas from the GBD report. Funding for selected health areas needs to be interpreted with caution because these data are mutually exclusive for some funders (Wellcome, MRC, CIHR, EC FP7 health, CDMRP), but not for others (NIH, NHMRC) See Table 4  Differences between funding organizations: the need for more evaluation of funding distribution models First, there is considerable diversity between funding organizations in terms of what they fund and how they distribute those funds. This begs the question: do some funding distribution models have more impact than others? The impact of different approaches to funding health research is regularly discussed in the literature, for example, for intramural versus extramural funding [23], for targeted versus untargeted funding [37], for 'people grants' versus project grants [36,38], for small grants versus large grants [10], and for competitive versus non-competitive research funding [39]. However, comparative evaluations of the impact of various funding models are scarce [10,23,38], even though approaches to measure the impact of health research are available [40]. An exception has been the recent comparisons of 'people grants' versus projects grants in the United States, which compared HHMI with NIH researchers and NIH Pioneer Awards with NIH project grants [36,[41][42][43]. These comparisons have led the NIH to consider a broad shift toward 'people grants' , demonstrating the value and potential impact of such evaluations [36]. Evaluations of this kind provide new insights when comparing funding models across funding organizations, but given the different contexts in which funders operate, comparing the impact of different models within one funding organization is perhaps particularly valuable and should become more common practice. There is also a need for more debate about where the power to decide priorities for publicly funded health research should lie (with parliaments, ministries, funding agencies, or independent committees of experts). Such debate is needed because there are finite resources for investing in health research and thus priorities need to be set using fair and legitimate methods and using the best possible evidence [44]. In practice, public sector health research funding decisions are not only made on the basis of what research is needed, but are regularly influenced by other factors, such as political interests, advocacy and lobbying [2]. Thus, there is a need for transparency on who makes those decisions and to debate who should make them [2,13,[45][46][47]. Analysis of funding organizations' priority setting processes was not part of this study (see Limitations) but deserves to be a more frequent subject of research studies in the future.  [2,[6][7][8][9][10][11]13]. Yet, this study shows that these data are often not available. Through our study, we did not find a list of all public or philanthropic health research funders worldwide that included their annual health research expenditures (Additional file 1). Therefore, we have now established such a list ourselves at www.healthresearchfunders.org. On this website, we provide access to the data collected for this article and to information on more than 200 other public and philanthropic funders of health research that we have added to this website since the mapping for this article was completed.
Besides the absence of a global listing of funding organizations, we found that data on organizations' funding patterns and funding distribution mechanisms are often not available, and when they are, they are difficult to aggregate, owing to differences in funders' data classification systems. Notably, we only collected these data for the 10 largest funding organizations of health research. The absence of such information, and the difficulties in aggregating the data across funders, are likely to be more prominent when smaller funders are also included. There is currently no consensus on a framework for producing descriptive data on funders' funding patterns (both in terms of health areas and research types) nor on a framework for describing their funding distribution mechanisms [6,8,37]. In this article, we have proposed three frameworks for reporting data on health research funding: for reporting data on funding distribution mechanisms (Table 4), for reporting data on funding patterns in terms of health problems (the Global Burden of Disease classification [34]), and for reporting data on funding patterns in terms of research types (biomedical research, clinical research, epidemiological research or health systems research, as proposed by Frenk [35]). The adoption of standards for reporting funding data, including guidance on what data classification systems to use, by funding organizations, for example through collaborative initiatives such as the Heads of International Research Organizations, would substantially improve the quality and comparability of reported funding data [9].
Funding organizations are starting to support the goal of transparency and are increasingly recognizing the problems noted above and addressing them. At the 2014 World Health Summit in Berlin, several major funders of health research expressed interest to work together toward developing a common approach for mapping health research funding flows [12]. Another good example of a multi-funder collaboration to increase insight in health research investments is the World RePORT website [48]. On a national level, the United Kingdom has led the way in terms of harmonized reporting by showing it is feasible to collect comparable data on health research funding from all major public funding bodies and charities in a country [22]. Besides initiatives from funders themselves, there are also several promising initiatives from other parties to address the lack of data on global health research funding [1,16,[49][50][51]. The recent decision to establish a Global Observatory on Health R&D at WHO in particular may help to improve transparency in this area [1].

Limitations
Finally, we note that the mapping conducted for this article has had several limitations. First, we have excluded funding organizations in the private for-profit sector (these are listed elsewhere [30]). Second, national systems for funding health research vary. In many countries, a large amount of health funding is dispersed directly from governments to universities or research institutes via block grants. We excluded these block grants and therefore the public funding organizations that we report on do not all contribute the same share of all health research that is publicly funded in a country. Third, we had to make several generalizations in order to be able to report data across funders that were diverse in their funding distribution mechanisms and reporting systems. For instance, what we have termed 'targeted' research funding, is a grey area that ranges from broad prioritized research areas to specific research topics or questions [52]. Also, funders reported on their expenditures on health research in various formats. Although we have kept track of these varying reporting formats, they decrease comparability across funders. Fourth, we would have liked to exclude overhead costs within project funding (not operational costs of the funder, which we did exclude where possible, but overhead costs of the research organization), to measure only the amount of funding that went to research, but this was not feasible because it was mostly not reported. Fifth, our proposed framework for reporting on funders' funding distribution mechanisms (Table 4) lacks detail. It would have been interesting to also report on more detailed mechanisms, such as funders' grants for businesses and PDPs/PPPs, but we did not include such analyses because of a lack of comparable data across funders. Sixth, funding organizations frequently make adaptations to their funding strategies, and therefore our findings should be viewed as a snapshot of funders' expenditures, funding distribution mechanisms and funding patterns during the time of our data collection [53]. Seventh, in addition to reporting about funding organizations' funding distribution mechanisms and patterns, we would have liked to report on funding organizations' priority setting processes as part of this work (another important aspect of how funders decide what gets funded). However, we found that priority setting processes were generally not welldescribed and highly variable across funders, making it difficult to analyse and report our data. It deserves recommendation that such an analysis is conducted in the future, but the development of a framework for assessing priority setting processes at funders is needed first, potentially based on existing guidance for health research priority setting [44]. Lastly, and most importantly, our search strategy was limited in scope (see for more detail Additional file 1), was aimed only at finding the 10 largest funding organizations of health research in the world, and detailed data were only collected for those 10 organizations.

Conclusions
This study identified the 10 largest funding organizations of health research in the world and showed that these organizations together fund research for $37.1 billion, 40% of all public and philanthropic health research spending globally. It also mapped the funding patterns and funding distributions mechanisms of these funders and showed that there is considerable diversity between organizations in terms of what they fund and how they distribute those funds, highlighting the need for comparative evaluations of the impact of different funding distribution models. Moreover, because many of the data we tried to collect were not available, our study demonstrates that there is a need for increased transparency on who the largest funding organizations of health research are, what they fund, and how they decide what gets funded. As a first step in improving transparency in this area, we have proposed frameworks for reporting on funding patterns (in terms of health problems and research types) and for reporting on funding distribution mechanisms in this article and have established www.healthresearchfunders.org, where we list more than 250 public and philanthropic funders of health research and their annual health research expenditures. We will further expand and update this list of funding organizations in the future and welcome both suggestions and data from all who wish to help us make this database more accurate and more inclusive.