Global Research, Disease Burden, & US Funding Impact
Okay, here’s a breakdown of the methods described in the provided text, categorized for clarity. This summarizes how the researchers analyzed the relationship between research effort and disease burden.
1. Defining Research Types & Categorization
APT Score (Article Potential for Translation): A machine learning-derived score predicting the likelihood a publication will be cited in a clinical trial. Higher score = higher clinical relevance.
iCite Classification: Used PubMed publication types to identify clinical research.
Three Research Groups:
Basic Research: APT < 0.5 AND not classified as clinical by iCite.
Applied Research: APT ≥ 0.5 AND not classified as clinical by iCite.
Clinical Research: Classified as clinical by iCite.
comparison to DALYs: For each research group, they calculated the proportion of articles dedicated to each disease and compared it to the distribution of Disability-adjusted Life Years (DALYs) – a measure of disease burden. This was visualized in Extended Data Fig. 6.
2. Industry Involvement Analysis
ClinicalTrials.gov Linkage: Publications were linked to clinical trials registered on ClinicalTrials.gov.
Focus on Phase 3, industry-Sponsored Trials: A subset analysis was performed specifically on industry-sponsored, Phase 3 clinical trials. They found increased research-disease divergence in this subset. Industry-Affiliated Authors: They also analyzed publications with authors from industry-affiliated institutions to see if the divergence was solely due to direct industry sponsorship.Results were consistent with the main analyses.
3. Funding Acknowledgement Analysis
Funded vs. Unfunded: Articles published in 2008 or later were divided into those that explicitly acknowledged funding and those that did not.
DALY Comparison: The distribution of articles in each funding group across diseases was compared to the DALY distribution (Supplementary fig. 5).
4. Geographical Analysis
DALY Distribution by Region: Calculated the share of DALYs per world region for each disease cause. Herfindahl-Hirschman Index (HHI): Used to measure the concentration of diseases within regions. Higher HHI = more regionally concentrated disease.
* correlation with Fig.4: Compared HHI results to the findings in Fig. 4 to identify diseases that contributed to reducing the divergence between research and disease burden. They found that locally concentrated, mostly communicable diseases were associated with reduced divergence.
In essence, the researchers used a combination of bibliometric data (publication details), machine learning (APT score), clinical trial registries, author affiliations, and funding acknowledgements to quantify the alignment between research effort and the global burden of disease. They then broke down this analysis by research type, industry involvement, funding status, and geographical location.