About the project — HIV Funding Impacts

Why our project exists

In 2025 the United States government began replacing the earlier arrangements for United States global health assistance with country-by-country agreements under the America First Global Health Strategy. A total of 36 agreements have been signed so far and 33 cover HIV. The consequences for treatment, testing and prevention will show up in national programme data over several years, in different places and at different speeds.

Most of the figures needed to reflect the status quo already exist. The figures are scattered across donor reporting systems, international estimates and national programme reports, and the publishers differ in schedule, in vocabulary and in level of detail. Assembling the figures by hand for a single country takes days, and the hand-assembled figures are out of date by the time the assembly is finished. Our pipeline assembles the figures programmatically, and we ensure that the resulting figures can be checked by reporting the data origins (i.e., original data sources).

Who runs our project

Dr. Adrian Smith (Associate Professor, Director of Clinical Studies, Nuffield Department of Population Health, Oxford), who specialises in HIV epidemiology, recently presented how HIV funding cuts from the United States affect LMICs. Dr. Jason Hung, who is a technically enabled social and health data scientist partly trained in quantitative health sociology, relocated to the Global South upon finishing his PhD at Cambridge. Upon discussion and given the common interests of Adrian and Jason in HIV/STI research in Global South contexts, Adrian and Jason decided to jointly build and run our project. Adrian is responsible for developing the research design and contextual framing. Jason, on the other hand, is responsible for building the pipeline and the website.

The project is not published by, or affiliated to, any research institution.

Where our project has got to

We have already built the aggregator layer, meaning the figures from international and donor bodies that compile figures across many countries. The aggregator layer is complete and near-global, covering panel data from UNAIDS, the WHO Global Health Observatory, the World Bank, the OECD Creditor Reporting System, United States self-reported disbursements and PEPFAR programme results. On top of the aggregator layer, we successfully built 42 connectors, meaning programs which read a single national source, at the country layer, namely four for Southeast Asia, a single connector for Eastern and Southern Africa, six for Western and Central Africa, six for South Asia, seven for East Asia and the Pacific, nine for Eastern Europe and Central Asia, eight for Latin America and the Caribbean and a single connector for the Middle East and North Africa. The Fijian connector holds no figure yet, because the government domain hosting the Fijian files gave no network address from 25 to 27 October 2026.

We decided to choose Southeast Asia data collection as the pilot because Jason specialises in, and is most familiar with, data science research in Southeast Asian contexts. We already built the aggregator-level connectors, which cover all eight LMIC regions on the website. The country-level assessment now covers all eight LMIC regions as well, and closed for the Middle East and North Africa on 28 October 2026.

What comes next

Separate United States funding by the agency that disbursed the funding, so that changes at USAID can be told apart from changes at the Centers for Disease Control and Prevention.
Publish the pipeline, the codebook and the assembled panel so that the figures on the website can be regenerated independently.

Corrections

If a figure on the website does not match what the source publishes, the discrepancy is worth reporting and will be investigated against the stored retrieval record. The panel keeps the URL of the data source and the retrieval date for every row, so our project can usually establish whether the source changed the figure or the pipeline misread the figure.

How to cite