Methodological challenges, solution pathways, and future priorities in Mendelian randomization research: a bibliometric and content analysis

To characterize the global landscape, intellectual structure, major methodological challenges, dominant solution pathways, and future priorities in Mendelian randomization (MR) research.

Articles and reviews on MR within the predefined 2003-2025 eligibility window were retrieved from the Web of Science Core Collection (WOSCC) and Scopus to construct a primary corpus (A-corpus). After field harmonization and hierarchical deduplication, conventional bibliometric analyses were performed. A problem-driven B-corpus was constructed by applying a prespecified terminology dictionary to titles, abstracts, and keywords, followed by manual eligibility review of dictionary-positive records and independent double coding across four dimensions: methodological challenges, solution pathways, future priorities, and solution orientations.

The A-corpus comprised 21,346 publications retrieved under the predefined 2003-2025 eligibility window; a small number of eligible records carried 2026 early-access or indexing metadata in the source databases. Strict dictionary screening identified 10,598 candidate records; 10,291 were excluded after eligibility review and 307 publications were retained in the final B-corpus. In the B-corpus, horizontal pleiotropy was the most frequent methodological challenge (90/307, 29.3%), mixed sensitivity framework was the dominant solution pathway (107/307, 34.9%), and reproducible reporting pipeline (45/307, 14.7%), mechanism-aware inference (36/307, 11.7%), and omics integration (34/307, 11.1%) were the leading explicitly coded future-priority categories. Inter-coder reliability ranged from moderate to almost perfect across dimensions, with Cohen's kappa values of 0.434 for methodological challenges, 0.531 for solution pathways, 0.703 for future priorities, and 0.897 for solution orientations.

This combined bibliometric and content analysis identified major publication patterns, recurrent methodological challenges, commonly reported solution pathways, and explicitly stated future priorities in MR research. The findings show the continuing prominence of causal-validity concerns alongside attention to mechanistic interpretation, reproducibility, generalizability, and translation; they describe patterns in the published literature rather than establishing the comparative scientific validity or clinical value of individual MR methods.