-
Risk factors for pancreatic cancer in electronic health records: an umbrella review of systematic reviews and meta-analyses
Sarah F. Moore, Sarah Price, Judit Konya, Sophie Blummers, Fiona M. Walter, Richard D. Neal, Gary AbelSummary
Background:Pancreatic cancer has poor survival because of predominantly advanced-stage diagnosis. One strategy for improving outcomes is earlier identification, possibly achievable by enhanced surveillance or improved risk prediction modelling. This umbrella review updates previous evidence with a comprehensive assessment of factors which could inform risk assessments.
Methods:Database searches were performed in MEDLINE and EMBASE via Ovid and the Science Citation Index Expanded of the Web of Science Core collection from inception to March 2025. Systematic reviews and meta-analyses of factors associated with altered risk of pancreatic cancer, available in a coded electronic healthcare record, were included. Participants in component studies were adults, and we compared exposed/not exposed/differentially exposed participants. There was no geographical restriction. The main outcome was potential risk factors for pancreatic cancer, categorised by degree of association. The study was registered with PROSPERO, registration number CRD42024526338.
Findings:2386 abstracts and 449 full texts were dual screened, resulting in 168 studies included in the review, comprising 365 meta-analyses of individual risk factors or strata and >2,255,495 pancreatic cancer cases. 21 meta-analyses reported gender-disaggregated data which were extracted and reported separately.Of the 80 potential risk factors identified, 38 were associated with an increased pancreatic cancer risk, 11 with a protective effect, and 31 had no significant association with pancreatic cancer. Major newly found risk factors were autoimmune liver disease, BRCA gene mutation, co-infection with hepatitis B and C, and insulin use.
Interpretation:This comprehensive umbrella review of pancreatic cancer risk factors provides an up-to-date summary useful for identifying prevention and surveillance approaches, and for developing risk prediction models and directing future research.