**Background**
The World Health Organization (WHO) is the United Nations specialized agency dedicated to the 'attainment by all peoples of the highest possible level of health,' with 194 member states and more than 8,000 staff. The World Health Assembly (WHA) is the plenary decision-making body that determines the budget, strategic priorities, and program of work, annually adopting non-binding resolutions and decisions. While previous research has analyzed WHA resolutions for specific periods (1970–2013 and 13-year studies of member state contributions), no comprehensive analysis of all resolutions since the WHA's creation in 1948 had been conducted. This study aimed to address three questions: (1) What is the overall structure of the normative web of WHA health norms? (2) Do some issues form distinctive network patterns? (3) What topics have been addressed and how has the policy agenda evolved over time?
**Methods**
The researchers used web scraping (Python) to retrieve all resolutions from the WHO Institutional Repository for Information Sharing (IRIS) database, covering 1948 to 2022. Documents were preprocessed, and optical character recognition (OCR) was applied where needed. All resolutions were imported into ATLAS.ti V.22 for qualitative analysis. Regular expressions ('WHA\\d*[.]\\d*') were used to automatically code citations between resolutions, initially finding 7,584 occurrences. Manual review of all resolutions identified and corrected 275 errors, including OCR inaccuracies and misidentified resolution numbers. After removing duplicated edges and self-loops, the final dataset contained 3,242 edges. Metadata including resolution titles and Medical Subject Headings (MeSH) terms were extracted from IRIS. Network analysis was performed using Cytoscape V.3.9.1, computing centrality metrics and using the Leiden algorithm for community detection. The network comprised 3,194 nodes (resolutions) and 3,242 edges (citations).
**Key Results**
The overall network had an average degree of 2.030 (median 1, SD 3.634), with a heavily right-skewed degree distribution—only a handful of nodes had many connections. Nearly half of the components (1,538 nodes, 48.15%) had no links at all. The total number of resolutions per year peaked in 1950 (123 resolutions) and declined thereafter, while citations per year showed an increasing trend, peaking at 124 in 1989. The ratio of cumulative edges to nodes consistently increased over time, from 0.0863 in 1953 (under Director-General Chisholm) to 1.0150 in 2022 (under Director-General Tedros). A giant component formed by 1,026 nodes connected by 2,573 edges was identified. The global clustering coefficient grew markedly from the tenures of Candau to Mahler (1953–1988) and stabilized from 2000 onwards.
Community detection revealed three distinct structural patterns: (1) Chain-like clusters for topics addressed year after year (e.g., leprosy control, cardiovascular diseases, health problems of seafarers), suggesting siloed governance; (2) Radial-like clusters centered on a few resolutions, all related to procedural and financial issues (scale of assessment, payment arrears), with 154 resolutions connecting to just 7 central resolutions; and (3) Densely connected clusters on contested health issues including child health, essential drugs, HIV-AIDS, health systems strengthening, breast-milk substitutes, and traditional medicine. Resolutions with the highest betweenness centrality included WHA62.12 on primary healthcare, WHA48.13 on communicable diseases prevention, WHA54.13 on strengthening health systems, and WHA66.10 on non-communicable diseases prevention.
Thematic analysis of 8,337 MeSH terms from 3,136 resolutions (1948–2019) showed that the most frequent terms were 'budgets' (n=569), 'financial management' (n=500), and 'organization and administration' (n=482). Procedural resolutions were more likely found outside the giant component (0–5% in the giant component), while thematic resolutions on nutrition policy, health planning, and primary healthcare were 83–93% likely to be in the giant component. Malaria received the most disease-specific attention (n=55 resolutions). Over time, the WHA agenda remained relatively stable, dominated by procedural matters, WHO budgets and finance, international cooperation, health planning, communicable diseases, and health systems.
**Clinical Implications**
This study demonstrates that WHA resolutions form an increasingly interconnected normative web, reflecting the complexity of global health diplomacy. The identification of bridging resolutions (e.g., on primary healthcare and NCDs) highlights potential leverage points for norm entrepreneurs seeking to advance shared international health norms. The finding that siloed governance patterns (chain-like clusters) coexist with highly interconnected clusters suggests that the WHO's criticized siloed approach may stem from earmarked voluntary funding rather than collective WHA will. The network approach offers a quantitative method to understand how health issues are linked, potentially informing more effective prioritization and policy coherence in global health governance. The authors call for standardized data curation across multilateral institutions to enable further computational diplomacy research.