The ‘mechanistic interaction test’ (MIT) is a powerful method for testing mechanistic interaction in long-term follow-up studies, according to a new study by researchers at National Taiwan University (NTU). The study was published online on March 26 in PLoS ONE.
Conducted by Dr. Jui-Hsiang Lin, doctoral students at NTU, and his advisor Dr. Wen-Chung Lee, professor of Institute of Epidemiology and Preventive Medicine in the College of Public Health, this study proposed a novel mechanistic interaction test for censored data.
In follow-up studies, epidemiologists/clinicians often misinterpret a significant multiplicative interaction from Cox model as a genuine mechanistic interaction. Though indices specific to mechanistic interaction have been proposed, those indices assume no loss to follow up and no competing death in a study. In this study, the proposed MIT can maintain reasonably accurate type I error rate in censored data when the hazard curves are proportional to, non-proportional to, or even crossing over one another. ‘‘It also has far greater powers than the previous methods!’’ said Dr. Lee.
As the researchers of the study are both based in Taiwan. Their proposed MIT method is truly Made In Taiwan.