The team searched Dimensions on 13 February 2024 and screened 1,521 grant records. They analysed 802 grants with a recorded funding amount that started in 1999–2023, which together accounted for USD 318.6 million. The authors proportionally shared grants across recipient institutions when one award listed multiple recipients. For topics, they applied latent Dirichlet allocation; team gender was inferred from first names. The manuscript noted that 272 additional relevant grants lacked reported amounts and were included in sensitivity analyses, and that grant-level data cannot be redistributed under Dimensions licensing—reconstruction requires the published query and a subscription.
Grant counts grew at an estimated 12.2% per year. Recorded funding increased 16.6% per year, which the authors adjusted to 14.1% after inflation. Despite that rapid percentage growth, the field averaged roughly USD 13 million a year in recorded funding. Growth was steady across the observation window, yet the absolute dollar base remained modest. The study team used log-linear models to estimate those trends and reported aggregated yearly trajectories rather than per-grant temporal breakdowns in the main text.
Funding was heavily concentrated. US funders provided 47.7% of recorded funding. One institute, NIAID, accounted for 21.0% of all recorded dollars and rose to 23.3% for the 2016–2023 period. The top 10% of recipient institutions received 53.4% of funding. Low- and lower-middle-income countries received only 0.33% of recorded funding. When counting grant counts rather than dollar values, the Global South received 17.5%—a discrepancy the authors attributed to several Global South funders reporting grants without dollar amounts. Collaboration between Global North and Global South institutions was rare: the dataset contained seven grants linking recipients across those regions.
Topic modeling indicated a change in thematic emphasis over the study interval. Funding explicitly linked to antibiotic-resistance research rose from about 3% of funding early in the series to roughly one-quarter by the end. The authors did not provide a per-grant breakdown of how topic assignment correlated with award size in the main manuscript, but they reported the broad reallocation of funding toward antibiotic-resistance topics and made aggregate tables available in the project repository.
The investigator distribution reflected narrow investigator engagement. Eighty-four percent of researchers in the dataset held a single grant. Team-level gender comparisons showed that female-only teams received grants that were 19% smaller than male-only teams after adjustment; the reported 95% confidence interval was -32 to -2. The authors presented this as an adjusted association; they did not claim causal mechanisms and did not report detailed covariate lists for every adjusted model in the article text.
The authors described multiple sensitivity analyses: they added the 272 grants without recorded amounts, they merged annual records into multi-year projects in alternate specifications, and they varied key definitions. Headline results held in 37 of 40 sensitivity checks. To measure concentration they used the Herfindahl-Hirschman index and the Gini coefficient. The code, country classification table, and every aggregate table and figure are openly available at the project GitHub link provided in the manuscript, and an interactive dashboard is hosted by the authors. Grant-level records remain subject to Dimensions terms and cannot be redistributed by the authors.
The dataset depends on funders’ reporting to Dimensions and on whether funders disclosed grant amounts; the authors noted that several funders—particularly some in the Global South—report grants without dollar values, producing divergent results when counting grants versus dollars. The manuscript did not report detailed per-grant topic-specific dollar allocations or full covariate lists for every adjusted model in-line; interested readers must consult the GitHub repository for aggregate tables and code. The team also acknowledged that grant-level data reconstruction requires a Dimensions subscription and the published query because redistribution is prohibited by the data provider. Finally, the authors argued that broader funding and more complete national reporting will matter as phage therapy moves from compassionate use toward regulated clinical use—areas that remain open for policy action and more complete investment tracking.