Global Imbalances in Funding Sources for HIV Randomized Control Trials

This article aims to assess how funding of recent research on HIV/AIDS varies according to where the study is conducted and where publication authors are located. Studies have consistently documented global imbalances in scientific resources, including research capacity and representation on editorial boards. ¹ Authorship within HIV/AIDS research is disproportionately skewed toward higher income countries, particularly the U.S. ² Despite overall increases in the amount of resources dedicated to HIV/AIDS before the middle of the past decade, ³ the opportunity to apply for and obtain research funding may not be equally distributed worldwide. Access to grant funding is particularly important for HIV/AIDS where resource-intensive interventions are common, and impacts disproportionately affect economically disadvantaged populations. ⁴ These imbalances impede translating research findings into meaningful changes for the most impacted communities. ^5,6 To provide initial descriptive evidence on how types of grant support were distributed, we focused on a subset of publications. We included randomized control trials (RCTs) with outcomes measuring HIV status or viral load because of their policy relevance for goals such as UNAIDs' 90-90-90 target. ⁷

Although other outcomes are also important, the population-level impact, feasibility, and cost of new public health standards, such as universal voluntary HIV counseling and testing, are often assessed through RCTs with these endpoints. ⁸ Moreover, RCTs are often referred to as the gold standard for understanding causal effects in health-related fields. ⁹ We have also focused on publications rather than grant awards because it provides us with the ability to collection information on a clearly defined population of interest (publications with a measurable set of characteristics) from public sources. We emphasize that publication output is distinct from grant awards, not least because of lag times in publishing.

We identified relevant studies published in 2019 and 2020 using PubMed. We aimed to provide an assessment of recent publications over a two calendar-year period to ensure at least 100 articles. Data extraction was finalized in April 2021. We considered other databases, including Econlit and Psychinfo; however, initial screening did not identify additional studies, nor did screening of citations and references. Searching titles and abstracts identified 116 studies, for which we recorded if funding was received, and the source. We also recorded the countries where the study was conducted, along with countries of the first and last author's affiliations.

We produced descriptive statistics for numbers and proportions of articles receiving funding grouped according to the following categories: U.S. government, biotech/pharmaceutical/consulting companies, universities/research institutions/hospitals, private/independent foundations, and others. We then analyzed cross-tabulations of funding category by where the study was conducted, and authors' countries. To provide easily interpretable summary statistics while maintaining sufficient studies in each category, our main analysis grouped countries according to income levels (national income per capita). We tested our research hypothesis that funding categories were equally distributed across location using chi-squared tests.

Of the 116 studies, 65% had a single funding source, with remaining publications receiving more than one type, for a total of 178 funding sources. Of 178 funding sources listed, 28% were U.S. government agencies, 25% were biotech/pharmaceutical/consulting companies, 16% received university/research institute/hospital funding, 11% received private/independent foundations funding, with the remaining 1% receiving no listed funding. Figure 1 summarizes cross-tabulations examining how funding categories are distributed according to study location and countries of the first and last authors' affiliations.

OPEN IN VIEWER

FIG. 1.

Distribution of funding type by study location and authors' affiliations. Bar height shows the proportion of publications in that panel that received funding from each type (U.S. government agency, biotech/pharmaceutical/consulting company, institution/university/hospital, private foundations, and others). The first row compares funding types for publications that included a study location in a higher income country, classified according to World Bank data, with publications that did not include a study location in a higher income country. The second row compares funding types for publications whose first authors had an affiliation in a higher income country with publications whose first authors did not have an affiliation in a higher income country. The third row compares funding types for publications whose last authors had an affiliation in a higher income country with publications whose last authors did not have an affiliation in a higher income country.

Around a third of the 116 publications (35%) involved a study site in more than one country; therefore, we focused our analysis on whether any of the research sites was a higher income country, but also conducted secondary analysis of publications involving a single country. Figure 1 shows clear differences in the frequencies with which publications in these two categories received different funding types.

For the 55 publications involving research in at least one higher income country, the most common funder was a biotech, pharmaceutical, or consulting company (60%). In contrast, for the 61 studies that did not involve research in at least one higher income country, the most common funder was a U.S. government agency (56%). Other funding types also occurred at different frequencies. For publications with at least one study site in a higher income country, 27% received U.S. government funding, 16% received institutional funding, and 9% received foundation funding. For publications with no study sites in a higher income country, 30% received institutional funding, and 23% received foundation funding. We rejected the null hypothesis of uniform distributions for funding types according to whether the study locations included at least one higher income country or not (χ ²  = 23, p < .001).

We found similar differences in funding distributions according to first author's affiliation (Fig. 1 row 2), and last author's affiliation (Fig. 1, row 3). We assessed whether the way we categorized publications, funders, and countries affected conclusions, but found similar results.

Overall, these findings correspond with previous research documenting that biomedical RCTs conducted in sub-Saharan Africa were more likely to be funded by entities based in higher income countries. ¹⁰ We find no evidence that the comparative reliance of RCTs conducted in Africa on funding from government agencies based outside Africa has changed since analyses conducted in the early 2000s. ¹¹

There are several potential explanations for these differences. First, they could reflect variation in funders' priorities. ¹² Our finding that publications with study locations in higher income countries are more likely to receive funding from biotech or pharmaceutical companies corresponds with previous research that found funding from private industry was associated with decreased emphasis on diseases relevant to Africa. ¹³ We found that even for first and last authors whose affiliation was not in a higher income country, the most common source of funding was a U.S. government agency.

This may reflect in part the role of the U.S. in supporting international research on HIV/AIDS, such as through PEPFAR. ¹⁴ Second, opportunities to apply for funding may mean access is not equally available depending on where the study is located or the affiliations of the researchers conducting it. Even beyond eligibility, access to funding may require resources such as physical infrastructure or be greatly facilitated by affiliation with certain scientific networks. ^{15 –17} Without equitable funding access, researchers and affected populations based in lower income settings will continue to be disadvantaged, further exacerbating already skewed distributions of scientific infrastructure and resources. ¹⁸

Our study has several limitations. First, although this analysis aimed to provide a snapshot of funding patterns in recent RCTs measuring HIV status or viral load, we are unable to assess whether findings generalize to other study designs or outcomes. Exploring other areas such as treatment-related measures and mental health is an important next step. Second, these findings are descriptive and do not identify causes of variation in funding types.

Third, included studies will have received funding before publication, sometimes well in advance, and not all funded studies result in publication. Future analyses should aim to follow researchers prospectively throughout the scientific process, and collect information on grant applications as well as amounts of funding requested. One possibility could be to analyze grant databases instead of publications, such as those provided by some national funding agencies. This would allow for assessment not just of which types of studies are funded, but what characteristics are predictive of receiving a higher funding amount.