Methods: A multidisciplinary team with expertise in informatics, statistics, epidemiology, systematic reviews, cancer biology, and nutrition was assembled. Five 1-day workshops were held to brainstorm ideas; in the intervening periods we carried out searches and applied our methods to a case study to test our ideas.
Results: We have developed a two-stage framework, the first stage of which is designed to identify mechanisms underpinning a specific exposure–disease relationship; the second stage is a targeted systematic review of studies on a specific mechanism. As part of the methodology, we also developed an online tool for text mining for mechanism prioritization (TeMMPo) and a new graph for displaying related but heterogeneous data from epidemiologic studies (the Albatross plot).
Conclusions: We have developed novel tools for identifying mechanisms and carrying out systematic reviews of mechanistic studies of exposure–disease relationships. In doing so, we have outlined how we have overcome the challenges that we faced and provided researchers with practical guides for conducting mechanistic systematic reviews.
Impact: The aforementioned methodology and tools will allow potential mechanisms to be identified and the strength of the evidence underlying a particular mechanism to be assessed.